International Journal of Advances in Engineering & Technology (IJAET) Volume 3 Issue 1

International Journal of Advances in Engineering & Technology (IJAET)
March Issue
Volume-3,Issue-1
URL
: http://www.ijaet.org
E-mail : editor@ijaet.org
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963
Table of Content
S. No.
Article Title & Authors (Vol. 3, Issue. 1, March-2012) Robust Fault Tolerant Control with Sensor Faults for a Four-Rotor Helicopter Hicham Khebbache1, Belkacem Sait1 and Fouad Yacef
Page Nos
1.
1-13
2.
Scrutinize the Abrasion Phenomenon with the Brunt Forces in the Joints of the Robots and Proposed A New Way to Decrease That Khashayar Teimoori and Mahdi Pirhayati
14-20
3.
Knowledge Management in Success of ERP Systems Usman Musa Zakari Usman, Mohammad Nazir Ahmad
21-28
4.
A New Proposal Erica+ Switch Algorithm For Traffic Management Ehab Aziz Khalil
29-40
5.
Automated Test Jig for Uniformity Evaluation of Luminaries Deepa Ramane, Jayashri Bangali, Arvind Shaligram
41-47
6.
Tracking of Intruders on Geographical Map using IDS Alert Manish Kumar, M. Hanumanthappa, T.V. Suresh Kumar
48-54
7.
Design and Analysis of Multidielectric Layer Microstrip Antenna with Varying Superstrate Layer Characteristics Samir Dev Gupta, Amit Singh
55-68
8.
Low Cost Intelligent Arm Actuator Using Pneumatic Artificial Muscle with Biofeedback Control Salman Afghani, Yasir Raza, Bilal Haider
69-75
9.
Two Stage Discrete Time Extended Kalman Filter Scheme for Micro Air Vehicle Sadia Riaz and Ali Usman
76-83
10.
Identification of Suspicious Regions to Detect Oral Cancers at an Earlier Stage A Literature Survey K. Anuradha and K. Sankaranarayanan
84-91
11.
Waveform Analysis of Pulse Wave Detected in the Fingertip with PPG
92-100
Subhash Bharati & Girmallappa Gidveer

12. Slotted PIFA with Edge Feed for Wireless Applications 101-110
Vol. 3, Issue 1, pp. i-vi
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 T. Anita Jones Mary, T. Joyce Selva Hephzibah, C. S. Ravichandran 13. A Novel Design of Multiband Square Patch Antenna Embeded with Gasket Fractal Slot for WLAN & WiMAX Communication Amit K. Panda and Asit K. Panda 14. Implementation of the Triple-DES Block Cipher using VHDL Sai Praveen Venigalla, M. Nagesh Babu, Srinivas Boddu, G. Santhi Swaroop Vemana 15. Computational Investigation of Performance Characteristics in a C-shape Diffusing Duct Prasanta. K. Sinha, A. N. Mullick, B. Halder and B. Majumdar 16. Guidelines in Selecting a Programming Language and a Database Management System Onkar Dipak Joshi, Virajit A. Gundale, Sachin M. Jagdale 17. Audio Steganalysis of LSB Audio using Moments and Multiple Regression Model Souvik Bhattacharyya1 and Gautam Sanyal 18. Analysis and Comparison of Combinational Circuits by using Low Power Techniques Suparshya Babu Sukhavasi, Susrutha Babu Sukhavasi, Vijaya Bhaskar M, B Rajesh Kumar 19. High Quality Design and Methodology Aspects to Enhance Large Scale Web Services Suryakant B Patil, Sachin Chavan, Preeti Patil 20. Energy Efficient Cluster Based Key Management Technique for Wireless Sensor Networks T. Lalitha and R. Umarani 21. On Product Summability of Fourier Series using Matrix Euler Method B. P. Padhy, Banitamani Mallik, U. K. Misra and Mahendra Misra 22. Material Handling and Supply Chain Management in Fertilizer Production A Case Study T. K. Jack 23. VLSI Implementation of Error Tolerance Analysis for Pipeline based DWT in JPEG 2000 Encoder Rajamanickam. G & Jayamani. S 200-208 197-199 191-196 186-190 175-185 161-174 145-160 137-144 129-136 117-128 111-116
ii
24.
Location Based Services in Android Ch. Radhika Rani, A. Praveen Kumar, D. Adarsh, K. Krishna Mohan, K. V. Kiran
209-220
25.
A Robust Design and Simulation of Efficient Micromechatronic System using FSV Controlled Auxiliary Damped PMLSM Sarin CR, Ajai M, Santhosh Krishnan, Arul Gandhi
221-232
26.
Influence of Filler Material on Glass Fiber/ Epoxy Composite Laminates During Drilling M.C.Murugesh1 and K. Sadashivappa
233-239
27.
LFSR Test Pattern for Fault Detection and Diagnosis for FPGA CLB Cells Fazal Noorbasha, K. Harikishore, Ch. Hemanth, A. Sivasairam, V. Vijaya Raju
240-246
28.
Fuzzy Based Rotor Ground Fault Location Method for Synchronous Machines Mohanraj. M, Rani Thottungal and Manobalan. M
247-254
29.
Turning Parameter Optimization for Surface Roughness of ASTM A242 Type-1 Alloys Steel by Taguchi Method Jitendra Verma, Pankaj Agrawal, Lokesh Bajpai
255-261
30.
Prediction of Heat-Release Patterns for Modeling Diesel Engine Performance and Emissions B.VenkateswaraRao and G. Amba Prasad Rao
262-269
31.
Multi-Objective Optimization of Cutting Parameters for Surface Roughness and Metal Removal Rate in Surface Grinding using Response Surface Methodology M. Janardhan and A. Gopala Krishna
270-283
32.
Harmonic Reduction in Cascaded Multilevel Inverter with Reduced Number of Switches using Genetic Algorithms C. Udhaya Shankar, J. Thamizharasi, Rani Thottungal, N. Nithyadevi
284-294
33.
Fingerprint Based Gender Identification using Frequency Domain Analysis Ritu Kaur and Susmita Ghosh Mazumdar
295-299
34.
Applying the Genetic Algorithms of Sorting the Elitist Non-Decisive Solutions in the Case Study "Resource Allocation"
300-305
iii
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 Khashayar Teimoori 35. Power Conditioning in Battery Chargers using Shunt Active Power Filter through Neural Network P. Thirumoorthi, Jyothis Francis and N.Yadaiah 36. Data Leakage Detection Archana Vaidya, Prakash Lahange, Kiran More, Shefali Kachroo & Nivedita Pandey 37. Extraction of Visual and Acoustic Features of the Driver for Real-Time Driver Monitoring System Sandeep Kotte 38. Review on Theory of Constraints CH. Lakshmi Tulasi, A. Ramakrishna Rao 39. Turbo Brake Assists TURBOTOR: Evaluation G. Ramkumar and Amrit Om Nayak 40. Structural Refinement by Rietveld Method and Magnetic Study of Nanocrystalline Cu-Zn ferrites K. S. Lohar, S. M. Patange, Sagar E. Shirsath, V. S. Surywanshi, S. S. Gaikwad, Santosh S. Jadhav and Nilesh Kulkarni 41. Implementation and Control Different Multilevel Inverter Topologies for Current Waveform Improvement T. Shanthi and K. Karthikadevi 42. Performance Evolution of High Speed Network Carrying Multimedia Ehab Aziz Khalil 43. Application of Newton-Based OPF in Deregulated Power Systems T. Nireekshana, G. Kesava Rao and S. Siva Naga Raju 44. Secured Energy Efficient Hybrid Routing in Wireless Sensor Networks Deepika Srikumar and Seethalakshmi Vijaykumar 45. Co-operative P2P Information Exchange (cPIE) using Clustering Approach in Wireless Network S. Nithya1 & K. Palanivel 46. Performance Evaluation of Object Recognition Using Skeletal Shock Graph: Challenges and Future Prospects Pankaj Manohar Pandit, Sudhir Gangadharrao Akojwar iv Vol. 3, Issue 1, pp. i-vi 418-426 405-417 395-404 381-394 371-380 362-370 354-361 Conceptual Development and 345-353 334-344 322-333 315-321 306-314
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 47. An Integrated Color and Hand Gesture Recognition Control for Wireless Robot Deepali Suryawanshi and C. S. Khandelwal 48. Conventional Ethanol Reforming Technology Developments for the Production of Hydrogen Prashant Tayade, Vilas Sapkal, Chandrasekhar Rode, Rajendra Sapkal 49. An Efficient Bit Rate Performance of Serial-Serial Multiplier with 1s Asynchronous Counter P. Rajesh and B. Gopinath 50. Controller Area Network Data Extraction for Automobile Tejal Farkande and S. N. Pawar 51. Comparative Study of Experimental and Theoretical Load Carrying Capacity of Stone Column with and without Encasement of Geosynthetics Kameshwar Rao Tallapragada & Golait Y. S. 52. A Novel Face Recognition Approach using a Multimodal Feature Vector Jhilik Bhattacharya, Nattami Sekhar Lakshmi Prabha, Somajyoti Majumder, Gautam Sanyal 53. Influence of Geofabrics in the Construction of Pavements on Expansive Clayey Subgrades A. V. Narasimha Rao, D. Neeraja 54. E-Learning System Analysis using Smart Aided Tools through Web Services M. Balakrishnan and K. Duraiswamy 55. The Impact of Electronic Customer Relationship Management on Consumers Behavior Usman Musa Zakari Usman, Abdullah Nabeel Jalal, Mahdi Alhaji Musa 56. Performance & Emission Studies on a Single Cylinder DI Diesel Engine Fueled with Diesel & Rice Bran Oil Methyl Ester Blends Bhabani Prasanna Pattanaik, Basanta Kumar Nanda and Probir Kumar Bose 57. Power Flow Analysis of Three Phase Unbalanced Radial Distribution System Puthireddy Umapathi Reddy, Sirigiri Sivanagaraju, Prabandhamkam Sangameswararaju 514-524 505-513 500-504 494-499 487-493 477-486 466-476 460-465 451-459 436-450 427-435
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 58. Hyperspectral Image Classification using M-Band Wavelet Transform K. Kavitha, S. Arivazhagan, R. Dhivya Priya 59. Modeling and Controller Design of an Industrial Oil-Fired Boiler Plant Orosun Rapheal. and Adamu Sunusi Sani 60. Internet Based Remote Monitoring and Control System Monita N. Jadhav and G. R. Gidveer 61. Advanced Control Techniques in Variable Speed Stand Alone Wind Turbine System V. Sharmila Deve and S. Karthiga 62. Analysis of Unified Power Quality Conditioner during Power Quality Improvement B. Sridhar , Rajkumar Jhapte 63. Design of Multi Bit LFSR PNRG and Performance Comparison on FPGA using VHDL Amit Kumar Panda, Praveena Rajput, Bhawna Shukla 64. Numerical Analysis of the Viscoelastic Fluid in Plane Poiseuille Flow N. Khorasani and B. Mirzalou 65. Laser Diode Modelling for Wireless Power Transmission Emanuele Scivittaro and Anna Gina Perri 66. Effect of Process Parameters on the Physical Properties of Wires Produced by Friction Extrusion Method A. Hosseini, E.Azarsa, B.Davoodi, Y.Ardahani 67. Fabrication of PP/Al2O3 Surface Nanocomposite via Novel Friction Stir Processing Approach Shahram Alyali1, Amir Mostafapour2, Ehsan Azarsa 68. Failed Steam Traps: First Steps to Replacement Kant E Kanyarusoke and Ian Noble-Jack Members of IJAET Fraternity A-H 606-617 598-605 592-597 581-591 572-580 566-571 558-565 559-557 542-548 534-541 525-533
vi
ROBUST FAULT TOLERANT CONTROL WITH SENSOR FAULTS FOR A FOUR-ROTOR HELICOPTER
Hicham Khebbache1, Belkacem Sait1 and Fouad Yacef 2
Automatic Laboratory of Setif (LAS), Electrical Engineering Department, Setif University, Algeria 2 Electrical Engineering Study and Modelling Laboratory (LAMEL), Automatic Control Department, Jijel University, Algeria
1
ABSTRACT
This paper considers the control problem for an underactuated quadrotor UAV system in presence of sensor faults. Dynamic modelling of quadrotor while taking into account various physical phenomena, which can influence the dynamics of a flying structure is presented. Subsequently, a new control strategy based on robust integral backstepping approach using sliding mode and taking into account the sensor faults is developed. Lyapunov based stability analysis shows that the proposed control strategy design keep the stability of the closed loop dynamics of the quadrotor UAV even after the presence of sensor failures. Numerical simulation results are provided to show the good tracking performance of proposed control laws.
KEYWORDS:
Backstepping control, Dynamic modelling, fault tolerant control (FTC), Nonlinear systems, Robust control, quadrotor, Unmanned aerial vehicles (UAV).
I.
INTRODUCTION
Unmanned aerial vehicles (UAV) have shown a growing interest thanks to recent technological projections, especially those related to instrumentation. They made possible the design of powerful systems (mini drones) endowed with real capacities of autonomous navigation at reasonable cost. Despite the real progress made, researchers must still deal with serious difficulties, related to the control of such systems, particularly, in the presence of atmospheric turbulences. In addition, the navigation problem is complex and requires the perception of an often constrained and evolutionary environment, especially in the case of low-altitude flights. In contrast to terrestrial mobile robots, for which it is often possible to limit the model to kinematics, the control of aerial robots (quadrotor) requires dynamics in order to account for gravity effects and aerodynamic forces [12]. In [8-15], authors propose a control-law based on the choice of a stabilizing Lyapunov function ensuring the desired tracking trajectories along (X,Z) axis and roll angle only. The authors in [4], do not take into account frictions due to the aerodynamic torques nor drag forces. They propose acontrollaws based on backstepping, and on sliding mode control in order to stabilize the complete system (i.e. translation and orientation). In [18-19], authors take into account the gyroscopic effects and show that the classical model-independent PD controller can stabilize asymptotically the attitude of the quadrotor aircraft. Moreover, they used a new Lyapunov function, which leads to an exponentially stabilizing controller based upon the PD2 and the compensation of coriolis and gyroscopic torques. While in [11] the authors develop a PID controller in order to stabilize altitude. The authors in [5] propose a control algorithm based upon sliding mode based on backstepping approach allowed the tracking of the various desired trajectories expressed in term of the center of mass coordinates along (X,Y,Z) axis and yaw angle. In [10], authors used a controller design based on backstepping approach. Moreover, they introduced two neural nets to estimate the aerodynamic components, one for aerodynamic forces and one for aerodynamic moments. While in [9] the authors propose a hybrid backstepping control technique and the Frenet-Serret Theory (Backstepping+FST) for attitude stabilization, that includes estimation of the desired angular acceleration (within the control law) as a
Vol. 3, Issue 1, pp. 1-13
function of the aircraft velocity. However, all control strategies previously proposed do not take into account the failures affecting the sensors of quadrotor, wich makes them very limited and induces undesired behavior of quadrotor, or even to instability of the latter after occurence of sensor faults. In this paper, we consider the stabilization problem of the quadrotor aircraft in presence of sensor failures. The dynamical model describing the quadrotor aircraft motions and taking into a account for various parameters which affect the dynamics of a flying structure such as frictions due to the aerodynamic torques, drag forces along (X,Y,Z) axis and gyroscopic effects is presented. Subsequently, a new control strategy based on robust integral backstepping approach using sliding mode taking into account the sensor faults is proposed and compared with a classical backstepping approach. This control strategy includes two compensation techniques, the first technique using an integral action, the second, is to use an another term containing "sign" function to compensate the effect of sensor faults. Finally all synthesized control laws are highlighted by simulations, a comparison with control strategies developed in this paper is also performed, these simulations show the inefficiency of classical backstepping approach after occurrence of sensor faults. Otherwise, fairly satisfactory results despite the presence of this sensor faults are given by the new control strategy.
II.
MODELLING
2.1. Quadrotor dynamic modelling The quadrotor have four propellers in cross configuration. The two pairs of propellers {1,3} and {2,4} as described in Fig. 1, turn in opposite directions. By varying the rotor speed, one can change the lift force and create motion. Thus, increasing or decreasing the four propellers speeds together generates vertical motion. Changing the 2 and 4 propellers speed conversely produces roll rotation coupled with lateral motion. Pitch rotation and the corresponding lateral motion; result from 1 and 3 propellers speed conversely modified. Yaw rotation is more subtle, as it results from the difference in the counter-torque between each pair of propellers. Let E (0,x,y,z) denote an inertial frame, and B (0,X,Y,Z) denote a frame rigidly attached to the quadrotor as shown in Fig. 1.
Figure 1. Quadrotor configuration
Using Euler angles parameterization [14], the airframe orientation in space is given by a rotation matrix R from RB to RE.
c c R = s c s
s s c s c c s c + s s s s s + c c cs s s c s c cc
(1)
where C and S indicate the trigonometrically functions cos and sin respectively. The dynamic equations based on the formalism of Newton-Euler are given by: & = && m = Ff + Ft + Fg & R = RS ( ) I = I M M + M & gh a f
(2)
S ( ) is a skew-symmetric matrix; for a given vector = [1 2 3 ] it is defined as follows:
T
0 S ( ) = 3 2
3 0 1
2 1 0
(3)
The approximate model of the quadrotor can be rewritten as: & = T 4 && m = R 0 0 (b i2 ) + K ft mge z i =1 & = RS ( ) R 4 T I = I J 0 0 ( 1)i +1 K 2 + M & r i fa f 1 where:
m =[x, y, z]T v
(4)
total mass of the structure position vector linear velocity roll angle pitch angle law angle angular velocity angular rotor speed i Ff resultant of the forces generated by the four rotors Ft resultant of the drag forces along (x,y,z) axis Fg gravity force Mgh resultant of torques due to the gyroscopic effects Ma resultant of aerodynamics frictions torques Mf moment developed by the quadrotor according to the body fixed frame Kft(x,y,z) translation drag coefficients Kfa(x,y,z) frictions aerodynamics coefficients I(x,y,z) body inertia Jr rotor inertia b lift coefficient d drag coefficient l distance from the rotors to the center of mass of the quadrotor aircraft The moment developed by the quadrotor according to the body fixed frame along an axis is the difference between the torque generated by each propeller on the other axis.
2 2 lb ( 4 2 ) Mf = lb ( 32 12 ) 2 2 2 2 d 1 2 + 3 4 The complete dynamic model which governs the quadrotor is as follows [5-6]:
(5)
I Iz ) K l && ( y && J & = r r & fax 2 + u 2 Ix Ix Ix Ix && && & = ( I z I x ) + J r K fay & 2 + l u 3 r Iy Iy Iy Iy ( I x I y ) & K faz & 2 + 1 u & && = 4 Iz Iz Iz x = K ftx x& + 1 u x u 1 && m m K y = fty y + 1 u y u 1 && & m m K cos( ) cos( ) && u1 z = ftz z& g + m m
with
(6)
u x = ( cos cos sin + sin sin ) u y = ( cos sin sin sin cos )
The systems inputs are posed u 1 , u 2 , u 3 , u 4 and r a disturbance, obtaining:
2 2 u1 = b 12 + 2 + 32 + 4 2 2 u 2 = lb (4 2 ) 2 2 u 3 = lb (3 1 ) 2 2 2 2 u 4 = d 1 2 + 3 4 r = 1 2 + 3 4
(7)
)
(8)
From (7) it easy to show that :
d = arcsin (u x sin ( d ) u y cos ( d ) ) (u x cos ( d ) + u y sin ( d ) ) d = arcsin cos (d )

2.2. Rotor dynamic model
(9)
The rotors are driven by DC motors with the well known equations: di v = Ri + L dt + k e (10) d k i = J +Cs + k r2 r m dt As we a small motor with a very low inductance, then we can obtain the voltage to be applied to each motor as follows [11]:
vi =
with:
& (
+ 0i2 + 1i + 2 , i [1,..., 4]
(11)
0 =
k k C kr k , 1 = e m , 2 = s and = m Jr JrR Jr JrR
where, vi : is the motor input, : is the angular speed, kr : is the load torque constant, ke, km : are respectively the electrical and mechanical torque constants, and Cs : is the solid friction.
III.
CONTROL STRATEGIES OF QUADROTOR
3.1. Backstepping control of quadrotor The model (6) presented in the first part of this paper can be written in the state-space form: X& = f ( X ,U ) with X n is the state vector of the system, such as: T T & & & & & X = [ x 1 ,..., x 12 ] = , , , , , , x , x , y , y , z , z& From (6) and (13) we obtain:
x2 2 a1 x 4 x 6 + a 2 x 2 + a3 r x 4 + b1u 2 x4 2 a 4 x 2 x 6 + a5 x 4 + a6 r x 2 + b 2u 3 x6 a7 x 2 x 4 + a8 x 62 + b 3u 4 x8 1 a9 x 8 + u x u 1 m x 10 1 a10 x 10 + u y u 1 m x 12 a x + cos( ) cos( ) u g 11 12 1 m
(12) (13)
(14)
with
I y Iz K fax J , a3 = r a1 = , a2 = Ix Ix Ix a = I z I x , a = K fa y , a = J r 4 6 Iy 5 Iy Iy (15) K faz K ft x K ft y K ft z Ix I y , a9 , a10 = , a11 = , a8 = a7 = Iz m m m Iz l l 1 b1 = ,b2 = ,b3 = Ix Iy Iz The adopted control strategy is summarized in the control of two subsystems; the first relates to the orientation control, taking into account the position control along (X,Y) axis, while the second is that of the attitude control as shown it below the synoptic:
zd xd yd ux uy
Attitude Control
u1 u2 u3 u4
Quadrotor
d
Passage
Position Control
Orientation Control
d
Figure 2. Synoptic scheme of the proposed control strategy
Using the backstepping approach for the control-laws synthesis, we simplify all the stages of calculation concerning the tracking errors and Lyapunov functions (Refer to [6] for more details) in the following way: x id x i i [1,3,5,7,9,11] (16) ei = & x (i 1)d + k (i 1)e(i 1) x i i [ 2,4,6,8,10,12] and 1 2 i [1,3,5, 7,9,11] 2 ei (17) Vi = 1 2 V + e i [ 2, 4, 6,8,10,12] i 1 2 i such as: k i > 0, i [1,...,12] The synthesized stabilizing control laws are as follows: 1 2 && u 3 = ( a4 x 2 x 6 a5 x 4 a6 r x 2 + d + k 3 ( k 3e 3 + e 4 ) + k 4e 4 + e 3 ) b2 1 2 && u 4 = ( a7 x 2 x 4 a8 x 6 + d + k 5 ( k 5e 5 + e 6 ) + k 6e 6 + e 5 ) b3 m && /u 1 0 u x = ( a9 x 8 + x d + k 7 ( k 7e 7 + e 8 ) + k 8e 8 + e 7 ) u1 m && /u 1 0 u y = ( a10 x 10 + y d + k 9 ( k 9e 9 + e10 ) + k 10e10 + e 9 ) u1 m && u 1 = ( g a11x 12 + z d + k 11 ( k 11e11 + e12 ) + k 12e12 + e11 ) cos( x 1 ) cos( x 3 )
(18)
3.2. Robust fault tolerant control with sensor faults of quadrotor

The choice of this method is not fortuitous considering the major advantages it presents:
It ensures Lyapunov stability. It ensures the handling of all system nonlinearities. It ensures the robustness and all properties of the desired dynamics in presence of sensor faults.
The complete model resulting by adding the sensor faults in the model (11) can be written in the statespace form: X& = f ( X ,U ) (19) Y = CX + E s Fs with Y p is the measured output vector, Fs q is the sensor faults vector, C p n and
E s p q are respectively, the observation matrix and the sensor faults distributions matrix, such as :
y = [ x1 x 2 + f s1 x 3 x 4 +f s 2 x 5 x 6 +f s 3 x 7 x 8 +f s 4 x 9 x10 + f s 5 x11 x12 +f s 6 ]
T
(20)
Remark 1: In our contribution, only the velocity sensor faults are considered, they are assumed to be bounded and slowly varying in time (i.e. f si f s 0 and f&si 0 ). Using this control strategy as a recursive algorithm, one can synthesize the control laws forcing the system to follow the desired trajectory. For the rst step we consider the tracking-error: (21) e 1 = x 1d x 1 And we use the Lyapunov theorem by considering the Lyapunov function e1 positive denite and its time derivative negative semi-denite: 1 1 (22) V 1 = e 12 + d%12 2 2
And its time derivative is then:
& & V&1 = e1 ( x 1d ( y 2 f s 1 ) ) + d%1 ( 1 )

The stabilization of e1 can be obtained by introducing a new virtual control y 2
(23)
( y 2 )d 1 = &d + k 1e1 + 11
(24)
In order to compensate the effect of the velocity sensor fault of roll motion, an integral term is introduced which can eliminate the tracking error. We take: (25) 1 = e1dt It results that:
k e1 %T % V&1 = e1 k 1e1 1d%1 + d%1 ( e1 ) = e1 d%1 1 1 = e1 Q1e1 1 0 d%1 k 1 and 1 are chosen so as to make the definite matrix positive Q1 , which means that, V&1 0 let us proceed to a variable change by making: & e 2 = x 1d + k 1e1 + 11 y 2 For the second step we consider the augmented Lyapunov function: 1 2 V 2 =V 1 + e 2 2 & = e k e d% + e + d% ( e ) + e ( b u ) V
(26)
(27)
(28) (29) (30) (31)
1 1
s1
1 2
Such as
2 && s 1 = ( x 1d + k 1 (k 1e1 + e2 ) 1e1 a1 y 4 y 6 a2 y 2 a3r y 4 ) + s 1
and if:
s 1 = k 11d% + a1 (f s 3 y 4 + f s 2 y 6 f s 2 f s 3 ) + a2 (f s 1 y 2 f s 2 ) + a3 r f s 2 1 1
with: ( s1) is the unknown part related to velocity sensor faults. The control input u 2 is then extracted satisfying V&2 0 1 && 2 u 2 = (d + k 1 ( k 1e1 + e 2 ) + (1 1 )e1 + k 2e 2 a4 y 2 y 6 a5 y 4 a6 r y 2 + 2sign (e 2 ) ) b1 The term k 2e 2 with k 2 > 0 is added to stabilize (e1 , e 2 ) . The same steps are followed to extract u 3 , u 4 ,u x , u y and u1 .
(32)
1 && d + k 3 ( k 3e 3 + e 4 ) + (1 3 )e 3 + k 4e 4 a4 y 2 y 6 a5 y 42 a6 r y 2 + 4sign (e 4 ) ) u 3 = b2 1 2 && u 4 = ( d + k 5 ( k 5e 5 + e 6 ) + (1 5 )e 5 + k 6e 6 a7 y 2 y 4 a8 y 6 + 6 sign (e 6 ) ) b3 (33) m && /u 1 0 u x = ( x d + k 7 ( k 7e 7 + e 8 ) + (1 7 )e 7 + k 8e 8 a9 y 8 + 8sign (e 8 ) ) u1 m && /u 1 0 u y = ( y d + k 9 ( k 9e 9 + e10 ) + (1 9 )e 9 + k 10e10 a10 y 10 + 10sign (e10 ) ) u1 m && u 1 = ( z d + k 11 ( k 11e11 + e12 ) + (1 11 )e11 + k 12e12 a11 y 12 + g + 12sign (e12 ) ) cos( x 1 ) cos( x 3 )
with
x id x i i [3,5,7,9,11] ei = x (i 1)d + k (i 1)e(i 1) + (i 1)(i 1) y i i [ 4,6,8,10,12] &

and
(34)
(35) sign (e i ) i [ 4, 6,8,10,12] The corresponding lyapunov functions are given by: 1 2 1 %2 i [3,5, 7,9,11] and j [ 2,..., 6] 2 ei + 2 d j (36) Vi = 1 2 V + e i [ 4, 6,8,10,12] i 1 2 i such as f sj % d j = d j i = i i [3, 5, 7, 9,11] and j [ 2,..., 6 ] i k i j (37) >0 i [3, 5, 7, 9,11] and j [ 2,..., 6 ] Q j = 1 0 k > 0 i [ 4, 6,8,10,12 ] i To synthesize a stabilizing control laws in the presence of velocity Sensor faults, the following necessary condition must be verified: (38) si = si sin i i [ 2,...,6]
i =
e i dt
i [3,5, 7,9,11]
( si ) represent the unknowns parts related to velocity sensor faults.

IV. SIMULATION RESULTS
In order to evaluate the performance of the controller developed in this paper, we performed two tests simulations. In the first test, a four sensor faults f sj , j [1, 2,3,6] are added in angular velocities and linear velocity of attitude Z of our system, with 100% amplitudes of these maximum values (i.e. max( x i ), i [ 2, 4, 6,12 ] ) at instants 15s, 20s, 25s and 30s respectively. In the second test, the same sensor faults are added at the same previous instants, but their amplitudes are increased of 200%. The data of this tested quadrotor are reported in appendix [11].
8 x 10
-3 -3
dp
y 2-Backstepping y 2-RFTCS Angular velocity of pitch [rad/s]
x 10
dp
y 4-Backstepping y 4-RFTCS Angular velocity of yaw [rad/s]
0.2
6 Angular velocity of roll [rad/s]
0.15
dp
y6-Backstepping y6-RFTCS
0.1
0.05
-2
-2
-4
10
15
20
25 30 temps[sec]
35
40
45
50
-4
10
15
20
25 30 temps[sec]
35
40
45
50
-0.05
10
15
20
25 30 temps[sec]
35
40
45
50
(a) evolution of angular velocity of roll

0.15 0.1 Linear velocity along X axis [m/s]
(b) evolution of angular velocity of pitch

0.15 0.1 Linear velocity along Y axis [m/s]
(c) evolution of angular velocity of yaw

0.35 0.3 linear velocity of Z attitude [m/s] 0.25 0.2 0.15 0.1 0.05 0 -0.05 -0.1 z dp y 12-Backstepping y 12-RFTCS
0.05
0.05
-0.05
-0.05
-0.1 x dp y8-Backstepping y8-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50
-0.1 y dp y 10-Backstepping y 10-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50
-0.15
-0.15
-0.2
-0.2
10
15
20
25 30 temps[sec]
35
40
45
50
(d) evolution of linear velocity along X axis
(e) evolution of linear velocity along Y axis
(f) evolution of linear velocity along Z axis
Figure 3. Tracking simulation results of the angular and linear velocities, Test 1.
12 10 8 6 4 2 0 -2 -4 x 10
-3
dp
y 2-Backstepping y 2-RFTCS Angular velocity of pitch [rad/s]
12 10 8 6 4 2 0 -2 -4
x 10
-3
dp
y 4-Backstepping y 4-RFTCS Angular velocity of yaw [rad/s]
0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 -0.05
dp
y 6-Backstepping y 6-RFTCS
Angular velocity of roll [rad/s]
10
15
20
25 30 temps[sec]
35
40
45
50
10
15
20
25 30 temps[sec]
35
40
45
50
10
15
20
25 30 temps[sec]
35
40
45
50
(a) evolution of angular velocity of roll

0.15 0.1 Linear velocity along X axis [m/s]
(b) evolution of angular velocity of pitch

0.15 0.1 Linear velocity along Y axis [m/s]
(c) evolution of angular velocity of yaw

0.6 z dp y 12-Backstepping y 12-RFTCS 0.5 linear velocity of Z attitude [m/s]
0.05
0.05
0.4
0.3
-0.05
-0.05
0.2
-0.1 x dp -0.15 y8-Backstepping y8-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50
-0.1 y dp -0.15 y 10-Backstepping y 10-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50
0.1
-0.2
-0.2
-0.1
10
15
20
25 30 temps[sec]
35
40
45
50
(d) evolution of linear velocity along X axis
(e) evolution of linear velocity along Y axis
(f) evolution of linear velocity along Z axis
Figure 4. Tracking simulation results of the angular and linear velocities, Test 2.
Figure 3 and Figure 4 shows at outset a very good tracking of the desired velocities, but upon the appearance of sensor faults, the measurements of angular velocities and linear velocity of attitude Z (illustrated respectively by (a), (b), (c), and (f)) are deviated to their desired velocities, with 100% amplitude of her maximum values in Test. 1, and 200% amplitude of these maximum values in Test. 2 for both control strategies developed in this paper, which gives us a wrong information of the velocities of our system.
10 8 6 4 Pitch angle[rad] Yaw angle[rad] Roll angle[rad] 2 0 -2 -4 -6 -8 2 0 -2 0.2 -4 -6 -8 0.6 x 10
-3
x 10
-3
1.2
d -Backstepping -RFTCS
6 4
1
0.8
0.4
10
15
20
25 30 Time[sec]
35
40
45
50
10
15
20
25 30 Time[sec]
35
40
45
50
-0.2
10
15
20
25 30 Time[sec]
35
40
45
50
(a) evolution of roll angle

0.3 0.3 0.2 0.2
(b) evolution of pitch angle

6 5
(c) evolution of yaw angle

Zd Z-Backstepping Z-RFTCS
0.1 X position [m] Y position [m]
0.1 Z attitude [m] Yd Y-Backstepping Y-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50
-0.1
-0.1
-0.2 Xd X-Backstepping X-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50
-0.2
-0.3
-0.3
-0.4
-0.4
-1 0
10
15
20
25 30 Time[sec]
35
40
45
50
(d) evolution of x position
(e) evolution of y position
(f) evolution of z position
Figure 5. Tracking simulation results of trajectories along roll ( ), pitch ( ), yaw angle ( ) and (X,Y,Z) axis, Test 1.
0.015 12 x 10
-3
1.2
d
0.01
-Backstepping -RFTCS
10 8 6 Pitch angle[rad]
1
0.8 Yaw angle[rad]
Roll angle[rad]
0.005
4 2 0 -2
0.6
0.4
0.2
-0.005
-4 0 -6
-0.01
10
15
20
25 30 Time[sec]
35
40
45
50
-8
10
15
20
25 30 Time[sec]
35
40
45
50
-0.2
10
15
20
25 30 Time[sec]
35
40
45
50
(a) evolution of roll angle

0.3 0.3 0.2 0.2
(b) evolution of pitch angle

6 5
(c) evolution of yaw angle

Zd Z-Backstepping Z-RFTCS
0.1 X position [m] Y position [m]
0.1 Z attitude [m] Yd Y-Backstepping Y-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50
-0.1
-0.1
-0.2 Xd X-Backstepping X-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50
-0.2
-0.3
-0.3
-0.4
-0.4
-1 0
10
15
20
25 30 Time[sec]
35
40
45
50
(d) evolution of x position
(e) evolution of y position
(f) evolution of z position
Figure 6. Tracking simulation results of trajectories along roll ( ), pitch ( ), yaw angle ( ) and (X,Y,Z) axis, Test 2.
According to Figure 5 and Figure 6, there is a very good tracking of the desired trajectories for both control strategies at outset, but after the occurrence of sensor faults in angular velocities and linear velocity of the attitude Z, the actual trajectories along roll angle, pitch angle, yaw angle, and z position (illustrated respectively by (a), (b), (c), and (f)) corresponding of classical backstepping approach are deviated to their desired values, which also explains the inefficiency of this control approach after occurrence of velocity sensor faults. However, the actual trajectories corresponding of this new control strategy converge to their desired trajectories after transient peaks with low amplitudes caused by the appearance of velocity sensor faults.
5.1 u1-RFTCS 5.05 5 3 4.95 Input control u [N.m] Input control u [N]
1
5 u1-Backstepping 4
x 10
-4
u2-RFTCS u2-Backstepping
4.9 4.85 4.8 4.75 4.7 4.65 4.6
2 1 0 -1 -2 -3 0
10
15
20
25 30 Time[sec]
35
40
45
50
10
15
20
25 30 Time[sec]
35
40
45
50
(a) evolution of input control u1

4 3 2 Input control u [N.m] 1 0 -1 -2 -3 -4 0 Input control u [N.m] x 10
-4
(b) evolution of input control u2

6 x 10
-4
u3-RFTCS u3-Backstepping 4
-2
-4
-6
10
15
20
25 30 Time[sec]
35
40
45
50
-8
10
15
20
25 30 Time[sec]
35
40
45
50
(c) evolution of input control u2
(d) evolution of input control u4
Figure 7. Simulation results of all controllers, Test 1.
10
5.1 5.05 5 3 4.95 Input control u [N.m] Input control u [N]
1
5 4
x 10
-4
4.9 4.85 4.8 4.75 4.7 4.65 4.6
2 1 0 -1 -2 -3 0
10
15
20
25 30 Time[sec]
35
40
45
50
10
15
20
25 30 Time[sec]
35
40
45
50
(a) evolution of input control u1

4 x 10
-4
(b) evolution of input control u2

1 x 10
-3
u3-RFTCS 3 u3-Backstepping 0.5 2 Input control u [N.m] Input control u [N.m] 0
-0.5
-1 -1 -2
-3
10
15
20
25 30 Time[sec]
35
40
45
50
-1.5
10
15
20
25 30 Time[sec]
35
40
45
50
(c) evolution of input control u2
(d) evolution of input control u4
Figure 8. Simulation results of all inputs control, Test 2.
From Figures 7 and 8, it is clear to see that the classical backstepping approach give a smooth inputs control, with transient peaks during the appearance of sensor faults in angular velocities and linear velocity of the attitude Z, However, the inputs control corresponding to RFTCS are characterized at outset by very fast switching caused by using of the discontinuous compensation term "sign", also with transient peaks during the appearance of velocity sensor faults, but this chattering gone after the appearance of sensor fault in linear velocity of the attitude Z.
3D position 3D position
Z coordinate [m]
Z coordinate [m] Reference trajectory Real trajectory using Backstepping Real trajectory using RFTCS 0 -0.2 Y coordinate [m] -0.4 -0.4 -0.2 X coordinate [m] 0.2 0
-2 0.4 0.2
-2 0.4 0.4 0.2 0 -0.2 Y coordinate [m]
Reference trajectory Real trajectory using Backstepping Real trajectory using RFTCS 0.2 0 -0.2 -0.4 -0.4 X coordinate [m]
0.4
(a) evolution of position along (X,Y,Z) axis, Test 1
(b) evolution of position along (X,Y,Z) axis, Test 2
Figure 9. Global trajectory of the quadrotor in 3D.
The simulation results given by Figure 9 show the efficiency of the robust fault tolerant control strategy developed in this paper, which clearly shows a good performances and robustness towards stability and tracking of this control strategy with respect to the backstepping approach after the occurrence of velocity sensor faults.
V.
CONCLUSION AND FUTURE WORKS
In this paper, we proposed a new strategy of fault tolerant control based on backstepping approach and including the velocity sensor faults. Firstly, we start by the development of the dynamic model of the quadrotor taking into account the different physics phenomena which can influence the evolution of our system in the space, and secondly by the synthesis a stabilizing control laws in the presence of
11
velocity sensor faults. The simulation results have shown that the backstepping approach renders the system unable to follow his reference after the appearance of velocity sensor faults, which also explains the sensitivity of this control technology towards a sensor failures. However, these simulation results also shows a high efficiency of the robust fault tolerant control strategy developed in this paper, she preserves performance and stability of quadrotor during a malfunction of these velocity sensors. As prospects we hope to develop other fault tolerant control strategies in order to eliminate the chattering phenomenon in the inputs control, while maintaining the performance and stability of this system with implementation them on a real system.
APPENDIX
m g l b d Jr I(x,y,z) Kfa(x,y,z) Kft(x,y,z) 0 1 2 0,486 kg 9,806 m/s2 0,25 m 2,9842 105 N/rad/s 3,2320 107 N.m/rad/s 2,8385 105 kg.m2 diag (3,8278; 3,8288; 7,6566) 103 kg.m2 diag (5,5670; 5,5670; 6,3540) 104 N/rad/s diag (5,5670; 5,5670; 6,3540) 104 N/m/s 0,0122 6,0612 189,63 280,19
REFERENCES
[1] [2] [3] [4] [5] M. Blanke, R. Azadi-Zamanabadi, SA. Bgh, & CP. Lunau. (1997) Fault tolerant control systems, Control Eng. Prcatice, pp. 693-702. M. Blanke, M. Kinnaert, J. Lunze, & M. Staroswiecki. (2003) Diagnosis and fault-tolerant control, Springer-Verlag. S. Bouabdellah, P.Murrieri, & R.Siegwart. (2004) Design and control of an indoor micro quadrotor, Proceeding of the IEEE, ICRA, New Orleans, USA. S. Bouabdellah, & R.Siegwart. (2005) Backstepping and sliding mode techniques applied to an indoor micro quadrotor, Proceeding of the IEEE, ICRA, Barcelona, Spain. H. Bouadi, M. Bouchoucha, & M. Tadjine. (2007) Sliding Mode Control Based on Backstepping Approach for an UAV Type-Quadrotor, International Journal of Applied Mathematics and Computer Sciences, Barcelona, Spain, Vol.4, No.1, ISSN 1305-5313. H. Bouadi, M. Bouchoucha, & M. Tadjine. (2007) Modelling and Stabilizing Control Laws Design Based on Backstepping for an UAV Type-Quadrotor, Proceeding of 6 the IFAC Symposium on IAV, Toulouse, France. M. Bouchoucha, M. Tadjine, A. Tayebi, & P. Mllhaupt. (2008) Step by Step Robust Nonlinear PI for Attitude Stabilization of a Four-Rotor Mini-Aircraft, 16th Mediterranean Conference on Control and Automation Congress Centre, Ajaccio, France, pp. 1276-1283. P. Castillo, A. Dzul, & R. Lozano. (2004) Real-Time Stabilization and Tracking of a Four-Rotor Mini Rotorcraft, IEEE Transactions on Control Systems Technology, Vol. 12, No. 4, pp. 510-516. J. Colorado, A. Barrientos, , A. Martinez, B. Lafaverges, & J. Valente. (2010) Mini-quadrotor attitude control based on Hybrid Backstepping & Frenet-Serret Theory, Proceeding of the IEEE, ICRA, Anchorage, Alaska, USA. A. Das, F. Lewis, K. Subbarao (2009) Backstepping Approach for Controlling a Quadrotor Using Lagrange Form Dynamics. Journal of Intelligent and Robotic Systems, Vol. 56, No. 1-2, pp. 127- 151. L. Derafa, T. Madani, & A. Benallegue. (2006) Dynamic modelling and experimental identification of four rotor helicopter parameters, IEEE-ICIT , Mumbai, India, pp. 1834-1839. N. Guenard, T. Hamel, & R. Mahony. (2008) A Practical Visual Servo Control for an Unmanned Aerial Vehicle, IEEE Transactions on Robotics, Vol. 24, No. 2, pp. 331-340. T. Hamel, R. Mahony, R. Lozano, & J. Ostrowski. (2002) Dynamic modelling and configuration stabilization for an X4-flyer, IFAC world congress, Spain. W.Khalil, Dombre. (2002) modelling, identification and control of robots, HPS edition.
[6]
[7]
[8] [9]
[10] [11] [12] [13] [14]
12
[15] [16] R. Lozano, P. Castillo, & A. Dzul. (2004) Global stabilization of the PVTOL: real time application to a mini aircraft, International Journal of Control, Vol 77, No 8, pp. 735-740. A. Mokhtari, & A. Benallegue. (2004) Dynamic Feedback Controller of Euler Augles and Wind Parameters Estimation for a Quadrotor Unmanned Aerial Vehicle, Proceeding of the IEEE, ICRA, New Orleans, LA, USA, pp. 2359-2366. J.J.E. Slotine, & W. Li. (1991) Applied nonlinear control, Prentice Hall, Englewood Cliffs, NJ. A. Tayebi, & S. Mcgilvray. (2004) Attitude stabilisation of a four rotor aerial robot, IEEE Conference on Decision and Control, Atlantis Paradise Island, Bahamas, pp. 1216-1217. A. Tayebi, & S. McGilvray (2006) Attitude stabilization of a VTOL Quadrotor Aircaft, IEEE Transactions on Control Systems Technology, Vol. 14, No. 3, pp. 562-571.
[17] [18] [19]
Authors Hicham KHEBBACHE is Graduate student (Magister) of Automatic Control at the Electrical Engineering Department of Setif University, ALGERIA. He received the Engineer degree in Automatic Control from Jijel University, ALGERIA in 2009. Now he is with Automatic Laboratory of Setif (LAS). His research interests include Aerial robotics, Linear and Nonlinear control, Robust control, Fault tolerant control (FTC), Diagnosis, Fault detection and isolation (FDI).
Belkacem SAIT is Associate Professor at Setif University and member at Automatic Laboratory of Setif (LAS), ALGERIA. He received the Engineer degree in Electrical Engineering from National Polytechnic school of Algiers (ENP) in 1987, Magister degree in Instrumentation and Control from HCR of Algiers in 1992, and Ph.D. in Automatic Product from Setif University in 2007. His research interested include discrete event systems, hybrid systems, Petri nets, Fault tolerant control (FTC), Diagnosis, Fault detection and isolation (FDI).
Fouad YACEF is currently a Ph.D. student at the Automatic Control Department of Jijel University, ALGERIA. He received the Engineer degree in Automatic Control from Jijel University, ALGERIA in 2009, and Magister degree in Control and Command from Military Polytechnic School (EMP), Algiers, in November 2011. His research interests include Aerial robotics, Linear and Nonlinear control, LMI optimisation, Analysis and design of intelligent control systems.
13
SCRUTINIZE THE ABRASION PHENOMENON WITH THE BRUNT FORCES IN THE JOINTS OF THE ROBOTS AND PROPOSED A NEW WAY TO DECREASE THAT
Khashayar Teimoori and Mahdi Pirhayati
Department of Mechanical and Aerospace Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
ABSTRACT
In this paper the influences of the brunt forces on the Robots' joints is studied. Moreover, the efficacious ways that we can use to have a little corrosion in the joints are suggested. We know that through the movements of the robots we have a lot of friction phenomenon between the main body and the articulation. We propose this way that: when we use one viscoelastic material between the articulation and the main body, we can decrease the corrosion phenomenon when brunt or sudden hit forces have occurred. To avoid the exasperation of this problem we already studied on the forces of unpredictable and we shown that how can we eradicate this problem. At the end of this article we conclude that viscoelastic materials have more useful traits to use them in many cases especially at the joints of the robots, because one of the best natures of them are rebounded if any forces act on those.
KEYWORDS: Articulation, Brunt Forces, Viscoelastic Material (VEM), Joint, Torque, Shoulder roll
I.
INTRODUCTION
In any robots we have many articulations that are rotating between the joints, and through this processes, abrasion phenomenon has occurred. Especially when we have brunt or hit forces in the legs or hands of the robots corrosion phenomenon is going to be increased.
p
x
y
Figure.1: Species forces and torques can be applied in the joints [1], [2, 6]
All the hit forces that are happened in the form of unpredictable event can be jeopardous for the articulations, because friction should be happen and cause of amputate the members of the body. To exasperate this problem, at the first we try to realize forces and their working, after that we proposed a new way that is use of VEM. We exemplify our new way with the ANSYS software result and finite element analysis.
I.1
Joints and their structures:
Fig.1. shows the joints and their working in the normal position. [7] In the static state, all the forces and the torques that we have (related to the dynamic and static positions) can be showed by
14
Vol. 3, Issue 1, pp. 14-20
Five equations that are clearly introduced the works of articulations, when they are rotating, but these equations are stated just for static analysis of the joints.
z
P
w
v
u
y
Figure.2: Rotation with respect to the origin [1,6]

n
1.
2. 3. 4. 5.
F
i =1
xi
=0,
n xi
Eq.1 : shows the sum of the forces in x direct
M
i =1 j =1 k =1
M yj M zk = 0 ,
Eq.2 : shows the Torques in the free space Eq.3 : situation of each points in the spaces Eq.4 : as the same with 3 in the different coordinate Eq.5 : way of translate of two coordinates
Z
r Pxyz = p x i x + p y jy + p z k z , r Puvw = pu i u + pv jv + pw k w ,
Pxyz = RPuvw .
6.
= (1 , 2 , 3 , 4 , 5 , 6 ,L n )
7.
Y = ( x , y , z ,O , A ,T )
X
Figure.3: End-effectors position and orientation, Formula
15
Vol. 3, Issue 1, pp. 14-20
Figure.4: Way of the rotation, translation about the coordinates with the frame hand [7, 4]
And rotation of the hands or legs of the robots respect to the origin can be represented by Fig.3. [3]
Cartesian: PPP
Cylindrical: RPP
Spherical: RRP
Articulated: RRR
SCARA: RRP
(Selective Compliance Assembly Robot Arm)
Hands coordinate: N: normal vector; s: sliding vector; A: approach vector, normal to the tool mounting plate
Figure.5: Robot Configuration with the positions of coordinates [4]
Now we can scrutiny in the distances between the articulations and the main wall of the body that is important to study on the abrasion phenomenon. Spaces between two fingers or two joints are illustrated in the Fig. 4. In every rotation or any sudden hit we have a lot of excess forces that are occur between them, and we should prevent from that problem to arise. [7] This occurrence is extremely affected on the articulations of the robots. At this moment we should focus on the abrasion phenomenon. To decrease this problem in this position we should proposed to design one of the materials for filling the lacuna and after that trying multifarious Materials, Finally we chose Viscoelastic material in many reasons. Now we try to show why VEM is a good material for using it between the joints. After try to design one sample joint in ANSYS software we pictured Fig.5 from the result of the program to find out that we have maximum principal elastic strain in the brunt or excess hit forces, when we utilize VEM. This way can produced more convenient conditions to rotating and decreased friction and excess torques while they can cause of the exhaustion.
16
Vol. 3, Issue 1, pp. 14-20
Picture1. An articulated robot, that is showed to illustrated a situation of the joints [6].
Figure.6: Finite element analysis. The figure symbolize the simulated strain field or brunt field for the shoulder roll
To illustrated more facts about of the advantages of this material we Designed one sample arm (shoulder roll) that contains it joint's. In the 4 section Fig. 6. the positions of the arm (shoulder roll) has showed.[5,6]
17
Vol. 3, Issue 1, pp. 14-20
In these pictures we use one VEM between the joints and then we can conclude from 4 sections (a-d) it has more convenient to rotating with this material but if we didnt use this material maybe we had a friction or break in this joint (but this problem is showed after many movements for example the break event in this joint maybe happened after 2 year or 20000 moving around the all orientations). [6]
(a)
(b)
(c)
(d)
Figure.7: Sample positions of the shoulder roll joint in a free field of hit: a), d) rotation of the joints about x-y plane, b) rotation of the joint about the x-y plane, c) rotation of the joint about y-z plane
I.2
RESULTS:
Figure.8: Effect of the Viscosity of the main material that we use in the spaces of the joints [8]
18
Vol. 3, Issue 1, pp. 14-20
Finally we plotted Fig.7 to show how the viscosities of the materials effects on the joints can decrease the abrasion phenomenon with increase the friction when the brunt forces acting on those. At the moment we should mention this advantage point that is the critical point (RV=0), when it shows us while the rates of the viscosity exceeded the mentioned level, it would be effect less, and we should reconsider to choose the better one according to our response in any situation.[9]
II.
ANALYSIS OF RESULTS AND DISCUSSION
As a brief conclusion of our working we should explicated our new achievement in one Table that is clearly demonstrated the application of our proposed material (VEM). As an example we introduced four models that are shows us this fact: how can we choose the best material to have a few shear rates and shear stresses in the articulations rotation, while we don't have any way to check this exactly. In this diagram differences between the effects of the three models are showed. Then it has caparisoned between the rheological constants through Newton's, Bingham plastic, Power law and HerschelBulkley models that all of them can be choose as a best model.
Table.1: Effects of the Rheological Constants on the shear rates and shear stresses that are occur when the articulations are going to rotating. [9]
Newton
Bingham plastic
Power law
Hrschel-bulkley
= 0.022 pas
y = 3.57 Pa p = 0.0167 Pas

k = 0.671
y = 1.15Pa k = 0.362
We can use a lot of models to choose the best material for our working, Table 1. says each joints of the robots can easily simulated by ANSYS software and it is very comfortable to use one of the above models in that software for analyzing the shear rates and shear stresses that are occurs when the Brunt forces or sudden Forces acting on the joints. [10] By doing this way we can choose our suitable material to decrease the abrasion phenomenon between the joints. [11]
III.
CONCLUSIONS AND FUTURE SCOPE
In any rotations of an articulations about the joints of the robots, we have a lot of corrosion phenomenon and this problem is going to be increase in every movement, but we can easily decrease this problem even if going to be zero in many situations by using one viscoelastic material between the spaces that are exists between the joints and bulbs. But if we have a good outcome to get to this target, we should design an applicatory space between the main body and joints, or we should try to find reasonable materials to accommodate in them, and this action should be scrutiny exactly to have a few abrasion phenomenons. Finally, in this paper we attempted to illustrate many facts about that we can use Viscoelastic materials, because they can prepare more convenient to rotating and it is more reasonable way to use them.
ACKNOWLEDGEMENT
We wish to thank Dr. N Ashrafi for helpful discussion.
REFERENCES
[1]. S. C. Jacobsen, E. K. Iversen, D, F. Knutti, R. T. Johnson, and K.B. Biggers, Design of the Utah/M.I.T. Dextrous Hand, Proc. IEEE International Conference on Robatics and Automation, pp. 15201532.1986 [2]. J. Zhang, G. Guo and W.A. Gruver, Optimal Design of a Six-Bar Linkage for an Anthropomorphic Three-Jointed Finger Mechanism,Proceedings of the ASME Mechanisms Conference, Phoenix,vol. DE-45, pp. 299304, 1992.
19
Vol. 3, Issue 1, pp. 14-20
[3]. N. Dechev, W.L.Cleghorn and S.Naumann, Multiple Finger, Passive Adaptive Grasp Prosthetic Hand, Mechanism and Machine Theory, Vol. 36, No. 10, pp. 11571173, 2001. [4]. A.Bicchiand,G.Tonietti,Fastandsoftarmtactics[robotarmdesign],Robotics&AutomationMagazine,I EEE,vol.11,no.2,pp.2233,June2004.[Online].Available: http://dx.doi.org/10.1109/MRA.2004.1310939 [5]. Machine Intelligence & Robotic Control, Vol.4, No. 3, 1-11 (2002) [6]. Ashrafi, N., "Statistical Analysis of the Articulating Surfaces of the Elbow" IASTED, Aug. 1997, Singapore. [7]. Ashrafi N., Analysis of Articulation of the Natural Joints by Contact Kinematics Second International Conference on Mechanical Engineering, May 1996, Shiraz, Iran. [8]. Ashrafi, N., "Analytical investigation of Articulation Incorporating Elasticity and Contact Kinematics," International Conference of Mechanics in Biology, June 30 - July 4, 1996, Slovenia. [9]. N. Ashrafi and H. Karimi Haghighi, " Stability of Shear thickening and Shear thinning Fluids in Narrow Gap between Rotating Cyllinders", MAJLESI Journal of Mechanical Engineering, Vol.4No.2-2011 (Serial No. 14)- P. 65-76. [10]. Siciliano, Khatib (Eds.), Springer Handbook of Robotics, ISBN: 978-3-540-23957-4,
Springer-Verlag Berlin Heidelberg, 2008, p. 371-378.

[11]. Chowdhury D, Prediction of Standpipe Pressure Using Real, Time Data, MSc Thesis Report (for the course TPG 4920),IPT, NTNU, Trondheim, 2009
Authors
Khashayar Teimoori was born in Tehran, Iran in 1992. Currently he is pursuing B.Tech-final year degree in the field of mechanical engineering from Islamic Azad University, Science and Research Branch of Tehran-Iran. Currently he is A manager of the Backstretch-team (In the fields of Robotics) in the address of www.backstretch-team.info, and he is a manager of webdesigning in the Rheosociety Group researches that it can be see in the address of www.rheosociety.com . Now he is a member in the technical society as ASME, ISME (Iranian society of mechanical Engineering), and IMS (Iranian Mathematical Society). His special interests are Computational Mechanics, Rheology, Viscoelasticity, and robotics. Mahdi Pirhayati was born in Tehran, Iran in 1985. He received his Bachelor degree in Mechanical Engineering from Shahid Chamran University, Ahvaz,Iran in 2009; MEng in Mechanical Engineering from Science and Research Branch, Islamic Azad University, Tehran, Iran in 2012. His research interests include hybrid vehicle, green vehicle and robot.There are several projects he is working on hybrid vehicle such as energy generation and suspension system.
20
Vol. 3, Issue 1, pp. 14-20
KNOWLEDGE MANAGEMENT IN SUCCESS OF ERP SYSTEMS

Usman Musa Zakari Usman, Mohammad Nazir Ahmad Department of Information Systems, Faculty of Computer Science and Information Systems Universiti Teknologi Malaysia (UTM), Skudai, Johor Bahru 81310, Malaysia
ABSTRACT
Special attention to critical success factors in the implementation of Enterprise Resource Planning systems is evident from the bulk of literature on this issue. In order to implement these systems,which are aimed at improving the sharing of enterprise-wide information and knowledge, organizations must have the capability to effectively share knowledge to start with. Based on a review of the literature on knowledge management in enterprise system implementation, this paper identifies two major areas of concern regarding the management of knowledge in this specific type of project: managing tacit knowledge, and issues regarding the process-based nature of organizational knowledge viewed through the lens of organizational memory. The more capable an organization is in handling these issues, the more likely it is that the implementation will result in competitive advantage for the organization. The competitive advantage arises from the organizations capabilities in internalizing and integrating the adopted processes with the existing knowledge paradigms and harmonizing the new system and the organizational culture towards getting the most out of the implementation effort.
KEYWORDS: Knowledge, Knowledge Management, Enterprise Resource Planning, Project Management
I.
INTRODUCTION
Enterprise resource planning (ERP) systems are popular among enterprises, with many organizations wanting to implement an ERP system, yet the rate of failure is quite high. Enterprise resource planning software presents a framework for an organization to help them improve their business processes. It consists of a wide range of software products supporting daily organizational business operations and decision making. ERP systems automate operations in supply chain management, inventory control, manufacturing scheduling, sales support, customer relationship management, financial and cost accounting, human resource and other business functional areas within an organization. Knowledge management (KM)is playing an important role in society, and becoming compelling issue within enterprises.In this article, we report on a systematic review of empirical studies of knowledge management in enterprise resource planning projects. Our main goal is to provide a clear overview of empirical studies within the ERP research field, identifying the concepts that have been explored in ERP projects, the main findings, and the research methodsthat have been used within this area. The target readership of the review is four groups which we expect will be interested in an overview of empirical research on knowledge management in ERP projects: (1) academic and enterprise researchers on knowledge management in general, who would be interested in making comparisons ERP projects; (2) practitioners within enterprises, who will be interested in learning about knowledge management initiatives in ERP project implementation; (3) knowledge management researchers who are interested in designing studies to address important research gaps in this field; and (4) researchers who are interested in identifying the relevant studies, and the major findings and their implication within the field. The structure of our research paper is as follows. Section 2 presents the background and general theories on knowledge management for ERP projects. Section 3 describes the research method that we use to select and review the data material for our research, and presents our chosen framework for
21
Vol. 3, Issue 1, pp. 21-28
analysis. Section 4 presents the results of the systematic review according to our chosen framework. In Section 5, we discuss and conclude the findings and their implications. For the implications for research, we identify what we believe are the most important research gaps. For practitioners, we provide advice on how to use the results in practice.
II.
BACKGROUND
In this section, a brief introduction to enterprise resource planning and our research focus together with our research question are presented. The remainder of the article presents the overview of the current work on knowledge management in ERP projects.
2.1 Knowledge Management

Knowledge is derived from data and information [20]. Knowledge management is the management of information and knowledge and their usage in organizational business processes within the organization. The main focus of knowledge management is steering strategy and, identifying and communicating the various types of knowledge that reside in processes, people, products and services in order to support integration to improve productivity and efficiency.Based on the knowledge management literature review, the conclusion can be made that knowledge resides in organizational resources, employees and external partnerships. Knowledge is categorized to pursue different research interests, [17] namely,the tacit and explicit dimensions of personal knowledge and processes required for managing to create organizational knowledge. Three knowledge types are identified by Petrashs framework [18]Based on the knowledge management literature, knowledge management processes are studied in accordance with distinctive and various types of knowledge and organizational objectives. It has traditionally been assumed that there are three broad types of knowledge processing: generation, transfer, and utilization. For example, Probst et al.[20] identified six knowledge processes required for managing organizational knowledge. Knowledge integration is viewed asan important process for innovation and building organizational capability [10],[23]. Coombs and Hull [5] identified ten distinctive processes,namely, identification, transfer, utilization, creation, acquisition, retention, codification, validation, developing, and integration of knowledge.
2.2 Knowledge Management from the ERP Project Perspective

An ERP system allows an organization to have a convergent and integrated view of the organizational information by means of centralized databases and integrated business processes across the lines of different divisions and departments [9,23,19]. It could be said that as a result of enterprise system implementation, the organizational information and knowledge converges across different divisions and departments on an organization-wide scope. IT experts need to know more about the business processes and business process experts need to leverage their knowledge about the IT systems in place in their organization. Eventually, the overlap between the knowledge of different divisions increases and the knowledge on the organizational scale follows a converging pattern. However, this convergence on the organizational level tends to turn into divergence as we move down to the individual level [2,17]. A broader knowledge of the organization is required for end users of enterprise systems compared to the traditional legacy systems that were adapted to each island of automation. As the view changes from the taskfocus to the process focus by implementing enterprise systems, employees need to know how their tasks fit into the overall process and how that process contributes to the achievement of organizational objectives. For example, an employee working in the customer billing section will need to know more about the IT systems as well as other business areas such as production and accounting. Similarly, the IT experts need to know more about different subject areas to adapt the new system to the areas requirements and configure the enterprise system to operate optimally. Therefore, as the organizational view of knowledge regarding the tasks and processes that are conducted in the organization tends to converge by the use of the enterprise system, the individual knowledge must diverge to accommodate the changes posed by enterprise system implementation [2,50, 55]. One major implication of such a view of enterprise system projects is that knowledge sharing needs to be significant across organizational boundaries to allow for the maximum sharing of observations and experiences among employees from different divisions with different mindsets about how the
22
Vol. 3, Issue 1, pp. 21-28
business is done along the lines of process. Knowledge sharing in enterprise system projects exists along different lines of interaction among organizational members, the ERP team, and external consultants which echoes the need for improved knowledge sharing along different organizational dimensions and in different levels of engagement with the implementation project. The next section reviews different lines of ERP specific knowledge sharing in more detail.
2.3 Knowledge Management in ERP Projects

The simultaneous implementation of enterprise resource planning and knowledge management systems in organizations implies some sort of contradiction by its nature. Enterprise systems are meant to increase the organizational efficiency by enhancing the information processing capability of the enterprise [15,19, 60, 62]. This capability enhancement is enabled by the systematization and centralization of information management and the adoption of standard approaches to the codification and processing of information. On the other hand, knowledge management initiatives aim at mobilizing the knowledge through organized knowledge repositories of explicit knowledge and communities of practice as a means of sharing and creating tacit knowledge, having their overall focus on improving innovation capabilities by increasing flexibility [4,7, 22, 24]. While it is traditionally believed that it is impossible for an organization to focus on both efficiency and flexibility, Newell et al. [19] show, by analyzing a case, that enterprise system and knowledge management initiatives are complementary rather than contradictory. Assuming enterprise systems are integrated databases of organizational information and explicit knowledge, as opposed to knowledge management initiatives being methods of managing tacit knowledge, their findings suggest that a balanced perspective of ERP and KM systems can assist in exploiting explicit knowledge as well as exploring and sharing tacit knowledge simultaneously. In other words, utilizing the respective strength of the enterprise system and KM in tandem enables the alignment of organizational capabilities in information processing, knowledge exploration and exploitation [19]. Knowledge management techniques are used over the course of enterprise system implementation and during different steps of implementation projects to facilitate this knowledge sharing [8]. A detailed view of how the knowledge of ERP project members evolves during these different stages is discussed next.
III.
METHOD
In this research, we use the systematic review approach based on [61-62]. Guided by, the research question, we identify the relevant research, carry out a selection process, appraise, synthesize and draw inferences. Finally, we address each of them.
3.1. Review Planning

We begin by providing a protocol for the systematic review, specifying the process and methods that areused. We aim to use the protocol to specify the research questions, research strategy and method of synthesis.
3.2. Research Identification

This systematic research started with the identification of the keywords and search terms. We used general keywords to search for many and various relevant papers regarding knowledge management and ERP projects. The search strategy for the review was directed towards finding published papers in archival journals, conference proceedings and technical reports from the contents of four electronic databases, namely, ACM portal, Elseviers Science Direct, IEEE Xplore and Springer-Verlags Link. The search terms used were: software selection criteria, software evaluation techniques, software selection methodologies, evaluating and selecting software packages, method for evaluating and selecting software packages, criteria for evaluating and selecting software package, software evaluation criteria, systems/tools for evaluation and selection of software packages, knowledge-based systems for software selection, framework for evaluating and selecting software packages, and software selection process. Other relevant journals found while searching the articles on this topic are Information and Management, Information and Software Technology, and European Journal of Operational Research. Articles published in proceedings of the IEEE on Software Engineering, Springer-Verlag,
23
Vol. 3, Issue 1, pp. 21-28
International Conference on COTS-Based Software System were also found relevant to this topic. The series of articles on evaluating software engineering methods and tools (part 5 to part 8), ACM SIGSOFT, is one of the major contributions to this topic.
3.3. Paper Selection

Our selection process had two parts: (i) an initial selection from the search results, based on reading the abstract of the papers; and (ii) final selection from the initially selected list of papers, based on reading the entire paper. The initial list consisted of 130 papers which we found relevant to the topic and potential candidates for inclusion in our review. Initial selection of the paper was done jointly by both the authors on the basis of reading the title and abstract of the paper. The first author then read all 130 papers in detail and considered 62 papers to be included in the final list for review. In the second phase of selection, we eliminated 4 papers that did not give any useful information on evaluation criteria, evaluation technique, selection methodology, and systems/tools for software selection. The second author cross-checked whether the papers in the final list considered for review addressed the research question and contributed to the basic purpose of the review. A random sample of 25 papers was selected for the cross-checking. There was no disagreement on the final selection of papers. The search began in early 2006 and was completed in early 2007.
3.4. Data Extraction

In the data extraction phase, the first author read every selected paper and extracted information about the attributes as set out in Table 1. The extracted data were then cross-checked by the second author by random selection of 20 papers, that is,about 30% of the total. During the data extraction phase we found that four papers did not give any useful information on software selection methodology, evaluation criteria, evaluation technique, and systems/tools for software selection. Therefore, those papers werenot considered when presenting the results of the review.
3.5. Synthesis
For the synthesis, we chose to only use the papers classified as empirical studies in our framework, in order to avoid problems associated with lessons learned reports stemming from their lack of scientific rigor. We extracted concepts covered, main findings and the research method for each article. One researcher (the first author) focused on the studies in the technical schools, while the other researcher (the second author) focused on the behavioural schools.
IV.
RESULTS
This section describes the analysis of the data extracted from our selected studies. The contribution of the reviewed literature in the field of knowledge management in ERP projects is presented, whichfocuses on the dimensions that should be considered when implementing an ERP projectItshows clearly that various areas of knowledge have been acquired from the literature review. There are similarities between the areas of knowledge, and the consistent expression of the need for this knowledge from the case studies emphasizes that this knowledge should be made explicit. These areas of knowledge are organized to a more manageable form in the following section. From the literature reviewed, three dimensions of knowledge are clearly identified for the successful implementation of an ERP system. The three dimensions to be considered for successful implementation are: i. Project management knowledge ii. Business and management knowledge iii. Technical knowledge. Project management knowledge refers to the knowledge required to manage the entire implementation process as a single project. Business and management knowledge refers to the knowledge about issues and knowledge to deal with these issues during and after implementation. These issues are often people-related and occur on a higher management level.Technical knowledge refers tothe knowledge required to install and implement the ERP system [58].
24
Vol. 3, Issue 1, pp. 21-28
Table 1. Knowledge management for ERP project success Stage of ERP Selecting ERP for implementation Implementation process Using ERP after implementation Changing ERP when needed KM Types Business and management/ project management Technical Role of KM Identifying the right package Creating good integration b/w business and system Using, transferring and storing data Re/using Successful Output of KM Business/organizational satisfaction Business/organizational impact Flow quality information System quality/ Business
Business and management Business and management/Project management
V.
DISCUSSION AND CONCLUSION
By comprehensive review of the literature on enterprise system knowledge management, this paper investigated the major concerns of the different lines. The first area concerns the effects and implications of the tacit category of ERP-specific knowledge. The subject of tacit knowledge management is addressed extensively in the literature and different issues along with their respective mitigating solutions are provided in various research works [16, 5, 10]. These solutions include the presence of tacit knowledge sharing facilitators during enterprise system implementation [28, 20, 10], and paying attention to the structure of team interactions and the atmosphere of the team. Proper utilization of each method can assist the adopting organization to overcome the difficulties of tacit knowledge sharing. Organizing communities of practice composed of the different groups involved in different stages of the enterprise system lifecycle is another way to overcome the difficulties of transferring such knowledge from where it resides to where it is needed. In the case of running the enterprise system project on distant locations [18], virtual communities centered on company intranets or the internet act as the facilitating bridge among separate bodies of knowledge across the entire enterprise. The process-based nature of organizational knowledge is the second area of concern in enterprise system knowledge management which was examined from the lens of organizational memory [29,30]. Organizational processes embed substantial knowledge of the organizations history and can be regarded as the organizational memory. Viewing the ERP knowledge through the lens of organizational memory sheds light on some interesting issues of concern in ERP implementation projects [31, 33]. Arranging powerful core enterprise system implementation teams and effective utilization of external consulting were identified to be among the most preferred methods of dealing with the knowledge barriers connected with enterprise system configuration caused by difficulties associated with organizational memory. The standardization which results from adopting the same best practices of enterprise system packages by many organizations might give rise to concerns about losing competitive advantage. In particular, the two subjects reviewed here are very illustrative. Finally, managing ERP-related knowledge across its lifecycle (pre-, implementation and postimplementation) is also an interesting area. For example, exploiting the contribution from disciplines such as ontology engineering into this area would give benefits within the context of ontology-based applications for enterprise systems. This may enhance the whole performance of ERP lifecycle knowledge management activities. An initial insight into this direction is systematically presented in [56] and an example is available from previous work such as [57]. REFERENCES
[1] S.A. Ajila,& Z. Sun, Knowledge management: impact of knowledge delivery factors on software product development efficiency, in: Proceedings of the IEEE International Conference on Information Reuse and Integration, Las Vegas, NV, United States, 2004, pp. 320325.
25
Vol. 3, Issue 1, pp. 21-28
[2] A.J. Al-Shehab, R.T. Hughes,& G. Winstanley, Facilitating organisational learning through causal mapping, in: Proceedings of the Seventh InternationalWorkshop on Learning Software Organizations, Springer Verlag, Kaiserslautern, Germany, 2005, pp. 145154. M. Alavi, D&.E. Leidner, Review: knowledge management and knowledge management systems: conceptual foundations and research issues, MISQuarterly 25 (1) (2001) 107136. N. Angkasaputra, D. Pfahl, E. Ras, &S. Trapp, The collaborative learning methodology CORONETtrain: implementation and guidance, in: Proceedings of the Fourth International Workshop on Learning Software Organizations, Springer Verlag, Chicago, IL, USA, 2002, pp. 1324. J. Arent, &J. Nrbjerg, Software process improvement as organizational knowledge creation: a multiple case analysis, in: Proceedings of the HawaiiInternational Conference on System Sciences, Maui, USA, 2000, p. 105. J. Arent, J. Nrbjerg, &M.H. Pedersen, Creating organizational knowledge in software process improvement, in: Proceedings of the 2nd Workshop on Learning Software Organizations, Oulu, Finland, 2000, pp. 8192. L. Argote, B. McEvily, &R. Reagans, Managing knowledge in organizations: an integrative framework and review of emerging themes, Management Science 49 (4) (2003) 571582. C. Argyris, Overcoming Organizational Defences: Facilitating Organizational Learning, Prentice Hall, Boston, 1990. C. Argyris, &D.A. Schn, Organizational learning II: theory, method and practice, Organization Development Series, Addison Wesley, Reading, MA, USA, 1996. A. Aurum, R. Jeffrey, C. Wohlin,& M. Handzic, Managing Software Engineering Knowledge, Springer Verlag, Berlin, 2003. M.D.O. Barros, C.M.L. Werner,& G.H. Travassos, Supporting risks in software project management, Journal of Systems and Software 70 (12) (2004) 21 35. V.R. Basili, G. Caldiera, F. McGarry, R. Pajerski, &G. Page, The software engineering laboratory an operational software experience factory, in: Proceedings of the 14th International Conference on Software Engineering, 1992, pp. 370381. V.R. Basili, G. Caldiera, &H.D. Rombach, The experience factory, in: J.J. Marciniak (Ed.), Encyclopedia of Software Engineering, 1, John Wiley, New York, 1994, pp. 469476. R. Baskerville,& J. Pries-Heje, Knowledge capability and maturity in software management, (1999). A. Birk, A Knowledge Management Infrastructure for Systematic Improvement in Software Engineering, Dr. Ing thesis, University of Kaiserslautern, Department of Informatics, 2000. F.O. Bjrnson,& T. Dingsyr, A study of a mentoring program for knowledge transfer in a small software consultancy company, in: Lecture Notes inComputer Science 3547, Springer Verlag, Heidelberg, 2005, pp. 245256. F.O. Bjrnson, Knowledge Management in Software Process Improvement, PhD thesis, Norwegian University of Science and Technology, Department of Computer and Information Science, 2007. F.O. Bjrnson,& T. Stlhane, Harvesting knowledge through a method framework in an electronic process guide, in: Proceedings of the Seventh International Workshop on Learning Software Organizations, Springer Verlag, Kaiserslautern, Germany, 2005, pp. 8690. P. Brssler, Knowledge management at a software engineering company an experience report, in: Proceedings of the 1st Workshop on Learning Software Organizations, Kaiserslautern, Germany, 1999, pp. 7786. A.F. Buono, &F. Poulfelt, Challenges and Issues in Knowledge Management, Information Age Publishing, Greenwich, CT, USA, 2005. B. Chatters, Implementing an experience factory: maintenance and evolution of the software and systems development process, in: Proceedings of IEEE M. Alavi, &D.E. Leidner, Review. Knowledge management and knowledge management systems: conceptual foundations and research issues, MISQuarterly 25 (1) (2001) 107136. R. Baskerville, S. Pawlowski,& E. McLean, Enterprise resource planning and organizational knowledge: patterns of convergence and divergence, in: Proceedings of the 21st ICIS conference, 2000. M. Beer, &N. Nohria, Cracking the code of change, Harvard Business Review 78 (3) (2000) 133141. M. Earl, Knowledge management strategies: toward a taxonomy, Journal of Management Information Systems 18 (1) (2001) 215233. T.L. Griffith, J.E. Sawyer, &M.A. Neale, Virtualness and knowledge in teams: managing the love triangle of organizations, individuals, and information technology, Mis Quarterly 27 (2) (2003) 265 287. V. Grover, &T.H. Davenport, General perspectives on knowledge management: fostering a research agenda, Journal of Management Information Systems 18 (1) (2001) 521.
[3] [4]
[5]
[6]
[7] [8] [9] [10] [11] [12]
[13] [14] [15] [16]
[17] [18]
[19]
[20] [21] [22] [23] [24] [25] [26]
[27]
26
Vol. 3, Issue 1, pp. 21-28
[28] C. Holland, &B. Light, A stage maturity model for enterprise resource planning systems use, The DATA BASE for Advances Information Systems 32 (2) (2001). [29] J. Huang, S. Newell,& S. Pan, Knowledge integration processes within the context of enterprise resources planning (ERP) systems implementation, in: Proceedings of the 9th ECIS Conference, 2001. [30] M. Jones, Tacit knowledge sharing during ERP implementation: a multi-site case study, Information Resource Management Journal 18 (2) (2005) 123. [31] M. Jones, &R. Price, organizational knowledge sharing in ERP implementation: lessons from industry, Journal of Organizational and End User Computing 16 (1) (2004). [32] M.C. Jones, M. Cline, &S. Ryan, Exploring knowledge sharing in ERP implementation: an organizational culture framework, Decision SupportSystems 41 (2) (2006) 411434. [33] J. Kallinikos, Deconstructing information packages, Information Technology and People 17 (1) (2004). [34] A. Kwang-Tat, T. James, &Y., Chee-Sing, IT implementation through the lens of organizational learning: a case study of insuror, in: Proceedings of the Eighteenth International Conference on Information Systems, 1997. [35] J. Lee, K. Siau, &S. Hong, Enterprise integration with ERP and EAI, Communications of the ACM 46 (2) (2003) 5460. [36] Z. Lee, &J.Y. Lee, An ERP implementation case study from a knowledge transfer perspective, Journal of Information Technology 15 (4) (2000) 281288. [37] Y. Malhorta, integrating knowledge management technologies in organizational business processes: getting real time enterprise to deliver real business performance, Journal of Knowledge Management 9 (1) (2005) 728. [38] M.L. Markus, C. Tanis, &P.C. van Fenema, Multisite ERP implementations, Communications of the ACM 43 (4) (2000) 4246. [39] S. Newell, J. Huang, &R. Galliers, Implementing enterprise resource planning and knowledge management systems in tandem: fostering efficiency and innovation complementarity, Information and Organization 13 (2003). [40] S. Newell, C. Tansley, &J. Huang, Social capital and knowledge creation in an ERP project team, in: Proceedings of the 7th AMCIS, 2001. [41] I. Nonaka, &H. Takeuchi, Knowledge-Creating Company, Oxford University Press, 1995. [42] S. Newell, J. Huang, &R. Galliers, Implementing enterprise resource planning and knowledge management systems in tandem: fostering efficiency and innovation complementarity, Information and Organization 13 (2003). [43] D. Robey, J.W. Ross, &M.C. Boudreau, Learning to implement enterprise systems: an exploratory study of the dialectics of change, Journal of Management Information Systems 19 (1) (2002) 1746. [44] U. Schultze, &D.E. Leidner, Studying knowledge management in information systems research: discourses and theoretical assumptions, Mis Quarterly 26 (3) (2002) 213242. [45] J.E. Scott, Post implementation usability of ERP training manuals: the users perspective, Information Systems Management 22 (2) (2005) 6777. [46] J.E. Scott, &I. Vessey, Managing risks in enterprise systems implementations, Communications of the ACM 45 (4) (2002) 7481. [47] C. Soh, S.S. Kien, &J. Tay-Yap, Cultural fits and misfits: is ERP a universal solution?, Communications of the ACM 43 (4) (2000) 4751. [48] D. Stenmark, Leveraging tacit organizational knowledge, Journal of Management Information Systems 17 (3) (2000) 924. [49] E. Stijin, &A. Wensley, organizational memory and completeness of process modeling in ERP systems, Business Process Management Journal 7 (3) (2001). [50] M. Sumner, Risk factors in enterprise-wide/ERP projects, Journal of Information Technology 15 (4) (2000) 317327. [51] S.W. Sussman, &W.S. Siegal, Informational influence in organizations: an integrated approach to knowledge adoption, Information Systems Research14 (1) (2003) 4765. [52] P. Weill, &R. Woodham, Dont Just Lead Govern, Implementing Effective IT Governance, MIT Sloan School of Management Working Paper, 2002. [53] L. Willcocks, &R. Sykes, The role of the CIO and IT function in ERP, Communications of the ACM 43 (4) (2000) 3238. [54] J. Worley, K. Chatha, &R. Weston, implementation and optimization of ERP systems: a better integration of processes, roles, knowledge and user competencies, Computers in Industry 56 (2005). [55] L. Pries-Heje, &Y. Dittrich, "ERP Implementation as Design: Looking at participatory design for means to facilitate knowledge integration," Scandinavian Journal of Information Systems: Vol. 21: Iss. 2, Article 4, (2009).
27
Vol. 3, Issue 1, pp. 21-28
[56] [57] [58] M. N., Ahmad,N. H. Zakaria., &D. Sedera,Ontology-based Knowledge Management for Enterprise Systems. International Journal of Enterprise Information Systems (IJEIS), 7(4), 64-90, (2011). G. C., Peng,&M. P. Nunes,Surfacing ERP exploitation risks through a risk ontology, .Industrial Management & Data Systems, 109 (7), 926-942, (2009) R. Chan.,J. Esteves., J. Pastor.,&M. Rosemann..An Exploratory Study of Knowledge types Relevance along Enterprise Systems Implementation Phases. In Proceedings of the 4th European Conference on Organizational Knowledge and Learning Capabilities (OKLC), pp. 1-14, Barcelona,(2003). D. Sedera., G. Gable., &T. Chan.Measuring Enterprise Systems Success: A preliminary model in proceedings of the 2003 Aericas Conference on Information Systems (AMCIS 2003), Tampa, Florida, USA, August 4-6. (2003). M. Easterby-Smith,&M.A. Lyles, The Blackwell Handbook of Organizational Learning and Knowledge Management, Blackwell Publishing, Malden, MA, USA, (2003) B. Kitchenham, Procedures for Performing Systematic Reviews, in Technical Report TR/SE-0401, Keele University, (2004) B. Kitchenham, R. Pretorius,D. Budgen,P. Brereton,M. Turner,&M. Niazi. Systematic literature reviews in software engineering,A tertiary study. Information and Software Technology, 52, 792805, (2010)
[59]
[60] [61] [62]
Authors
Usman Musa Zakari Usman is an M.Sc. research student at the Faculty of Computer Science and Information Systems of University Technology Malaysia (UTM), Skudai, Johor Malaysia. He holds his B.Sc. in Business Information Technology from Limkokwing University of Creative Technology Cyberjaya Malaysia (LUCT). His research interests are innovative solutions for "knowledge-based" information systems that span several areas applying ontology and knowledge management for interoperating information systems, software engineering and enterprise systems.
Mohammad Nazir Ahmad is currently working in the Faculty of Computer Science and Information Systems at the Universiti Teknologi Malaysia (UTM), Skudai, Johor Malaysia. Nazir holds a PhD from the University of Queensland in Information Systems and a Masters degree in Information Systems from the Universiti Teknologi Malaysia. He holds a Bachelor degree in Industrial Computing from the Universiti Kebangsaan Malaysia (UKM). His main research interests are innovative solutions for "knowledge-based" information systems that span several areas applying ontology and knowledge management for interoperating information systems, software engineering and enterprise systems. Currently, he is an Editorial Board/Review Member of the International Journal of Knowledge Management Practice (JKMP), International Journal of Computer Science and Emerging Technologies (IJCSET) and International Journal of Information, Knowledge and Management (IJIKM). Moreover, he is a member of the Association for Information Systems (AIS) and the International Association for Ontology and its Applications (IAOA).
28
Vol. 3, Issue 1, pp. 21-28
A NEW PROPOSAL ERICA+ SWITCH ALGORITHM FOR TRAFFIC MANAGEMENT

Ehab Aziz Khalil
Department of Computer Science & Engineering, Faculty of Electronics Engineering, Menoufiya University, Menouf-32952, EGYPT
A BSTRACT
A new proposal to the Explicit Rate Indication for Congestion Avoidance+ (ERICA+) switch algorithm for traffic management is present. The new proposal can be used for enhancing quality of service of multimedia; the using of non_zero MCR is very useful for carrying multimedia over ATM network. We have adopted continuous event driven simulation methodology to evaluate performance of integrated video and data traffics on the ATM network when using ABR service. The study confirms that the system parameters (e.g., dynamic/constant queue control functions, ICRs for sources, number of video and data traffic intensity) have sensitive effects on the performance characteristics of the network. The method we have used depends on a separate queues for each traffic types to isolate them from overlapping, so the delay will reduce especially for video traffic, however, the new proposal algorithm gives better performance than the original algorithm, its promising enough. K EYW ORDS: ERICA+ switch algorithm, ATM-ABR service, Performance of Video and Data traffics.
I.
INTRODUCTION
It is well known that ATM (Asynchronous Transfer Mode) has emerged as most promising technology which can provide high speed networks with the capability of sending all types of traffic including video and data, and provides high speed communications for different types of data [1]. ATM supports multiple Quality of Service (QoS), which include Constant Bit Rate (CBR), Variable Bit Rate (VBR), Available Bit Rate (ABR), and Unspecified Bit Rate (UBR). These services share a common link and thus not all of them can get the bandwidth they require. In ATM networks, the ABR service and UBR service are used to support non-delay sensitive data applications. ABR normally uses the available bandwidth. This is often the left-over of the higher priorities services, which are CBR and VBR. Though the current standards for ABR service do not require the cell transfer delay and cell loss ratio to be guaranteed, it is desirable for switches to minimize the delay and loss as much as possible. The ABR service requires network switches to constantly monitor their load and feed this information back to the sources, which in turn dynamically adjust their input into the network. This is mainly done by inserting Resource Management (RM) cells into the traffic periodically and getting the network congestion state feedback from the returned RM cells, which may contain congestion information reported by the switches and destinations. Depending upon the feedback, the source is required to adjust its transmission rate. Obviously, that the congestion control mechanisms are essential for the support of ABR service to provide efficient and fair bandwidth allocation among ABR applications [2-19]. Figure 1 shows an ABR traffic management model. The RM cell contains an Explicit Rate (ER) field. The switches along the path put some information to indicate the rate that the source should use after the receipt of the RM cell. ABR users are allowed to declare a Minimum Cell Rate (MCR), which is guaranteed to the Virtual Connection (VC) by the network. Most VCs use zero as the default MCR value. However, for an ABR with higher MCR, the connection may be denied if sufficient bandwidth is not available. Both ABR data traffic and the available bandwidth for ABR are variable.
29
Vol. 3, Issue 1, pp. 29-40
Figure 1 Traffic Model in ABR
If there is not enough buffer the bursty traffic (either from VBR, which requires more bandwidth), too many losses will result in low performance. This paper presents a new proposal algorithm to the switch depending on the General weighted Fair ERICA+ (GWFairERICA+) switch algorithm were described in [20,21]. As shown in Figure 2, we have considered two types of traffic (video and data), each traffic has n sources (Source 1,.,Source n) accommodates in one queue and the service of these queues occurs at different levels of priorities, that is to enhance throughput guarantees to support multimedia applications.
Figure 2
The output link bandwidth divided dynamically between these queues according to the level of priority. It is to mention here that the use of separate queues resulting in reduces the delay of the cells transmission. Also, we assume that the Average Interval (AI) period for each traffic type is different and depends on the RM cells, for example if RM cell of data source is sending every X period an data cell and RM cell of video source is sending every 50X period a video cell, that indicates that the AI for video sources much more than that of data sources 50 times, i.e., when one feedback arrives at video sources there are several feedback arrive at data sources. The remaining of the paper organized as the following, section 2 gives brief overview of ERICA+ switch algorithm and then section 3 discusses the new proposal algorithm. In section 4 the configuration parameters, and simulation are discussed. The results are discussed in section 5. Section 6 gives discussion and finally section 7 presents the conclusion.
II.
OVERVIEW OF THE ORIGINAL ERICA+ SWITCH ALGORITHM
At the beginning it is well known that the main advantages of ERICA are its low complexity, fast transient response, high efficiency, and small queuing delay [22-25], also, in ERICA, the time is divided into consecutive equal-sized slots called switched averaging intervals [26]. The ERICA+ algorithm is concerned with the fair and efficient allocation of the available bandwidth to all contending sources. Like any dynamic resource algorithm, it requires monitoring the available capacity and the current demand on the resources. There, the key resource is the available bandwidth at a queuing point. In most switches, output buffering is used, which means that most of the queuing happens at the output ports. Thus, the ERICA+ algorithm is applied to each output port. Assuming that measurements do not suffer from high variance, the above algorithm is sufficient to converge to efficient operation in all cases and to the max-min fair allocations in most cases [20, 27-32]. As mentioned above that the ERICA+ operates at the output port of a switch. It periodically monitors the load, active number of VCs and provides feedback in the backward RM (BRM) cells. The measurement period is called the Averaging Interval. The measurements are done
30
Vol. 3, Issue 1, pp. 29-40
in the forward direction and feedback is given in the reverse direction. The complete description of ERICA+ algorithm and its performance in one of these references [20,33-35], and some related research in [36-44].
III.
THE NEW PROPOSAL SWITCH ALGORITHM [45, 46]
In the new proposal switch algorithm, two types of traffic and two queues (as shown in Figure 2) are used instead of one type of traffic and one queue which have been developed in several switch algorithm to compute the feedback to be indicated to ABR sources in RM cells [31-33, 48, 49]. Also, the service of these queues depends on different levels of priorities that will enhance throughput guarantees to support multimedia applications and dividing the output link bandwidth between these queues dynamically depending on the traffics priority level. The separate queues also protect each traffic type from overlapping , so the delay of both traffic will reduce. When using more than one queue as shown in Figure 2, the treatment will be different because the case of one queue does not bother about the status of the traffic within the network, for example may be the videos queue is full and at the same time the datas queue is empty. Also, there are two functions of queue length, one for each traffic, Fqv for video traffic, and FqD for data traffic. Each queue function defines the feedback for the source dealing with that queue and the queue functions are independent from each other. We assume that the AI (Averaging Interval ) period for each traffic type is different and depends on the Resource Management (RM) cells. Instead of using one queue function there are two queues and two functions which may be dynamic or static functions. These functions operate independently and the total bandwidth which divided on the active sources will depend on the two applied functions. A simple choice is to use a Constant Queue control Function (CQF), where the queue Factor is set to a value less than one. The (1-Factor) is used for queue draining. Another choice is to use a Dynamic Queue control Function (DQF) [40]. In case of DQF, the Factors value equals one for the short queue length and drops sharply by increasing of the queue length. ERICA+ switch algorithm uses a hyperbolic or inverse hyperbolic function for calculating the value of the DQF factor [49]. GWFairERICA+ and ERICA+ Switch algorithms were described in [20,21,26,50-52] using target ABR capacity which is obtained by multiplying the total available ABR capacity by a fraction term. Fraction amount of the link capacity is used to drain the queue[26]. Fraction can be either a constant less than one or dynamic function of the switch queue length (Fq). The using of one queue for all traffic with dynamic function of the switch queue length (Fq), resulting in the queue length will be very important to define the feedback for each active source and the status of the network depends on that queue, when the queue is full beyond threshold2 value (represents the transient point from steady state to over load ) the network is congested and if between threshold1 ( represents the transient point from under load to steady state ) and threshold2 values, the network is in steady state and if less than threshold1, the network is under load. The GWFairERICA+ Switch algorithm can operate with the new proposal because of the using of that algorithm to weight function which used to distribute the excess bandwidth among sources depending on their weights. In this paper we have followed the same general weighted function as in [48] with the new proposal N Gi = Ui + Wi (A-U) / Wj where j =1 to N 1 Gi = GW fair allocation for connection i. Ui = MCR of connection i. Wi= Preassigned weight associated with the connection i.
31
Vol. 3, Issue 1, pp. 29-40
U = Sum of MCRs of active connections Bottlenecked at this link. A = Excess bandwidth, to be shared by Connections bottlenecked on this link. N = total bottleneck sources. Pseudo code of the New Proposal Algorithm At the end of the averaging interval for data: For data sources ( target ABR data capacity = data factor total ABR data capacity . data input rate=sum of all inputs of data queue . data over load factor = data input rate target ABR data capacity . ) repeat for video sources in case of the end of the AI for video. For each source calculate weight (cost+ mcr) for each source . End . For each source Excess fair share = target capacity source_ weight sum of weights for certain traffic type. End. For each queue Queue fair share = target capacity sum of weights for sources dealing with this queue sum of weights for all traffic type. End. When a BRM is received: For each source virtual channel share (Vcshare)=max(0,source_ rate - mcr) over load factor . explicit rate (ER) = mcr + max(Excess fair_ share, Vcshare) . ER in RM cells=min(ER in RM,ER, Target rate). End The main different between the original GWFairERICA+ algorithm and the new proposal algorithm is the calculating of the variables of each traffic type independently and divide the available output link bandwidth dynamically among the queues depending on the sum of all weights of traffic types. Dividing the output link bandwidth among queues will happened at the end of the Averaging Interval period. We assume that only one feedback is given in each averaging interval to the sources. This avoids unnecessary conflicting feedbacks to the sources. May be its very difficult to all sources to reach a steady state region at the same time because of the independency among queues but each source will obtain its need from the available bandwidth, and that will realize our main goals to maximize link utilization, minimize queuing delays, achieve fair allocation, reduce transient response time and achieve stable and robust operation.
IV.
SIMULATION CONFIGURATION AND PARAMETERS
In this section, the simulation configuration and parameters are discussed. We use the common original configuration shown in Figure 3, to test the performance of the new proposal switch algorithm. We assume that the sources are greedy, i.e., they have infinite cells to send at Allowed Cell Rates (ACRs). In the configuration the traffic is unidirectional, from source to Destination. If bi-directional traffic is used, similar results will be achieved, except that the convergence time will be longer since the RM cells in the backward direction will travel along with cells from destination to source. In this configuration cells are traveling from the sources to the destinations through the two switches (SW1 and SW2) and the bottleneck link. We assume that only one feedback is given in each averaging interval to the sources, that is to avoid unnecessary conflicting feedbacks to the sources. The common original configuration is used to confirm that the new proposal switch algorithm can achieve the general fairness for different set of weight functions.
32
Vol. 3, Issue 1, pp. 29-40
Figure 3 N Source N Destinations Configuration
Definition of the Parameters within the Configuration N infinite sources sends to N destinations. The direction of the traffic is unidirectional. The Initial Cell Rate (ICR) values of all the sources are chosen randomly in the range between (0, link rate). All links are of length 1000 Km, which correspond to the propagation delay of 5 ms. All links have a bandwidth of 149.76Mbps(155.52Mbps less the Synchronous Optical NETwork SONET) overhead ). The sources start at random time in the range between (0, RTT), where RTT is the Round Trip Time. RTT = 30 ms for the mentioned above configuration. Hyperbolic function parameters for dynamic queues: a = 1.15, b = 1.05, where a and b are the parameters which control the degree of curvature of the hyperbolic function. QDLF (the Queue Drain Limit Factor) = 0.5. TCP Maximum Segment Size (MSS) of 512 bytes. Weight = Cost + MCR. Using Motion Picture Experts Group (MPEG-2)[53] to generate video frames and using a Leaky Bucket shaper [54] to smooth out the traffic at the sources. See the Simulator Flow chart at the end of the paper. In the next section, we shall explore the simulation results of the new proposal algorithm. The performance studies of different rates, queue lengths and utilization are present. All the performance studies are done within the switch (SW1), our future research will look after the two switches (SW1 and SW2) within the configuration..
V.
SIMULATION RESULTS
At the beginning the video and data queues at the switch grow depending on the Initial Cell Rates (ICRs). So the maximum queue depends on the Initial Cell Rates (ICRs) and Round Trip Time (RTT) and is independent of the queue control function used. Influence of ICRs appear only during first Round Trip Time. The feedback information reaches the sources and the sources adjust their rates accordingly.
33
Vol. 3, Issue 1, pp. 29-40
Figure 4: Video and Data ACRs Vs Time
Figure 4 shows the variance in rates for video and data traffics during 350 milliseconds where the initial cell rates are 70, and 50 Mbps for video and data sources respectively. In this case the sum of initial rates (70 + 50 = 120 < ABR capacity. Moreover the two sources achieve the General Weighted (GW) fairness rates. The weight function used in this case is Cost + MCR (25+30 for video traffic and 5+10 for data traffic). The left over capacity (149.76-30-10 = 109.76), is divided proportional to (55,15). Hence the GW fair for each source is (30+55/70*109.76, 10+15/70*109.76) = (116.24,33.52) Mbps. All sources enter to a steady-state region during the first 100 milliseconds from the simulation period.
Figure 5 Video and data queues Vs Time using a CQF
Figure 6 Video and Data Queues Vs Time Using DQF
Figures 5 and 6 show the video and data queues lengths during 350 milliseconds respectively. Results here in two cases, when using a Dynamic Queue Control Function (DQF) and a Constant Queue Control Function (CQF). The GW fair for video and data sources are 116.24,and 33.52 Mbps respectively in case of using a DQF while when using a CQF are 104.47, and 30.31 Mbps respectively. All values and parameters like costs, weights, and ICRs used in this case is same as in figure 4. Comparing results when using a DQF with results when using a CQF, better values obtained in case of a CQF while the link utilization when using a DQF is better than a CQF (see Figure 7).
Figure 7: Link Utilization Vs Time
Figure 8: Video and Data Queues Vs Time using a CQF
34
Vol. 3, Issue 1, pp. 29-40
Figure 9: Video and Data Queues Vs Time using a DQF
Figure 10: Link Utilization Vs Time
Figures 8, 9, and 10 show queues lengths and utilization when using a CQF and DQF if there are four active sources, two video sources and two data sources during 350 milliseconds. Where the initial cell rates are (35,40) Mbps for video and (15,20) Mbps data sources. The weight function used in this case is Cost + MCR (25+30, 25+35 for video traffic and 10+10, 10+15 for data traffic). When number of active video and data sources increased as in this case the performance is acceptable as seen in figures 8, 9, and 10. Also confirm that our proposed algorithm and simulator work efficiently.
VI.
DISCUSSION
This section presents a comparison discussion between our proposal and previous algorithms. Most of the prior studies were used the ERICA+ switch algorithm with one queue carrying multimedia traffic over ATM-ABR service [49, 53, 54], but in our proposal we used separate queues. The main different between the prior studies results and our study results can be summaries in the following points: 1- In almost all studies each source obtains its fair share rate. 2- In our study the separate queues protect each traffic type from overlapping, so the delay reduces. However, the using of one queue for more than one traffic type resulting in the overlapping may be occurs, so the delay will be longer. 3- In case of three sources sending to ABR switch. Our results when using Constant Queue Function (CQF) are identical to that results when using GWFairERICA+ switch algorithm with one queue of multimedia traffic studied in [49]. But it is to mention here that, our results when using Dynamic Queue Function (DQF) are more steady than the results when using one queue of multimedia traffic studied in [49]. 4- The weakness here is the cost in which it will increase by using several queues in ABR switch hardware.
VII.
CONCLUSION
The paper has discussed a new proposal algorithm that can be used for enhancing quality of service of multimedia application when using ABR service in ATM network. Using a non_zero MCR was very useful for carrying multimedia over ATM network. The new proposal algorithm depends upon the use of an independent queue for each traffic type to protect the traffic from overlapping that makes sure for reducing the delay particularly for the traffics sensitive to the delay such as video and audio traffics. This is will be very good to the video cells which they are very sensitive to the delay resulting in continuity increases of the throughput of the switch. Also, the new proposal algorithm divides the output link bandwidth dynamically among different queues. This method depends on summing of all the weights of the sources which dealing with a specific queue and divide it on the total weight of all the active sources and the result will be the ratio of the bandwidth which the queue will use of it to transmit the cells. Dividing the output bandwidth among the queues will happen at the end of the averaging interval period which different for each queue type. The simulation results indicate that the using of general weighted fair ERICA+ switch algorithm (GWFairERICA+) with separate queues will maximize link utilization, minimize queuing delays, achieve stable and robust operation, achieve fair allocation, and reduce transient response. Obviously
35
Vol. 3, Issue 1, pp. 29-40
that, the ATM network using ABR service can effectively handle multimedia traffic in real-world network environment.
REFERENCES
[1] Ning L 46370961, Yue Xu (Debbie) 44676963, TCP over ATM , http://www.cs.ubc.ca/spides/dux/course_project/527.html. [2] V. Singhal, A. K. Vatsa, "A Novel Congestion Control Mechanism With Accelerating Effect," International Journal of Computer Applications (0975-8887), Vol.22, No.5, May 2011. [3] Yuxing Wang, "TCP-FIT: An Improved TCP Congestion Control Algorithm and its Performance," Proceedings of IEEE/INFOCOM'11, pp.2894-2902, 10-15 April, 2011. [4] Zhang Mu, "Research on FAST TCP Congestion Control Algorithm," Proceedings of Future Information Technology and Management Engineering (FITME'10), pp. 464-466, 9-10 Oct.2010. [5] Sahin I., and Simaan M. A.,: "Competitive Flow Control in General Multi-Node Multi-Link Communication Networks," International Journal of Communication Systems, Vol.21, No.2, pp.167-184, Feb. 2008. [6] Ignaciuk P., and Bartoszewicz A. :, "Congestion Control Protocol for Connection Oriented Networks With a Periodic Feedback and Non-Persistent Sources," Theoretical and Applied Informatics, Vol.19, No.3, pp.217-233, 2007. [7] Arjan Durresi, Leonard Barolli, Raj Jain, and Makoto Takizawa, "Congestion Control Using Multi Level Explicit Congestionn Notification," IPSJ Digital Courier, Vol.3, pp.42-54, 2007. [8] Ijaz Haider Naqvi, and Tanguy Perennou, "A DCCP Congestion Control Mechanism for WiredCum-Wireless Environments," IEEE Communication Society, and in WCN 2007 Proceedings. [9] R.S. Deshpande, Dr. P.D. Vyavahare, Recent Advances and a survey of congestion control mechanisms in ATM networks, IE(I) Journal, Vol. 88, pp. 47-54, , 2007. [10] W Li, Z Che, Y Li, Research on the congestion control of Broadband Integrated Service Digital Network based on ATM, Proceedings of the fifth international conference on Machine Learning and Cybernetics, Daliaan, pp 2510-2512, 2006. [11] S. Floyd, E. Kokler, "Profile for Datagram Congestion Control Protocol (DCCP) Congestion Control ID 2:TCP-Like Congestion Control, RFC 4341, March 2006. [12] Ignaciuk P.,and Bartoszewicz A. :, "Congestion Control in Connection-Oriented Communication Networks With Unisochronic Feedback," Proceedings of International Conference on Signals and Electronic Systems, Lodz, Poland, pp.445-448, Sept. 2006. [13] Bruni C., Delli Priscoli F., Kock G., and Vergari S.,: "Traffic Management in a Band Limited Communication Network: An Optimal Control Approach," International Journal of Control, Vol.78, No.16, pp.1249-1264, Nov. 2005. [14] Alpcan T., and Basar T.,:" A Globally Stable Adaptive Congestion Control Scheme for InternetStyle Networks With Delay," IEEE/ACM Transactions on Networking, Vol.13, No.6, pp.12611274, Dec., 2005. [15] Minseok Kwon, and Sonia Fahmy, "On TCP Reaction to Explicit Congestion Notification," Journal of High Speed Networks, Vol.13, No.2, pp.123-138, 2004. [16] E Al-Hammadi and M M Shasavari, Engineering ATM networks for congestion avoidance, Mobile Networks and Application, Vol. 5, pp.157163, 2000. [17] A Hac, H.Lin, Congestion Control for ABR traffic in an ATM network, International Journal of Network Management, Vol. 9, pp. 249-264, 1999. [18] Su C. F. Veciano G., and Walrand J., "Explicit Rate Flow Control For ABR Services in ATM Networks," IEEE/ACM Transactions on Networking, Vol.8, No.3, pp.350-361, June 2000. [19] Zhao Y., Li S. Q., and Sigarto S., : "A Linear Dynamic Model for Design of Stable ExplicitRate ABR Control Schemes," Proccedings of the IEEE INFOCOM'97, Kobe, Japan, Vol.1, pp.283-292, April 1997. [20] B. Vandalore, S. Fahmy, R. Jain , R. Goyal , and M. Goyal , A Definition of General Weighted Fairness and its Support in Explicit Rate Switch Algorithms, Proceedings of 6th International Conference on Network Protocols 1998 (ICNP98), Austin, Texas, USA, pp.22-30, Oct.13-16, 1998, [21] B. Vandalore, S. Fahmy, R. Jain, R. Goyal , and M. Goyal , General Weighted Fairness and its Support in Explicit Rate Switch Algorithms, Journal of Computer Communication, Vol.23, Issue 2, , pp.149-161, January 2000. [22] M. Sreenivasulu, E.V. Parasad, and G.S.S. Raju, "Performance Evaluation of EFCI, and ERICA Schemes for ATM Networks," IJCTA, Vol.2 (4), pp. 981-986, July-August, 2011. [23] M. Sreenivasulu, E. V. Parasad, and G.S.S. Raju, " Performance Evaluation of Rate Based Congestion Control Schemes for ATM Networks," IJCSNS, Vol.11,No.6, pp.190-196, June,2011.
36
Vol. 3, Issue 1, pp. 29-40
[24] M. Sreenivasulu, E. V. Parasad, and G.S.S. Raju, " Enhanced ERICA Congestion Control Scheme for ATM Networks," IJCSNS, Vol.11, No.5, pp.133-140, May 2011. [25] Sonia Fahmy, Raj Jain, Rohit Goyal and Bobby Vandalore, On Determining the Fair Bandwidth Share for ABR Connections in ATM Networks, Journal of High Speed Networks, Vol.11, Issue 2, pp.121-135, 2002 [26] Shivkumar Kalyanaman, Raj Jain, Sonia Fahmy, Rohit Goyal, and Bobby Vandalore, "The ERICA Switch Algorithm For ABR Traffic Management in ATM Networks," IEEE/ACM Transactions on Networking, Vol. 8, No. 1, pp 87-98, February 2000. [27] Y. T. Hou, H. H. Y Tzang, S. S. Panwer, A Generalized max-min Rate Allocation Policy and Its Distributed implementation using the ABR Flow Control Mechanism, Proceedings of INFOCOM98, April 1998. [28] S. P. Abraham and A. Kumas, A Stochastic Approximation Approach for a max-min Fair Adaptive Rate Control of ABR sessions with MCRs, Proceedings of INFOCOM98 , April , 1998. [29] Y. T. Hoa , H. Tzeng , and S. S. Panwar, A Simple ABR Switch Algorithm for The Weighted max-min Fairness Policy, Proceeding IEEE ATM97 workshop, Lisbon, Portugal, pp.329-338, May 25-28, 1997. [30] Y. Yin ,Max-min Fairness vs. MCR Guarantee on Bandwidth Allocation For ABR", Proceeding of IEEE ATM96 , workshop, San Francisco, CA , August 25-27, 1996. [31] D. H. K. Tsang and W. K. F. Wong, A New Rate Based Switch Algorithm for ABR Traffic to Achieve max-min Fairness with Analytical Approximation and delay Adjustment", Proceedings IEEE INFOCOM96, pp.1174-1181, March 1996. [32] L. Kalampoukas, A. Varma, and K. K. Ramakrishman, An Efficient Rate Allocation Algorithm for ATM Networks Providing max-min Fairness, Proceedings of the 6th IFIP International Conference on High Performance Networking, Sept 1995. [33] R. Jain, L. Kalampoukas, R. Goyal, S. Fahmy, and R. Viswanathan, "ERICA Switch Algorithm: A Complete Description," ATM Forum/96-1172, August, 1996. [34] R. Jain, L. Kalampoukas, S. Fahmy, S. Kalyanaraman, and R. Goyal, "ABR Switch Algorithm Testing: A Case Study With ERICA," ATM Firum/96-1267, October 1996. [35] Sonia Fahmy, Raj Jain, Rohit Goyal, and Bobby Vandalore, " Design and Simulation of ATMABR End System Congestion Control, " Transactions of the Society for Computer Simulation, Volume 78, Issue 3, March/April 2002. [36] A Subramani, A Krishnan, "Doubly Finite Queues (DFQ) Supporting for ABR Traffic Load in ATM Networks Using MSVDR Algorithm,"IEEE International Advance Computing Conference IACC, pp.13-19, 2009. [37] Su Bing, Yu Haiyang, Lu Jieru, Ma Zhenghua,"Traffic Optimization on the Dynamic Switching of ABR for OSPF Networks," 2009 International Conference on Information Technology and Computer Science, pp.429-432, 2009. [38] X Li, Y Zhou, G M Dimirovski, Y Jing, "Simulated Annealing Q-learning Algorithm for ABR Traffic Control of ATM Networks," 2008 American Control Conference, pp.4462-4467, 2008. [39] K. G. Shin, D. Saha, D. D. Kandlur, "Scalable Flow Control for Multicast ABR Services in ATM Networks," IEEE/ACM Transactions on Networking, pp.67-85, 2002. [40] N. Ghani, J. W. Mark, "Enhanced Distributed Explicit Rate Sllocation for ABR Services in ATM Networks," IEEE/ACM Transactions on Networking, pp.71-86, 2000. [41] B. K. Kim, C. Thompson, "ABR Traffic Control in ATM Networks Using Optimal Control Theory," 1998 ICATM98 1998 1st IEEE International Conference on ATM, pp.327-333, 1998. [42] Hiroyuki Ohsaki, Masayuki Murata, Hideo Miyahara, "Designing Efficient Explicit-Rate Switch Algorithm with Max-Min Fairness for ABR Service Class in ATM Networks," In Proc. of the IEEE ICC97, pp.182-186, 1997. [43] A Pitsillides, P Ioannou, "An Integrated Switching Strategy for ABR Traffic Control in ATM Networks," in Proceedings Second IEEE Symposium on Computer and Communications (1997) Issue: July, Publisher: IEEE Comput. Soc., pp. 501-506, 1997. [44] Nasir Ghani, John W Mark,"Dynamic Rate-Based Control Algorithm for ABR Service in ATM Networks," In Proc of the IEEE/GLOBECOM'96, pp.1074-1079, 1996. [45] E. A. Khalil, I. Z. Morsi, and M. Mashem, "New Proposal to the ERICA+ Switch Algorithm," Accepted for publication in the 3rd International Conference on Networking (ICN'04), Mrach 1-4, 2004, Pointe-a-Pitre, Guadeloupe, Franch Canbbean. [46] E. A. Khalil, I. Z. Morsi, and M. Mashem, "Achieving QoS For TCP Multimedia Traffic Over rd ATM-ABR Services," Accepted for publication in the 3 International Conference on Networking (ICN'04), Mrach 1-4, 2004, Pointe-a-Pitre, Guadeloupe, Franch Canbbean.
37
Vol. 3, Issue 1, pp. 29-40
[47] Y. Afek, Y. Mansour ,and Z. Ostfeld, "Phantom: A Simple and Effective Flow Control Scheme," Proccedings of ACM SIGCOMM'96, August 1996. [48] K. Siu, and T. Tzeng, "Intelligent Congestion Control for ABR Service in ATM Networks," Computer Communication, Review, Vol. 24, No. 5, pp. 81-106, October 1995. [49] K-Yeung Siu, and Hong-Yi Tzeng, " Performance of TCP Over ATM with Time-Varying Available Bandwidth," Computer Communication, Vol. 19, pp.927-936, 1996. [50] Bobby Vandalore, "Traffic Management to Enhance Quality of Service (QoS) of Multimedia Over Available Bit Rate (ABR) Service in Asynchronous Transfer Mode (ATM) Networks," Ph.D. Dissertation, the Ohio Sate University, June 2000.
[51] Bobby Vandalore, Raj Jain, Rohit Goyal, Sonia Fahmy, " Dynamic Queue Control Functions For ATM ABR Switch Scheme: Design and Analysis," Computer Networks, Vol. 31, Issue 18, pp. 1935-1949, August 1999.
[52] B. Vandalore, R. Jain, R. Goyal, and S. Fahmy, Design and Analysis of Queue Control Functions for Explicit Rate Switch Schemes, Proceedings of ICCCN98, pp. 780-786, October 98. [53] Christos Tryfonas, MPEG-2 Transport Over ATM Networks, M.S Thesis, UC Santa Cruz, Sept 1996. [54] M. Graf, VBR video over ATM: Reducing Network Resource Requirement through end System traffic shaping,, Proceedings IEEE INFOCOM97, Kobe, Japan, pp. 48-57, Apr. 7-11 1997.
Flow charts
38
Vol. 3, Issue 1, pp. 29-40
39
Vol. 3, Issue 1, pp. 29-40
Author
Ehab Aziz Khalil, (B.Sc78 M.Sc.83 Ph.D.94), Ph.D. in Computer Network and Multimedia in the Dept. of Computer Science & Engineering, Indian Institute of Technology (IIT) Bombay-400076, India in July 1994, Research Scholar from 1988-1994 with the Dept. of Computer Science & Engineering, Indian Institute of Technology (IIT) Bombay-400076, India, M.Sc in the Systems and Automatic Control, Faculty of Electronic Engineering, Minufiya University, Menouf 32952, EGYPT, Oct. 1983, B.Sc. in the Dept. of Industrial Electronics, Faculty of Electronic Engineering, Minufiya University, Menouf 32952, EGYPT, May 1978. Since July 1994 up to now, working as a Lecturer, with the Dept. of Computer Science & Engineering, Faculty of Electronic Engineering, Minufiya University, Menouf 32952, EGYPT.. Participated with the TPC of the IASTED Conference, Jordan in March 1998, and With the TPC of IEEE IC3N, USA, from 2000-2002. Consulting Editor with the Whos Who? in 2003-2004. Member with the IEC since 1999. Member with the Internet2 group. Manager of the Information and Link Network of Minufiya University, Manager of the Information and Communication Technology Project (ICTP) which is currently implementing in Arab Republic of EGYPT, Ministry of Higher Education and the World Bank. Published more than 85 research papers and articles review in the international conferences, Journals and local newsletter.
40
Vol. 3, Issue 1, pp. 29-40
AUTOMATED TEST JIG FOR UNIFORMITY EVALUATION OF LUMINARIES

Deepa Ramane1, Jayashri Bangali2, Arvind Shaligram3
1
Department of Electronic science ,Dr. D. Y. Patil Arts, Commerce and Science College, Pimpri, Pune, Maharashtra, India 2 Department of Electronics, Kaveri College of Science and Commerce, Erandwane, Pune, Maharashtra, India 3 Department of Electronics Science, University of Pune, Pune, Maharashtra, India
ABSTRACT
Uniformity of illumination is of prime importance in the designing of lighting system. Often it is achieved by measuring illuminance values on the target plane at predefined locations using luxmeter. To speed up the experimentation and to automate the measuring procedure, a portable test-jig is developed at work place. It acquires illuminance data from sixteen predefined locations on target surface simultaneously. The data is used for computing uniformity of illumination. The test-jig consists of photo sensors, Data Acquisition System (DAS) and control system. The photo sensors are fixed on a flex sheet which can be rolled. The sheet is spread over the surface whose uniformity is to be measured. The portable, automated test-jig gives the uniformity of illumination quickly. The test-jig is used to evaluate the performance of three luminaries viz. incandescent bulb, CFL and LED. Results confirm that LED bulbs are energy efficient than others giving more illuminance without hampering illumination uniformity. LED luminaire of 6 W illuminates 1m X 1m target surface having uniformity in the range of 0.4 0.73 for target-source distance of 50cm 170cm. Dependence of uniformity on source wattage and view angle is also reported.
KEYWORDS:
lighting system, uniformity of illumination, photo sensor, DAS (Data Acquisition System)
I.
INTRODUCTION
Light Emitting Diodes (LEDs) have been the subject of growing interest of lighting designers over recent years. LEDs are replacing conventional light sources such as incandescent bulb, compact fluorescent tube etc. in almost all of the illumination applications [1, 2]. Wide range of illumination applications demand different specifications such as recommended lux level, uniformity of illumination, minimum glare, low cost, less energy consumption etc. The recommended values of these specifications are provided in IESNA handbook and are achieved by proper luminaire design. Uniformity is one of the major specifications in many applications. Before installation of luminaire, lighting designer has to verify whether the proposed illumination system design, fulfils the specified value of uniformity along with the recommended value of illuminance level. Computation of uniformity needs illuminance data on target surface from number of locations. The data provides maximum, minimum and average illuminance values. Acquisition of illuminance data from number of locations simultaneously and accurately is a challenging task. The proposed automated, portable test-jig measures the illuminance in lux at different points on the target surface and computes the uniformity of illumination which can be used as a metric to evaluate the performance of the luminaire. The illumination uniformity evaluation methods reported in literature are reviewed in the beginning. The test-jig developed for automatic acquisition of illuminance data is presented in subsequent section followed by description of experimental setup. Using the test-jig performance of three different
41
Vol. 3, Issue 1, pp. 41-47
luminaries viz. incandescent bulb, CFL and LED is evaluated. The results of this testing and discussion on the obtained results are given at the end.
II.
LITERATURE REVIEW
Lot of research is being done to design luminaire with improved uniformity. Majority efforts claim to provide optimal solution so as to have uniform illumination over a planer surface by optimizing number of parameters like number of source elements, their geometrical placement, optical characteristics of sources, source to target plane distance etc [3-8]. Papers have reported use of secondary optics such as diffuser, lenses, and reflectors for further improvement in uniformity [9]. To verify the feasibility of LED luminaire design, optical simulation programs either based on ray tracing or on analytical equations are used [10-13]. Experimentally, performance of luminaire can be evaluated by three different methods: illuminance, luminance, and small target visibility [14]. Among these methods, the illuminance measurement method is mostly used to evaluate the luminaire performance. Zeljko et. al. have measured illuminance with automated inspection of ceramic tiles using dot-method[15]. Many times the uniformity is computed by capturing image of illuminated surface by CCD camera. The image is used to plot iso-contours using MATLAB software which are used as metric of uniformity [7]. Simple method to measure illuminance levels is using handheld luxmeter. The illuminance data is acquired from different points on the target surface by manually positioning the luxmeter at desired locations. The data is used to determine maximum, minimum and average illuminance values which predict uniformity. This method of manual collection of data is sometimes troublesome and labour-intensive. In applications like evaluation of roadway lighting systems, safety of operators is key issue. Zhou et . al. have reported a measurement system consisting of light meter, a distance measurement system, a computer and software for roadway lighting system [14]. Considering need for development of portable, handy and speedy automated test-jig for uniformity evaluation, an attempt has been made.
III.
TEST JIG DEVELOPMENT
The block diagram of the test-jig developed for this task is shown in figure 1. It consists of a luminaire under test, the control system, 16 channel Data Acquisition System (DAS) and a personal computer to store, analyze and display the captured data. The idea is to sense the illuminance on the target surface at number of predetermined points for the computation of uniformity as per IESNA guideline. Further illumination analysis is carried out in MATLAB version 7.8. The development of test-jig involves the following tasks: hardware development for automatic collection of illuminance data; software development to collect, store and analyze illuminance data.
3.1 Hardware Development

Development of hardware involves designing of control system and 16 channel DAS. The control system comprises of photo sensors, signal conditioners, an analog multiplexer, a counter and a pulse generator. The photo sensors are fixed on the flex sheet of 1m X 1m dimension at predetermined locations. The sheet is spread over the target plane of which uniformity is to be evaluated. This photo sensor sheet can be folded or rolled so that it becomes handy for carriage and for storage. Here photo sensors used are LDR whose output resistances are calibrated against intensity values using luxmeter. The locations are dependent on number of luminaires for which guideline is given in IESNA handbook [16]. The photo sensors capture illuminance values. The outputs of these photo sensors are connected to the input channels of the analog multiplexer. The analog multiplexer sequentially passes channel illuminance values to output of DAS. To automate the selection of the channels, 4-bit counter is used which generates binary sequence from 0000 to 1111. The rate of channel selection is controlled by the frequency of the pulse generator. The output of the multiplexer is connected to Data Acquisition System of Rishabh company.
42
Vol. 3, Issue 1, pp. 41-47
Figure 1. Block Diagram of Test-jig for Uniformity Evaluation of Illumination System
3.2 Software Development

Through USB port the resistance data of photo sensors are transferred to Rishabh system software one after the other. The equivalent illuminance data values are retrieved in excel sheet as a function of position of target point. The sampling frequency of data acquisition can be adjusted in the software. For the results produced in the next section sampling frequency is kept as 1 second. From each target point five data samples have been taken. Average of these data samples is taken as illuminance of that target point. Further the uniformity is computed using formula:
Uniformity of illumination =
Average illuminance Maximum illuminance
In this case average illuminance on target plane is the average of illuminance values at n target points.
n
Illuminance values
Average illuminance =
1
at target point
.... 2
Perfect uniform illumination is said to be achieved when maximum and average illuminance levels match. Under this condition uniformity ratio become one. Designer tries to achieve uniformity value as maximum as possible. For general illumination applications uniformity ratio upto 0.6 is tolerable while task illuminations need uniformity ratio greater than 0.9. Illuminance data at n target points is imported in MATLAB 7.8. The 3-D plots of illuminance distribution helps in further analysis.
43
Vol. 3, Issue 1, pp. 41-47
IV.
EXPERIMENTAL SETUP
Experimental setup used for uniformity measurement is given in figure 2. It consists of source assembly, pulley arrangement and a luxmeter. The luxmeter is used to note down illuminance readings to validate the results of the developed test-jig. The pulley arrangement is used for source mounting and source-target height adjustment. The source assembly is mounted on a wooden board along with its driving circuitry. The source height is adjusted by moving the wooden board assembly up and down by pulling the rope. The target surface is at a distance of h meter from the source. The flex sheet is spread over the illuminated target surface. Thirteen photo sensors are placed on the flex sheet as shown in figure 3. The placement of photo sensors is as per the guidelines given in reference 16 for average illuminance measurement for regular area with symmetrically located single luminaire. Illuminance values are measured using test jig and also by luxmeter.
rope Pulleys LED Source Incandescent source Compact fluorescent source
Luxmeter Target Surface Illumination Pattern
Figure 2 Experimental setup for uniformity measurement
Figure 3 Placement of photo sensors on the flex sheet
44
Vol. 3, Issue 1, pp. 41-47
V.
RESULTS AND DISCUSSIONS
The test-jig is used to evaluate the uniformity of illumination when the target surface of 1m X 1m dimension is illuminated by incandescent, CFL and multiple LED source luminaire at different heights. Considering LED as future of illumination, experiments are performed for multiple LED source luminaire with variable source geometry and with optical characteristics.
Uniformity values are computed using illuminance results obtained using test-jig as well as by actual measurement with luxmeter. The results of experimentation are tabulated in table 1 and 2. Table 1 compares the performance of light sources at different heights. Figure 4 shows 3-dimensional graphs of illuminance distribution over flex sheet, plotted in MATLAB.
Table 1 Uniformity results for source comparison ( T = test-jig results; E = luxmeter results )
Light source LED, 6 W, CFL, 11 W Incandescent, 60 W Uniformity at height 50 cm T E 0.41 0.42 0.30 0.44 0.46 0.32 Uniformity at height 80 cm T E 0.52 0.51 0.41 0.57 0.53 0.45 Uniformity at height 110 cm T E 0.61 0.62 0.51 0.64 0.65 0.53 Uniformity at height 140 cm T E 0.65 0.63 0.61 0.67 0.67 0.64 Uniformity at height 170 cm T E 0.70 0.72 0.80 0.73 0.76 0.84
(a) LED luminaire : 6 W
(b) CFL luminaire : 11 W
(c) Incandescent luminaire : 60 W
Figure 4 : 3-dimensional view of Illuminance distribution over flex sheet
3-D graphs show greater peak illuminance at center of target plane for all luminaries. The peak values are 61.8, 29.5 and 41.6 lux for LED, CFL and Incandescent luminaire respectively. The illuminance is greater for LED luminaire in spite of lower wattage. Table 1 results show that uniformity of LED bulb is at par with other two sources. So one can say that LED bulbs are energy efficient giving more illuminance without hampering uniformity. LED luminaire performance is evaluated for increased in wattage and increasing view angles at different heights. Table 2 summarizes the uniformity results for the same.
Table 2 Uniformity results for LED luminaire ( T = test-jig results; E = luxmeter results )
LED luminaire wattage with spatial distribution of each LED 6W, 135 6 W, 60 12 W, 60 Uniformity at height 50 cm Uniformity at height 80 cm Uniformity at height 110cm Uniformity at height 140 cm Uniformity at height 170 cm
T 0.41 0.34 0.41
E 0.44 0.38 0.44
T 0.52 0.41 0.63
E 0.57 0.44 0.67
T 0.61 0.54 0.65
E 0.64 0.58 0.66
T 0.65 0.60 0.70
E 0.67 0.62 0.72
T 0.70 0.63 0.81
E 0.73 0.65 0.83
45
Vol. 3, Issue 1, pp. 41-47
Readings show that as height increases uniformity of illumination improves for all three luminaires. For same height wider angle LED luminaire gives better uniformity. If one increases source flux then also uniformity greatly improves.
VI.
CONCLUSION
The paper reports an automatic test-jig using 16-channel DAS useful for the illumination uniformity measurement. The photo sensors are mounted on flex sheet of size 1m X 1m and the paper is spread over the area on the illuminated surface of which analysis is to be done. Using developed test-jig illuminance values on target plane are measured at predefined locations. It simultaneously acquires illuminance data from sixteen predefined locations. The portable, automated test-jig speeds up the illuminance measurement procedure and hence useful for computing uniformity of illumination on the target plane. The jig is tested for three types of conventional sources and for LED luminaire with variable wattage and view angle. The developed test-jig results are compatible with the results obtained using luxmeter. The developed system is useful for evaluation of uniformity of illumination for all types of luminaire and thus helps the designer to optimize luminaire design.
REFERENCES
[1]. Evans D.L., High Luminance LEDs Replace Incandescent Lamps in New Applications, SPIE, vol. 3002, 142-153. [2]. Peon R, Doluweera G, Platonova, Halliday D., (2005) Solid state lighting for the developing world- the only solution, SPIE, 5th international conference on solid state lighting, vol. 5941,59410N-1 to 59410N-15. [3]. Moreno I., (2004) Configuration of LED arrays for uniform illumination, Proc. SPIE, Vol. 5622, pp 713-718. [4]. Kopparapu S.K., (2006) Lighting design for machine vision application, Image Vision Computation, Vol. 24, pp 720-726. [5]. Moreno I., (2006) Design of LED spherical lamps for uniform far-field illumination, Proc. SPIE, Vol. 6046, pp 60462E 1- 60462E7. [6]. Wittels, Norman, Gennert, Michael A., (1994) Optimal lighting design to maximize illumination uniformity, SPIE, Vol. 2348, pp 46-56. [7]. Hongming Yang, Jan W. M. Bergmans, Tim C. W. Schenk, Jean-Paul M. G. Linnartz, and Ronald Rietman, (2009) Uniform Illumination Rendering Using an Array of LEDs: A Signal Processing Perspective, IEEE Transactions on Signal Processing, Vol. 57, No. 3, pp 1044 1057. [8]. Gennert, Michael A., Wittels, Norman, Leatherman, Gary L, (1993) Uniform frontal illumination of planer surfaces: where to place the lamps, Optical Engg. Vol. 32 (06), pp 1261-1271. [9]. Fei Chen, Sheng Liu, Kai Wang, Zongyuan Liu, Xiaobing Luo, (2009) Free-form lenses for high illumination quality light-emitting diode MR16 lamps, Optical Engg, Vol. 48 (12), pp 123002-7. [10]. Zong Qin, Kai Wang, Fei Chen, Xiaobing Luo, Sheng Liu, (2010) Analysis of condition for uniform lighting generated by array of light emitting diodes with large view angle, Optics Express, Vol. 18 (16), pp 17460-76. [11]. Allen Whang, Yi-Yung Chen, Yuan-Ting Teng, (2009) Designing Uniform Illumination Systems by surface-tailored Lens and configurations of LED arrays, Journal of Display Technology, Vol. 5 (3), pp 94-103. [12]. Jin-Jia Chen, Kuang-Lung Huang, Te-Yuan Wang, Yi-Chih Wang, Chuen-Ching Wang, Tsung-Yi Guo, (2010), LED lighting module design based on a prescribed candle-power distribution for uniform illumination, Proc. of SPIE, Vol. 7849, pp 78490X-1 to 78490X8. [13]. Ramane D.V., Shaligram A.D., (2011) Optimization of multielement LED source for uniform illumination of plane surface, Optics Express, Vol. 19, No. S4, A639-A648.
46
Vol. 3, Issue 1, pp. 41-47
[14]. Huaguo Zhou, Fatih Pirinccioglu, Peter Hsu, (2009) A new roadway lighting measurement system, Transportation Research Part C, vol.17, 274-284. [15]. Zeljko Hocenski, Adriana Dizar, Verica Hocenski, (2008), Illumination Design of a control system for visual inspection of ceramic tiles, IEEE, pp 1093-97. [16]. Measurement of Light and other Radiant Energy , IES Lighting handbook , 1981: reference volume, section 4.
Authors Biographies
D.V. Ramane born in 1970, is Asst. Prof. in the Department of Electronics, Dr. D.Y. Patil Arts, Commerce and Science College, Pimpri, Pune, India. She received her M. Sc. in 1992 and pursuing her Ph.D. at University of Pune. She has more than 10 publications in the National and International Journals/Conferences. She has authored 11 books on Electronics subjects. Her research interest includes solid state lighting, mathematical modeling and simulation and PC/Micro controller based Instrumentation.
J.A. Bangali born in 1974, is Asst. Prof. in the Department of Electronics, Kaveri College of Science and Commerce, Erandwane, Kothrud, Pune, India. She received her M. Sc. in 1996 and M.Phil. in 2009 from University of Pune. She has around 10 publications in the National and International Journals/Conferences. Her research interest includes smart space automation, lighting research, PC/Micro controller based Instrumentation and sensors.
A. D. Shaligram born in 1960, is professor and Head of the Department of Electronic Science, University of Pune, Pune, India. He received his M. Sc. in 1981 and Ph. D. in 1986 from the same University. He has more than 70 publications in the National and International Journals. His research interest includes fiber optic and optical waveguide sensors, PC/Micro controller based Instrumentation and Biomedical Instrumentation and sensors.
47
Vol. 3, Issue 1, pp. 41-47
TRACKING OF INTRUDERS ON GEOGRAPHICAL MAP USING IDS ALERT

Manish Kumar1, M. Hanumanthappa2, T.V. Suresh Kumar3
Asst. Professor, Dept. of Master of Computer Applications, M. S. Ramaiah Institute of Technology, Bangalore-560 054, India 2 Dept. of Computer Science and Applications, Jnana Bharathi Campus, Bangalore University, Bangalore -560 056, India 3 Professor & Head, Dept. of Master of Computer Applications, M. S. Ramaiah Institute of Technology, Bangalore-560 054, India
1
ABSTRACT
Intrusion Detection System (IDS) is the most powerful system that can handle the intrusions of the computer environments by triggering alerts to make the analysts take actions to stop this intrusion. Knowing from where an intrusion originated or where a spammer or suspect intruder is located is key information to identify security threats to your system and confidential information. In this paper we are discussing the method which tracks the intruders on the geographical map. The technique is based on the source IP address of Intruders. In current Internet communication world, validity of the source of IP packet is an important issue. The problems of IP spoofing alarm legitimate users of the Internet. IP spoofing is a technique used to gain unauthorized access to computers, whereby the intruder sends messages to a computer with an IP address indicating that the message is coming from a trusted host. This paper also discusses some of the technique to defense the IP spoofing problem.
KEYWORDS: Intrusion Detection, Geolocation, IP Spoofing.
I.
INTRODUCTION
A computer intrusion is, An intentional event where an intruder gains access that compromises the confidentiality, integrity, or the availability of computers, networks, or the data residing on them. The amount of damage done by an intruder to a system can vary greatly. Some intruders are malicious in nature and others are just curious and want to explore what is on a local network. Computer users must protect themselves from intrusion. While there are no 100% effective methods of eliminating intruders completely, some methods must be used to reduce intrusions. In the event that an intrusion has taken place the last line of defense is an intrusion detection system. An intrusion detection system can alert the system administrator in the event that the system has been breeched. Once the intrusion detection system has detected an event, an intrusion investigation should be conducted to note the extent of the intrusion and any damages that may have occurred and to locate the source of the attack. While intrusion prevention, detection and tolerance all play an important role in solving the problem of todays network-based intrusion, they are all passive and not adequate to solve the intrusion problem [22]. One fundamental problem with existing intrusion prevention, detection and tolerance is that they do not effectively eliminate or deter network-based intrusions. The best they can do is to avoid being victims of network-based intrusions temporarily. Because they do not address the root cause of intrusions intruders, those intruders can always explore new system vulnerabilities, find new accomplices potentially insiders, and launch new attacks from different places. Because intrusion prevention, detection and tolerance do not effectively address the problem of compromised
48
Vol. 3, Issue 1, pp. 48-54
system recovery, intruders can use those compromised system as new base for further intrusion. What we need is an effective way to hold network-based intruders accountable for their intrusions. So far most computer security research regarding network-based intrusions has focused on prevention and detection. Intrusion response has been an afterthought and is generally limited to logging, notification and disconnection at local host. Any further response usually involves manual interactions such as off-line analysis, reporting incidents to CERT and installing fixes. Given todays high-speed network, many network-based intrusions can be very fast and short across wide areas of networks. Current ad hoc and manual intrusion response process neither provides the needed real-time intrusion response nor scales with network. Furthermore, because current automated intrusion responses are passive and lack network-wide response, they do not, however, eliminate or even effectively deter network-based intrusions. In this paper, we propose Tracing-the intruders on geographical map using the log report IP address to address the problem of network-based intrusions. It differs from existing intrusion defense approaches in that it addresses the intrusion problem by targeting the root cause of the problem: intruders. It collaborates with IDS and will trigger automatically when intrusions are detected. By actively tracing intrusions at real-time, it helps to apprehend the intruders on the spot and hold them accountable for the intrusions. Therefore it is likely a more effective deterrent against further intrusion attempts.
II.
INTRUSION DETECTION SYSTEM IN PRACTICE
IDS have historically been categorized as network, host, anomaly or misuse (signature) based. This simple categorization is, however, no longer adequate. IDS can also be distributed or centralized, can be passive or reactive, can be application-specific or general-purpose, can focus on real time or afterthe-event analysis [9] [14]. The five IDS described later are not intended to be exemplifiers of these or other various categories, but are presented as an indication of how major trends have developed. Almost all IDS will output a small summary line about each detected attack [15]. This summary line typically contains the information fields shown below. 1. Time/date; 2. Sensor IP address; 3. Vendor specific attack name; 4. Standard attack name (if one exists); 5. Source and destination IP address; 6. Source and destination port numbers; 7. Network protocol used by attack. Other more general information is also often provided; information such as a textual description of the attack, identification of the software attacked, information that identifies the patches required to fix the vulnerability, and advisories regarding the attack. Attacks on systems and networks have increased as rapidly changing technology, systems integration, global networks, information warfare and hacker boredom have become prevalent. For a long time, enterprises have relied upon solutions like intrusion detection to alert them of potential attacks. As the Internet and e-business have evolved, so has the need for a more proactive solution. In this paper we are discussing the techniques for detecting the intrusion, tracking the intruders IP address and showing the intruders location on geographical map The most challenge part is to track the intruders IP address. Many time intruders use the spoofed IP address to hide their identity. We need to first track the correct IP address of the intruders. The following section discusses the various techniques for tracking the intruders IP address.
III.
METHODS OF IP TRACEBACK
The purpose of IP traceback is to identify the true IP address of a host originating attack packets [21]. Normally, we can do this by checking the source IP address field of an IP packet. Because a sender can easily forge this information, however, it can hide its identity. If we can identify the true IP address of the attack host, we can also get information about the organization, such as its name and the network administrators e-mail address, from which the attack originated. With IP traceback technology, which traces an IP packets path through the network, we can find the true IP address of
49
Vol. 3, Issue 1, pp. 48-54
the host originating the packet. To implement IP traceback in a system, a network administrator updates the firmware on the existing routers to the traceback support version, or deploys special tracing equipment at some point in the network. Existing IP traceback methods can be categorized as proactive or reactive tracing.
3.1 Proactive Tracing

Proactive tracing prepares information for tracing when packets are in transit. If packet tracing is required, the attack victim (or target) can refer to this information to identify the attack source. Two proactive tracing methods packet marking and messaging is explained. Packet Marking: - In packet marking, packets store information about each router they pass as they travel through the network. The recipient of the marked packet can use this router information to follow the packets path to its source. Routers must be able to mark packets, however, without disturbing normal packet processing. With IPs record route option, for example, the IP packet can store router addresses in its option field. In another proposed approach, the router writes its identifier probabilistically in the packets IP header identification field. Each marked packet contains information in its identification field about only one or two routers on the attack path. In a flooding-style attack, however, the target network receives many attack packets and can collect enough information to identify the attack path. The identification field is used to reassemble fragmented packets. Because few fragments are created on the Internet, however, modifying the identification field rarely affects normal packet processing. Messaging: In messaging approaches, routers create and send messages containing information about the forwarding nodes a packet travels through. The Internet Engineering Task Forces proposed method, the Internet control message protocol (ICMP) traceback message. A router creates an ICMP traceback message, which contains part of a traversing IP packet, and sends the message to the packets destination. We can identify the traversed router by looking for the corresponding ICMP traceback message and checking its source IP address. Because creating an ICMP traceback message for every packet increases network trafc, however, each router creates ICMP traceback messages for the packets it forwards with a probability of 1/20,000. If an attacker sends many packets (for example, in a ooding-style attack), the target network can collect enough ICMP traceback messages to identify its attack path.
3.2 Reactive Tracing

Reactive tracing starts tracing after an attack is detected. Most of the methods trace the attack path from the target back to its origin. The challenges are to develop effective traceback algorithms and packet-matching techniques. Various proposals attempt to solve these problems.
3.3 Hop-by-hop tracing.

In hop-by-hop tracing, a tracing program, such as MCIs DoS Tracker, logs into the router closest to the attacked host and monitors incoming packets. If the program detects the spoofed packet (by comparing the packets source IP address with its routing table information), it logs into the upstream routers and monitors packets. If the spoofed flooding attack is still occurring, the program can detect the spoofed packet again on one of the upstream routers. This procedure is repeated recursively on the upstream routers until the program reaches the attacks actual source.
3.4 Hop-by-hop tracing with an overlay network.

In hop-by-hop tracing, the more hops there are, the more tracing processes will likely be required. As a result, a packet will take longer to trace, and necessary tracing information might be lost before the process is complete. To decrease the number of hops required for tracing, one approach builds an overlay network by establishing IP tunnels between edge routers and special tracking routers and then reroutes IP packets to the tracking routers via the tunnels. Hop-by-hop tracing is then performed over the overlay network.
3.5 IPsec authentication.

Another proposed reactive tracing technique is based on existing IP security protocols. With this method, when the IDS detects an attack, the Internet key exchange (IKE) protocol establishes IPsec
50
Vol. 3, Issue 1, pp. 48-54
security associations (SAs) between the target host and some routers in the administrative domain (for example, autonomous system boundary routers). Routers at the SA ends add an IPsec header and a tunnel IP header containing the routers IP address to traversing packets. If the attack continues and one of the established SAs authenticates a subsequent attack packet, the attack must come from a network beyond the corresponding router. The receiver checks the source IP address of the tunnel IP header to find out which routers the attack packet traversed. Repeating this process recursively, the receiver finally reaches the attack source. Because this technique uses existing IPsec and IKE protocols, implementing a new protocol for tracing within an administrative domain is unnecessary. To trace beyond the administrative domain, however, a special collaboration protocol is needed. The IETF Intrusion Detection working group (IDWG) is discussing such a protocol. Traffic Pattern Matching. A fourth proposed technique traces an attack path by comparing traffic patterns observed at the entry and exit points of the network with the network map.
IV.
CURRENT ACTIVE RESEARCH ON IP TRACEBACK
Two network tracing problems are currently being studied: IP traceback and traceback across stepping-stones (or a connection chain). IP traceback is to identify the origins of sequences IP packets (e.g., identify the origin of DDOS packets) when the source IP addresses of these packets are spoofed. IP traceback is usually performed at the network layer, with the help of routers and gateways. Traceback across stepping-stones is to identify the origin of an attacker through a chain of connections (e.g., connections established with telnet, rlogin, or ssh), which an attacker may use to hide his/her true origin when he/she interacts with a victim host. Traceback across stepping-stones is beyond the network layer, since at each intermediate host the data is transmitted to application layer in one connection, and then resent to the network in the next connection. Research on IP trace back has been rather active since the late 1999 DDOS attacks [17,24,18]. Several approaches have been proposed to trace IP packets to their origins. The IP marking approaches enable routers to probabilistically mark packets with partial path information and try to reconstruct the complete path from the packets that contain the marking [10,16,4]. DECIDUOUS uses IPSec security associations and authentication headers to deploy secure authentication tunnels dynamically and trace back to the attacks origins [19,1]. ICMP traceback (iTrace) proposes to introduce a new message ICMP trace back (or an iTrace message) so that routers can generate iTrace messages to help the victim or its upstream ISP to identify the source of spoofed IP packets [2]. An intention-driven iTrace is also introduced to reduce unnecessary iTrace messages and thus improve the performance of iTrace systems [3]. An algebraic approach is proposed to transform the IP traceback problem into a polynomial reconstruction problem, and uses techniques from algebraic coding theory to recover the true origin of spoofed IP packets [13]. An IP overlay network named CenterTrack selectively reroutes interesting IP packets directly from edge routers to special tracing routers, which can easily determine the ingress edge router by observing from which tunnel the packets arrive [12]. A Source Path Isolation Engine (SPIE) has been developed; it stores the message digests of recently received IP packets and can reconstruct the attack paths of given spoofed IP packets [11,7]. There are other techniques and issues related to IP traceback (e.g., approximate traceback [8], legal and societal issues [23], vendors solutions [5]). An archive of related papers can be found at [20]. Though necessary to make attackers accountable (especially for DDOS attacks where there are a large amount of packets with spoofed source IP addresses), IP traceback has its own limitations. In particular, IP traceback cannot go beyond the hosts that send the spoofed IP packets. Indeed, a typical attacker will use a fair number of steppingstones before he/she finally launches, for example, a DDOS attack. Thus, only identifying the source of IP packets is not sufficient to hold the attackers responsible for their actions. Similar to IP traceback, there have been active research efforts on tracing intruders across stepping-stones. In general, approaches for traceback across a connection chain can be, based on the source of tracing information, divided into two categories: host-based and network-based. In addition, depending on how the traffic is traced, traceback approaches can be further classified
51
Vol. 3, Issue 1, pp. 48-54
into either active or passive. Passive approaches monitor and compare all the traffic all the time, and they do not select the traffic to be traced. On the other hand, active approaches dynamically control when, where, what and how the traffic is to be correlated through customized processing. They only trace selected traffic when needed.
V.
GEOLOCATION TRACKING OF INTRUDERS IP ADDRESS
After tracking the IP address of intruders, our next objective is to find the geolocation of the intruders. IP to geolocation tracking is the technique of determining a user's geographic latitude, longitude and, by inference, city, region and nation by comparing the user's public Internet IP address with known locations of other electronically neighboring servers and routers IDS can detect the intrusion. We can find the IP address of intruders but barely having a IP address, it do not give the idea that from which place attack is generated.
5.1 Advantage of Geolocation Tracking

Tracking the intruders IP address and plotting the trace on geogrphical (Figure:- 1) map gives a clear picture that whether the attack is distributed and initiated from multiple country or it is initiated from one specific country our region. This information may be the vital information for the organization to take any further action or any precaution measures.
Figure 1: Intruders IP address tracking on Map
5.2 System Architecture

The overall system (Figure 2) works on IDS alert analysis. Each alerts generated by IDS is passed to IDS alerts log report. All the alerts from IDS log report is further analyzed for tracking the Intruders source IP address. Once the correct source IP address of the intruders is confirmed, it is passed to the API which map the source IP address on geographical map.
Intrusion Detection System IDS Alerts Log Report Intruders IP Address Validatior Mapping the Intruders IP Address on Geographical Map
Figure 2: Architecture of Intruders Geographical Location Mapping System
5.3 Implementation Detail

We have implemented the system using Snort and Google API for geolocation mapping of intruders. Snort is the well known open source IDS software which detect the intrusion event. Snort log this report in alert file. The intruders IP address is analyzed and traced back. The traced IP address is passed to Google Geolocation API which enables a web application to: Obtain the user's current position, using the getCurrentPosition method Watch the user's position as it changes over time, using the watchPosition method Quickly and cheaply obtain the user's last known position, using the lastPosition property The Geolocation API provides the best estimate of the user's position using a number of sources (called location providers). These providers may be onboard (GPS for example) or server-based (a
52
Vol. 3, Issue 1, pp. 48-54
network location provider). The getCurrentPosition and watchPosition methods support an optional parameter of type PositionOptions which lets you specify which location providers to use.
VI.
EVALUATION
Geolocation of intruders are obtained by tracking the IP addresses of intruders using databases that map Internet IP addresses to geographic locations. Google uses MaxMinds database for mapping IP addresses to a geographical location [6]. They claim it is 99% accurate. What is in the fine print, is that it is 99% accurate in determining the country, but pinpointing the exact position is still a challenging issues which need to be addressed.
VII.
CONCLUSIONS
Our system is able to trace the intruders geographical map but our whole system is depended on the IP traceback. Traceback has several limitations, such as the problem with tracing beyond corporate firewalls. To accomplish IP traceback, we need to reach the host where the attack originated. Another limitation relates to the deployment of traceback systems. Most traceback techniques require altering the network, including adding router functions and changing packets. To promote traceback approaches, we need to remove any drawbacks to implementing them. Moreover, even if IP traceback reveals an attacks source, the source itself might have been used as a stepping-stone in the attack. IP traceback methods cannot identify the ultimate source behind the stepping-stone; however, techniques to trace attacks exploiting stepping-stones are under study. Some operational issues must also be solved before IP traceback can be widely deployed. To trace an attack packet through different networks, for example, there must be a common policy for traceback. We also need guidelines for dealing with traceback results to avoid infringing on privacy. Furthermore, we need to consider how to use information about an attack source identified by IP traceback. In the future, we will likely need to focus on the authenticity of results from IDSs and IP traceback systems.
REFERENCES
[1] Computer Emergency Response Team. (2000) CERT Advisory CA-2000-01 Denial-of Service Development. http:// www.cert.org/advisories/CA-2000-01.html. [2] Computer Emergency Response Team. (1999) Results of the Distributed-Systems Intruder Tools Workshop. http://www.cert.org/reports/dsit_workshop.pdf. [3] Computer Security Institute. Annual CSI/FBI Computer Crime and Security Survey. (2001) http://www.gocsi.com/prelea_000321.htm. [4] H. Y. Chang, R. Narayan, S.F. Wu, B.M. Vetter, X. Y. Wang et al. (1999) DecIdUouS: Decentralized Source Identification for Network-Based Intrusions, In Proceedings of 6th IFIP/IEEE International Symposium on Integrated Network Management. [5] H. Jung, et al. Caller Identification System in the Internet Environment. (1993) In Proceedings of 4th USENIX Security Symposium. [6] http://netsolutions.net.au/web-design/geo-targeting-by-ip-address/ ( Accessed on 28/01/2012) [7] J. D. Howard. (1997) An Analysis of Security Incidents on The Internet 1989 - 1995, PhD Thesis, http://www.cert.org/research/JHThesis/Start.html. [8] J. Ioannidis and M Blaze. (1993) The Architecture and Implementation of Network-Layer Security under Unix. In Proceedings of 4th USENIX Security Symposium. [9] K2. ADMmutate README. ADMmutate source code distribution. Version 0.8.4. URL: http://www.ktwo.ca/c/ADMmutate-0.8.4.tar.gz (Jan 2002) [10] K. L. Calvert, S. Bhattacharjee and E. Zegura. (1998) Directions in Active Networks. IEEE Communication Magazine. [11] L. T. Heberlein, K. Levitt and B. Mukherjee. (1992) Internetwork Security Monitor: An Intrusion-Detection System for Large-Scale Networks. In Proceedings of 15th National Computer Security Conference. [12] M. B. Greenwald, S. K. Singhal, J. R. Stone and D. R. Cheriton. (1996) Design an Academic Firewall: Policy, Practice and Experience with SURF. Internet Society Symposium on Network and Distributed System Security (NDSS 96). [13] N.G. Duffield and M. Grossglauser. (2000) Trajectory Sampling for Direct Traffic Observation. Proceedings of the ACM SIGCOMM 2000. [14] P. Stephenson The Application of Intrusion Detection Systems in a Forensic Environment, Proceedings of the RAID 2000 Conference, Toulouse, France, 2000.
53
Vol. 3, Issue 1, pp. 48-54
[15] Peter Sommer Intrusion Detection Systems as Evidence, Presented in RAID 98 Conference, Louvain-laNeuve, Belgium. [16] R. H. Campbell, Z. Liu, M. D. Mickunas, P. Naldurg and S. Yi. (2000) Seraphim: Dynamic Interoperable Security Architecture for Active Networks. In Proceedings of IEEE OPENARCH2000. [17] S. M. Bellovin. (2000) ICMP Traceback Messages. Internet Draft: draft-bellovin-itrace-00.txt. [18] S. Bhattacharjee, K. L. Calvert and E. W. Zegura. (1997) An Architecture for Active Networking. High Performance Networking (HPN97), White Plans, NY. [19] S. Staniford-Chen, L. T. Heberlein. (1995) Holding Intruders Accountable on the Internet. In Proceedings of IEEE Symposium on Security and Privacy. [20] S. Kent, R. Atkinson. (1998) Security Architecture for the Internet Protocol. IETF RFC 2401. [21] Tatsuya Baba and Shigeyuki Matsuda, Tracing Network Attacks to Their Sources, IEEE INTERNET COMPUTING, MARCH - APRIL 2002 [22] Wang, X., Reeves, D., & Wu, S. F. (n.d.). Tracing based active intrusion response. Retrieved from http://arqos.csc.ncsu.edu/papers/2001-09-sleepytracing-jiw.pdf [23] W. Jansen, P. Mell, T. Karygiannis, D. Marks. (1999) Applying Mobile Agents to Intrusion Detection and Response. NIST Interim Report (IR) 6416. [24] W. Bender, D. Gruhl, N. Morimoto and A. Lu. (1996) Technique for Data Hiding. IBM Systems Journal, Vol. 35, Nos. 3&4
AUTHORS
Manish Kumar is working as Asst. Professor in Department of Master of Computer Applications, M. S. Ramaiah Institute of Technology, Bangalore, India. His areas of interest are Cryptography and Network Security, Computer Forensic, Mobile Computing and eGovernance. His specialization is in Network and Information Security. He has also worked on the R&D projects relates on theoretical and practical issues about a conceptual framework for E-Mail, Web site and Cell Phone tracking, which could assist in curbing misuse of Information Technology and Cyber Crime. He has published several papers in International and National Conferences and Journals. He has delivered expert lecture in various academic Institutions. M Hanumanthappa is currently working as a Associate Professor in the Department of Computer Science and Applications, Bangalore University, Bangalore, India. He has over 15 years of teaching (Post Graduate) as well as Industry experience. His areas of interest include mainly Data Structures, Data Base Management System, Data Mining and Programming Languages. Besides, he has conducted a number of training programs and workshops for computer science students/faculty. He is also the member of Board of Studies /Board of Examiners for various Universities in Karnataka, India. He is also guiding the research scholars in the field of Data Mining and Network Security. T V Suresh Kumar is working as Professor and Head, Department of Master of Computer applications, M S Ramaiah Institute of Technology, Bangalore. He delivered lectures at various organizations like Honeywell, SAP Labs, Wipro Technologies, DRDO, Mphasis, Indian Institute of Science (Proficience), HCL Technologies, L&T Infotech, Nokia and various Universities/Academic Institutions. His areas of interest are Software Performance Engineering, Object Technology and Distributed Systems etc. He has published several research papers in various National and International Conferences and Journals. He has carried out software projects for various organizations. He is life member of ISTE and member of IEEE.
54
Vol. 3, Issue 1, pp. 48-54
DESIGN AND ANALYSIS OF MULTIDIELECTRIC LAYER MICROSTRIP ANTENNA WITH VARYING SUPERSTRATE LAYER CHARACTERISTICS
Samir Dev Gupta1, Amit Singh2
1
Department of Electronics and Communication Engineering, JIIT Noida, U. P., India. 2 Agilent Technologies, Manesar, Haryana, India
A BSTRACT
The multidielectric layer microstrip antenna structure involves addition of a superstrate layer over the substrate. It is important that the superstrate layer must act as a part of the antenna. Design of the multidielectric layer microstrip patch antenna based on different thickness and permittivity of the superstrate layer has significant effect in gain and antenna efficiency. The designer must however ensure that the superstrate layer does not adversely affect the performance of the antenna. With proper choice of the thickness of substrate and superstrate layer, significant increase in gain can be achieved for practical applications. This paper discusses a set of closed form expressions for the resonant frequency for the general case of multidielectric layers. Multidielectric layer microstrip antenna designed for applications where various physical properties of antenna viz. permittivity, patch dimensions, height of the substrate and superstrate layer parameters which significantly affect accuracy of the resonant frequency is analyzed. Considering the effect of superstrate layer, a method for accurately determining the resonant frequency of such structures have been obtained using variation of the patch dimension. The antenna performances have been evaluated for variety of cases of permittivity and thickness of the superstrate layer.
K EYW ORDS:
Layer
Microstrip Antenna, Multidielectric Layer, Resonant Frequency, Permittivity, Superstrate
I.
INTRODUCTION
Microstrip antennas have inherent limitation of narrow bandwidth. When a microstrip antenna is covered with a superstrate (cover) dielectric layer, its properties like resonance frequency, gain and bandwidth are changed which may seriously degrade the system performance [1- 4]. By choosing the thicknesses of substrate layers and the superstrate layer, a very large gain can be realized [5-9]. ShunShi Zhong, Gang Liu, and Ghulam Qasim [1] have described the significance of determining accuracy of the resonant frequency in the design of a microstrip antenna with multidielectric layers. Therefore in view of the inherent narrow bandwidth of the microstrip patch antennas, the antenna with multidielectric layered structure must be designed to ensure that there is a minimum drift in the resonant frequency [10]. Theoretical methods for calculating the resonant frequency of such structures have been reported using the variation technique, the multiport network approach, the spectral domain analysis and other full-wave analysis methods. Numerical methods are highly accurate but too laborious and time consuming for direct use in CAD programs. Generalization of the transmission line model treats a rectangular microstrip antenna with several dielectric layers as a multilayer microstrip line. With quasi-TEM wave propagating in the microstrip line, a quasistatic value of the effective permittivity eff is derived by means of the conformal mapping technique. Relatively simple expressions based on the conformal mapping technique and the transmission line model is therefore used even for more complicated multidielectric structures. The conformal transformation used by Wheeler [11] and by Svacina [4] has been used. For the general case of multidielectric layers, it has been suggested to obtain a set of closed form expressions for the resonant frequency. The frequency
55
Vol. 3, Issue 1, pp. 55-68
dependence of eff and the open-end extension of a patch, results in determination of resonant frequency of such antenna structures with good accuracy. Samir Dev Gupta et. al [12] have shown that the dimension of substrate and the patch of the microstrip antenna, and the corresponding calculated resonant frequency are such that a very small variation in the dimension of the antenna parameters results into a very significant change in the actual resonant frequency. Since these calculations are repetitively carried out, the errors are cumulative at every step, thereby resulting in a notable change in the frequency. To minimize the compounding errors an algorithm has been used [12]. The algorithm minimizes errors at each step thereby providing a result which is highly accurate. In the following sections the discussions and analysis are devoted to (i) The design of a multidielectric layer microstrip antenna at 10 GHz. The muldielectric layer design considers the effect of the cover layer on antenna performance. (ii) Studies on performance analysis of the multidielectric antenna based on parameters involving combination of the superstrate layer permittivity and thickness. (iii) Finally analyzing the characteristics of the designed antenna with/without superstrate layer. (iv) Analysis related to axial ratio on the basis of superstrate thickness. In addition improvement in bandwidth with changes in the superstrate thickness.
II.
EFFECT OF CHANGING SUPERSTRATE LAYER THICKNESS ON THE ANTENNA PARAMETERS
The microstrip antenna under consideration is designed to operate at a frequency of 10 GHz. The design is based on various selection criteria such as the thickness of the substrate and the superstrate, width and length of the element. Effects on antenna parameters with respect to the change in thickness of the superstrate layer have also been analyzed in the following subsections.
2.1 Substrate selection in the design of the patch antenna

Suitable dielectric substrate of appropriate thickness and loss tangent is chosen. A thicker substrate is mechanically strong with improved impedance bandwidth [13]. However it will increase weight and surface wave losses. The substrates dielectric constant r plays an important role similar to that of the substrate thickness. A low value of r for the substrate will increase the fringing field of the patch and thus the radiated power. A high loss tangent increases the dielectric loss and therefore reduces the antenna efficiency. The substrate parameters so chosen are as follows: The top layer chosen is RT Duroid 5880 having a thickness of 0.787mm, permittivity r = 2.2 and the loss tangent tan = 0.0009. The bottom layer dielectric is RT Duroid 5870 with a thickness of 0.787mm, permittivity r = 2.33 and the loss tangent tan = 0.0012.
2.2 Element width and length

The selection criteria for an efficient radiator with patch size which is not too large are: (i) A low value of the patch width W and (ii) The ratio between the width and the length of the patch leading to antenna with a good radiation efficiency. For the antenna to be an excellent radiator, the ratio between W and L should lie between 1< W
< 2 [14], [15]. The patch dimensions determine the
resonant frequency. Various parameters in design of microstrip antenna are critical because of the inherent narrow bandwidth of the patch. Using the algorithm [12], we first calculate antenna design parameters. Calculated length and width of the patch obtained is 8.69638mm and 9.6 mm respectively. Effective permittivity obtained is 2.15644, for the net height of the substrate 1.574 mm. So accordingly, the calculated Rin (the value of the patch resistance at the input slot) comes out to be 326. 8508ohms. The condition h <
1 is taken into account, we obtain the values of self conductance 10

1 1 (k 0 h )2 24
G1 and susceptance B1 [16] using equation (1) and (2) respectively.
W G1 = 120 0
(1)
56
Vol. 3, Issue 1, pp. 55-68
W B1 = 120 0 (1 0.636 log e (k 0 h ))
(2)
Treating element as two narrow slots, one at each end of the line resonator, the interaction between the two slots is considered by defining a mutual conductance. Considering far fields expressions, the directivity of a patch and the mutual conductance between patches are calculated [17].
W cos sin k 0 2 sin 3 d cos 0 G3 = 2 120
... (3)
W cos sin k 0 1 2 (J (k L sin )sin 3 )d G12 = 2 0 0 cos 120 0 1 Therefore the calculated input resistance of the patch is Rin = 2(G3 + G12 )
... (4) ... (5)
120 h ( A + B ) 100 log1 + 4 0.5 We 2 2 0.5 ( r + 1) Z= 100

1 1+ eff where We = W + We and We = W 2
We obtain W from the following equation
... (6)
... (7)
e t W = log 4 2 1 2 t + h W + 1 .1 t
Also the parameters A and B are obtained using the following equations.
... (8)
57
Vol. 3, Issue 1, pp. 55-68
14 + 8 eff A= 11
where eff
0. 5 1 1 + eff 4 h and B = A2 + 2 2 We 2 r + 1 r 1 1 + 0.041 W = + 1 h 2 2 2 1 + 12 h W
... (9)
... (10)
Thus applying the formula for the impedance matching we get the matching line impedance to be (326.8508*90) ^0.5=171.512 ohm. Calculation of the strip impedance process based on the microstrip patch antenna parameters width W, height h of the substrate having relative permittivity r and t the thickness of the patch involves the following sets of equations. Considering the condition W/h>1, the characteristic impedance of the strip is obtained based on the equation 6. The ADS based calculator determines the impedance matching feed line width which works out to be around 0.31 mm. Design changes incorporated based on the calculated values of the patch length and width is 8.69638 mm and 9.6 mm respectively.
2.3 Impedance Calculation

The matching transformer impedance =178.813 ohm and width = 0.26 mm, therefore the calculated Patch ohm. The input impedance is obtained from the ADS design. The patch with width 9.6 mm and length 8.58 mm corresponds to input impedance= 324.0673 ohm. Corresponding to the effective permittivity
(Impedance matching transformer impedance) 2 (178.813) Impedance= = =325.09025 98.3545 Feed line impedance
2
eff is equal to 2.15644, the input resistance of the patch is equal to 326.8508 ohm. Impedance of the port used is 90 ohm. Hence the impedance line parameters are: width = 0.26 mm and height = 7.25 mm, Feed line impedance=98.3545ohm and width = 1.5 mm, this comes fairly same as the calculated value obtained through the design.
The multidielectric microstrip antenna designed is seen to radiate 1.034 mW power, having directivity of 6.77 dB and gain of 5.95 dB thereby achieving antenna efficiency of 87.88%. Antenna is seen to resonate at the designed frequency of 10 GHz at a return loss of -31.797 dB as depicted in Figure 1. Considering the frequency of resonance at the designed frequency along with significantly low return loss as shown in Figure 1, it confirms accuracy of feed design to achieve impedance match along with perfect radiation pattern E and E as shown in Figure 2 and Figure 3 respectively. In addition, veracity of algorithm used [12], ensures multidielectric layer microstrip antenna design to accuracy overcomes anomalies likely to occur during fabrication along with measurement errors. Importantly operation in X-band at 10 GHz for defence application demands design accuracy for operational efficacy.
58
Vol. 3, Issue 1, pp. 55-68
Figure 1. Return Loss of the Multidielectric Antenna without Superstrate layer
Figure 2. Radiation Pattern E both front and back
III.
EFFECT OF CHANGING SUPERSTRATE LAYER THICKNESS ON THE ANTENNA PARAMETERS
3.1 Superstrates Selection
59
Vol. 3, Issue 1, pp. 55-68
The addition of a cover layer over the substrate can also result in structural resonance referred to as the resonance gain method [18]. Superstrates are selected to compare the effects of its permittivity and its thickness on various antenna parameters. The two superstrate selected are from the data sheet of Rogers Corporation are: High Permittivity: RT/Duroid 6010LM having relative permittivity of 10.2 and loss tangent = 0.0023. Low Permittivity: RT/Duroid 5880LZ having relative permittivity of 1.96 and loss tangent = 0.0019. Low and high thickness of the substrate under consideration are 0.254mm and 2.54mm respectively. Analysis of the antenna structure is based on method of moments utilizing Momentum tool in Advanced Design System (ADS) of Agilent Technologies. The Momentum based optimization process varies geometry parameters automatically to help us achieve derived antenna structure.
3.2 Analysis based on Superstrate Layer Properties

The effect of the superstrate layer on antenna parameters including radiation pattern involves selection of combination of superstrate layer viz. high/low permittivity and thick/thin superstrates.
Table 1. Comparative Chart Depicting Effect of Superstrate Layer on Antenna Parameters Parameter Cover thickness Frequency Return Loss Power Radiated Directivity Gain Efficiency High Permittivity Thick Superstrate 2.54 mm 8.537 GHz -3.574 dB 0.1929 mW 8.842 dB 3.329 dB 37.65 % Low Permittivity Thick Superstrate 2.54 mm 9.392 GHz -14.405 dB 0.8413 mW 7.443 dB 5.883 dB 79.04 % High Permittivity Thin Superstrate 0.254 mm 8.607 GHz -21.331 dB 0.9895 mW 7.1255 dB 6.1425 dB 86.20 % Low Permittivity Thin Superstrate 0.254 mm 9.721 GHz -35.041 dB 1.015 mW 6.876 dB 5.977 dB 86.92 %
The resonance gain method for the practical application has been studied using moment method [19]. This resonance gain method involves a limited structural geometry, resonant frequency drift [18]. As described in section 3.1, superstrate relative permittivity chosen is either
10.2 or 1.96 corresponding to high or low relative permittivity respectively. Thickness of the superstate considered 2.54 or .254 mm corresponds to thick or thin superstrate respectively. Table 1 shows the effect on antenna parameters due to change in permittivity and thickness of superstrate layer. 3.2.1 Case 1 High Permittivity Thick Superstrate has effect on antenna parameters. Poor gain accompanied by very low antenna efficiency of the order of 37.65%. In addition for the Case 1, return loss is also very poor and at frequency of 8.537 GHz, it is -3.754 dB as shown in Figure 4.
Figure 4. Return Loss for High Permittivity Thick Superstrate Antenna
60
Vol. 3, Issue 1, pp. 55-68
The lossy nature of the antenna combined with poor return loss is substantiated by the radiation pattern in both and plane highlighting minor lobes and distorted pattern as can be seen in Figure 5 and Figure 6. It is therefore concluded that combination of thick superstrate of high relative permittivity will result in antenna behavioural pattern not conforming to the design. 3.2.2 Case 2 Antenna parameters in case of Low Permittivity Thick Superstrate show improvement in antenna gain and efficiency. Return loss at 9.392 GHz is -14.405 dB shows marginal improvement as seen in Figure 7. Radiation patterns seen in Figure 8 and Figure 9 shows significant improvement so also radiated power output as compared to Case 1.
61
Vol. 3, Issue 1, pp. 55-68
Figure 7. Return Loss for Low Permittivity Thick Superstrate Antenna
62
Vol. 3, Issue 1, pp. 55-68
3.2.2 Case 3
High Permittivity Thin Superstrate when used, shows a significant improvement in antenna parameters viz. antenna gain and efficiency. Return loss at 8.607 GHz is -21.331 dB, shows good improvement as seen in Figure 10. Radiation pattern seen in Figure 11 and Figure 12, shows similar radiation plots as seen in Case 2. Marginal increase in radiated power output is seen as compared to Case 2.
Figure 10. Return Loss for High Permittivity Thin Superstrate Antenna.
63
Vol. 3, Issue 1, pp. 55-68
3.2.4 Case 4 Low Permittivity Thin Superstrate used as shown in Table 1 shows drop in antenna gain and marginal increase in efficiency. Return loss at 9.721 GHz, is -35.041 dB, shows significant improvement and as seen in Figure 13. Figure 14 and Figure 15, shows radiation plots. Radiation pattern is seen to be with perfect null in both the plane. Increase in radiated power output is 1.01mW which is the best among all the other three cases discussed above. Hence for antenna to resonate close to the desired frequency with return loss better than -30 dB, radiated power is around 1 mW and pattern with no sidelobes and perfect null. Case 4 viz. Low Permittivity Thin Superstrate is the best choice for multi-dielectric antenna design.
Figure 13. Return Loss for Low Permittivity Thin Superstrate Antenna
Plots shown in Figures 4,7,10 and 13 indicate changes in resonant frequency, effect on return loss. There are variations in antenna directivity, gain, efficiency and finally the radiated power due to change in superstrates characteristics. To minimize losses and resonate close to the desired designed frequency, choice of thin and low permittivity superstrate having sufficient mechanical strength to withstand stress and weather vagaries is recommended for aerospace applications.
64
Vol. 3, Issue 1, pp. 55-68
3.2.5
Combined Result
Table 2 gives the birds eye view to analyse multidielectric antenna and effect of relative permittivity and thickness of Superstate (Cover) layer on antenna parameters and radiation pattern in both the planes.
Table 2. Multidielectic Layer Antenna Parameters with and without Superstrate Layer Permittivity Superstrate Types Without Superstrate Low Permittivity Thin Superstrate High Permittivity Thin Superstrate Low Permittivity Thick Superstrate High Permittivity Thick Superstrate Resonant Frequency (fz) (GHz) 10 9.721 8.607 S11 (dB)/ (Normalised) -31.8/(0.91) -35/(1.0) -21/(0.61) Power Radiated (milliwatt) 1.03 1.02 1.0 Gain (dB)/ (Normalised) Directivity (dB)/ (Normalised) 6.77/(0.77) 6.88/(0.78) 7.13/(0.81) Efficiency / (Normalised)
5.95/(0.97) 5.98/(0.97) 6.14/(1.0)
87.89/(1.0) 86.92/(0.99) 86.20/(0.98)
9.392
-14.4/(0.4)
0.84
5.88/(0.96)
7.44/(0.84)
79.04/(0.90)
8.537
-3.6/(0.1)
0.2
3.24/(0.53)
8.84/(1.0)
36.63/(0.42)
Though the results are reasonably attractive for low permittivity dielectric of the superstrate layer thickness of the order of 0.254mm, this choice may lead to fragile structure. Hence it is desirable to go for designs with low permittivity superstrate layer thickness of 2.54mm with good gain, antenna efficiency and radiated power output implying low losses.
IV.
Effect on Axial Ratio due to Superstrate Thickness & Improvement in Bandwidth
A very important parameter is the polarization of an antenna. The axial ratio helps to quantify the polarization. The axial ratio is the relationship between major and minor axes of an elliptically polarized wave and it varies between one and infinity. A linearly and a circularly polarized antenna, the axial ratio tends to infinity and 1 respectively. Aspect ratio observed in case of multidielectric antenna without and with superstrate layer is of the order of 1. It is a linearly polarized multidielectric antenna. Similarly with the superstrate layer incorporated in a multidielectric layer microstrip antenna, it is observed that there is a bandwidth enhancement of the order of 25% to 43.8% with a
65
Vol. 3, Issue 1, pp. 55-68
exception related to case 1 (viz. Thick superstrate with high permittivity dielectric constant). Figure 16 shows combined plot of all antenna parameters normalized. The multidielectric layer antenna with low permittivity thin superstrate achieves best result vis--vis all other combinations of permittivity and superstrate thickness. However a discussed in the previous sections, the combination with next best results that suits for practical applications is the multidielectric layer microstrip antenna with low permittivity and thick superstrate.
Table 3. Aspect Ratio and Bandwidth variations in Multidielectic Layer Antenna with and without Superstrate Layer Permittivity Superstrate Types Without Superstrate Low Permittivity Thin Superstrate High Permittivity Thin Superstrate Low Permittivity Thick Superstrate High Permittivity Thick Superstrate Aspect Ratio(dB) Normalized 1 0.988949735 0.980851123 0.899375385 0.416799696 Bandwidth (MHz)/ (Normalized) 280/(0.7) 400/(1.0) 350/(0.88) 400/(1.0) 0/(0)
Figure 16. Antenna Parameters plot for Multidielectric Antenna without and with Superstrate Layers.
V.
CONCLUSION
Parameters of microstrip antenna which inherently limits the gain, directivity, returns loss and radiated power is improved upon. Considering the effect of superstrate layer method for accurately determining the resonant frequency of such structures have been reported using the variation of patch dimension. To overcome the time consuming and laborious accurate numerical methods a direct use in algorithm for the design of the antenna is suggested. Data obtained from simulation with variation of the height of the transformed antenna and its effect can be used to predict the antenna parameters including resonant frequency, return loss, power radiated, directivity and gain for a multilayer microstrip antenna subjected to the limits for the thickness of the superstrate layer (0.254mm-2.54mm). Gain of a multilayered structure increases as the height of the cover layer is decreased. As regard thin cover layer dielectric, conductor losses are dominant while for thicker cover layer surface wave losses
66
Vol. 3, Issue 1, pp. 55-68
are significant. It is found choice of low permittivity dielectric both thin and thick as cover layer is suitable for applications requiring high antenna efficiency.
REFERENCES
[1] Shun-Shi Zhong, Gang Liu & Ghulam Qasim, (1994) Closed Form Expressions for Resonant Frequency of Rectangular Patch Antennas With Multi-dielectric Layers, IEEE Transactions on Antennas and Propagation, Vol. 42, No. 9, pp. 1360-1363. H. A. Wheeler, (1964) Transmission line properties of parallel wide strips by a conformal mapping approximation, IEEE Trans. Microwave Theory Tech., Vol. M1T-12, pp. 280-287. M. Kirschning & R. H. Jansen, (1982) Accurate model for effective dielectric constant of microstrip with validity up to millimeter-wave frequencies, Electronic Letter., Vol. 18, pp. 272-273. J. Svacina, (1992) Analysis of multilayer microstrip lines by a conformal mapping method, IEEE Trans. Microwave Theory Tech., Vol. 40, No. 4, pp.769- 772. N. G. Alexopoulos, & D. R. Jackson, (1984) Fundamental superstrate (cover) effects on printed circuit antennas, IEEE Trans. Antennas Propagation, Vol. AP-32, pp. 807816. G. Alexopoulos & D. R. Jackson, (1985) Gain enhancement methods for printed circuit antennas, IEEE Trans. Antennas Propagation, Vol. AP-33, pp. 976987. H. Y. Yang & N. G. Alexopoulos, (1987) Gain enhancement methods for printed circuit antennas through multiple substrates, IEEE Trans. Antennas Propagation, Vol. AP-35, pp. 860863. X. Shen, G. Vandenbosch & A. Van de Capelle, (1995) Study of gain enhancement method for microstrip antennas using moment method, IEEE Trans. Antennas Propagation, Vol. 43, pp. 227231. X. Shen, P. Delmotte & G.Vandenbosch, (2001) Effect of superstrate on radiated field of probe fed microstrip patch antenna, Proc. Inst. Elect.Eng.-Microwave Antennas Propagation, Vol. 148, pp. 141 146. Zhong, S.Z., Liu, G. & Qasim, G. (1994) "Closed Form Expressions for Resonant Frequency of Rectangular Patch Antennas with Multidielectric Layers," IEEE Transactions on Antennas and Propagation, Vol. 42, No. 9, pp. 1360-1363. H. A. Wheeler, (1964) Transmission line properties of parallel wide strips by a conformal mapping approximation, IEEE Trans. Microwave Theory Tech., Vol. M1T-12, pp. 280-287. Samir Dev Gupta, Anvesh Garg & Anurag P. Saran (2008) "Improvement in Accuracy for Design of Multidielectric Layers Microstrip Patch Antenna, International Journal of Microwave and Optical Technology (IJMOT), Vol.3, No. 5, pp 498-504. U.K. Revankar & K.S.Beenamole, (2003) Low Sidelobe Light Weight Microstrip Antenna Array For Battlefield Surveillance Radars, IEEE Radar Conference, pp. 97-101. Richards W.F., Y.T. Lo & D.D. Harrison, (1981) An Improved Theory for Microstrip Antennas and Applications, IEEE Trans on Antennas and Propagation, Vol. AP-29, pp. 38-46. Lo Y.T., D. Solomon & W.F. Richards, (1979) Theory and Experiment on Microstrip Antennas, IEEE Trans. on Antennas and Propagation, Vol. AP-27, pp. 137-145. C. A. Balanis, (1997) Antenna Theory Analysis and Design, John Willy & Sons, 2nd Edition, Chapter 14, pp.730-734. Anders G. Derneryd, (1978) A Theoretical Investigation of the Rectangular Microstrip Antenna Element, IEEE Transactions on Antennas and Propagation, Vol. AP-26, No. 4, pp. 532-535. Chisang You & Manos M. Tentzeris, (2007) Multilayer Effects on Microstrip Antennas for Their Integration With Mechanical Structures, IEEE Transactions on Antennas and Propagation, Vol. 55, No. 4, pp. 1051-1058. X. Shen, G. Vandenbosch & A. Van de Capelle, (1995) Study of gain enhancement method for microstrip antennas using moment method, IEEE Transactions on Antennas and Propagation., Vol. 43, pp. 227 231.
[2] [3] [4] [5] [6] [7] [8] [9]
[10]
[11] [12]
[13] [14] [15] [16] [17] [18]
[19]
Authors Samir Dev Gupta received his B.E. (Electronics) from U.V.C.E. Bangalore, M.Tech (Electrical Engg.) from I.I.T. Madras and M.Sc.(Defence Studies) from Madras University. His current area of research is Conformal Microstrip Antenna Design. Experience of 19 years in teaching profession at Post Graduate and Graduate level including three years teaching at Institute of Armament Technology, Pune (Defence Research and Development Organisation) then affiliated to Pune University, now Defence Institute of Advanced Technology (Deemed University). His areas of specialization include Antennas, Microwave Communication and Radar Systems. He was a recognized Post Graduate Teacher in Microwave Communication at Pune University.
67
Vol. 3, Issue 1, pp. 55-68
Work experience of about 15 years in maintenance, operations and modifications of Radar, Microwave Communication, Aircraft Simulators and Avionics related systems. Amit Singh received B.Tech degree in Electronic and Communication Engineering from the Jaypee Institute Of Information Technology, Noida, India, in 2010. He is a Research and Development Engineer at the EEs of, Agilent Technologies, Gurgaon, India. His main research interests include electromagnetic and its applications, in particular conformal microstrip antennas.
68
Vol. 3, Issue 1, pp. 55-68
LOW COST INTELLIGENT ARM ACTUATOR USING PNEUMATIC ARTIFICIAL MUSCLE WITH BIOFEEDBACK CONTROL
Salman Afghani, Yasir Raza, Bilal Haider
Department of Research & Development Electronics Engg., APCOMS, Rawalpindi Pakistan
ABSTRACT
Hard Exoskeletons and arm actuators have already been designed and available in industry but they are expensive, heavy and cumbersome so to reduce these drawbacks a low cost lightweight intelligent arm actuator has been designed, constructed and evaluated. It consists of a pneumatic artificial muscle, constructed using a football bladder and covered by parachute material to limit its expansion. Pneumatic artificial muscle is under the control of micro controller that keep monitoring and controlling the air pressure of artificial muscle using biofeedback amplifier provided to micro controller from cuff bladder - which detects the muscle tension.
KEYWORDS
Intelligent Arm actuator, less price pneumatic Artificial Muscle, biofeedback control, low cost Arm actuator.
I.
INTRODUCTION
A pneumatic artificial muscle is non linear and contractile type actuators, they are not commonly used in robotic or man machine systems. At present Electromagnetic motors such as AC motors, DC motors, stepper motors, and linear motors are widely used which are not as lightweight as a pneumatic artificial muscle, also a pneumatic artificial muscle has a direct transmission capability whereas electromagnetic motors has not. Electric motors has power to weight ratio of about 100 (W/kg) and their torque to weight ratios range is within 1-10 (Nm/Kg)[1]. Whereas power to weight ratios of PAM actuators is reported in excess of 1000 (W/kg)[2]. Pneumatic artificial muscles are not use widely due to difficulty in controlling them accurately although they have high power to weight ratio but with less controlling capabilities as compared to electric motors In next section after Pneumatic Artificial Muscle we explained our system design considering the feedback control system and new design for detection of muscle tension.
II.
PNEUMATIC ARTIFICIAL MUSCLE
In order to develop PAM, initially we used motorcycle and cycle tyre tubes but displacement produced from tubes wasnt enough to load object. So for more displacement we used football bladder which gave us our desired displacement. Lightweight parachute material was used to make cover for football bladder, to limit the expansion of bladder to some specific extent. Parachute material was sewed in folding to more strengthen the muscle. Special input nozzle for bladder was designed to insert and release air more easily. Hooks were installed on both sides of muscle to easily hook with body. Following formula can be used to calculate the generated force of PAM if energy losses and energy needed to contact PAM is neglected.
F = p
dV dl
(1)
69
Vol. 3, Issue 1, pp. 69-75
In above stated formula p represents the gauge pressure, V represents the volume of PAM and l represents the length of PAM after contraction as shown in fig 1 and dV/dl is the muscles effective area [3].
Figure 1. PAMs concept of operation
Figure 2. Testing of PAM
III.
FEEDBACK CONTROL SYSTEM
Feedback of PAM is taken and provided back to microcontoller after signal amplification, where micro controller reads and processes the feedback signal and control PAM accordingly. Two types of system were design and experimented Three state system and two state system
3.1
Three State System
Figure 3 System Block diagram
70
Vol. 3, Issue 1, pp. 69-75
Where Vref1 is the voltage nominal voltage level of Bio feedback amplifier and Vref2 is the voltage where PAMs expansion limit has reached. E1 and E2 give us the difference of voltages between Vref1 to Bio feedback voltage and Vref2 to PAM feedback voltage respectively. Fuzzy controller reads the value of E1 and produces a 3 state system considering trapezoidal logic.
State # State 1 State 2 State 3 Table 1 Three state values Biofeedback Voltage E1 Bio feedback Fuzzy Range (v) Representation 0.67 1.50 Positive Positive Feedback 1.35 3.35 0 Nominal Value 2.00 4.20 Negative Negative Feedback PAM status Stretched Hold position Inflated Full
Figure 4 Fuzzy logic diagram
3.1.1
Three State Diagram
Figure 5 Three State Diagram
71
Vol. 3, Issue 1, pp. 69-75
3.1.2. Flow Diagram for Three State system
Figure 6 Flow Diagram for Three State system
3.2
Muscle tension sensing unit/ Bio feedback amplifier
Automatic control of PAM needs a system which automatically detects the muscle tension and provide to micro controller. A new idea of muscle tension detection is used i-e a cuff bladder is attached at muscle of a person and tension is detected by detection the change in air pressure of cuff bladder occurred due to expansion or contraction of persons muscle, using air pressure sensor[4]. To achieve this we did back engineering of digital blood pressure monitor which measures blood pressure by detecting the change in pressure using pressure sensor so we used its bladder, pump and pressure sensor and controlled all of them from micro controller.
Cuff Bladder
Air Pressure Sensor
Instrumentation Amplifier with signal conditioning
Vout
Figure 7 Block diagram of sensing unit
We can sense muscle tension at two different locations of our Arm i-e Bicep and Wrist. Experiments are done on both but shown for wrist configuration.
72
Vol. 3, Issue 1, pp. 69-75
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 3.2.1 Instrumentation Amplifier
We performed many experiments on instrumentation amplifier to get our optimal value as the output of the sensor was very low value of voltage. We used different values of resistors in our circuit to get maximum output. During experiments we came to know that when we adjust the values of resistors at 10k we gain much better output but yet its value was low but as compare to our other experimental values of resistors it was much batter so fixed the value of 10k except feedback resistors R1 and Rgain. For R1 we again perform different experiments to increase its gain, and by placing the value of 670k we got the maximum output from our circuit and along with Rgain where we place the resistor of value 570 ohm we achieved the best and the maximum output value of voltage as compare to any other values of resistors. Our final range of voltage was from 0.64v to 4.20v.
Figure 8 Instrumentation Amplifier Schematic
3.3
Two state system
In case of operational amplifier, there is a trade off between linearization and Amplification. Output voltage of a pressure sensor was very less so we also tried another type of system by increasing amplification and decreased linearization which finally gave us almost a digital output value of Instrumentation Amplifier i-e 0.67V in relaxed position and 4.5V at tensed position of muscle. This reduced our system to only 2 states as shown in fig 7.
3.3.1
Two State Diagram
Figure 9 Two State Diagram
73
Vol. 3, Issue 1, pp. 69-75
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 3.3.2
Flow Diagram for Two State System
Figure 10 Flow Diagram
IV.
DISCUSSION AND RESULTS
We considered this two state system as well as three state system for this arm actuator and both performed pretty well, Two state system just reduced the complexity involved in fuzzy logic control and our system became simple. Table 2 contains the values and results obtained from two state system. Whereas three state system includes complexity of system but give more smooth and intelligent behaviour of system. Table 1 contains values and results of three state system. As mentioned above that we can sense muscle tension from two different locations of a body but when we used bicep location for sensing, we observed disturbance in muscle tension sensor due to expansion of PAM. Whereas it works fine for wrist configuration.
Table 2 Two state values State # State 1 State 2 Muscle tension detected? No Yes Inst Amp Voltage (V) 0.67 4.5 Bio feedback binary representation 0 1 PAM status Stretched Inflated
V.
CONCLUSION AND EXPERIMENT
Low Cost Arm Actuator with Pneumatic Artificial Muscle can be designed using a simple football bladder and parachute material and also it can easily be controlled and made intelligent using biofeedback control. Biofeedback can be taken from two locations of Arm i-e Bicep or Wrist. Biofeedback amplifier can be modified in two types of systems Three states system and two states system. Three states system is more intelligent due to its fuzzy logic control but its more complex than two states system. We can also enhance and design our project to whole body of a person to
74
Vol. 3, Issue 1, pp. 69-75
make a complete soft exoskeleton lieu of hard exoskeletons that has already been developed throughout the world.
PAM
Sensing Unit Lightweight harness
REFERENCES
[1]. F. Daerden, Conception and Realization of Pleated Pneumatic Artificial Muscles and Their Use as Compliant Actuation Elements, PhD Dissertation, Vrije Universiteit Brussel, July 1999 [2]. Frank Daerden, Dirk Lefeber THE CONCEPT AND DESIGN OF PLEATED PNEUMATIC ARTIFICIAL MUSCLES Vrije Universiteit Brussel [3]. H. M. Paynter, "Hyperboloid of revolution fluid-driven tension actuators and methods of making", US Patent No. 4 721 030, 1988. [4]. DIGITAL BLOOD PRESSURE MONITOR Journal of Applied Research and Technology, december, lamb / vol. 2, number 003 University National Autonomous Mexico, Federal District, Mxicopp. 224-229
Authors Salman Afghani Was born in Pakistan in 1958. Professor, PhD, Advanced man machine
systems, MPhil, Industrial automation, Ms Mech Engg. Research Interest: Man Machine systems, robotics intelligent all terrain vehicles & submersibles, Zoological and Botanical computers, Non-intrusive botanical genetics using computer simulated time acceleration technique.
Yasir Raza was born in Rawalpindi, Pakistan in 1989. He did BSEE in Electrical Engineering
from University of Engineering and Technology, Taxila Pakistan
Bilal Haider Was born in Rawalpindi, Pakistan in 1987. He did BSEE in Electrical
Engineering from University of Engineering and Technology, Taxila Pakistan
75
Vol. 3, Issue 1, pp. 69-75
TWO STAGE DISCRETE TIME EXTENDED KALMAN FILTER SCHEME FOR MICRO AIR VEHICLE
Sadia Riaz and Ali Usman
Department of Mechanical Engineering, NUST College of EME, Rawalpindi, Pakistan
ABSTRACT
Navigation of Micro Air Vehicle (MAV) is one of the most challenging areas of twenty first centurys research. Micro Air Vehicle (MAV) is the miniaturized configuration of aircraft with a size of six inches in length and below the weight of hundred grams, which includes twenty grams of payload as well. Due to its small size, MAV is highly affected by the wind gust and therefore the navigation of Micro Air Vehicle (MAV) is very important because precise navigation is a very basic step for the control of the Micro Air Vehicle (MAV). This paper presents two stage cascaded discrete time Extended Kalman Filter while using INS/GPS based navigation. First stage of this scheme estimates the Euler angles of Micro Air Vehicle (MAV) whereas the second stage of this scheme estimates the position of Micro Air Vehicle (MAV) in terms of height, longitude and latitude. As the system is considered as non-linear, so Extended Kalman Filter is used. On-board sensors in first stage included MEMS Gyro, MEMS Accelerometer, MEMS Magnetometer whereas second stage includes GPS.
KEYWORDS:
Micro Electro Mechanical System (MEMS), Micro Air Vehicle (MAV), Measurement Covariance Matrix, Process Covariance Matrix
NOMENCLATURE
PN PE WN WE V air p q r
m m m Q R
ox oy oz
Inertial North Position of MAV Inertial East Position of MAV Wind from North Wind from East Total Airspeed Angular Rate about x-axis Angular Rate about y-axis Angular Rate about z-axis Roll Angle Pitch Angle Yaw Angle As superscript shows the rate of change Northern Magnetic Field Component Eastern Magnetic Field Component Vertical Magnetic Field Component Process Covariance Noise Matrix Measurement Covariance Noise Matrix
I.
INTRODUCTION
Micro Air Vehicle (MAV) is that type of aircraft, which is smaller enough than the other aircrafts. According to DARPA (Defence Advanced Research Project Agency), Micro Air Vehicle (MAV) are small aircraft usually defined to be less than fifteen centimeter ( six inches) in length and below the weight of hundred grams of weight which include twenty grams of payload as well [1]. In developing these airborne spies the primary concerns are the size, weight, sensors, communication and the computational power. These are remote controlled, semi-autonomous or the autonomous aircrafts, which can be used in military, surveillance and reconnaissance, and civilian, search and rescue [2].
76
Vol. 3, Issue 1, pp. 76-83
Micro Air Vehicle (MAV) mainly have three types which include Fixed Micro Air Vehicle (MAV), Flapping Micro Air Vehicle (FMAV) and Rotary Micro Air Vehicle (MAV). Fixed Micro Air Vehicles (MAV) are the most efficient because they have higher loitering time [3], but they cannot hover. Flapping Micro Air Vehicle (FMAV) can hover but they have very short flight time [4]. Rotary Micro Air Vehicle (MAV) can hover in any direction but at the cost of flight duration [5]. Figure 1 shows different types of Micro Air Vehicles which includes Fixed Wing Micro Air Vehicle, Flapping Wing Micro Air vehicle, Wing Morphing Micro Air Vehicle and Rotary Wing Micro Air Vehicle. o
Figure 1: Types of Micro Air Vehicles
Applying Micro Electromechanical Systems (MEMS) inertial sensors for the Navigation of an autonomous Micro Aerial Vehicle (MAV) is an extremely challenging area [6]. The Global rial Positioning System (GPS) can provide long term stability with high accuracy and world1wide long-term coverage. Since the performance of the low cost micro GPS receiver can be easily degraded in high maneuvering environments, fusing the navigation data with other sensors such as a magnetometer or barometer is necessary [7]. Two stage discrete time Extended Kalman Filter is used in this paper to Two-stage estimate the Euler angles in first stage and the position in the second stage. To combine MEMS gyro, MEMS accelerometer and MEMS magnetometer in one stage is beneficial because they support each other in estimating the precise states of Micro Air Vehicle (MAV).
II.
DISCRETE TIME EXTENDED KALMAN FILTER FORMULATION

x k = f k 1 (x k 1 , u y k = h h (x k , v k wk vk
k 1
The system and measurement equations are given as follows:
, w k 1
(0 , Q k ) (0 , R k )
Initialize the filter as follows:
) ) P0+ = E x0 x0+ x0 x0+
[(
) x0+ = E ( x0 )
)( )
)]
T
77
Vol. 3, Issue 1, pp. 76-83
For k = 1, 2, 3 . . . . Compute the following partial derivative matrices [8]:
Fk 1 =
f k 1 x
)+ x k 1
L k 1 =
f k 1 w
)+ x k 1
Perform the time update of the state estimate [9] and estimation error covariance [10] as follows:
Pk = F k 1 Pk+1 F kT1 + L k 1Q k 1 LT 1 k ) )+ x k = f k 1 x k 1 , u k 1 , 0
Compute the following partial derivative matrices
HK =
hk x
) xk
MK =
hk v
) xk
Perform the measurement update [11] of the state estimate and estimation error covariance as follows
T T T K k = Pk H k H k Pk H k + M k Rk M k ) ) ) xk+ = xk + K k yk hk xk ,0
))
Pk+ = (I K K H k )Pk
The following formatting rules must be followed strictly. This (.doc/.docx) document may be used as a template for papers prepared using Microsoft Word. Papers not conforming to these requirements may not be published in the journal.
III.
TWO STAGE EXTENDED KALMAN FILTER ESTIMATION SCHEME
Because the heading update equation uses the same inputs as the Pitch and Roll equations, it is not unnatural to lump them together in the same estimation block. The exclusiveness of the heading state lies in the output equations. There is not a sensor output equation which will relate heading to accelerometer readings, which is why it was convenient to split heading estimation into its own stage as mentioned earlier. One of the merits of heading and attitude at the same time is that magnetometer information may be beneficial in the estimation of pitch and roll, since no maneuver will upset the earths magnetic field, as they may the accelerometer readings. And, depending on the attitude and heading of the MAV, projecting pitch and roll onto the magnetic field vector may refine the pitch and roll estimates. Figure 2 shows a schematic for Two Stage Discrete Time Extended Kalman Filter Scheme.
Figure 2: Two stage Cascaded Discrete time Extended Kalman Filter
78
Vol. 3, Issue 1, pp. 76-83
First stage state variables and inputs can be shown as following:
p acc x , u = q , y = acc x= y r acc z Vair Stage two state variables and inputs can be shown as following:
GPS N PN GPS P E x = E , u = , y = GPS Velocity W N Vair GPS Heading WE
In two stage discrete time Extended Kalman Filter state estimation scheme, the states can be related to the inputs in the following two steps as shown below: Stage 1 consists of the following:
= f ( x , u ) =
Linearization through jacobian:
p + q sin tan + r cos tan q cos + r sin sin cos q +r cos cos
q sin + r cos cos 2 0 0 0 0
q cos tan sin tan ) f (x , u ) A= = q sin + r cos x (q cos r sin )sec

Process Covariance Matrix is given as follows:
(q sin + r cos )sec tan
Q = E wwT
Where,
w w = w w
And
wT = [w
w ]
w w w2 w w w w w w 2 w
Vol. 3, Issue 1, pp. 76-83
so,
2 w Q = E w w w w
79
Noise is uncorrelated:
w2 Q = E 0 0
0 w2 0
0 0 2 w
The output equations for attitude estimator are as follows:
Vair q sin + sin g Vair ( r cos p sin ) ) cos sin h( x , u ) = g V airq cos cos cos g
Linearization through jacobian :
0 ) h( x , u ) Hk = = cos cos x cos sin
Vair q cos + cos g Vair (r sin + p cos ) + sin sin g Vair q sin + sin cos g
0 0 0
Measurement Covariance Matrix is given as follows:
R = E (vvT )
Where,
vaccx v = vacc y vacc z

v = vacc x
And
vacc y
vacc z
]
v acc x v acc z v acc y v acc z 2 v acc z
So,
v 2 acc x R = E v acc y v acc x v v acc z acc x

v acc x v acc y
2 v acc y
v acc z v acc y
80
Vol. 3, Issue 1, pp. 76-83
v 2 acc x R = E 0 0
0
2 v acc y
0 0 2 v acc z
Stage 2 consists of the following:
PN Vair cos WN P Vair sin WE E = 0 WN 0 W E

Process Covariance Matrix is given as follows:
Q = E wwT
Where,
wP w= N wPE
And
wT = wPN
2 wPN Q = E 0
wPE
0 wPE
Output equations by inertial estimator:
GPS N PN ) h( x , u ) = = GPS E PE
Linearization through Jacobian:
) h (x , u ) 1 = x 0
81
0 1
Vol. 3, Issue 1, pp. 76-83
Measurement Covariance Matrix is given as follows:
Where,
R = E (vvT )
and
GPSN GPS E v= GPSVelocity GPSHeading
vT = GPS N
So
GPS E
GPSVelocity
GPS Heading
]
GPS N GPS Velocity GPS E GPS Velocity 2 GPS Velocity GPS Heading GPS Velocity GPS N GPS Heading GPS E GPS Heading GPS Velocity GPS Heading 2 GPS Heading
2 GPS N GPS E GPS N R = E GPS Velocity GPS N GPS Heading GPS N
GPS N GPS E 2 GPS E GPS VElocity GPS GPS Heading GPS
E E
The noise is uncorrelated
GPS 0 R = E 0 0
2 N
0 GPS 0 0
2 E
0 0 GPS
2 Velocity
0 0 0 GPS
2 Heading
IV.
CONCLUSIONS
The MAVs are playing a significant role military surveillance and reconnaissance, and civilian search and rescue. To control MAV, it is important to navigate it properly. Kalman Filter is one of the techniques, which can be used for the navigation of MAV. Three different schemes can be used for the discrete time Extended Kalman filter as shown in the paper above. Work is in process on the simulations based on this mathematical model.
ACKNOWLEDGEMENTS
The authors are indebted to the College of Electrical and Mechanical Engineering (CEME), National University of Sciences and Technology (NUST), Higher Education Commission (HEC) and Pakistan Science Foundation (PSF) for having made this research possible.
REFERENCES
[1]. James M. McMichael (Program Manager Defense Advanced Research Projects Agency) and Col. Michael S. Francis, USAF (Ret.) (Defense Airborne Reconnaissance Office), Micro air vehicles Toward a new dimension in flight dated 8/7/97
82
Vol. 3, Issue 1, pp. 76-83
[2]. Application of Micro-Technology to Integrated Navigation of Autonomous Micro Air Vehicles by S. Technology Navigation Winkler_, M. Buschmann_, T. Kordes_, H. W. Schulz_, and P. V orsmanny Institute of Aerospace H.Systems, TU Braunschweig, D 38108 Braunschweig, Germany. CANEUS 2004 D-38108 2004--Conference on Micro-Nano-Technologies 2004, Monte Technologies Monterey, California [3]. Micro Air Vehicle: Configuration, Analysis, Fabrication, and Test by Huaiyu Wu, Member, IEEE, Dong Sun, Member, IEEE, and Zhaoying Zhou, Senior Member, IEEE. IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 9, NO. 1, MARCH 2004. [4]. http://www.ornithopter.net/history_e.html [5]. A Fixed-Wing Aircraft for Hovering in Caves, Tunnels, and Buildings William E. Green and Paul Y. Wing Oh, Drexel Autonomous Systems Lab (DASL) Drexel University, Philadelphia, PA[william.e.green,paul.yu.oh]@drexel.edu] [william.e.green,paul.yu.oh]@drexel.edu] [6]. Discrete time adaptive control for a MEMS gyroscope by Sungsu Park1 and Roberto Horowitz2 1 Department of Aerospace Engineering, Sejong University, Seoul, Korea. 2 Department of Mechanical Engineering, University of California, Berkeley, CA 94720, U.S.A. INTERNATIONAL JOURNAL f OF ADAPTIVE CONTROL AND SIGNAL PROCESSING Int. J. Adapt. Control Signal Process. 2005; 19:485503 Published online 1 April 2005 in Wiley InterScienc (www.interscience.wiley.com) 503 DOI:10.1002/acs.868 [7]. Estimation with Applications to Tracking and Navigation by Yaakov Bar Bar-Shalom, X-Rong Li, Thiagolingam Kirubarajan. A Wiley Interscience Publication. John Wiley & sons. INC.. Wiley[8]. State estimation for micro air vehicles by Randal W. Beard, Department of Electrical and Computer Engineering, Brigham Young University, Provo, Utah. Studies in Computational Intelligence (SCI) 70, 173199 (2007). Springer-Verlag Berlin Heidelberg 2007 Verlag [9]. Applied optimal estimation Gelb, A. Cambridge, MA: MIT Press, 1974. [10]. Kalman filtering theory and practice by Grewal, M. and A. Andrews. Englewood Cliffs, NJ: Kalman Prentice-Hall, 1993.
Authors
Sadia Riaz is a Mechanical Engineer from National University of Sciences and Technology (NUST). She did her MS in Mechanical Engineering from National University of Sciences Engineering and Technology (NUST). Now she is post graduate student in National University of post-graduate Sciences and Technology (NUST). Her area of research is state estimation with the application area of Micro Air Vehicles (MAVs).
Ali Usman is Mechatronics Engineer from Air University. He did his MS in Mechanical Engineering from National University of Sciences and Technology. Now he is post graduate post-graduate student in National University of Sciences and Technology. His area of research is classification of Wing design and navigation of Micro Air Vehicles (MAVs). .
83
Vol. 3, Issue 1, pp. 76-83
IDENTIFICATION OF SUSPICIOUS REGIONS TO DETECT ORAL CANCERS AT AN EARLIER STAGE A LITERATURE SURVEY
K. Anuradha1 and K.Sankaranarayanan2
2
Research Scholar, Karpagam University, Coimbatore Dean, Electrical Sciences, Easa College of Engg. and Technology, Coimbatore
ABSTRACT
Oral cancer is a significant health problem throughout the world. Most oral cancers are identified at a later stage where, treatment becomes less successful. It is very important to detect such types of cancer at an earlier stage. Early detection helps surgeons to provide necessary therapeutic measures which also benefit the patients. Dental Radiographs are the X rays that help to identify problems with teeth, mouth and jaw. In this paper, a detailed survey has been done on various methods analyzed by the researchers for the detection of oral cancers at an earlier stage. A comparison is made between various methods for the identification and classification of cancers. An overview of algorithms in each step of the cancer detection algorithms is given. K EY W ORDS: Oral cancer, Dental Radiographs, X rays.
I.
INTRODUCTION
Oral cancer ranks as the fourth most frequent cancer among men and eighth for women worldwide and may affect the tongue, cheeks, peridontium or any other part of the oral pharynx. Oral cancer is a significant health problem throughout the world. Tumors that arise from odontogenic (tooth forming) tissues are referred to as odontogenic tumors. Tumors are either benign or malignant. Malignant tumors are cancerous. Oral cancer can affect any area of the oral cavity including the lips, gum tissues, tongue, cheek lining and the hard and soft palate. This paper focuses on identifying oral cancers at an earlier stage. Oral cancer detection at an earlier stage saves lives [1]. Beyond oral cancer, many problems can occur within the bones of the mouth. Systemic problems those that affect the entire body many times appear in the mouth first. In general, the mouth is a good indicator of whats going on in the body, which is why the physicians for generations have asked patients to open their mouth. X Rays are an essential part of dental care. Although X rays are effective diagnostic tools, some dental practices particularly those that handle a large number of dental implant cases, are using more advanced imaging techniques to ensure an even higher degree of accuracy. Oral cancers are often located on the tongue, MR images may become blurry because of moving artifacts induced by the moving tongue and jaw. So an efficient image processing algorithm is needed to identify the suspicious region in the cancer area with high accuracy. Dental radiographs are used for screening oral pathologies continuously and it is often a difficult task to detect early stage cancer tissues in a dental radiograph. Unlike other types of cancers, oral cancers are visibly seen with the naked eye, some cancers are located internally in the mouth, making their detection difficult. And also some non cancerous tissues are not harmful. The organization of the chapter is as follows: In section 2, a literature study has been performed for cancer. Some of the recent methods for cancer detection are presented in Sub Section 2.1. Sub Section 2.2 is devoted to feature extraction and cancer classification methods. Discussions about the techniques are made in Section 3. Finally Section 4 summarizes and concludes the section.
84
Vol. 3, Issue 1, pp. 84-91
II. LITERATURE SURVEY If oral cancer is detected at an earlier stage it is curable. The exact cause of cancer is not known, however, there are certain risk factors, which may trigger this disease in individuals. Among the common factors are chemicals, which may be in the form of tobacco or chemicals present in food, air, water etc [2]. These chemicals are known as Carcinogens. Detection and Diagnosis of cancer has become one of the most significant areas of research in the Medical Imaging and Image Processing Communities. Symptoms: The symptoms for an oral cancer at an earlier stage [3] are: 1) Patches inside the mouth or on lips that are white, red or mixture of white and red 2) Any sore or area in the mouth which does heel for discolored more than 14 days 3) Bleeding in the mouth 4) Difficulty or pain when swallowing 5) A lump in the neck. These symptoms identify the suspect for a cancer. The system identifies that a cancer has occurred with image processing Techniques. A surgeon who suspected the presence of cancer in a patient has few options: X rays studies to find the cancers exact location, providing proper medications can recovery from cancer, excise a portion of the unhealthy tissue for biopsy, remove the cancer, explore the surrounding tissues to determine if the cancer has spread. Over the last few decades, imaging technology refinements have substantially widened the range of medical options: The Tests now provide much clearer and more detailed pictures of organs and tissues. New imaging technology allows us to do more than simply view anatomical structures such as bones, organs and tumors. Functional imaging - the visualization of physiological, cellular, or molecular processes in living tissues enables us to observe activity such as blood flow, oxygen consumptions or glucose metabolism in real time. Imaging technology already has had lifesaving effects on our ability to detect cancer early and more accurately to diagnose the disease. Image Processing [4] algorithms have been continuously applied to get better results. The literature survey carried out has revealed that a fair amount of research has been put in the areas of cancer imaging.
2.1 Cancer Detection Techniques
[Banumathi.A et al 2009] [5] have proposed cyst detection and severity measurement of cysts using image processing techniques and neural network methods. The suspicious cyst regions are diagnosed using Radial Basis Function Network. The severity of the cysts is calculated using circularity values and the results show the part of the cysts extracted. [Woonggyu Jung et al] [6] has proposed a technique in oral cancer detection using Optical Coherence Tomography. For the imaging depth of 2-3 mm, OCT is suitable for oral mucosa. They also detected oral cancer in 3-D volume images of normal and precancerous lesions. [Simon Kent] [7] has proposed a diagnosis method for oral cancer using Genetic Programming. The Technique proposed solved many complex problems. The comparison between a Genetic Programming system and Neural Network model was provided. The Genetic Programming system played a major advantage in diagnosing the tumor. [Ranjan Rashmi Paul et al] [8] has proposed a detection methodology to detect oral cancer using wavelet - neural networks. The wavelet coefficients of TEM images of collagen fibers from normal oral sub mucosa and OSF tissues have been used in order to choose the feature vector which, in turn, used to train the Artificial Neural Network. The trained network could satisfy the normal and precancer stages. [Ireaneus Anna Rejani.Y, Dr. S. Thamarai Selvi 2009] [9] have proposed a technique to detect breast cancer at an earlier stage using SVM Classifier. The proposed system focused on two problems. One is to detect tumors as suspicious regions with a very weak contrast to their background and another is to extract features which categorize tumors. They used Gabor filtering technique for image enhancement and Top hat transform operation for image extraction. [Saheb Basha.S, Dr. K. Satya Prasad 2009] [10] have proposed an automatic detection of Breast cancer mass in mammograms using Morphological operations and Fuzzy C-means clustering. They
85
Vol. 3, Issue 1, pp. 84-91
developed an algorithm to distinguish masses and micro calcifications from background tissue using morphological operators and fuzzy C- means clustering (FCM) algorithm has been implemented for intensity based segmentation. The proposed technique showed better results. [Man Kin Derek Ho 2007] [11] have proposed an algorithm to segment images for Medical confocal images. The results provide accurate count (95% with 6.2% standard deviation) of cells or droplets. [Ghassan Hamarneh, Artur Chodorowski, Tomas Gustavsson, 2000] [12] have proposed the application of active contour models for the segmentation of oral lesions in medical color images acquired from the visual per of the light spectrum. The proposed work also classifies cancerous and non cancerous lesions. The automatic segmentation algorithm simplifies the analysis of oral lesions and can be used in clinical practice to detect potentially cancerous lesions. [S. Murugavalli, V. Rajamani, 2007] [13] proposed an improved implementation of brain tumor detection using segmentation based on neuro fuzzy technique. Various tissues like white matter, gray matter, cuff and tumor were detected. The Fuzzy C means algorithm is used to classify the image, layer by layer. The neuron fuzzy technique shows that MRI brain tumor segmentation using HSOM FCM also perform more accurate one. [Mohammed.M.M et al 2003] [14] has proposed Prostate cancer diagnosis based on Gabor filter texture segmentation of ultrasound image. Multichannel filtering is an excellent method for prostate texture investigation. Using human visual system (HVS), medical doctors use three features for texture analysis, mainly repetition, directionality and complexity. A set of Gabor filters that is well distributed to cover the entire frequency plane is designed to mimic the HVS and therefore it is an excellent tool that can be used for prostate texture segmentation. [Seshadri.H.S, Kandasamy.A] [15] has proposed detection of breast cancer at an early stage by Digital Mammogram Image analysis. The gradient of the preprocessed image is calculated and finally the segmentation algorithm is applied to the image. They have tested the proposed method on digital mammograms taken from mini MIAS database and found the lesion segmentation algorithm closely matches the radiologists outlines of these lesions. Varsha.H. Patil et al, [16] proposed an automated system for detecting breast tumor at an earlier stage. The system uses super resolution technique to display the necessary information for boosting the physicians diagnosis. CAD tools are used to design the system. [Poulami Das, et al, 2009] [17] have proposed a method to identify abnormal growth cells in breast tissue and suggested further pathological tests. They compared normal breast tissue with malignant invasive breast tissue by a series of image processing steps. Features of cancerous breast tissue are extracted and analysed with normal breast tissue. [Shekar Singh et al 2011][18] have proposed the detection of breast cancer and classification of histopathological images. They used Feed Forward Back Propagation Neural Network for classification of benign and malignant breast cancers. The breast cancers were classified into type 1, type 2, type 3. They extracted eight features after cancer detection. Feed forward Neural Network gives fast and accurate classification of breast cell nuclei. Computer aided diagnosis systems for detecting malignant texture in biological study have been investigated using several techniques. Vijay Kumar.G et al [19] proposed an approach in computer aided diagnosis for early prediction of brain cancer using Texture features and neuro classification logic. Nine distinct invariant features with calculation of minimum distance for the prediction of tumor in a given MRI image. A neuro fuzzy approach is used for the recognition of the extraction region. Yan Zhu and Hong [20] suggested the Hopfield neural network for the detection of brain tumor boundaries which was based on an active contour model. This is more suitable for real time applications. Automated detection of tumors in different medical images is motivated by the inevitability of high accuracy when we deal with human life. L. Jeba sheela et al [21] proposed a system using imaging techniques to categorize the images as normal or abnormal and then classify the tissues of the abnormal brain MRI to identify brain related diseases. 2.2 Feature extraction and cancer classification techniques [DSVGK Kaladhar, B. Chandana, P. Bharath Kumar, 2011] [22] have predicted oral cancer survivability using Classification algorithms. The classification algorithms used are CART, Random
86
Vol. 3, Issue 1, pp. 84-91
Forest, LMT, and Nave Bayesian. The algorithms classify the cancer survival using 10 fold cross validation and training data set. Out of the other techniques, the Random Forest classification technique correctly classified the cancer survival data set. The absolute relative error is less when compared to other methods. [Xiaowei Chen et al 2006] [23] have proposed Automated segmentation, classification and tracking of cancer cell nuclei in time lapse microscopy. Existing computational imaging methods are rather limited in analyzing and tracking such time lapse datasets and manual analysis is unreasonably time consuming and subject to observer variances. An automated system that integrates a series of advanced analysis methods to fill this gap. The cellular image analysis methods can be used to segment, classify and track individual cells in a living cell population over a few days. Experimental results show that the proposed method is efficient and effective in cell tracking and phase identification. [Yung nien Sun et al 2010] [24] have proposed an automatic color based feature extraction system for parameter estimation of oral cancer from optical microscopic images. Parameter comparisons between four cancer stages are conducted, and only the mean parameters between early and late cancer stages are statistically different. The proposed system provides a useful and convenient tool for automatic segmentation and evaluation for stained biopsy samples of oral cancer. [Yung nien Sun et al 2007] [25] have proposed a new color based approach for automated segmentation and classification of tumor tissues from microscopic images. The algorithm is evaluated by comparing the performance of the proposed fully automated method against semi automated procedures. The experimental results shows consist agreement between the two methods. The proposed algorithm provides an effective tool for evaluating oral cancer images. It can be applied to other microscopic images prepared with the same type of tissue staining. [Neha Sharma, Nigdi Pradhikaran, Akurdi 2011] [26] have compared the performance of data mining techniques for oral cancer prediction. The two data mining techniques used are Multilayer Perceptron Neural Network model and tree Boost model. For Training data as well as validation data, Multilayer Perceptron Neural Network and Tree Boost indicates the same specificity and sensitivity. Misclassification of data is not seen in both training and validation data in Multilayer Perceptron Neural Network as well as tree boost model. Also the most important variable for the prediction of malignancy is "Presence of Lymph Node" as seen on USG. As per the study, Tree Boost Classification Model and Multilayer Perceptron Neural Network model both are optimal for predicting malignancy in patient. [M. Muthu Rama Krishnan, Chandran Chakraborthy, Ajoy Kumar Ray, 2010] [27] have proposed a wavelet based texture classification for oral histopathological sections. As the conventional method involves in stain intensity, inter and intra observer variations leading to higher misclassification error, a new method is proposed. The proposed method, involves feature extraction using wavelet transform, feature selection using Kullback Leibler (KL) divergence and diagnostic classification using Bayesian Approach and Support Vector Machines [A. Chodorowski et al, 1999] [28] have proposed a method for oral lesion classification using true color images. Five different color representations were studied and their use for color image analysis of mucosal images evaluated. Four common classifiers (Fishers Linear Discriminant, Gaussian quadratic, KNN nearest neighbor and Multilayer perceptron) were chosen for the evaluation of classification performance. Classification accuracy was estimated using resubstitution and 5 fold cross validation methods. The best classification methods were achieved in HIS system and linear discriminant function. [A. Ji Wan Han 2008] [29] have investigated the classification of radicular cysts and odontogenic keratocysts. The classification was made using cascaded haar classifier. Three separated classifiers were trained respectively for each type of cyst to process unseen histological images in turn, to return a statistical count of the number of each corresponding cyst nuclei type present. The experimental results show the success of these classifiers in locating individual cells nuclei and in classifying the cyst types. [Laine. A.F et al] [30] have proposed Mammographic feature enhancement by multiscale analysis. Methods of contrast enhancement are described based on three over complete multiscale representations: 1) the dyadic wavelet transform (separable), 2) the transform (non separable,
87
Vol. 3, Issue 1, pp. 84-91
nonorthogonal), and 3) the hexagonal wavelet transform (nonseparable), Multiscale edges identified within distinct levels of transform space provide local support for image enhancement. [Sebastian Steger, Marius Erdt, Gianfranco Chiari and Geirgious Sakas] [31] have proposed a method for novel image feature extraction approach that is used to predict oral cancer reoccurrence. Several numeric image features that characterize tumors and lymph nodes are also proposed. In order to automatically extract those features Registration and supervised segmentation of CT/MR images form the base of automated extraction of geometric and texture features of tumor and lymph nodes. Higher accuracy and robustness is achieved compared to todays clinical practice. Literature survey reveals that cancer imaging in one of the active areas of research today. According to researchers, it is important to detect, segment and classify cancers at an earlier stage. The researchers working in this area have contributed towards development of algorithms in cancer detection, segmentation, classification etc.
III. COMPARISON OF METHODS

Table 1. Comparisons of various cancer detection methods
The above referred research works are classified as cancer detection methods and cancer classification methods. A comparative study is made between the detection methods and the classification methods separately.
3.1 Comparison of Cancer Detection Techniques

In [13], a neuro fuzzy model was used to achieve a higher value of tumor pixels. The algorithm used classifies the image layer by layer. Brain Tumor was detected using a Neuro Fuzzy model. The performance of MRI image in terms of weight vector, execution time and tumor pixels detected and compared the results with the existing one. A higher value of detected tumor pixel than any other was achieved. Ghassan Hamasneh et al, [12] applied snakes for semi-automatic segmentation of oral lesions in color images of the human oral cavity. Snakes reduced the need for edge linking compared to traditional edge based segmentation and lead to small segmentation errors. But operator interaction was needed due to large variability of the objects and images in this application. In [8], the first procedure is
88
Vol. 3, Issue 1, pp. 84-91
determining the seed regions. The Fuzzy C means clustering algorithm is used a segmentation strategy to function as better classifier and aims to classify data into separable groups according to their characteristics. As the number of clusters increases, more and more information is obtained about the tissue which cannot be identified by pathologists. Varsha.H. Patil et al, [16] proposed an automated system for detecting breast tumor at an earlier stage. The system was online and interactive, hence faster and accurate than manual process. The system uses super resolution technique to display the necessary information for boosting the physicians diagnosis. In [15], the proposed approach provided promising segmentation results. However, several control parameters are not automatically defined and the identification of lesions needs further development. In future, a new methodology is proposed to extract various parameters characterizing each basin. These parameters will be used in view to automatically identify the suspect of lesions. Woong et al [6] used 2D and 3D OCT for early detection and diagnosis of oral premalignancy and malignancy. 3D images provide detailed structural information at any location, and may be viewed at any desired by the clinician. OCT has the potential to become a powerful method for early oral cancer detection. 3.2 Comparison of Cancer Classification methods Muthu Rama Krishnan et al, [27], classified oral tumor using Bayesian Classification and Support Vector Quantization. All wavelet family has been used as an input to classifier to determine the signification of measurement. 48 gabor wavelet features and 9 wavelet features of epithelium are extracted. The signification of each feature is tested using KL divergence. Classification accuracy with wavelet and Gabor wavelet based texture features is also made. Wavelet family with gabor texture features leads to 92% average overall classification accuracy for Support Vector Quantization and 76.83% accuracy for Bayesian one. Ji Wan Han et al [29] used Haar cascade classifier for classification. The classifiers were able to find the individual cell nuclei, but there were much false positive detection. These false detections have a negative influence on the overall classification results of the technique. However, the performance of this technique against that of [30] is based on lesser information. Landini [32] analysed epithelial lining architecture in radicular cysts and odontogenic keratocysts applying image processing algorithms to follow traditional cell isolation based approach. Ireaneus Anna Rejani.Y et al [9] used a thresholding method for segmentation. The classification of breast cancer is done by SVM classifier. The method was tested on 75 mammographic images, from the mini MIAS database. The methodology achieved a sensitivity of 75%. A. Chodorowski et al, [28], proposed a method for oral lesion classification using true color images. Classification accuracy was estimated using res resubstitution and 5- fold cross validation methods. The best classification results were achieved in HSI color system and using linear discriminant function 94. % of accuracy was achieved. The comparisons of various techniques are tabulated in Table 1.
IV. DISCUSSION
There are many techniques for detecting cancers. Some researchers have suggested neuro fuzzy models for classifying cancers. Many methods aim for high accuracy, more features and enhancements. In [19], the considerable iteration time and the accuracy level is found to be about 50 60% improved in recognition compared to the existing neuro classifier. In [20], the desired detection strongly depends on active contour model. Hence in this work adaptive active contour model was used. The accuracy and speed of detection can be further modified by modifying model and neural network training approach. These papers focus more on accuracy. But the approaches applied for breast cancers or brain cancers cannot be applied directly for oral cancers, because of the moving artifacts induced by the moving tongue and jaw.
V.
CONCLUSIONS
In this paper, various methods to detect cancers are analyzed. The proposed work will identify oral cancer at an earlier stage which helps surgeons to provide medications and other treatments necessary
89
Vol. 3, Issue 1, pp. 84-91
for the particular cancer type. The proposed work will explore different enhancement techniques to improve the quality of images capturing devices like Ultra Sonography (US), Positron Emission Tomography (PET), Single photon Emission Computed Tomography (SPECT), Optical Imaging (OI), Computed Tomography (CT), X ray, Ultrasound and MRI. This will benefit the patients suffering from oral cancer.
ACKNOWLEDGMENT
The authors would like to thank Dr. T.P.Swamy, MDS (Orthodontist, Annai Orthodontist Centre, Coimbatore) for his valuable suggestions about the occurrence and diagnosis of oral cancer. The authors devote their sincere thanks to the Management and Staff of Karpagam College of Engineering for their constant support and motivation.
REFERENCES
[1] [2] [3]
[4] [5]
[6]
[7] [8]
[9] [10]
[11] [12]
[13] [14]
[15] [16]
[17]
[18]
[19]
Arlene Guagliano, Dental Tribune Middle East and African Edition,June August 2011. National Cancer Institute Website, www.cancer.gov Crispian Scully, Jose.V. Bagan, Colin Hopper, Joel. B. Epstein, Oral Cancer: Current and future diagnostics techniques A review article, American Journal of Dentistry, Vol. 21, No.4, pp 199 209, , August 2008. R.C. Gonzales and R.E. Woods, Digital Image Processing, Addison Wesley 1992. A. Banumathi, J. Praylin Mallika, S. Raju, V. Abhai Kumar, Automated Diagnosis and Severity Measurement of cyst in dental X ray images using Neural Network, International Journal of Biomedical Soft Computing and Human Sciences, Vol.14, No.2, pp 103 107, 2009. Woonggyu Jung, Jun Zhang, Jungrae Chung, Petra Wilder Smith, Matt Brenner, J. Stuart Nelson and Zhongping Chen, (2005) Advances in Oral Cancer Detection using Optical Coherence Tomography, IEEE Journal of Selected Topics in Quantum Electronics, Vol. 11, No.4.pp 811 817. Simon Kent, Diagnosis of oral cancer using Genetic Programming A Technical Report, CSTR -96-14. Ranjan Rashmi Paul, Anirban Mukherjee, Pranab K. Dutta, Swapna Banerjee, Mousumi, Pal, Jyotirmoy Chatterjee and Keya Chaudhuri, A novel wavelet neural network based pathological stage detection technique for an oral precancerous condition , Journal of Clinical Pathology, Vol.58, Issue.9, pp 932 938, 2005. Ireaneus Anna Rejani.Y, S. Thamarai Selvi, Early Detection of Breast cancer using SVM Classifier Technique, International Journal of Computer Science and Engineering, Vol. 1(3), pp 127-130, 2009. S. Saheb Basha, K.Satya Prasad, Automatic Detection of Breast cancer mass in mammograms using morphological operators and fuzzy C means clustering, Journal of Theoretical and Applied Information Technology, pp 704 708. Man Kin Derek Ko, Watershed Segmentation for Medical Confocal Image Analysis Towards in Vivo Early Cancer Detection, Journal of Biological Applications, pp 14 15, 2007. Ghassan Hamarneh, Artur Chodorowski, Tomas Gustavsson, Active Contour models: Application to oral Lesion detection in color images, IEEE Conference on Systems, Man, and Cybernetics in IEEE Conference in Systems, Man and Cybernetics, Nashville, TN , USA, October 2000, pp 2458 2463. S. Murugavalli, V. Rajamani, An Improved Implementation of Brain Tumor Detection using Segmentation based on Neuro Fuzzy Technique, Journal of Computer Science, Vol.3, Issue. 11, pp 841 846, 2007. M.M. Mohamed, T. K. Abdel-galil, M. M. A. Salama, A. Fenster, K. Rizkalla, D. B. Downey, Prostate Cancer Diagnosis Based on Gabor Filter Texture Segmentation of Ultrasound Image, IEEE Canadian Conference on Electrical and Computer Engineering (CCECE03), Montreal, Canada, Vol. 3 , pp. 1485 1488, May 2003. H. S. Seshadri, A. Kandasamy, Detection of breast cancer tumor based on morphological watershed algorithm, International Journal on Graphics, Vision and Image Processing, Vol.5, Issue. 5, 2005. Varsha H. Patil , Dattatraya S. Bormane, Vaishali S. Pawar, An automated computer aided breast cancer detection system, International Journal on Graphics, Vision and Image Processing, Vol.6, Issue. 1, 2006, pp 69 72. Poulami Das, Bebnath Bhattacharya, Samir. K. Bandyopadhyay, Tai hoon Kim, Analysis and Diagnosis of Breast cancer, International Journal of u- and e- Service, Science and Technology, Vol.2, September 2009. Shekar Singh, Dr.P.R.Gupta, Manish Kumar Sharma, Breast cancer detection and classification of histopathological images, International Journal of Engineering Science and Technology, Vol 3, No.5, May 2001, pp 4228 4232. G. Vijay Kumar, Dr.G.V.Raju, Biological early brain cancer detection using Artificial Neural Network, International Journal of Computer Science and Engineering, Vol. 2, No.8, pp 2721 2725, 2010.
90
Vol. 3, Issue 1, pp. 84-91
[20] [21] [22]
[23]
[24]
[25]
[26]
[27]
[28] [29]
[30] [31]
[32]
Yan Zhu, Hong, Computerized Tumor boundary detection using Hop Field Network, IEEE Transactions on Medical Imaging, 1997. L. Jeba Sheela,Dr. V. Shanthi, Image mining Techniques for classification and segmentation of Brain MRI Data, 2005 2007, JATIT. DSVGK Kaladhar, B. Chandana and P. Bharath Kumar, Predicting cancer survivability using Classification algorithms, International Journal of Research and Reviews in Computer Science (IJRRCS) Vol.2, No.2, pp 340 343, April 2011. Xiaowei Chen, Xiaobo Zhou and Wong, S.T.C, Automated segmentation, classification, and tracking of cancer cell nuclei in time-lapse microscopy, IEEE Transactions on Bio medical Engineering, Vol. 53, Issue.4, pp 762 766, 2006. Yung nien Sun, Yi-ying Wang, Shao-chien Chang, Li-wha Wu and Sen tien Tsai, Color based tumor segmentation for the automated estimation of oral cancer parameters, Microscopy Research and Technique, Vol. 73, Issue. 1, pp 5- 13, 2010. Yung nien Sun, Yi-ying Wang, Shao-chien Chang, Li-wha Wu and Sen tien Tsai, A color based approach for automated segmentation in Tumor Tissue Classification, Proceedings of the 29th Annual International conference of the IEEE EMBS, 2007. Neha Sharma, Nigdi Pradhikaran, Akurdi, Comparing the performance of data mining techniques for oral cancer prediction, Proceedings of the 2011 International Conference on Communication, Computing & Security (ICCCS11), ISBN: 978-1-4503-0464-1, New York, USA, 2011. M. Muthu Rama Krishnan, Chandran Chakraborthy, Ajoy Kumar Ray, Wavelet based texture classification of oral histopathological sections, International Journal of Microscopy, Science, Technology, Applications and Education, pp 897-906. A. Chodorowski, U. Mattsson, T. Gustavsson, Oral Lesion classification using true color images, Proceedings of SPIE, Vol. 3661, ISBN. 978081943132, pp 1127 1138, 1999. A. Ji Wan Han, Toby Breckon, David Randell, Gabriel Landini, Radicular cysts and odontogenic keratocysts epithelia classification using cascaded Haar classifiers A review article published by University of Birmingham. Laine, A.F. Schuler, S. Jian Fan Huda, W. Mammographic feature enhancement by multiscale analysis IEEE Transactions on Medical Imaging. Vol.13, Issue.4, pp 725 740, December 1994. Sebastian Steger, Marius Erdt, Gianfranco Chiari and Georgios Sakas, Feature Extraction from Medical Images for an oral cancer reoccurrence prediction environment, World Congress on Medical Physics and Biomedical Engineering, September 7 - 12, 2009, Munich, Germany. G. Landini, Quantitative analysis of the epithelial lining architecture in radicular cysts and odontogenic keratocysts, Head and Face medicine, Vol.2, February 2006.
Authors K. Anuradha completed her B.Sc (Applied Science) from PSG College of Technology, Coimbatore in 2000 and MCA from Bharathiar University in the year 2003. She is currently pursuing her Ph.D (Computer Science) at Karpagam University, Coimbatore. She has published a paper in the International Journal. She has presented 5 papers in National and International Conferences. She has got 8 years of teaching experience and 1 year industrial experience. Her areas of interest include Medical Image Processing, Computer Graphics and Distributed Computer Systems.
K. Sankaranarayanan, born on 15.06.1952, completed his B.E. (Electronics and Communication Engineering) in 1975 and M.E. (Applied Electronics) in 1978 from P.S.G. College of Tech., Coimbatore under University of Madras. He did his Ph.D. (Biomedical Digital Signal Processing and Medical Expert System) in 1996 from P.S.G.College of Technology, Coimbatore under Bharathiar University. He has got more than 32 years of teaching experience and 10 years of research experience. He has guided 10 candidates for Ph.D and presently guiding 18 candidates for Ph.D. His areas of interest include Digital Signal Processing, Computer Networking, Network Security, Biomedical Electronics, Neural Networks and their applications and Opto Electronics. He has published more than 50 papers in National Journals and International Journals.
91
Vol. 3, Issue 1, pp. 84-91
WAVEFORM ANALYSIS OF PULSE WAVE DETECTED IN THE FINGERTIP WITH PPG

Subhash Bharati & Girmallappa Gidveer
Jawaharlal Nehru Engineering College, Aurangabad, M.S., India
ABSTRACT
Photoplethysmography is one of the biomedical Instruments and when combined with pulse wave analysis it becomes information rich. Photoplethysmography is a non-invasive techniques that measures relative blood volume changes in the blood vessels close to the skin. This instrument was specially designed for the analysis healthy, diabetes an arthritis person. Many problems were created while designing PPG instrument such as light source, skin thickness of fingertip, signal amplification, storing of signals, heart rate and respiratory rate. We present the results of analysis of photoplethysmography (PPG) signal for healthy and patient with cardiovascular disorder (diabetes and arthritis). PPG signal of 21 subjects were recorded from the fingertip. The analysis indicates the content of PPG signal is different for healthy and cardiovascular patients. We also investigated heart rate and respiratory rate using PPG signal. In the present method of PPG analysis, the aim of this study to analyze the waveform in relation with diabetes, arthritis and healthy persons. The peripheral pulse has a steep rise and notch on falling slope in the subjects and a more gradual rise and fall and very small dicrotic notch were observed. The analysis with falling slope indicates the type of diseases.
KEYWORDS: Pulse Wave, Photoplethysmograph, Diabetes, Arthritis.
I.
INTRODUCTION
Human skin plays an important role in various physiological processes including thermoregulation, neural reception, and mechanical and biochemical protection. The heart-generated blood-pressure waves propagate along the skin arteries, locally increasing and decreasing the tissue blood volume with the periodicity of heartbeats. The dynamic blood volume changes basically depend on the features of the heart function, size and elasticity of the blood vessels, and specific neural processes. Therefore direct monitoring of skin blood pulsations may provide useful diagnostic information, especially if realized non-invasively. Optical technologies are well suited for non invasive monitoring of skin blood pulsation. Radiation of the red to near infrared spectral region penetrates several millimeter under the skin surface. Skin blood pumping and transport dynamics can be monitored at different body location (e.g. fingertip, earlobe, and forehead) with relatively simple and convenient PPG contact probes. Simultaneous data flow from several body locations the multi channel PPG technique increases the reliability of clinical measurements also allowing us to study heart beat pulse wave propagation in real time and to evaluate the vascular blood flow resistance an important physiological parameter for vascular diagnostics. In general each recorded PPG pulse contains useful information can be obtained by analysis of the PPG signal sequence recorded [1]. Plethysmograph is a combination of the Greek word. Plethysmos meaning is increase and Graph is the words for write [2]. It is an instrument used mainly to determine and register the variation in blood volume or blood flow in the body. We used photo electric type plethysmograph. Hence is known as Photoplethysmograph [3]. Pulse wave analysis helps to study diabetes & arthritis & it is unique for each individual so it would also give unique identification as biometric identification [4]. Pulse wave analysis also helps to study large artery damage& an abnormality in the cardiovascular disease which is one of the common causes of high mortality rate.PPG analysis emphasizes the importance of early evaluation of the diseases [5]. Several studied conducted by various groups of population showed that in PPG, the
92
Vol. 3, Issue 1, pp. 92-100
reflectance of light from in vivo tissue is described as a function of light for the wavelength in the range from 420 to 940 nm [6]. The electrical signal from PPG is related to blood volume changes in tissue. This signal provides a means of determining the diseases related to cardiac cycle & changes in arthritis & diabetes. The suggested PPG method is reliable, simple, low cost & noninvasive which could become an effective new screening tool for the early detection of diabetes neuropathic foot [7].
II.
LITERATURE SURVEY
2.1. Methods of Photoplethysmograph (PPG)

2.1.1. Reflected Photoplethysmograph Reflection PPG method uses the back scattered Optical signals for analysis of skin blood volume pulsation [8]. 2.1.2 Transmitted PPG In the transmission method, an optical signal change according to its absorption at the pulsation as oxygenated allows red wavelength more and deoxygenated blood allows infrared wavelength. It employs the principle that oxygenated blood is bright red. Whereas reduced or deoxygenated blood is dark red so combination of red and near infrared LEDs and photo sensors can be used to monitor the colour of blood [8].
2.2. Issues Regarding PPG

2.2.1. First Issues Regarding PPG The first issue concerns the contact and noncontact PPG, in which both are has nearly same potential only difference in the amplitude of the received signal and clarity. In the noncontact PPG signals are not so cleared as compared to contact type PPG [2]. 2.2.2 Second Issues Regarding PPG The second issue concerns the dynamic range of the detected signal. The detected pulsatile (AC) signal is very small compared to the non-pulsatile (DC) signal as shown in figure1 [2]. 2.2.3 Third Issues Regarding PPG The third issue is ambient light artifact. The detector will receive increased ambient light due to the probe separation from the tissue bed. Introducing close packaging of finger bed with detector could reduce this effect [2].
Figure 1. Breakdown of the component of the detected PPG signal
III.
MEASUREMENT SYSTEM
3.1. Block Diagram 93 Vol. 3, Issue 1, pp. 92-100
We used red light source as sensor, i.e. LED. LED gives brighter light at low power as compare to other. Also by using LED, the problem of localized heating can be avoided [9]. The PPG device incorporates a light source and detector, which are placed against the surface of skin [10]. Incident light passes through the tissue and blood and diffuses in the tissue bed. Variation in the intensity of light arises from changes in the blood fractional volume of arteries and tissues, which alter light absorption. This variation occurs with arterial pulsation. The absorption coefficient of tissue bed at various wavelengths for pulsation blood it has maximum output at wavelength 420 to 900 nm[6].For detectors we used LDR whose dark resistance is 10 K when the cell is not illuminated (i.e. dark resistance). The spectral response of LDR is good for visible region and we used red colored source. In our project we used LDR as detector which gives output change in resistance and this change in resistance is directly proportional to light incident on its surface. Coupling capacitor used which blocks the dc component present in our signal. As in previous work we showed that the detected signal consist ac and dc components.PPG signal = DC blood and tissues + AC blood modulates. Thus dc components can be filtered out [2]. The wave form different specimens (i.e. different male/females) depend on the skin thickness to overcome these problem we used variable gain (pot) to set the gain as per requirement. Lastly we used comparator to compare to shift the waveform as per subject requirement, thus added comparator due to which the dc level can be shifted, because of skin thickness and different artifacts i.e. movements etc. the wave form may be shifted. Thus we get final output which is PPG signal. Once we get PPG wave forms then our main aim is to show the effect of diabetic or arthritis on the waveforms. This depends upon the dicrotic notch.
Figure 2. Block Diagram
3.2. Physiological Measurement

Photoplethysmographic signal measurements were obtained from 21 subjects (13 male, 8 female). Subjects were in the age of 22-70. Measurements were performed in a laboratory. Each subject was asked to relax and sit on the chair and rest the forearm on the lab table to help the entire hand keep steady. An operator then attached the finger sensor to the forth fingertip of the left hand. It is important to have a comfortable arm position in order to keep the finger relatively motionless for a stable and repeatable recording. The length of the recorded signal was 10 seconds.
3.3. Analysis
The analysis of the healthy, diabetes, arthritis person is as shown in figure 3, 5, 7. From figure 3, we observed number of peaks; from this peak we calculated heart rate and respiration rate. Heart rate = No. of peaks 12 Heart rate = 6 12=72 Respiratory rate=Heart rate2 Respiratory rate=722=36 3.3.1. Analysis of Healthy Person
94
Vol. 3, Issue 1, pp. 92-100
Figure 3.Healthy subject status
Analysis of healthy person is as shown in figure 3. Figure 4 shows the SPPPG (single period photoplethysmograph) and animated signal of healthy person.
Figure 4. SPPPG and animated signal of healthy person
3.3.2. Analysis of Diabetes Person
95
Vol. 3, Issue 1, pp. 92-100
Figure 5. Diabetes subject status
Analysis of diabetes person is as shown in figure 5. Figure 6 shows the SPPPG and animated signal of diabetes person.
Figure 6. SPPPG and animated signal of diabetes person
3.3.3. Analysis of Arthritis Person
Figure 7. Arthritis subject status
96
Vol. 3, Issue 1, pp. 92-100
Analysis of arthritis person is as shown in figure 7. Figure 8 shows the SPPPG and animated signal of arthritis person.
Figure 8. SPPPG and animated signal of arthritis person
IV.
DISCUSSION AND CONCLUSIONS
The main differences in the PPG with healthy, diabetes and arthritis patients were observed with the presence or absence of dicrotic notch & trailing edge slope. As vessels get stiffer during aging process, the reflected wave returns faster and due to the summation of wave the resultant pulse wave changes [11]. After filling in the patient data AC-component of his/her PPG signal is detected and stored. The analysis of healthy, diabetes and arthritis person is illustrated at figure 3, 5 and 7, which shows that the health status, heart rate and respiratory rate. The SPPPG and animated signal is illustrated at figure 4, 6 and 8.The shape of the single PPG pulse detected at the periphery (e.g. fingertip) can differ significantly from that at the magistral arteries; it primarily depends on resistance of the vascular system. If the vessel resistance is abnormally high due to diabetes, arthritis or other vascular pathology that narrow the vessels, velocity of blood flow from big arteries to small capillaries decreases. In the summary comparisons of the PPG wave form shows that it can provide a simple non-invasive means of studying diabetic & arthritic patients [13].
V.
RESULTS
The main objective of this work was to show the relation of trailing edge slope with diabetes and arthritis patient shows the change in slope than healthy subject i.e. the difference in the pulse shape changes as a function of disease which can be well observed visually. In the diabetic patients PPG waveform shows very small dichrotic notch & slope is less, where as arthritis patients PPG waveform shows very sharp slope and no dichrotic notch were observed. Single period PPG signal comprises a fast raising part or anacrota and subsequent falling part or catacrota. Anacrota reflects the stretching of the blood vessel walls under the increased blood pressure after each heartbeat and catacrota relaxation processes of the blood vessel wall in-between each two heartbeat. The catacrota can be variously shaped depending on the vascular condition; it normally contains so called predicrotic dip, and secondary peak (notch) caused by elastic reflection in the arterial system. A typical healthy persons SPPPG signal shape is presented at figure 4.The propagating blood pressure pulse wave becomes broadened and delayed and may completely lose its secondary (dicrotic) peak, when the periphery is reached. SPPPG signals were bell-shaped without any secondary peak at the catacrota. A typical diabetes persons SPPPG signal shape is at figure 6.A typical SPPPG signal of arthritis persons is as shown in figure 8.This peak is a clear evidence of increased blood flow via the damaged vessels.
ACKNOWLEDGEMENT
The author would like to thank Dr. Mali for pathological support and patient co-ordination. The author would also like to thank Dr. S. D. Deshmukh Principal, Prof. J.J. Rana H.O.D; Jawaharlal Nehru Engineering College, Auranagbad (M.S.) India. The author would also like to thank reviewers
97
Vol. 3, Issue 1, pp. 92-100
for their valuable suggestions. This work was supported by grants from the Department of Electronics and Telecommunication, Jawaharlal Nehru Engineering College, Auragabad (M.S.), India.
REFERENCES
[1].Janis Spigulis, Optical Non-Invasive Monitoring of Skin Blood Pulsations. Applied Optics Vol. 44, No.10.April 2005.pp.1850-1857. [2]. Peck Y.S.Cheng and P.R.Smith, An Overview of Non-Contact Photoplethysmography, Dept.of Electronics & Electrical Engineering, Loughborough University, LE 1 1 3TU, UK, pp57-59. [3].Leslie Cromwell, Fred J.Weibell, Erich A. Pfeiffer, Biomedical Instrumentation and Measurement. Second Edition, Prentice-Hall of India Private Limited, New Delhi, pp.150-163. [4].M.H.Sherebin, R.Z. Sherebin, Frequency Analysis of Peripheral Pulse Wave Detected in the Finger with Photoplethysmograph.IEEE Transaction on Biomedical Engineering, Vol.37No.3, March 1999. [5].K.Meigas, R.Kattai, M.Nigul, Comparisons of Signal of Pulse Profile as Skin Surface Vibration PPG and Doppler Spectrogram for Continuous Blood Pressure Monitoring. Proceeding of The International Federation for Medical and Biological Engineering, Vol. 3, 2002.pp. 510-511. [6]. Weijia Cui, Lee E. Ostrander & Bok Y. Lee, In Vivo Reflectance of Blood and Tissue As a Function of Light Wavelength, EEE Transaction on Biomedical. Engg. Vol - 37, No. 6 June 1990, pp.1-1. [7].Deon Won Kim, Sung Woo Kim, Detection of Diabetic Neuropathy Using Blood Volume Ratio of Finger and Toe by PPG. Engineering in Medicine and biology Society, 2007.EMBS 2007.29TH Annual international Conference of the IEEE. pp. 2211-2214. [8]. Vincent P.Crabtree, Prospective Venox Feasibility Study, Dept. of EEE, Loughborough University.pp.2727. [9].Joydeep Bhattacharya, Partha Pratim Kangilal, Analysis & Characterization of Photoplethysmography Signal, IEEE Transaction on Biomedical Engg.Vol.48, No.1, Jan.2001.pp1-1 [10]. Stephen A.M., H.H. Asada, Photo-Plethysmograph Finger Nail Sensors for Measuring Finger Forces without Haptic Obstruction, IEEE Transaction On Robotics and Automation, Vol.17, No.5, Oct.2001.pp.1-1. [11].Irina Hlimonenko, Kalju Meigas, Rein Vahisalu, Waveform Analysis of Peripheral Pulse Wave Detected in the fingertip with Photoplethysmograph. Tallinn Technical University, Biomedical Engg, Center, Tallinn, Estonia. Measurement Science Review,Vol.3 Section 2,2003.pp.49-52. [12]. Sijung H; P.R. Smith, Comparison of Pulse Interval in Contact & Noncontact Photo Plethysmography, Dept. of EEE, Loughborough University.pp.39-42. [13]. Jonis Spigulis, Indulis Kukulis, Potentials of Advanced Photo-Plethysmography Sensing for NonInvasive Vascular Diagnostics and Early Screening, University of Latvia, Dept. of Physics, Riana Blvd.19, Riga, LV-1586, Latvia.pp1-5.
APPENDIX
#include <REG51.H> #define adc_port P1 sbit rd = P3^7; sbit wr = P3^6; sbit cs = P3^4; sbit intr = P3^5; //ADC Port //Read signal P1.0 //Write signal P1.1 //Chip Select P1.2 //INTR signal P1.3
void conv(); //Start of conversion function void read(); //Read ADC function void delay(unsigned int); unsigned char serial_read(); void serial_send(unsigned char); void serial_init(); unsigned char adc_val;
98
Vol. 3, Issue 1, pp. 92-100
void main() { intr = 1; serial_init(); while(1) { //Forever loop conv(); //Start conversion read(); //Read ADC serial_send(adc_val); delay(10); } } void conv() { cs = 0; wr = 0; wr = 1; cs = 1; while(intr == 1); } void read() { cs = 0; rd = 0; adc_val = adc_port; rd = 1; cs = 1; }
//Make CS low //Make WR low //Make WR high //Make CS high //Wait for INTR to go low
//Make CS low //Make RD low //Read ADC port //Make RD high //Make CS high
void delay(unsigned int count) { unsigned int i; while(count) { i = 115; while(i > 0) i--; count--; } } void serial_init() { TMOD = 0x20; SCON = 0x50; TH1 = 0xFD; TL1 = 0xFD; TR1 = 1; } void serial_send(unsigned char dat) { SBUF = dat; while(TI == 0); TI = 0; } unsigned char serial_read()
99
Vol. 3, Issue 1, pp. 92-100
{ while(!RI); RI = 0; return SBUF; }
Authors:
Subhash Bharati has 10 years experience in Academic. He had worked as Lecturer in Gangamai College of Engineering, Dhule, M.S., India. At present he is as M.E. Scholar at Jawaharlal Nehru Engineering College, Aurangabad, M.S., India. He has successfully published National & International Research Paper.
Girmallappa Gidveer is professor in Electronics and Telecommunication Engineering Dept. in Jawaharlal Nehru Engineering College, Aurangabad, M.S., India. And has 35 years experience in Academic. He has published many Research Papers in various Journal and Conferences.
100
Vol. 3, Issue 1, pp. 92-100
SLOTTED PIFA WITH EDGE FEED FOR WIRELESS APPLICATIONS

T. Anita Jones Mary1, T. Joyce Selva Hephzibah2, C.S.Ravichandran3
Department of Electronics and Communication, Karunya University, Coimbatore, Tamil Nadu, India 3 S.S.K. Engineering College, Coimbatore, Tamil Nadu, India
1,2
A BSTRACT
In order to overcome the shortcoming of narrow bandwidth of conventional PIFA, a novel bandwidth enhancement approach is proposed in this paper. A handset antenna technique that combines a parallel excitation of PIFA and slot radiators is presented. To improve the bandwidth of handset antennas in both low and high bands, slots can be introduced into the radiating patch of different shapes such as rectangular and Hshape. Comparison between the PIFA with U-slot on the radiating patch of these shapes can provide that which shape is suitable for enhanced antenna performance. The presented antenna can cover handheld satellite phones, WLAN, Bluetooth and Wimax frequency bands. From the simulation results obtained using FEKO, it is observed that the relative bandwidth of PIFA with U-slot on rectangular patch is 10.59 times and on H-shape patch is 24.3 times more than the conventional PIFA.
K EYW ORDS: Handset antenna, Planar Inverted F- Antenna(PIFA),Slot , Bandwidth, Patch
I.
INTRODUCTION
Recently modern wireless communication technologies are in the process of rapid development. Multi-system applications have been used explosively. Owing to this, it has been a necessity to design antennas with the characteristics of multiband and wideband for mobile terminals[8]. In modern mobile handsets, PIFAs are generally used as built-in antennas. Nowadays PIFA is being adopted extensively as handset antenna because of its advantages of compact structure, low profile, easy fabrication, low manufacturing cost and easy integration with portable devices. It has reduced backward radiation and enhanced antenna performance. However, a major disadvantage of the PIFA antenna is its narrow impedance bandwidth. Hence, it is desirable to find methods that can enhance the bandwidth of the PIFA antenna. Therefore the antenna designed should be as smaller as possible to be fitted in the handsets and should have acceptable performance. Phones are increasingly adding components and features such as large color screens, digital cameras, digital music players, digital and analog radio and multimedia broadcast receivers [1]. For conventional PIFA, only single band of operation is possible. To have dual band operation, some part of top plate has to be removed and another PIFA should be inserted. It will lead to more complexity. A rectangular slot is introduced into the ground plane of mobile chassis. A more compact design of a practical wideband PIFA antenna with trapezoidal feed has the bandwidth for S11 < -10 dB ranges from 1.67 GHz to 4.05 GHz and the relative bandwidth reaches 83% . The antenna height mainly influences the bandwidth, resonant frequencies, and return loss [3]. Changes in the width of the planar element can also affect the determination of the resonant frequency [7]. The central slot has good matching at low bands (17.4%) but at higher bands it still behaves similar to the reference antenna (8.3%) since the slot is in the cut-off state. This means that the slot does not operate as a parasitic element [5]. Resonant frequency decreases with the decrease in short
101
Vol. 3, Issue 1, pp. 101-110
circuit plate width. Unlike micro-strip antennas that are conventionally made of half wavelength dimensions, PIFAs are made of just quarter-wavelength. Analyzing the resonant frequency and the bandwidth characteristics of the antenna can be easily done by determining the site of the feed point at which the minimum reflection coefficient is to be obtained [6]. Besides the good bandwidth results obtained when the ground plane is tuned to the same frequency of the handset antenna. Analysis of the way to minimize the length of the feeding line maintenance as much as possible the advantages of the bandwidth benefits of the open-slots is explained [12]. PIFAbased internal multiband antenna that can support the following eight frequency bands: GSM ( 860 980 MHz), DCS (17101880 MHz), PCS (personal communication services, 18801990 MHz), UMTS ( 1.9 2.17 GHz), WiBro (23002390 MHz), Bluetooth (24002480 MHz), S-DMB (2630 2655 MHz), and WLAN (50006000MHz)[17]. A reduced height, multiband internal antenna has been proposed for wireless personal communication handsets to operate at GSM- 900, DCS, PCS, UMTS, WiBro, Bluetooth, S-DMB, and WLAN frequency bands. Radiation and return loss performances of the antenna are reasonable at all the frequency bands[15]. A prototype that can cover the frequency bands of GSM850, GSM900, DCS1800 and PCS1900 simultaneously was fabricated and measured. In addition, the measured impedance bandwidth (defined at -6 dB level) of the low band ranges from 0.7 GHz to 0.98 GHz and for the high band it is between 1.65 GHz and 2.02 GHz, which completely covers the required quadband operational bandwidth[2]. There are four methods available to enhance the bandwidth of conventional PIFA. They are introducing slots, capacitive loading, loading dielectric with high permittivity and adding chip resistor. This paper describes a modified U-slot coupled H-shaped antenna & U-slot coupled rectangular patch to reduce the area of antenna and achieve the termination of high order harmonic component behaviours. It also exhibits excellent efficiency, bandwidth enhancement and linearity. The paper is organized as follows. In section II, the survey about slots on the ground plane as well as on the radiating patch is discussed. It provides the information about the applications in which slotted antennas are used. Section III provides the design parameters for the conventional & slotted PIFA. Section IV presents the low profile design covering handheld satellite phones, WLAN, Bluetooth and Wimax frequency bands. The simulation results using FEKO are discussed briefly in section IV. Finally, Section V summarizes the work.
II.
RELATED WORK
When the rectangular slot is on the ground plane, enhancement in bandwidth mainly depends on the size of the ground plane [8]. The novel H-shaped antenna has many excellent advantages over the conventional rectangular microstrip antenna such as the harmonic mode suppression, bandwidth enhancement, and the increase of the whole transmitter efficiency.The U-slot on the H-shaped patch is discussed [1]. Though the bandwidth and the size of an antenna are generally mutually conflicting properties, that is, improvement of one of the characteristics normally results in degradation of the other [9], hence compact size antennas are needed for mobile handsets with improvement in bandwidth. By varying the location and size of the slot, the equivalent length of the ground plane can be adjusted to the optimal lengths of the low and high bands, which thus enhance the bandwidth of PIFA antennas for both the low and high bands [18]. The resonance frequency depends inversely on the slot length and feed point, while it increases with increasing the slot width and coaxial probe feed radius [16]. Transmission line feeding is provided in order to provide excitation to antenna and RF module. The feeding line follows long meander shape path to reach antenna feeding pad and RF module [14]. The operational bandwidth is increased because the electrical length of the ground plane increases when meandering or open-end slots are used on the ground plane, although its physical size is fixed [4].
III.
ANTENNA DESIGN
Antenna design for mobile handsets can be of two types-internal and external. One of the main disadvantage of external antenna is it is very close to the users head and the radiation is directly incident on the head making the absorption rate high. Internal antenna can be installed on the side of
102
Vol. 3, Issue 1, pp. 101-110
the PCB i.e. opposite to the human head thus avoiding the human interference. One of the techniques to obtain multiband behavior for handset antennas is to create several resonant paths. The specification of the internal antenna depends strongly on the design of mobile phones and changes have to be made for each design. Moreover the internal antenna is difficult to design than its counterpart because the designer must consider characteristics such as feed point, ground position, radiation pattern, etc. At present, PIFAs have attracted much interest due to their small size and appreciable electrical characteristics compatible with existing specification, making it a promising candidate for internal antennas. The PIFA structure in general is shown in Fig.1.
Figure.1. PIFA structure
Bandwidth enhancement and size reduction are not achieved at the same time, especially for resonant structures such as PIFA. Dual resonance is obtained by introducing slots parallel to the radiating edge of the patch. The resonant frequency of the antenna can be reduced by decreasing the stub width and also by introducing open slots. To reduce the PIFA size, it is necessary to shorten the antenna. But it affects the impedance of the antenna and the radiation resistance becomes reactive. It can be compensated using capacitive load. The shape and size of the slot plays significant role in improving the bandwidth. In case of the PIFA, the loop structure formed by the shorting pin and feeding line is modeled as a shunt inductance and impedance matching of the PIFA can be achieved by controlling the required amount of the shunt inductance, which is generated by the loop structure. In the case of the slot antenna on the ground plane, the above mentioned loop structure can be modelled as a series inductance in contrast with the shunt inductance in the PIFA. The design parameters for PIFA without slot on rectangular patch and H-shaped patch are given in Table1 & 2 respectively. The simulated result of PIFA as per Table.1 &2 are shown in Fig.2 & 3 respectively. The feeding voltage is given at one of its edge. The resonant frequency of PIFA can be approximated using the following equation(1). L1 + L2 = /4 (1) Here in Table 1, Lg, Wg are length and width of the ground plane respectively. Lp, Wp are length and width of the radiating patch respectively. Lf, Wf, are length and width of the feed respectively. Ls, Ws are length and width of the shorting pin respectively. H-shaped patch is made using 3 rectangles. Lp1, Wp1 are length and width of the first rectangle, Lp2, Wp2 are length and width of the second rectangle and Lp3,Wp3 are length and width of the third rectangle. The U-slot dimensions are given in Table 3. The resonant frequency of the PIFA mainly depends on the length of the ground plane and the radiating patch. The simulated results for PIFA with U-slot on rectangular and H-shaped patch are shown in Fig.4 & 5 respectively.
Table 1.Values of the Design Parameters (PIFA on rectangular patch) Parameter Lg Wg Lp Wp Lf Wf Ls Ws Value(mm) 100 40 16 30 2 3 4 3
103
Vol. 3, Issue 1, pp. 101-110
Table 2.Values of the Design Parameters (PIFA on H-shaped patch) Parameter Lg Wg Lp1 Wp1 Lp2 Wp2 Lp3 Wp3 Lf Wf Ls Ws Value(mm) 100 40 16 5 3 5 16 5 2 3 4 3
Table.3.Slot Dimensions U-slot(mm) L1 10 L2 6 W 2
The slot antennas of half wavelength structures are generally used to operate at the fundamental resonant mode. The electrical length of the proposed slot is shorter than a half of a guided wavelength. In the case of the slot antenna on the ground plane, the previously mentioned loop structure can be modeled as a series inductance in contrast with the shunt inductance in the PIFA. The slot antennas of half-wavelength structures are generally used to operate at the fundamental resonant mode. The electrical length of the slot is between a quarter and half of guided wavelength.
Figure.2. Conventional PIFA on rectangular patch in CADFEKO
Figure.3. Conventional PIFA on H-shaped patch in CADFEKO
104
Vol. 3, Issue 1, pp. 101-110
Figure.4.PIFA with U-slot on rectangular patch in CADFEKO
Figure.5.PIFA with U-slot on H-shaped patch in CADFEKO
Its half-wavelength resonance at the desired resonant frequency can be achieved because the impedance of the slot antenna is transformed to the series resonance, which is generated by the loop structure of the feed line. Thus, the amount of shunt and series inductance is dependent on the loop size formed by the feed line and the shorting pin. Hence, the resonant frequencies of the proposed hybrid antenna can be controlled by adjusting the distance between the feed line and the shorting pin. The distance between feed and shorting pin for the desired resonant frequency is chosen to be 1.5 mm.
IV.
RESULTS & DISCUSSIONS
The return loss and field distribution on the PIFA without & with slots has been computed using the FEKO. From the simulated results of Fig.6 & 7, it can be seen that PIFA with U-slot on rectangular patch has three resonant modes at 1.6GHz, 4.147GHz and 6.56GHz. Bandwidth can be calculated using the following equation(2), where fH, fL and fC are high frequency, low frequency and centre frequency respectively. (2) The calculated bandwidth for each of these antennas are shown in Table.4. Bandwidth can be improved by varying the slot width and also by varying the distance between feed and stub. Slot on the ground plane is more efficient than the slot on the patch. The optimal performances of the low and high bands are mainly depends on the slot location and size. Also, the optimal slot location and size of the low band are different from those of the high band. Hence, to broaden the bandwidths for the low and high bands simultaneously, a compromised solution for the slot location and size should be selected. A U-slot adds a capacitive component in the input impedance that compensates for the inductive component of coaxial probe. A single half wavelength slot resembles the half wave dipole in terms of gain and radiation except that there is a difference in polarization. The electric field across the slot is maximum at the centre and tapers off towards the edges as indicated. In general, rectangular and circular slots are easy to analyze and hence U shaped slots are analyzed in this paper. The introduction of an open slot adjacent to U-
105
Vol. 3, Issue 1, pp. 101-110
slot reduces the frequency, particularly of the higher band drastically. This is due to the fact that there are currents flowing at the edge of the U-shaped slot, therefore a capacitive loaded slot reduces the frequency and thus the antenna dimensions drastically. The impedance matching of the dual band can be obtained by positioning of the single feed and the shorting pin within the U-shaped slot, and by optimizing the space between feed and shorting pins. A shorting wall, which is placed at the edge of this PIFA, controls the isolation/separation between the two bands. The current follows larger path due to slot on the ground plane. Slots are not only useful to antenna design but also for damping undesired modes for EMC purposes. Maximum current distribution of the PIFA is obtained close to the shorting pin and decreases away from it. The slot is weakly excited at 900MHz compared to the excitation at 1720 & 2000MHz. The antenna performance is good when the feeding port and slot are arranged on the same side. Fringing fields are the radiating sources of PIFA. A dual band PIFA antenna has been used to analyze the bandwidth and efficiency improvement when a slot in the ground plane is introduced: the original dual-band antenna has been enhanced to introduce more frequency bands. Therefore, a multiband antenna with good bandwidths and efficiency response can be obtained with only adding a slotted ground plane.
PIFA on rectangular patch PIFA on Hshaped patch
Figure. 6. Simulated return loss against frequency for PIFA without slot on rectangular & H-shaped patch
PIFA with U-slot on rectangular patch PIFA with U-slot on H-shaped patch
Figure.7. Simulated return loss against frequency for PIFA with U slot on rectangular & H-shaped patch
106
Vol. 3, Issue 1, pp. 101-110
It can be noticed that introducing slots in the ground plane is a simple way to achieve a multiband antenna without modifying PIFA antenna geometry and increasing the handset volume. A dual band PIFA antenna has been used to analyze the bandwidth and efficiency improvement when a slot in the ground plane is introduced. From Fig.7, PIFA with U-slot provides three resonances. The separation between the resonances is sensitive to the dimension of the slot. The original dual-band antenna has been enhanced to introduce more frequency bands. The present technique is useful since it simplifies the length of the feeding transmission line. Therefore, a multiband antenna with good bandwidth and efficiency response can be obtained with only adding a slotted ground plane.
Table 4.Performances of the Proposed Broadband Antennas Antenna Type (Rectangular patch) PIFA without slot PIFA with U-slot on patch Bandwidth(%) Antenna Type (H-shaped patch) PIFA without slot PIFA with Uslot on patch Bandwidth(%)
6.16 10.59
16.5 24.3
If the S11 parameter is above -6dB, then it provides good impedance matching. From Fig.6, PIFA without slot on rectangular patch has return loss of -5.8dB and H-shaped patch has return loss of 13dB. From Fig.7, PIFA with U-slot on rectangular patch has return loss of -8dB and H-shaped patch has return loss of -16dB. Hence the impedance of PIFA with U-slot on patch is better than the impedance of conventional PIFA and also PIFA with U-slot on H-shaped patch is having more enhanced bandwidth than PIFA with U-slot on rectangular patch. By adjusting the shorting plate width, good impedance matching can be achieved. Hence more energy has been radiated from the PIFA with U-slot on patch to the load. Adding open slots in certain locations of the ground plane improves the bandwidth of the antenna. When the open slots are added, the feeding line length should be high. In order to minimize the length of the feeding line, open slots are avoided in this paper. To generate a slot resonating at the second band, slot length should be increased. The slot width was varied in the model to achieve a feed point input impedance of approximately 50 . Here enough power is transmitted from one end to other end. The slotted PIFA could be suitable to be used in wireless communication applications. By feeding the slot in its edge, triple resonance response is obtained with U-slot on H-shaped patch whereas with Uslot on rectangular patch double resonance response is obtained. By instead using a slot in the ground plane as the radiating element the antenna performance becomes independent of available height. In order to provide very broadband response, the slot should be placed in the center of the ground plane. It provides better quality factor to the antenna. The slot is located under the antenna projection in order to get a good coupling between the PIFA and the slot. It can also be used as a parasitic antenna. It means that the slot can serve as part of the directional antenna but has no direct connection to the receiver and transmitter. It reflects or reradiates the energy that reaches it. When the slot length is similar to 0.5 the slot acts as an effective parasitic element and a coupling effect between PIFA and slot may be obtained. The slot used in Fig.4&5 are too short to resonate at higher frequencies (18002000 MHz), this is why it only improves bandwidth at low bands. The ground surface waves produce spurious radiations or couple energy at discontuities leads to distortions in the main pattern or unwanted loss of power. All the antennas have the same radiation pattern at low frequencies. Therefore, adding some slots in the ground plane does not modify the antenna radiation structure. Introducing slots in the ground plane is a simple way to achieve a multiband antenna without modifying PIFA antenna geometry and increasing the handset volume. It is also observed that the first resonant frequency shifted upwards with the increasing of slot width. Again, with decreasing the slot width, second resonant frequency exhibits better resonant at the expense of reducing the upper edge frequency resulting in a bandwidth reduction. The slot is selected in an attempt to reduce the two orthogonal resonant modes to lower frequencies. Independent control of the two operating frequencies is possible with a U-shaped slot.
107
Vol. 3, Issue 1, pp. 101-110
The proposed antenna shows an interesting dual band resonant behavior with a wide range of the two resonant frequency ratio without changing the external dimensions of the combined slot structure. It is clear that slot can be either resonant or non-resonant. If it is resonant, the current along the edges of the slot introduces an additional resonance. If these additional resonances are near the patch resonance, adding them together causes enhancement in impedance bandwidth. Enhancement of impedance bandwidth and gain does not affect the nature of broadside radiation characteristics. It is found that lower and upper resonance frequency shifts towards lower side as the slot length increases and the bandwidth at both frequencies decreases. Higher resonance frequency shifts to higher side and lower resonance frequency shifts to lower side and the bandwidth for different slot width. It is also found that at lower resonance frequency the bandwidth depends inversely on the slot width whereas at upper resonance frequency the bandwidth depends directly on the slot width. It is clear that the resonance frequency decreases with increasing the feed position. The resonance frequency is highly dependent on the slot dimensions as well as the feed locations. From these discussions, it is noted that PIFA with U-slot on H-shaped patch enhances the bandwidth, whereas PIFA with U-slot on rectangular patch reduces the quality factor.
V.
CONCLUSION
It can be observed that the relative bandwidth of slotted PIFA is larger than conventional PIFA from the simulation results. The concept based on a PIFA-slot has been shown to be useful to design multiband handset antennas where the number of frequency bands is given by the sum of the bands given by each radiator. Moreover, said bands can be controlled independently which adds an extra freedom design. It is proposed that a novel hybrid PIFA with U-slot on rectangular and H-shaped patch for handheld satellite phones, WLAN, Bluetooth and Wimax applications. Its resonant frequency can be controlled by adjusting the inductance of the loop structure formed by the feed line and shorting pin. Thus, the PIFA for the lower band (1.6 GHz) and the slot for the higher band (11 GHz) operate using the same feeding voltage source. The relative bandwidth of PIFA with U-slot on H-shaped patch is greater than the conventional PIFA and PIFA with U-shaped slot on rectangular patch. Small volume, good electrical characteristics make slotted PIFA a promising candidate for the wireless applications. Main considerations for the design are Dimensions of conducting patch: They depend on the design frequency. The conducting patch should be of /4 dimension. Size of the ground plane: Ground plane affects the bandwidth to a greater extent; it should be optimized for the design frequency. The optimized value is 45% for the length and 25% for the width respectively. Position of the feed: It plays a major role in the impedance matching. The position of the feed should be as close to the short in order to get good impedance matching. Height of PIFA from ground plane: It will determine the bandwidth of PIFA. More the height, more the bandwidth will be.
REFERENCES
[1]. Amit A. Deshmukh, Priyanka Thakkar, Sneha Lakhani, Mayank Joshi, K. P. Ray, Formulation of
Resonance Frequency for Dual band Slotted H-shaped Microstrip Antenna, International Journal of Computer Applications, Proceedings of ICTSM 2011, Mumbai, India ,Feb. 25 27, 2011, pp.18-26 [2]. Jawad K. Ali, A New Dual Band E-shaped Slot Antenna Design for Wireless Applications, Progress In Electromagnetics Research, PIERS Proceedings, Suzhou, China, September 12-16, 2011 [3]. Jaume Anguera, Ivn Sanz, Josep Mumbr, and Carles Puente, Multiband Handset Antenna With a Parallel Excitation of PIFA and Slot Radiators, IEEE Transactions On Antennas And Propagation, Vol. 58, No. 2, Feb. 2010 [4]. Kekun Chang, Guan-Yu Chen, Jwo-Shiun Sun, and Y. D. Chen, PIFA Antenna with Coupling Effect for Bandwidth Enhanced Design and Measurement, Progress In Electromagnetics Research Symposium Proceedings, Xi'an, China, March 22-26, 2010
108
Vol. 3, Issue 1, pp. 101-110
[5]. Vinod K. Singh, Zakir Ali, Design and Comparison of a Rectangular-Slot-Loaded and C-Slot-Loaded Microstrip Patch Antenna, IJCSNS International Journal of Computer Science and Network Security, vol.10 No.4, April 2010 [6]. C. I. Lee, W. C. Lin, Y. T. Lin, and Y. T. Lee, A Novel H-shaped Slot-coupled Antenna for the Integration of Power Amplifier, Progress In Electromagnetics Research Symposium Proceedings, Cambridge, USA, July 5-8, 2010 [7]. Hussein Attia, Mohammed M. Bait-Suwailam, and O. M. Ramahi, Enhanced Gain Planar Inverted-F Antenna with Metamaterial Superstrate for UMTS Applications, Progress In Electromagnetics Research , PIERS Proceedings, vol. 6, No. 6, 2010, pp.585-588 [8]. Xingyu Zhang and Antti Salo, Design of novel wideband PIFA for mobile applications, Progress In Electromagnetics Research Symposium, Beijing, China, March 23-27, 2009 [9]. Sinhyung Jeon, Hyengcheul Choi, and Hyeongdong Kim,Hybrid planar inverted F antenna with a Tshaped slot on the ground plane, ETRI Journal, Vol. 31, Number 5, October 2009 [10]. Xingyu Zhang and Anping Zhao ,Enhanced bandwidth PIFA antenna with slot on the ground plane PIERS Proceedings, Beijing, China, March 23-27, 2009 [11]. C. Picher and J. Anguera, A. Cabedo, C. Puente, S. Kahng, Multiband handset antenna using slots on the ground plane, Progress In Electromagnetics Research Proceedings, Vol. 7, 95109, 2009 [12]. C. Lin and K. L. Wong, Internal hybrid antenna for multiband operation in the mobile phone, Microw. Opt. Tech. Lett., vol. 50, no. 1, pp. 3842, Jan. 2008. [13]. J A Ansari, Satya Kesh Dubey and Prabhakar Singh, R. U. Khan, Babau R. Vishvakarma Analysis of U-slot loaded patch for dual band operation, International Journal of Microwave and Optical Technology ,Vol. 3, No. 2, April 2008. [14]. S. Hong, W. Kim, H. Park, S. Kahng, and J. Choi, Design of an internal multiresonant monopole antenna for GSM900/DCS1800/USPCS/ S-DMB operation, IEEE Trans. Antennas Propag., vol. 56, no. 5, pp. 14371443, May 2008. [15]. S. Kahng, The rectangular power-bus with slits GA-optimized to damp resonances, IEEE Trans. Antennas Propag., vol. 55, no. 6, pp. 18921895, Jun. 2007. [16]. M. Cabedo, E. Antonino, A. Valero, and M. Ferrando, The theory of characteristic modes revisited: A contribution to the design of antennas for modern applications, IEEE Antennas Propag. Mag., vol. 49, no. 5, pp. 5268, Oct. 2007. [17]. Zhu, X. W., X. X. He, and J. Liu, Design of planar inverted-F antenna for mobile 3G applications," Mobile Communications, Vol. 31, 79-81, 2007. [18]. B. N. Kim, S. O. Park, Y. S. Yoon, J. K. Oh, K. J. Lee, and G. Y.Koo, Hexaband planar inverted-F antenna with novel feed structure for wireless terminals, IEEE Antennas Wireless Propag. Lett., vol. 6, pp. 6668, 2007. [19]. J. Anguera, I. Sanz, A. Sanz, A. Condes, D. Gala, C. Puente, and J.Soler, Enhancing the performance of handset antennas by means of groundplane design, presented at the IEEE Int.Workshop on Antenna Technology: Small Antennas and Novel Metamaterials (IWAT), New York, Mar. 2006. [20]. E. Antonino, C. A. Surez, M. Cabedo, and M. Ferrando, Wideband antenna for mobile terminals based on the handset PCB resonance,Microw. Opt. Technol. Lett., vol. 48, no. 7, pp. 14081411, Jul. 2006. [21]. H. Guodong, G. Changqing, "Study on Size Reduction of Quasi- Sierpinski Carpet Microstrip Antenna," Antennas, Propagation & EM Theory, 2006. ISAPE '06. 7th International Symposium on Oct. 2006. pp:1-4. [22]. M. A. Saed, Broadband CPW-Fed Planar Slot Antennas with various tuning stubs Electromagnetics Research, PIER 66, 199212, 2006 Progress In
[23]. Dimitrios Peroulis, kamal Sarabandi, Linda P. B. Katehi, Design of Reconfigurable slot Antennas, IEEE Transactions on Antennas and Propagation, Vol. 53, No.2, February 2005
109
Vol. 3, Issue 1, pp. 101-110
[24]. C. Di Nallo and A. Faraone, Multiband internal antenna for mobile phones, Electron. Lett., vol. 41, pp. 514515, 2005. [25]. S. Kumar, L. Shafai, and N. Jacob, Investigation of wide-band microstrip slot antenna, IEEE Trans. Antennas Propag., vol. 52, no. 3, pp. 865872, Mar. 2004. [26]. Hossa, R., A. Byndas, and M. E. Bialkowski, Improvement of compact internal antenna performance by incorporating open-end slots in ground plane," IEEE Microwave and Wireless Components Letters, Vol. 14, 283-285, 2004. [27]. Wang, F., Z. Du, Q. Wang, and K. Gong, Enhanced-bandwidth PIFA with T-shaped ground plane," Electronics Letters, Vol. 40, 1504-1505, 2004. [28]. Azadegan, R., Sarabandi, K., "A novel approach for miniaturization of slot antennas," Antennas and Propagation, IEEE Transactions on Vol. 51, Issue 3, March 2003 pp:421 - 429. [29]. Salonen, P., Effect of groundplane size on radiation efficiency and bandwidth of dual-band UPIFA, IEEE Antennas and Propagation Society International Symposium, Vol. 3, 7073, June 2003. [30]. M. F. Abedin and M. Ali, Modifying the ground plane and its effect on planar inverted-F antennas (PIFAs) for mobile phone handsets, IEEE Antennas Wireless Propag. Lett., vol. 2, pp. 226229, 2003. [31]. K. L. Wong, Planar Antennas for Wireless Communications. New York: John Wiley & Sons Inc., 2003. [32]. Vainikainen, P., J. Ollikainen, O. Kivekas, and I. Kelander,Resonator-based analysis of the combination of mobile handset antenna and chassis, IEEE Transactions on Antennas and Propagation, Vol. 50, No. 10, 14331444, October 2002. [33]. Pekka Salonen, Mikko Keskilammi, and Markku Kivikoski, Single-Feed Dual-Band Planar Inverted F Antenna with U-Shaped Slot, IEEE Transactions on Antennas and Propagation, Vol. 48, No. 8, August 2000. [34]. C. R. Rowell and R. D. Murch, A compact PIFA suitable for dual frequency 900/1800-MHz operation, IEEE Trans. Antennas Propag., vol. 46, no. 4, pp. 596598, Apr. 1998. [35]. Corbett R. Rowell and R. D. Murch, A Capacitively loaded PIFA for compact Mobile Telephone Handsets, IEEE Transactions on Antennas and Propagations, 1996.
Authors
T. Anita Jones Mary was born in India on 8th May 1976. She has received B.E degree in Electronics and Communication Engineering from Madurai Kamaraj University, in 1998. She has received M.E degree in Communication Systems from Madurai Kamaraj University, in 2000. Currently she is pursuing PhD degree in Design of MIMO antennas for wireless applications.
T. Joyce Selva Hephzibah was born in Tamilnadu, India on 21st July 1987. She has received B.E degree in Electronics and communication from Anna University, Chennai in 2008. Since July 2011, she has been a student of M.Tech in Communication systems in Karunya University, Coimbatore. Her research interest includes Design of PIFA with slots for wireless applications. C. S. Ravichandran was born in India on 16th March 1967. He has received B.E degree in Electrical and Electronics Engineering from Pondicherry University, Pondicherry in 1989. He has received M.E degree in Power System from Bharathiar University, Coimbatore in 1993. He has received Ph.D degree in Control System from Bharathiar University. He has published many technical papers in International and National Journals, and at National and International conferences. He has been approved as research guide by Anna University, Coimbatore. Currently he is guiding 12 Ph.D Scholars.
110
Vol. 3, Issue 1, pp. 101-110
A NOVEL DESIGN OF MULTIBAND SQUARE PATCH ANTENNA EMBEDED WITH GASKET FRACTAL SLOT FOR WLAN & WIMAX COMMUNICATION
Amit K. Panda1 and Asit K. Panda2
2
Department of ECE, Guru Ghashi Das Central University, Bilashpur, India Department of ECE, National Institute of Science & Technology, Berhampur, India
A BSTRACT
A compact multiband patch antenna embedded with gasket fractal slots is proposed in this paper. The structure consists of square patch element with modified gasket slots on both radiating edge side. The antenna is fed by 50 co-planar waveguide (CPW) to make the structure purely planar. The investigation took place ranges between 1-7.5 GHz using CST MWS electromagnetic simulator. There are 3 resonant frequencies appeared at 2.45GHz, 3.6GHz & 5.6 GHz. From the return loss plot it is seen the antenna achieved the IEEE Bluetooth / WLAN (2.4-2.484 GHz), WiMAX (3.4-3.69 GHz) & WIFI (5.1-5.825 GHz) frequency band with -10 dB return loss and also nearly omni-directional radiation patterns achieved. The peak realized antenna gain is around 5dB in all distinct bands.
K EYW ORDS: Self-similar, fractal, CPW, Sierpinski gasket, WiMAX, WLAN.
I.
INTRODUCTION
Due to rapid growth of wireless communication technology, the necessities to cover multiple applications with a single antenna element have been developing extensive research on it. Traditionally, each antenna operates at a single or dual frequency bands, where different antenna is needed for different applications. This will cause a limited space and place problem. In order to overcome this problem, multiband antenna can be used where a single antenna can operate at many frequency bands. So, this has initiated antenna research in various directions, one of which was using fractal shaped antenna element [1]. Fractal antenna [2] has very good features like small size & multiband characteristic. Most fractal objects have self -similar with different scaling and spacefilling geometrical properties [3-5]. The fractal shape carried out by applying the infinite number of iterations using multiple reduction copy machine (MRCM) algorithm [6].Till now, many different antennas have been designed using different configurations to design multiband antenna such as, sierpinski gasket [7], multiple ring [8], hexagonal fractal [9] and circular fractal slot antenna [10]. Electric Circuits are of three types 1D (e.g. transmission line), 2D (e.g. microstrip line) and 3D (e.g. waveguides). The patch and the ground are placed in the same plane (co-planar) which makes the design purely planar structure. In particular, a great interest in coplanar waveguide (CPW) fed antennas has been seen in the literature because of their many attractive features such as, simplest structure of a single metallic layer, no soldering point and easy integration with active devices or MMICs[11]. Here in this paper we have presented a square patch antenna embedded with a modified sierpinski gasket slot which exhibits a large reduction in size along with multiband operation. The antenna was fed by CPW-like matching section which suits for WLAN, WIMAX & WIFI applications. The design procedure of the proposed geometry was depicted in Fig. 1. The purposed fractal geometry was constructed using square patch element and gasket fractal slot. The initial geometry of antenna at 0th
111
Vol. 3, Issue 1, pp. 111-116
iteration has been taken a square patch element. Then at 1st iteration the square patch was etched with gasket fractal slot on both radiating edge side of patch. Similarly, third and fourth iteration were achieved with the reduced scale and overlapping portion was subtracted from the inner square metallization respectively. This process of dropping the central continues for nth iteration to generate purposed fractal structure. The iteration order was limited to 2nd iteration due to tolerance and complexity in fabrication. The scale factor will determined the high of each sub gasket and is given as:
hn 2 hn +1
(1)
(a)
(b)
(c)
Figure 1. Design procedure of purposed fractal structure geometry upto 2nd iteration
II.
ANTENNA CONFIGURATION
The proposed multiband antenna prototype was illustrated in Fig. 2. The entire dimensions of the antenna were 52mm 62 mm. The 50-SMA connector was used to feed the antenna at the CPW line. The design of the antenna starts with a single element using basic square patch microstrip antenna operating at frequency 1.8 GHz using an electromagnetic solver (CST MWS).The dimension of the antenna was determined from the equation of the patch antenna design equation. The antenna was designed up to 2nd iteration process. It was designed on FR-4 substrate with thickness of the substrate = 1.59 mm. (1/16), r = 4.21 & tan () = 0.019 respectively and where the radiating element was chosen as copper clad. The structure was fed using CPW Feed .The width (W) of the ground plane on either side of the CPW central strip was 19.5 mm and its length (Lg) is 28.5 mm. The spacing (g) between ground plane and central conductor is 0.5mm and the separation (h) between ground plane and patch was1 mm considered.
Figure 2. Square patch antenna embeded with gasket slots
The purposed multiband antenna structure was constructed using square and gasket slots on both radiating edges of square patch. For generation of gasket slot, it starts with a gasket patch element, and then the central Inverted gasket was removed with vertices that are located at the midpoints of the sides of the original Gasket. This process was repeated for the three remaining Sectoral until 2nd iteration in this particular case, so three scaled versions of the Sierpinski gasket were found on the antenna. The scale factor among the three gaskets was =2. Three scaled versions of the Sierpinski gasket, the first sub-gasket is 3rd order Sierpinski of height 13mm, the second sub-gasket was second order Sierpinski of height 26 mm and the third is 1st order Sierpinski of height 52 mm.
112
Vol. 3, Issue 1, pp. 111-116
III.
RESULT AND DISCUSSION
3.1 Return Loss Characteristics

The proposed antenna was simulated & analyzed using CST Microwave studio (CST MWS) between the frequencies 1-10 GHz. From the return loss plot as shown in Fig. 3, it was found that the antenna was matched in 3 resonant frequencies effectively appeared at 2.4 GHz, 3.61GHz, and 5.58 GHz respectively for 2nd iteration. The return losses in all 3 bands were quite good. The 1st impedance BW at 2.4 GHz was 105 MHz covering IEEE Bluetooth/WLAN band (2.2 2.56 GHz) with return loss 35 dB, the 2nd impedance BW at 3.6GHz was 112 MHz covering WiMAX band (3.35-3.50 GHz) with return loss -22 dB and the 3rd impedance BW was 150 MHz covering WIFI band (5.1-5.825 GHz) respectively.
Figure 3. Simulated return loss
3.2 Effect of Width of the CPW Feed (W1)

A parametric study was done by changing the width of CPW feed for better matching section. Fig. 4 depicts the simulated return loss curves for different CPW feed widths (W1=2.7, 2.925, 3.15, 3.375 and 3.6 mm).when Ws increased from 3 to 4 the impedance matching of the antenna gets better. This fact gives further indication that a better impedance matching can be obtained by optimizing the width of central conductor W1 =3.6mm. It was noticed that the resonant frequencies shift significantly for the five different W1. When W1 is narrowed, the resonant frequency decreased dramatically, leading to the variations of the operating bandwidth range of the antenna. When W1 was increasing, the resonant frequency as well as S11 increasing significantly. The return losses in all five bands are acceptable and all bandwidths are wider.
Figure 4. Simulated reurn loss curves for different width CPW feed
From the return loss plot it was seen that the optimized result for the antenna suited for IEEE Bluetooth/WLAN (2.4-2.484 GHz) & WIMAX (3.4-3.69 GHz) applications was attained at W1=3.6mm.
3.3 Gain vs. Frequency

The simulated peak gain of the proposed antenna is plotted in Fig. 5. It is seen that gain level was achieved throughout the band. As observed in fig.5, gain vs. frequency plot, it was found that the gain was around 4.2 dB at lower frequency band and were around 6 dB at higher band.
113
Vol. 3, Issue 1, pp. 111-116
Figure 5. Gain versus frquency
3.4 Current Distribution & Radiation Pattern

The current density and radiation patterns were analyzed using CST Microwave Studio. With a series of simulations it was seen that the magnetic current at the central gap & the electric current on the patch region of the antenna around the gap is crucial for resonance & radiation characteristics of such antenna. Simulation current density on the surface of the antenna at 2.4 GHz, 3.61GHz and 5.6 GHz were shown in Fig.6. It was observed as the number iterations was increased, multiple resonant frequencies obtained but the radiated power from the antenna was found to deteriorate
(a)
(b)
(c)
Figure 6. Current density distribution on antenna surface at (a) 2.4 GHz, (b) 3.61 GHz , (c) 5.6 GHz
114
Vol. 3, Issue 1, pp. 111-116
The simulated normalized radiation patterns at all distinct frequencies for = 00 &900 were shown in Fig. 7. It was observed that the H-plane patterns were reasonable over the entire operating bandwidth. The radiation patterns were consistent for the different resonant frequencies for 2nd iteration.
(a) E- for =900 at 2.4 GHz
(b) E- for =900 at 3.61 GHz
(c)E- for =900 at 5.6 GHz Figure 7. Simulated Normalized radiation patterns of purposed patch antenna for = 900 at (a) 2.4 GHz, (b) 3.61GHz (c) 5.6GHz
An interesting phenomenon was phenomenon is observed that with increasing resonant frequencies the patterns show more undulations. The pattern for lower bands are more omni directional & the cross polar isolation are very minimum but there are side lobes existing at higher frequency.
IV.
CONCLUSION
A novel compact CPW fed compact square patch antenna embedded with Gasket fractal slot was designed & simulated for multiband operations. The simulated results indicate that the antenna exhibits a good return loss, and the antenna gain was above 5 dB at the designed frequencies and other multiband frequencies suitable for IEEE Bluetooth/WLAN (2.4-2.484 GHz), WiMAX (3.4-3.69 GHz) & WIFI (5.1-5.825 GHz) wireless communication applications The design was implemented by using CST MWS electromagnetic simulation tools. The self similarity in the structure for the 2nd iteration leads to multiband operation of the antenna. The key parameters that influence antenna performance have been analyzed to gain an insight into antenna operation. Hence a good antenna performance over the operating frequencies over the whole band was obtained.
115
Vol. 3, Issue 1, pp. 111-116
ACKNOWLEDGMENT
The authors would like to thank, CST Company, India for their support in CST EM tool. The authors are grateful to the anonymous reviewers for their constructive & helpful comments & suggestions.
REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
[11]
B. B. Mandelbrot, The Fractal Geometry CfNature, Freeman, 1983. H.Jones, et al, Fractals and chaos, A.J.Crilly, R.A.Earnsshaw, and H.Jones, Eds. , Springer-Verleg., Newyork , 1990. C. Puente, J. Romeu, R. Pous, A. Cardama, On the behavior of the Sierpinski multiband antenna, IEEE Trans. Antennas Propagat., vol. 46, pp. 517-524, Apr. 1998. Werner D.H., Mittra R, Frontier of electromagnetic, Wiley-IEEE Press, Newyork, 1999. D. H. Werner, S. Ganguly, "An overview of Fractal Antenna Engineering Research", IEEE Antennas and Propagation Magazine, vol. 45, pp.38-57, 2003. H.O.Peitgen., et al, Chaos and Fractals, A.J.Crilly, R.A.Earnsshaw,and H.Jones, Eds. , SpringerVerleg., Newyork , 1990. C. Borja and J. Romeu, Multiband Sierpinski fractal patch antenna, Antennas and Propagation Society International Symposium, IEEE, vol. 3, pp. 1708-1711, July 2000. C. T.P. Song, Peter S. Hall and H. Ghafouri-Shiraz, Multiband multiple ring monopole antennas, IEEE Transactions Antenna & Propagation, vol. 51, pp. 722-729, Apr 2003. Kan Philip Tang and Parveen Wahid, Hexagonal Fractal Multiband Antenna, Antennas and Propagation Society International Symposium, IEEE, vol. 4, pp. 554-557, June 2002. Ji-Chyun Liu, Der-Chyuan Lou, Chin-Yen Liu, Ching-Yang Wu and Tai-Wei Soong, Precise Determinations of the CPW-FED Circular fractal slot antenna, Microwave and Optical Technology Letters, vol. 48, pp.1586-1592, Aug 2006. Ip, K.H.Y., Kan, T.M.Y., and Eleftheriades, G.V.: A single-layer cpw-fed active patch antenna, IEEE Microw. Guide. Wave Lett., 2000, 10, pp. 6466
Authors Biography
Amit K. Panda received his M.Sc in Electronics from Berhampur university& the Master of technology degree in Electronics design technology from Tezpur cental University. He is currently working as Asst. professor with the Department of electronics and communication Engineering, Guru Ghasidas Central University, Bilaspur.His main Research interests are in FPGA based system design, VLSI digital design, Network implementation on FPGA, RF and microwave control devices and semiconductor components. Amit K. Panda is a faculty in ECE department at National institute of science & Technology (NIST), Berhampur, India. He obtained his M.Tech in ECE from NIST under Biju patnaik Technical University (BPUT) in 2009, completed his B.Tech in ECE in 2003 from NIST,.Currently continuing his Ph.D work on metamaterial. His current area of interest are DNG material, metamaterial antenna, invisible cloaking, Multiband & wideband patch antenna, fractal antenna. He has presented 8 international conference papers.
116
Vol. 3, Issue 1, pp. 111-116
IMPLEMENTATION OF THE TRIPLE-DES BLOCK CIPHER USING VHDL

Sai Praveen Venigalla, M. Nagesh Babu, Srinivas Boddu, G. Santhi Swaroop Vemana
Department of Electronics & Communications Engineering, KL University, Vijayawada, A.P., India
ABSTRACT
This paper presents FPGA implementations of the DES and Triple-DES with improved security against power analysis attacks. The proposed designs use Boolean masking, a previously introduced technique to protect smart card implementations from these attacks. Triple DES was the answer to many of the shortcomings of DES. Since it is based on the DES algorithm, it is very easy to modify existing software to use Triple DES. It also has the advantage of proven reliability and a longer key length that eliminates many of the shortcut attacks that can be used to reduce the amount of time it takes to break DES. However, even this more powerful version of DES may not be strong enough to protect data for very much longer. The DES algorithm itself has become obsolete and is in need of replacement.DES encrypts data in 64-bit and it is a symmetric algorithm. The key length is 56-bits.
KEYWORDS: DES, Encryption, Decryption, Cryptography, Simulation, Synthesis, TDES, Cipher.
I.
INTRODUCTION
DES (the Data Encryption Standard) is a symmetric block cipher developed by IBM. The algorithm uses a 56-bit key to encipher/decipher a 64-bit block of data. The key is always presented as a 64-bit block, every 8th bit of which is ignored. However, it is usual to set each 8th bit so that each group of 8 bits has an odd number of bits set to 1.[1] The algorithm is best suited to implementation in hardware, probably to discourage implementations in software, which tend to be slow by comparison. However, modern computers are so fast that satisfactory software implementations are readily available. DES is the most widely used symmetric algorithm in the world, despite claims that the key length is too short. Ever since DES was first announced, controversy has raged about whether 56 bits is long enough to guarantee security. The key length argument goes like this. Assuming that the only feasible attack on DES is to try each key in turn until the right one is found, then 1,000,000 machines each capable of testing 1,000,000 keys per second would find (on average) one key every 12 hours. Most reasonable people might find this rather comforting and a good measure of the strength of the algorithm[9]. Those who consider the exhaustive key-search attack to be a real possibility (and to be fair the technology to do such a search is becoming a reality) can overcome the problem by using double or triple length keys. In fact, double length keys have been recommended for the financial industry for many years. Section II discusses about Triple DES, Section III discusses about the algorithm of Triple DES, Section IV represents Simulation Results, Section V describes Conclusion and finally section VI describes Scope and Future Development.
II.
TRIPLE DES
Use of multiple length keys leads us to the Triple-DES algorithm, in which DES is applied three times. Triple DES is simply another mode of DES operation. It takes three 64-bit keys, for an overall key length of 192 bits. In Private Encryption, you simply type in the entire 192-bit (24 character) key rather than entering each of the three keys individually[4]. The Triple DES DLL then breaks the user provided key into three sub keys, padding the keys if necessary so they are each 64 bits long. The procedure for encryption is exactly the same as regular DES, but it is repeated three times. Hence the
117
Vol. 3, Issue 1, pp. 117-128
name Triple DES, The data is encrypted with the first key, decrypted with the second key, and finally encrypted again with the third key. Triple DES, also known as 3DES. Consequently, Triple DES runs three times slower than standard DES, but is much more secure if used properly. The procedure for decrypting something is the same as the procedure for encryption, except it is executed in reverse[2]. Like DES, data is encrypted and decrypted in 64-bit chunks. Unfortunately, there are some weak keys that one should be aware of: if all three keys, the first and second keys, or the second and third keys are the same, then the encryption procedure is essentially the same as standard DES. This situation is to be avoided because it is the same as using a really slow version of regular DES[4]. Note that although the input key for DES is 64 bits long, the actual key used by DES is only 56 bits in length. The least significant (right-most) bit in each byte is a parity bit, and should be set so that there are always an odd number of 1s in every byte. These parity bits are ignored, so only the seven most significant bits of each byte are used, resulting in a key length of 56 bits. This means that the effective key strength for Triple DES is actually 168 bits because each of the three keys contains 8 parity bits that are not used during the encryption process.
Figure 1. Triple DES-Block Diagram
If we consider a triple length key to consist of three 56-bit keys K1, K2, K3 then encryption is as follows: Encrypt with K1 Decrypt with K2 Encrypt with K Decryption is the reverse process: Decrypt with K3 Encrypt with K2 Decrypt with K1 Setting K3 equal to K1 in these processes gives us a double length key K1, K2. Setting K1, K2 and K3 all equal to K has the same effect as using a single-length (56-bit key). Thus it is possible for a system using triple-DES to be compatible with a system using single-DES. DES operates on a 64 bit block of plaintext[3]. After an initial permutation the block is broken into a right half and left half, each 32 bits long. Then there are 16 rounds of identical operations, called Function f, in which the data are combined with the key. After the sixteenth round, the right and left halves are joined, and a final permutation (the inverse of the initial permutation) finishes off the algorithm. DES operates on a 64 bit block of plaintext. After an initial permutation the block is broken into a right half and left half, each 32 bits long. Then there are 16 rounds of identical operations, called Function f, in which the data are combined with the key. After the sixteenth round, the right and left halves are joined, and a final permutation (the inverse of the initial permutation) finishes off the algorithm. In each round the key bits are shifted, and then 48 bits are selected from the 56 bits of the key. The right half of the data is expanded to 48 bits via an expansion permutation, combined with 48 bits of a shifted and permuted key via an XOR, sent through 8 S- boxes producing 32- new bits, and permuted again.
118
Vol. 3, Issue 1, pp. 117-128
These four operations make up Function f. The output of Function f is then combined with the left half via another XOR. The results of these operations become the new right half; the old right half becomes the new left half. These operations are repeated sixteen times, making 16 rounds of DES.
Figure 2. Enciphering computation
In each round the key bits are shifted, and then 48 bits are selected from the 56 bits of the key. The right half of the data is expanded to 48 bits via an expansion permutation, combined with 48 bits of a shifted and permuted key via an XOR, sent through 8 S- boxes producing 32- new bits, and permuted again. These four operations make up Function f. The output of Function f is then combined with the left half via another XOR. The results of these operations become the new right half; the old right half becomes the new left half. These operations are repeated sixteen times, making 16 rounds of DES.
III.
ALGORITHM FOR TDES
3.1. Encryption
Step1: k1, K2, k3 are the keys in key expander with the selection function. Step2: If selection function is active i.e. 1 then encryption process is activated with key k1. And this encryption output is given to input of the decryption i.e. selection function is 0 with key K2. Step3: Decryption output is given to input of encryption i.e. if selection function is 1 with k3.
3.2. Decryption
Step4: It is the reverse process of encryption.
119
Vol. 3, Issue 1, pp. 117-128
Figure 3. TDES Algorithm
Figure 4. Single Round of DES
3.3. Initial permutation (IP)
Figure 5. Initial permutation
120
Vol. 3, Issue 1, pp. 117-128
Table 1. Initial permutation IP 58 60 62 64 57 59 61 63 50 52 54 56 49 51 53 55 42 44 46 48 41 43 45 47 34 36 38 40 33 35 37 39 26 28 30 32 25 27 29 31 18 20 22 24 17 19 21 23 10 12 14 16 9 11 13 15 2 4 6 8 1 3 5 7
Table 1 specifies the input permutation on a 64-bit block. The meaning is as follows: the first bit of the output is taken from the 58th bit of the input; the second bit from the 50th bit, and so on, with the last bit of the output taken from the 7th bit of the input. The initial permutation occurs before round one; it transposes the input block as described in table 1 this table, like all the other tables in this chapter , should be read left to right, top to bottom. For example, the initial permutation moves bit 58 of the plaintext to bit position 1, bit 50 to bit position 2, and so forth. The initial permutation and the corresponding final permutation do not affect DESs security.
3.4. Final permutation (IP-1)
Figure 6. Final permutation Table 2. Final permutation IP-1 40 39 38 37 36 35 34 33 8 48 7 47 6 46 5 45 4 44 3 43 2 42 1 41 16 15 14 13 12 11 10 9 56 55 54 53 52 51 50 49 24 23 22 21 20 19 18 17 64 63 62 61 60 59 58 57 32 31 30 29 28 27 26 25
121
Vol. 3, Issue 1, pp. 117-128
The final permutation is the inverse of the initial permutation; the table is interpreted similarly. This is shown in table 2
3.5. Expansion permutation (E)
Figure 7. Expansion permutation Table 3. Expansion permutation E 32 4 8 12 16 20 24 28 1 5 9 13 17 21 25 29 2 6 10 14 18 22 26 30 3 7 11 15 19 23 27 31 4 8 12 16 20 24 28 32 5 9 13 17 21 25 29 1
The expansion permutation is interpreted as for the initial and final permutations. Note that some bits from the input are duplicated at the output; e.g. the fifth bit of the input is duplicated in both the sixth and eighth bit of the output. Thus, the 32-bit half-block is expanded to 48 bits. This operation expands the right half of the data, RI, from 32-bits to 48 bits. Because this operation changes the order of the bits as well as repeating certain bits, it is known as an expansion permutation. This operation has two purposes: it makes the right half the same size as the key for the XOR operation and it provides a longer result that can be compressed during the substitution operation. However, neither of those is its main cryptographic purpose. By allowing one bit to affect two substitutions, the dependency of the output bits on the input bits spreads faster. This is called an avalanche effect. This is shown in table 3
3.6. Permutation (P)
Figure 8. Permutation
122
Vol. 3, Issue 1, pp. 117-128
Table 4. permutation P 16 29 1 5 2 32 19 22 7 12 15 18 8 27 13 11 20 28 23 31 24 3 30 4 21 17 26 10 14 9 6 25
The 32 bit output of the S box substitution is permuted according to a P box. This permutation maps each input bit to an output position; no bits are used twice and no bits are ignored. This is called a straight permutation or just a permutation. This is shown in table 4.
3.7. Substitution boxes (S-boxes)
Figure 10. Calculation of f(R, k)
After the compressed key is XORed with expanded block, the 48 bit result moves to a substitution operation. The substitutions are performed by eight substitution boxes, or S-boxes. Each S box has a 6-bit input and a 4-bit output, and there are eight different S-boxes. The total memory requirements for the eight DES S-boxes are 256 bytes. The 48 bits are divided into eight 6-bit sub-blocks. Each separate block is operated on by a separate S-box: The first block is operated on by S-box 1; the second block is operated on by S-box 2, and so on.
3.8. Rotations in the key-schedule

Before the round subkey is selected, each half of the key schedule state is rotated left by a number of places. This table specifies the number of places rotated. Triple DES has two attractions that assure its widespread use over the next few years[6]. First, with its 168-bit key length, it overcomes the vulnerability to brute-force attack of DES. Second, the underlying encryption algorithm in Triple DES is the same as in DES. This algorithm has been subjected to more scrutiny than any other encryption algorithm over a longer period of time, and no effective cryptanalytic attack based on the algorithm rather than brute-force has been found[5]. Accordingly, there is a high level of confidence that 3DES is very resistant to cryptanalysis. If security were the only consideration, then 3DES would be an appropriate choice for a standardized encryption algorithm for decades to come.
123
Vol. 3, Issue 1, pp. 117-128
Figure 11. Key schedule calculation
Figure 12. Feistel Decryption Algorithm
3.9. DES Decryption

i) Use same function
124
Vol. 3, Issue 1, pp. 117-128
ii) Key is the key Used in reverse order (K1,, K16 becomes K16,, K1) Right circular shift of 0-2 bits 0 1 2 2 2 2 2 2 1 2 2 2 2 2 2 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 (1 1 2 2 2 2 2 2 1 2 2 2 2 2 2 1) With DES it is possible to use the same function to encrypt or decrypt a block. The only difference is that the keys must be used in the reversed order. That is , if the encryption keys for each round are K1,K2,K3,K16, then the decryption keys are K16, K15, K14, ,K1.The algorithm that generates the key used for each round is circular as well. The key shift is shown above.
3.10. Applications
The DES3 core can be utilized for a variety of encryption applications including: Secure File/Data transfer Electronic Funds Transfer Encrypted Storage Data Secure communications
3.11. Features
FIPS 46-3 Standard Compliant Encryption/Decryption performed in 48 cycles(ECB mode) Up to 168 bits of security For use in FPGA or ASIC designs Verilog IP Core
3.11.1. Non Pipelined Version

Small gate count shared DES
3.11.2. Pipelined Version

Pipelined for maximum performance Encryption/Decryption performed in 1 cycle (ECB mode) after an initial latency of 48 cycles
IV.
SIMULATED RESULTS
Figure 13. Waveform of DES Block
125
Vol. 3, Issue 1, pp. 117-128
Figure 14. Waveform of Add Key
Figure 15. Waveform of Add left
Figure 16. Waveform of Expansion Table
Figure 17. Waveform of Pbox
126
Vol. 3, Issue 1, pp. 117-128
Figure 18. Waveform of S1box
Figure 19. Waveform of Sbox
V.
CONCLUSIONS
As DES will run through 16 iterations to achieve its desired cipher text (final output).With Triple DES, it will Encrypt-Decrypt-Encrypt the block and a completely different output is generated with a final combination. Its said that the security is 192 bit encryption, but also argued that regardless of the keys, the security is only 168 bit. This debate is clearly beyond the scope of this article/writer. If you wish to participate with the scientists in their discussions, its your humility at stake. It's a safe but that Triple DES is exponentially stronger than the previous DES. After that, AES may supplant Triple DES as the default algorithm on most systems if it lives up to its expectations. But Triple DES will be kept around for compatibility reasons for many years after that. So the useful lifetime of Triple DES is far from over, even with the AES near completion.
VI.
SCOPE AND FUTURE DEVELOPMENT
For the foreseeable future Triple DES is an excellent and reliable choice for the security needs of highly sensitive information. The AES will be at least as strong as Triple DES and probably much faster. It's the industry mandate from Visa and MasterCard that's requiring ATM deployers to upgrade and/or replace their legacy terminals. In a nutshell, it's all about three waves of encryption, and it's designed to make ATM transactions more secure.
REFERENCES
[1]. Data Encryption Standard, Federal Information Processing Standard (FIPS) 46, National Bureau of Standards, 1977. [2]. Federal Information Processing Standards Publication 140-1, Security Requirements for Cryptographic Modules, U.S. Department of Commerce/NIST, Springfield, VA: NIST, 1994
127
Vol. 3, Issue 1, pp. 117-128
[3]. B. Schneier, Applied Cryptography, Protocols, Algorithms, and Source Code in C, John Wiley & Sons 1994. [4]. D. C. Feldmeier, P. R. Karn, UNIX Password Security Ten Years Later, CRYPTO89, Santa Barbara, California, USA, pp. 44-63, 1989. [5]. Xilinx, San Jose, California, USA, Virtex, 2.5 V Field Programmable Gate Arrays, 2001, www.xilinx.com. [6]. NIST Special Publications 800-20, "Modes of Operation Validation System for the Triple Data Encryption Algorithm", National Institute of Standard and Technology, 2000. [7]. Ohjun KWON, Hidenori SEIKE, Hirotsugu KAJISAKI and Takakazu KUROKAWA, Implementation of AES and Triple-DES cryptography using a PCI-based FPGA board, Proc. of the International Technical Conference On Circuits/Systems Computers and Communications 2002, ITCCSCC-2002, Phuket, Thailand, July 16-19, 2002. [8]. Pawel Chodowiec, Kris Gaj, Peter Bellows, and Brian Schott, Experimental Testing of the Gigabit IPSec- Compliant Implementations of Rijndael and Triple DES Using SLAAC-1V FPGA Accelerator Board, Proc. Information Security Conference, Malaga, Spain, October 1-3, 2001, pp. 220-234. [9]. Herbert Leitold, Wolfgang Mayerwieser, Udo Payer, Karl Christian Posch, Reinhard Posch, and Johannes Wolkerstorfer, A 155 Mbps Triple-DES Network Encryptor, Proc. Cryptographic Hardware and Embedded Systems - CHES 2000, USA, August 17-18, 2000. Authors
Sai Praveen Venigalla was born in A.P, India. He received the B.Tech degree in Electronics & communications Engineering from Jawaharlal Nehru Technological University in 2009. Presently he is pursuing M.Tech VLSI Design in KL University. His research interests include FPGA Implementation, Low Power Design. M. Nagesh Babu was born in Kurnool, Kurnool (Dist.), AP, India. He received B.Tech in Electronics & Communication Engineering from JNTU Anantapur, AP, India and M.Tech from Hyderabad Institute of Technology and Management, R.R (Dist), AP, India. He is working as Associate Professor in Department of Electronics & Communication Engineering, K L University, Vijayawada, AP, India. He has 9 years of Industry experience and 9 years of teaching experience. He presented 2 papers in National conferences. Srinivas Boddu was born in A.P, India. He received the B.Tech degree in Electronics & communications Engineering from Jawaharlal Nehru Technological University in 2009. Presently he is pursuing M.Tech VLSI Design in KL University. His research interests include FPGA Implementation, Low Power Design.
G Santhi Swaroop Vemana was born in A.P, India. He received B.Tech degree in Electronics and Communication Engineering from Jawaharlal Nehru technological university in 2008. He worked as OFC engineer at united telecoms ltd at GOA during 2009-2010. Presently he is pursuing M.Tech VLSI Design in KL University. His research interests include FPGA Implementation, Low Power Design.
128
Vol. 3, Issue 1, pp. 117-128
COMPUTATIONAL INVESTIGATION OF PERFORMANCE CHARACTERISTICS IN A C-SHAPE DIFFUSING DUCT

Prasanta. K. Sinha1, A. N. Mullick2, B. Halder3 and B. Majumdar4
1 2,3
Department of Mechanical Engineering, KATM, Durgapur-713212, India Department of Mechanical Engineering, NIT, Durgapur 713 209, India 4 Department of Power Engineering, Jadavpur University, Kolkata, India
ABSTRACT
In the present investigation the distribution of mean velocity, static pressure and total pressure are experimentally studied on a C-shape diffuser of 45 angle of turn with an area ratio of 1.298 and centerline length was chosen as three times of inlet diameter. The experimental results then were numerically validated with the help of Fluent and then a series of parametric investigations are conducted with same centre line length and inlet diameter but with different area ratios varying from 1.25 to 3.75. The measurements were taken at Reynolds number 2.45 x 105 based on inlet diameter and mass average inlet velocity. Predicted results of coefficient of mass averaged static pressure recovery (52%) and coefficient of mass averaged total pressure loss (11%) are in good agreement with the experimental results of coefficient of mass averaged static pressure recovery (48%) and coefficient of mass averaged total pressure loss (10%) respectively. Standard k- model in Fluent solver was chosen for validation. From the parametric investigation it is observed that for the increase in area ratio static pressure recovery increases sharply upto area ratio 2 after that increment in lesser gradient up to 2.67 and it was maximum at area ratio 2.67. But when area ratio increases from 2.67 to 3.75, pressure recovery decreases steadily. The coefficient of total pressure loss almost remains constant with the change in area ratio and angle of turn for similar inlet conditions.
KEYWORDS:
Ar As CC Cpr CV
C-shape diffuser, Five-hole probe, Fluent solver and k- model.
NOMENCLATURE
Area ratio Aspect ratio Concave or inward wall Coefficient of pressure recovery Convex or outward wall D L Re Inlet diameter of the Diffuser Centerline length of the Diffuser Reynolds number Angle of turn of the center line
I.
INTRODUCTION
Diffusers are used in many engineering application to decelerate the flow or to convert the dynamic pressure into static pressure. Depending on application, they have been designed in many different shapes and sizes. The C-shape diffuser is one of such design and is an essential component in many fluid handling systems. C-shape diffusers are an integral component of the gas turbine engines of high-speed aircraft. It facilitates effective operation of the combustor by reducing the total pressure loss. The performance characteristics of these diffusers depend on their geometry and the inlet conditions. Part turn or curved diffusers are used in wind tunnels, compressor crossover, air conditioning and ventilation ducting systems, plumes, draft tubes, etc.
129
Vol. 3, Issue 1, pp. 129-136
The objective of the present study is to investigate the flow characteristics within a circular C-shaped diffuser. The research work on curved diffuser was initiated from the study of curved duct. The earliest experimental work on curved ducts has been reported by Williams et al. [1] in the beginning of the last century. It was reported that the location of the maximum axial velocity is shifted towards the outward wall of the curved pipe. The effect of the centrifugal force on the pressure gradient was studied by Dean [2, 3]. He established a relation between viscous force, inertia force and the curvature by a non dimensional number known as Dean Number. Experimental investigation on circular 90 and 180 turn curved ducts was carried out by Rowe et al. [4] and reported the generation of contra rotating vortices within the bends. Enayet et al. [5] investigated the turbulent flow characteristics through 90 circular curved duct of curvature ratio 2.8. It was observed that the thickness of the inlet boundary layer has a significant role on the generation of secondary motion within the duct. Kim and Patel [6] have investigated on a 90 curved duct of rectangular cross section. It was reported that the formation of vortices on the inner wall due to the pressure driven secondary motion originated in the corner region of curved duct. The earliest work on curved diffuser was reported by Stanitz [7]. The diffuser was designed based on potential flow solution for two-dimensional, invisid, incompressible and irrotational flow. The evaluation of the performance of these diffusers was on the basis of wall static pressure only. The first systematic studies on 2-D curved subsonic diffusers were carried out by Fox & Kline [8]. The centerline of the diffuser was taken as circular with a linearly varying area distribution normal to the centerline. They established a complete map of flow over a range of the L/D ratio and at different values of . Parson and Hill [9] investigated on three 2-D curved diffusers of As = 10 of various combination of ratio between the centerline length and inlet width. They observed that the streamline curvature affects the flow development substantially within the curved diffuser. A qualitative measurement of the mean flow quantities in a 40 curved diffuser of rectangular cross section of Ar = 1.32 and inlet As = 1.5 have been reported by McMillan [10]. The result clearly showed the development of strong counter rotating vortices between two parallel walls, which dominate the flow and performance characteristics. Majumdar et al. [11] experimentally studied the flow characteristics in a large area ratio curved diffuser with splitter vanes installed at different angles to the flow at the inlet of the diffuser. It was observed that splitter vanes deflect the flow towards the convex wall and a pair of contra rotating vertices generated at the flow passage. Yaras [12] experimentally investigated the flow characteristics of 90 curved diffuser with strong curvature having Ar =3.42 for different values of inlet boundary layer thickness and turbulence intensity. Measurements were taken by the help of seven-hole pressure probe. He observed that the performance parameters were almost independent by the variations in the inlet boundary layer. Majumdar et al. [13] experimentally studied the turbulent characteristics in a curved diffuser. They observed that the stream wise bulk flow shifted towards the outward wall in the downstream of the diffuser, which was mainly due to the influence of centrifugal force. Moreover, one pair of contra-rotating vortices was identified at 30 turns in the flow passage. The overall static pressure recovery was observed as 5l%. Majumdar et al. [14] conducted an experiment on 180 bend rectangular diffusing duct. They measured the wall pressure, velocity and turbulence intensity along the flow passage of the diffusing duct. The observation clearly showed the formation of vortical motions between the two parallel walls. The overall pressure recovery was found about 48% Sinha et al. [15] conducted an experiment on 30 curved annular diffuser. They measured the mean velocity, static pressure and total pressure along the flow passage of the diffuser. They are also conducted a series of parametric investigations with same centre line length and inlet diameter but with different area ratios. They observed that the high velocity fluids shifted and accumulated at the concave wall of the exit section. They also observed that with the increase in area ratio pressure recovery increases upto certain point than with further increase in area ratio Pressure recovery decreases. Sinha et al. [16] investigated an experiment on 37.5 annular diffusing duct. They measured the mean velocity, static pressure and total pressure along the flow passage of the diffuser. They are also conducted a series of parametric investigations with same centre line length and inlet diameter but with varying area ratio and different angle of turn. They observed that the high velocity fluids shifted and accumulated at the concave wall of the exit section. They also observed that with the increase in area ratio pressure recovery increases but with the increase in angle of turn Pressure recovery decreases. Sinha et al. [17] investigated an experiment on 42 C-shape diffusing duct. They measured the mean velocity, static pressure and total pressure along the flow passage of the diffuser.
130
Vol. 3, Issue 1, pp.
They are also conducted a series of parametric investigations with same centre line length and inlet diameter but with varying area ratio and with 42 angle of turn. They observed that the high velocity fluids shifted and accumulated at the concave wall of the exit section. They also observed that with the increase in area ratio pressure recovery increases upto certain point than with further increase in area ratio Pressure recovery decreases.
II.
EXPERIMENTAL FACILITY
A test rig for the present investigation has been constructed at Fluid Mechanics & Machinery Laboratory of Power Engineering Department Jadavpur University to investigate the flow characteristics within a circular cross sectioned C-shape diffuser. The geometry of the test diffuser is shown in Fig. 1 with co-ordinate system and measurement locations. The entire set up was fabricated from mild steel sheet except the test diffuser. The test diffuser was designed with increase in area from inlet to exit and it distributed normal to the centerline as suggested by Fox and Kline [8]. The test diffuser was designed based on an area ratio of 1.298 and centerline length of 235 mm. The test diffuser is made of fiber glass reinforcement plastic. Centerline was turned at 45 from inlet to exit with inlet diameter of 78 mm. In order to avoid the pressure losses and flow distortion at the inlet and exit, two constant area connectors were attached at the inlet and exit of the test diffuser. A precalibrated five-hole pressure probe was used to obtain detailed flow parameters like mean velocity and its components, total and static pressure and secondary motions along the entire length of the diffuser. Ambient air was used as working fluid. For measuring mean velocity and its components and static and total pressure surveys along the entire cross section of curved diffuser, the test piece was divided into four planes, one at the Inlet section one diameter upstream of the test diffuser, two planes, Section A and Section B at 15 and 30 turn along the length of the diffusing passages and the fourth plane, Section C is at the mid point of the exit duct. The details of measured planes are shown in Fig. 1. For measurement of flow parameters the five hole pressure probe was inserted through a 8 mm drilled hole provided at four locations, namely , 0, 45, 90, and 315 angle as shown detail in Fig.1.
Figure 1: Geometry of test diffuser and measuring locations
The pre-calibrated five hole pressure probes was mounted in a traversing mechanism and the probe inserted into the flow field through 8 mm diameter drilled hole provided at the wall. The probe was placed within 1 mm of solid surface for the first reading. The probe was then moved radially and placed at the desired location as shown in Fig. 1. Instrumentation for the present study was chosen such that the experimental errors are minimum and also to have quick response to the flow parameters. The pre-calibrated hemispherical tip five-hole pressure probe used for the present study. The probe was calibrated and using non null technique was used to measure the flow parameter. All the five sensing ports of the probe were connected to a variable inclined multi tube manometer. The readings were recorded with respect to atmospheric pressure. The mean velocity and components of mean velocity distribution have been drawn with the help of SURFER software The assessment of errors resulting from the readings of the present five hole pressure probe was made as a function of all incidence angles for all flow characteristics in all the probe sectors and discussed in details[18], [19].
131
III.
The flow characteristics have been evaluated by mass average mean velocity, between the curved walls, total pressure and static pressure of the flow at various cross sections. Measured flow quantities have been presented in the form of 2-D profiles. All the velocities and pressures were normalized with respect to the inlet mass average velocity and inlet dynamic pressure respectively.
3.1
Mean velocity contour
The normalized mean velocity distribution in the form of contour plots at various sections of the curved diffuser has been discussed here and shown in Fig.2. The mean velocity contour at Inlet Section is shown in Fig.2(a) and it indicates that the flow is nearly symmetrical in nature throughout the entire cross-sectional area. The mean velocity contour of Section A, shown in Fig.2(b), indicates the overall diffusion of velocity particularly near the concave wall of this section and also reveals the shifting of the high velocity fluid close to the convex wall. These are mainly due to combined effect of the inertia force and centrifugal action on the flow.The mean velocity distribution in Section B is shown in Fig.2(c). The figure shows that the overall diffusion takes place at this section. The flow pattern at this section has changed compared to the preceding section. An appreciable diffusion across the entire section is observed from this figure. It is also observed that the high velocity core has occupied major part of the central area of the cross section and this is mainly due to the tendency of the flow to move towards concave wall due to inertia. As a result, low velocity fluid is pushed towards the convex wall indicating the accumulation of low momentum fluid at this section.
(a) Inlet section
(b) Section A (c) Section B Figure 2: Mean velocity contour
(d) Section C
The mean velocity distribution of Section C is shown in Fig.2(d). The figure depicts that the high velocity core is shifted towards concave wall and it occupies a substantial portion of the cross sectional area. The low velocity fluid close to the convex wall, as compared with the observation at Section B, occupies more area at Section C. Comparing the velocity distribution lines about the mid horizontal plane, it can be observed that the flow is symmetrical between top and bottom surfaces; it indicates the development of similar types of flow about the vertical plane. In a closed conduit, this phenomenon is only possible if the directions of flow in the two halves are opposite in nature (counter rotating flows), which is a natural phenomenon of flow through curved duct of any cross-section. Over an above it, increase in the cross-sectional area has further complicated the flow development.
3.2
Pressure Recovery & loss Coefficient
The variation of normalized coefficients of mass averaged static pressure recovery and total pressure loss based on the average static and total pressures at different sections of C-shape diffuser are shown in Fig.3. The figure shows that the coefficient of pressure recovery increases continuously up to the Section A, subsequently the increase takes place with a lesser gradient up to the Section B, and then again increases in a steeper gradient. The overall coefficient of mass averaged static pressure recovery is nearly 48% for this test diffuser. The coefficient of mass averaged total pressure loss increases rapidly in the curved diffuser up to the Section A. Then it increases steadily and marginally up to Section C. The overall mean value of the coefficient of mass averaged total pressure loss is nearly 11% for this test diffuser.
132
Figure 3: Variation of mass average pressure recovery and loss coefficients
3.3
Numerical Validation
In the present study a preliminary investigation was carried out using different turbulence models available in FLUENT. Based on the Intensive investigation it was found that Standard k model of turbulence provides the best result and results obtained from computational analysis match both in qualitatively and quantitatively with the experimental results. It is to be noted here that the inlet profiles obtained during experiment are fed as an inlet condition during the validation with FLUENT. Some of the validation figures are shown in Fig. 4(a), Fig. 4(b) and Fig. 4(c) respectively. Experimental
(a) Section A
(b) Section B
(c) Section C and Section C obtained
Computational
Figure 4: Comparison of normalized velocity distribution at Section A, Section B through Computational and Experimental investigation.
All three figures indicate that the mass averaged mean velocity contours obtained by computational and experimental investigations, which shows a qualitative matching to each other. This could be due to the complicated nature of flow at those planes, which was not properly predicted by the process of computer simulation. The mean velocity distribution at the Section B and Section C are shown in Fig. 4(c) and Fig. 4(d) show a reasonably good agreement of the computational investigation with the experimental results. These agreements confirm that the CFD code using Standard k- model can predict the flow and performance characteristics reasonably well for similar geometries with similar boundary conditions. Figure 5 shows the comparison of performance parameters like coefficient of static pressure recovery and coefficient of total pressure loss obtained through experimental and computational investigation. From the figure it has been observed that coefficient of pressure recovery Cpr for the computational investigation was obtained as 52% compared to the experimental investigation, which obtained as 48%. Similarly the coefficient of pressure loss is obtained as 10% in
133
computation investigation compared to the 11% of experimental study. This shows very good matching of the predicted results with the experimental one. These agreements confirm that the CFD code using Standard k model can predict the flow and performance characteristics reasonably well for similar geometries with same boundary conditions
the similar observation is reported in the work by Dey et al [20].
Figure 5: Comparison of performance parameters obtained through computational and experimental investigation
3.4
Parametric investigation
To obtain a more insight of the performance parameters an intense parametric study of pressure recovery/loss coefficient for different area ratio diffusers with the angle turn 42 was carried out. For this purpose area ratios1.25, 1.5, 1.75, 2, 2.25, 2.75, 3, 3.25 and 3.5 with the angle of turns 45 Cshape diffusers have chosen. From this investigation it is observed from Fig.6 that for the increase in area ratio static pressure recovery increases sharply upto area ratio 2 after that increment in lesser gradient up to 2.67 and it was maximum at area ratio 2.67. But when area ratio increases from 2.67 to 3.75 pressure recovery decreases steadily. The coefficient of total pressure loss almost remains constant with the change in area ratio and angle of turn for similar inlet conditions.
Figure 6: Variation of mass average pressure recovery and loss coefficients.
IV.
CONCLUSION
Based on the present investigation following conclusion have drawn for the present paper. High velocity fluids shifted and accumulated at the concave wall of the exit section. The mass average
134
static pressure recovery and total pressure loss for the curved test diffuser is continuous from Inlet section to Section C. Performance parameter like coefficient of mass average static pressure recovery and coefficient of mass average total pressure loss are 48% and 10% respectively. A comparison between the experimental and predicated results for the annular curved diffuser show good qualitative agreement between the two. The coefficient of mass averaged static pressure recovery and total pressure loss are obtained as 52% and 11% in predicted results and in the experimental results their values value obtained as 48% and 10% respectively, which indicate a good matching between the experimental and predicted results From the parametric investigation it is observed that for the increase in area ratio static pressure recovery increases sharply upto area ratio 2.0, after that increment in lesser gradient up to 2.67 and it was maximum at area ratio 2.67. But when area ratio increases from 2.67 to 3.75 pressure recovery decreases steadily. Among the different turbulence models within the fluent solver a standard k- model shows the good results and predicts the flow and performance characteristics well for C-shape diffusing ducts with uniform flow at inlet.
REFERENCE
G. S. Williams, C.W. Hubbell, and G.H. Fenkell, Experiments at Detroit, Mich, on the effect of curvature upon the flow of water pipes, Trans. ASCE, ,Vol.47, 1902, pp.1-196. [2] W. R. Dean, Note on the Motion of Fluid of Curved Pipe, Philosophical Magazine, Vol.20, pp.208-223. [3] 1927,W. R, Dean, The Streamline Motion of Fluid in a Curved Pipe, Philosophical Magazine, Vol.30, 1928, pp.673-693. [4] M. Rowe, Measurement and computations of flow in pipe bends, Journal of Fluid Mechanics, 1970, Vol.43, pp.771-783. [5] M. M. Enayet, M.M. Gibson, A.M.K.P. Taylor, and M. Yianneskis, Laser Doppler measurements of Laminar and Turbulent Flow in a Bend, Int. Journal of Heat and Fluid Flow, 1982, Vol. 3, pp.211-217. [6] W.J. Kim, and V.C. Patel, Origin and Decay of Longitudinal Vortices in the Development of flow in a Curved Rectangular Duct, Trans ASME, Journal of Fluid Engineering, 1994 Vol116 Vol.116, pp.45-52 [7] J. D. Stanitz, Design of Two Dimensional Channels with Prescribed Velocity Distribution along the Channel Walls, NACA Report No. 115, 1953 [8] R.W. Fox, and S.J. Kline, Flow Regimes in Curves Subsonic Diffuser, Trans ASME, Journal of Basic Engineering, 1962,Vol. 84, pp. 303 316 [9] ] D.J. Parson and P.G. Hill, Effect of Curvature on Two Dimensional Diffusing flow, Trans. ASME, Journal of fluid engineering, 1973, Vol. 95. pp. 112. [10] O. J, Mcmillan, Mean Flow Measurements of the Flow Diffusing, NASA Contractor Report 3634, 1982. [11] B. Majumdar, S.N. Singh, and D.P. Agrawal, Flow characteristics in a large area curved diffuser, Proc. I Mech E, Journal of Aerospace Engineering, 1996, Vol. 210, pp. 65 75. [12] M. I. Yaras, Effects of Inlets Conditions in the Flow in a Fishtail Diffuser with Strong Curvature, Trans. ASME, Journal of Fluid Engineering, 1996, Vol. 118, pp 772 778. [13] B., Majumdar, R. Mohan, S.N. Singh, and D.P. Agrawal, Experimental study of Flow in a high Aspect ratio 90 Curved Diffuser, Trans. ASME, Journal of Fluids Engineering, 1998,Vol. 120, pp. 83 89. [14] B. Majumdar, S. N. Singh, and D.P. Agrawal, Flow Structure in a 180 Curved Diffusing Ducts, The Arabian journal of science and technology, 1999, Vol. 24, pp. 79 87. [15] P. K. Sinha, A. N. Mullick, B. Halder and B. Majumdar, A Parametric Investigation of Flow through an Annular Curved Diffuser, Journal of International Review of Aerospace Engineering, 2010,Vol. 3. No. 5, pp. 249-256. [16] P. K. Sinha, A. K. Das and B. Majumdar, Numerical Investigation of flow through Annular Curved Diffusing duct, International Journal of Engineering & Technology IJET-IJENS, 2011, Vol..03, No.03, pp.190-196. [17] P. K. Sinha., A. N. Mullick, B. Halder and B. Majumdar, Computational Analysis of Flow performance in a C- shape Subsonic Diffuser, International Journal of Engineering Research and Applications, Vol..01, Issue 3, pp.824-829, 2011. [18] D. Chowdhoury, Modelling and calibration of pressure probes, M.E Thesis, Jadavpur University, 2007. [19] S. Mukhopadhya, A. Dutta A. N. Mullick and B. Majumdar, Effect of Five-hole probe tip shape on its calibration, Journal of the Aeronautical Society of India, 2001, Vol.53, No.4, pp.271279. [20] R.K., Dey, S., Bharani, S. Singh, and V Sheshadri, Flow Analysis in an S-shaped Diffuser with Circular Cross-Section, Arabian Journal of Science and Engineering, Vol.27, 2002, pp. 197-206. [1]
135
Authors Biographies
P. K. Sinha received the B. Tech (Mechanical) and M.Tech. (Mechanical Designs of Machines) degrees both from N.I.T. Durgapur, India in 1984 & 1991 respectively. He is obtained his Ph.D from Jadavpur University in 2011. His research interests are Experimental Fluid Dynamics and Aerodynamics, Machine Design and Non conventional Energy. He has published thirty seven papers. Presently he is working in Department of Mechanical Engineering, Techno India GroupKanksa Academy of Technology & Management, Durgapur. He is a Fellow of Institution of Engineers (India) and Life member of fluid Mechanics and Fluid Power. A. N. Mullick received the B.E. (Mechanical) and M.E. (Mechanical) degrees both from Jadavpur University, India in 1979 & 2000 respectively. He is obtained his Ph.D from Jadavpur University in 2007. His research interests are Experimental Fluid Dynamics and Aerodynamics. He has published thirty eight papers. Presently he is working in Department of Mechanical Engineering, N.I.T. Durgapur, India. He is a Life Member of ASME, IEEE, IE(India), AeSI, LMFMFP, LMAMM B. Halder received the B. Tech (Mechanical) and M.Tech. (Mechanical) degrees both from N.I.T. Durgapur, India in 1982 & 1984 respectively. He is obtained his Ph.D from Indian Institute of Technology, Kharagpur in 1993. His research interests are Experimental Fluid Dynamics and Aerodynamics. He has published twenty four papers. Presently he is working in Department of Mechanical Engineering, N.I.T. Durgapur, India. He is a Member of Fellow IIM, Fellow IIE, LMISTE B. Majumdar received the B.E. (Mechanical) and M.E. (Mechanical) degrees both from Jadavpur University, India in 1981 & 1983 respectively. He is obtained his Ph.D from Indian Institute of Technology, Delhi in 1995. His research interests are Experimental Fluid Dynamics and Aerodynamics, Hydro Power and wind power energy. He has published Seventy two papers. Presently he is working in Department of Power Engineering, Jadavpur University, Kolkata, India. He is a Life Member of Institution of Engineers (India).
136
GUIDELINES IN SELECTING A PROGRAMMING LANGUAGE AND A DATABASE MANAGEMENT SYSTEM

Onkar Dipak Joshi1, Virajit A. Gundale2, Sachin M. Jagdale3
Assistant Professor, CSE Department, Sharad Institute of Technology College of Engineering, Yadrav Dist. Kolhapur India. 2 Professor, Mechanical Engineering Department, Sharad Institute of Technology College of Engineering, Yadrav Dist. Kolhapur India. 3 Assistant Professor, CSE Department, Sharad Institute of Technology Polytechnic, Yadrav Dist. Kolhapur India.
1
A BSTRACT
Software development is a very complex process. When the project developer is planning to build a project for any organization in that case it is obvious question for developer to select suitable PL (Programming Language) and DBMS (Database Management System). It is a notable issue in project development because if developer fails to select suitable programming language or database management system then it badly affects on project development. This paper discusses what factors play important role for developers to select the programming language as well as database management system. In many cases some researchers point out only about the database management system or about programming language with no thumb rule at all. No published information whatsoever is available on this subject too. This paper gives a working and practical guideline for the developer in selecting upon the programming language and database management system as crucial factors for project development.
K EYW ORDS: Software development, DBMS, PL, developer, project development
I.
INTRODUCTION
For building any software project there is need to get perfect understanding of the user requirements and accordingly the users and developers need to plan the project. While planning a project there are some basic facts need to be addressed. Those are user requirements, programming language, database management system, tools etc. If the project developer wants to develop an excellent project then he has to concentrate on these above mentioned things because it plays a pivotal role in the project development. In the case of user requirements it is a lengthy discussion process between the users about his need for the project so as one can go deep into the interactions you may get good requirements from the user. Software testing tool is the testing part of the project that comes after project coding. The most concern part in the project development from initial stage is selecting an appropriate programming language and database management system. It is because after some amount of time if the programming language or database management system becomes obsolete then this will jeopardize the whole work. This paper introduces some guidelines for choosing proper programming language as well as database management system. Project development in the IT service industry is the crucial factor because the whole IT industry profit depends upon the projects success. Figure 1 shows major causes of IT project failure, in which 29% of IT project fail due the inadequate coordination of resources in the project where programming language and database management system is one the major resources in the project. While discussing the project development some issues need to be sorted out in the early stages because IT projects time duration is very long and hence if we do not settle the issues early in the development stage then it become unavoidable.
137
Vol. 3, Issue 1, pp. 137-144
Figure 1 Major causes of project failure
Due to this we need to reopen or restart the project once again and hence it is wastage of money, man power and resources. If the developer can find the technology in the initial stages of the project development then it can achieve better time saving, cost reduction and utilization of the resources in the project development. Without proper decision of programming language and database management system it can take lot of time to build project. This paper guides how to select programming language and database management system as per the applicability of the project in different scenarios. IT industry has capacity to provide the services to all the businesses. Providing service is the same as providing a software project which fulfills the requirement of the business or user. Project development is a very complex process in which thousands of employees are involved for creating excellent project. Software projects depend upon the Requirement Analysis, Design, Coding and Testing. Sometimes programming language is also called front end of the project and database management system is backend for the project. The brief discussion for selecting programming language for an IT project is as follows:
II.
SELECTION OF PROGRAMMING LANGUAGE
Selection of a particular language should be made based on overall goals of the project. Before starting an important project, it would be important to create several independent testing code programs in various languages. This will describe which one will be best for your task and that will go to save the time over selecting a particular language without unnecessary headache. By considering the user requirements the developer should consider the language for example if user need an immediate project within short amount of time then developer may choose RAD i.e. Rapid Application Development supported languages that will provide drag and drop facilities for the controls and easily available inbuilt functions that saves time of developer from developing controls and functions. If developer is novice in project development then he may go for simpler languages like VB whereas if the project is large enough then using JAVA, .NET, C++ etc. is common. Another approach is to divide project into separate modules or small projects with different languages for each module and selecting the one which may be best suited for the project. Some programming languages need high end configurations whereas other needs common configurations, it is very important to consider the users machine configuration while selecting language. For selection of particular language, the developer needs to consider the platform dependency, employees who know about the language, popularity of the language, and unanimity of the project team about language. The selected language must be easily understandable by user as well as the developer. It is expected to select the language which can contribute to save time, money, efforts and suitability in project.
III.
SELECTION OF DATABASE MANAGEMENT SYSTEM
Database management system (DBMS) is set of software which takes part in controlling the creation, maintenance, and organization of a users database. DBMS allows organizations to easily develop
138
Vol. 3, Issue 1, pp. 137-144
databases for various applications. A database is a mainly collection of data records, files, and other objects. It gives concurrent access to different user application programs to database at single time. It provides variety of database models, such as network model, hierarchical model, relational model and object model to conveniently support different applications. It generally supports query languages which are also work as high-level programming language. Database management system provides facilities for controlling data access, enforcing data integrity, managing concurrency control, and recovering the database has capacity to restore data from failures with the help of various backup techniques as well as maintaining database security. IT project requires database management system to store, retrieve and organize data from the database. There are some different types of database management system viz. 1. Hierarchical Database 2. Network Database 3. Relational Database 4. Object oriented Database There are number of database types available today; this paper describes some commonly known database types which are frequently used.
3.1. Hierarchical DatabaseHierarchical Databases are mostly used for mainframe computers. These are very traditional methods of managing and storing data. This kind of database is organized in pyramid fashion, where all the leafs are extending from the roots as shown in the Figure 2 where the library collection node is at the root position and all other nodes are extending from the root. The advantage of hierarchical databases that data can be accessed and updated rapidly because it consist tree-like structure and the relationships between records are maintaining in advance.
Figure 2 Hierarchical database
The main disadvantage of this type is that each child in the tree may have only one parent, due to which relationships between children are not permitted, even if they make sense from a logical point. These databases are so complex in their design that addition of new field or record requires that entire database be redefined.
3.2. Network Database

There are a considerable differences between hierarchical and network databases. In this database type, children of database are called members and parents are called owners. The most important difference is each child or member can have more than one parent as shown in Figure 3
139
Vol. 3, Issue 1, pp. 137-144
Figure 3 Network database
where edu, com, org are the parents and they have multiple childs from different domains. Network databases are considered more flexible than hierarchical database because it has more than one parent. Limitations must be considered when using this kind of database is network databases must be defined in advance and limitation to the number of connections that can be made between records.
3.3. Relational Databases

In Relational database, relationship between data files is relational. Relational database connect data in different files or tables by using common data elements or a key field like primary or candidate key. Data in relational database is stored in different tables, each table having a key field that uniquely identifies each row as shown in figure 4.
Figure 4 Relational database
This database is now more popular for these major reasons. There is no need to get trained for using relational database only introductory knowledge about database is enough. Database entries can be modified without redefining the entire structure.
3.4. Object-oriented Databases

This type of database is able to handle many new data types such as graphics, photographs, audio, and video. Object-oriented databases are useful for small, reusable modules of software called objects. The instructions contained within the object are used to do something with the data in the object. Though object oriented also has some disadvantages such as cost for developing this database is very high. The benefits to object-oriented databases are ability to mix and match reusable objects provides incredible multimedia capability.
140
Vol. 3, Issue 1, pp. 137-144
Figure 5 Object oriented database
IV.
DBMS
Various Database management systems have their own advantages and disadvantages. After analyzing the dbms, developer has to decide which database type is appropriate for the project. This paper discusses some cases which provide useful information for novice developer in selecting programming language and database for the project. a) Suppose there is a need to build an parametric design application for mechanical engineering CAD product like UG, Catia, Solidworks, etc. where the design table which is a simple database is available in the excel table, where the programmer will just access this database and update existing values. While developing such kind of file there is no need to get knowledge of high end databases like oracle or Microsoft SQL. The same task is performed with the help of Microsoft Access or the excel file can also function as a basic database too. b) Civil engineers mostly work with the estimation, costing, measurement, etc this kind of application database dont need of any special kind of services like performance tuning, security it only needs to store numbers and retrieve wherever necessary. Here, there is no need of performance tuning or security then the Microsoft Access is the best choice because it is easily available. c) Consider that the developer is going to develop banking system application at novice level, any banking application needs performance tuning, security, Business Intelligence, scalability, reliability then Microsoft access is not going to satisfy the requirements and hence developer have to think another databases which provides such kind of requirements for e.g. ORACLE, Microsoft SQL. While selecting database developer has to analyze and understand the current as well as future requirements of the user. Nowadays, it is not possible for businesses to hire a person who will be only hired for handling the project but business need an employee who can simultaneously do the project operations as well as handle the given work. If the developer is going to develop project for the first time then Microsoft Access is the best solution, it is because Microsoft Access is easily available with Microsoft Office, easy to understand and can handle database transactions up to certain level. When the developer gets trained on Microsoft Access and able to understand the database transactions then by experience it will use advance database management system efficiently.
V.
RESULTS AND DISCUSSION
Proper selection of programming language and database management system for project is a crucial task. After studying different database management systems, this paper gives some basic differences between these dbms in Table 1 Software project development is project life cycle process which starts from requirement analysis from user to the testing of project and deploys it on to the user side.
Table 1 Difference between various Database Management Systems Microsoft SQL Oracle Postgre SQL Access Easily Available EXCELLENT GOOD GOOD EXCELLENT Ease to Understand EXCELLENT GOOD AVERAGE AVERAGE Security AVERAGE GOOD EXCELLENT AVERAGE Issues
141
Vol. 3, Issue 1, pp. 137-144
Scalability Portability Data Handling Database Administration Business Intelligence Basic Data Types Concurrent Access Costly Support Size of project Stores Procedure AVERAGE GOOD AVERAGE AVERAGE AVERAGE AVERAGE LIMITED NO SMALLMEDIUM NO VERY GOOD GOOD GOOD GOOD GOOD AVERAGE YES NO ALL YES EXCELLENT VERY GOOD EXCELLENT VERY GOOD GOOD AVERAGE YES YES ALL YES AVERAGE GOOD GOOD GOOD POOR VERY GOOD YES NO ALL NO
Without proper selection of database management system and programming language, the goals of project are not achieved. Due to which there will be a loss for the organization directly affecting the profit of the business. Table 1 shows the difference between the various database management systems which are commonly used. By analyzing the table it is very clear that Microsoft access and PostgreSQL dbms are most favorable for the developer. PostgreSQL is an open source database management system which is freely available but due to some of its technical issues it not famous as Microsoft access for novice level application development. Microsoft access is used for small or medium sized projects, where there is no need of security, scalability and efficiency. Microsoft access is used by the beginner or the developers who want to develop the module or small project in short time. Whereas other database management systems like SQL, Oracle and Postgre SQL are useful for all projects, where there is need of security, efficiency, scalability and concurrent access to database. But these databases are not affordable for the small industries.
\ Figure 6 Popularity of the databases according to the cases
If developer decides programming language and database management system in initial stage of the project then, risk management will be achieved in the project and resources are also properly utilized. By referring this paper developer will come to know the factors on which the programming language and database management system will be selected for project. Figure 1.5 shows use of the some popular database management systems. By observing the figure it can easily understand that for advance operations like storing procedure, efficiency, security there must be a need of Oracle, SQL, Postgre SQL etc database management systems. For the novice level database applications developer must have to work with Microsoft Access.
VI.
CONCLUSION
The primary goal of the developer is to develop a project which satisfies users need as well as the project will complete within time deadline. If the selection of programming language and database management system fail at initial stage in the project, then developer team need to go to basics and
142
Vol. 3, Issue 1, pp. 137-144
start the work again. It is wastage of resources for the organization. This paper provides guidelines for choosing a programming language and database management system for project development. If the developer working with unsuitable programming language and/or database management system then project will definitely not function properly and due to which the user is not satisfied with project. For developing the small and medium sized projects for the routine functionalities and there is no high expectations from project then developer may use Microsoft access as the backend. If there are tremendous needs from the project for security, scalability and performance in the project it is better to use SQL or Oracle like architectures. Cost of IT projects is in millions or more, it is not affordable for any company to invest such big amount again and again for project development only. If software developer considers the points discussed in this paper for selecting programming language and database management system then project will achieve the goals set by the developer as well as clients of project.
ACKNOWLEDGEMENT
I would like to express utmost gratitude to my Co-author Dr. Virajit A. Gundale for his continued support, encouragement and guidance provided in articulating this paper. I would also like to thanks Mr. Babaso Ghag who extended his help in preparation of this paper.
REFERENCES
[1] Silberschatz, korth & Sudarshan, Database System Concept. McGraw Hill publication,2010. ISBN 0-07-352332-1. [2] Pankaj Jalote, An Integrated approach to Software Engineering. Narosa, 2011. ISBN: 978-81-7319-702-4 [3] Ben, mernachem, marlis, Software Quality. CENGAGE, 2008. ISBN-13: 9788131509463 [4] Ramakrishnan & Gehrke, Database Management systems. McGraw Hill publication, 2003. ISBN-10: 0072465638 [5] Thomas Connolly & Carolyn Begg, Database Systems : A practical approach to design, implementation and management. Pearson Education, 2009. ISBN-13: 9780321523068 [6] Mikell P. Groover, Emory W. Zimmers, CAD/CAM: Computer Aided design and Manufacturing. PHI, 2008. ISBN 13: 9780131101302 [7] P.N.Rao, CAD/CAM: Principles and Applications, Tata McGraw-Hill Education 2004. ISBN 0-07058373-0 [8] M.G.Korgaonkar, Just in Time Manufacturing, Vedams eBooks (P) Ltd, ISBN 10: 0333926633 / ISBN 13: 9780333926635 [9] IT diagram, Retrieved on January 11, 2012 from, http://2.bp.blogspot.com/_hTYkhoBH1iI/S5Ef053jCZI /AAAAAAAAAAc/HETvx-erhBk/s400/IT-Business-Focus.jpg [10] Gints Plivna , How to choose database, Retrieved on January 07, 2012 from, http://www.gplivna.eu/papers/choose_database.htm [11] 11 advantages to information technology provider services, Retrieved on January 08, 2012 from, http://intinc.com/benefits/technical-advantages.htm [12] How do you choose what programming language to use, Retrieved on January 08, 2012 from, http://stackoverflow.com/questions/144691/how-do-you-choose-what-programming-language-to-use [13] Susan Desousas, Different types of databases Retrieved on January 08, 2012 from, http://www.my-project-management-expert.com/different-types-of-databases.html [14] Types of database management system Retrieved on January 08, 2012 from, http://www.personal.psu.edu/glh10/ist110 /topic/topic07/topic07_06.html [15] Types of database, Retrieved on January 08, 2012 from http://www.theukwebdesigncompany.com/articles/types-of-databases.php [16] Hierarchical Database Model Retrieved on January 11, 2012 from, http://www.ibm.com/developerworks/xml/library/x-matters8/hier.gif [17] Network Database Model Retrieved on January 11, 2012 from, http://docs.oracle.com/cd/B19306_01/server.102/b14231/img/admin041.gif [18] OODB Model Retrieved on January 11, 2012 from, http://www.cs.pitt.edu/~chang/156/images/fig1914.gif [19] Relational database model, Retrieved on January 11, 2012 from, http://4.bp.blogspot.com/NK4lb2DAQyk/TeI8OsraBEI/AAAAAAAAAAU/fwTEwGYo9Ew/s1600/relational+database.jpg [20] Comparison of Oracle, MySQL and Postgres DBMS Retrieved on January 29, 2012 from, http://www-ss.fnal.gov/dsg/external/freeware/mysql-vs-pgsql.html
143
Vol. 3, Issue 1, pp. 137-144
[21] Oracle10g vs PostgreSQL 8 vs MySQL 5(n.d.). Retrieved on January 29, 2012 from, http://it.toolbox.com/blogs/oracle-guide/oracle-10g- vs- postgresql-8-vs-mysql-5-5452
Authors
Onkar Dipak Joshi is presently working as the Assistant Professor at Sharad Institute of Technology, College of Engineering. Yadrav Dist. Kolhapur, India. He obtained his B.E. in Computer Science and engineering from Shivaji University with distinction and currently pursuing his M.S. by research in Computer Science and Engineering. His total teaching experience is more than 2 years.
Virajit A. Gundale is presently working as the Professor, Mechanical Engineering Department at Sharad Institute of Technology, College of Engineering. Yadrav Dist. Kolhapur, India. Apart from this he is well known consultant in the design and development of Submersible pumps and motor components. He has worked on various international projects in Bangladesh, Indonesia, Egypt, etc. He obtained his B.E. in Mechanical Engineering from Shivaji University and M.S. in Manufacturing Management from BITS, Pilani. He obtained his Ph.D. in Manufacturing Technology from UNEM, Costa Rica in 2010. He also works on various Software development projects related with Automation of CAD/CAM, ERP, etc. His total experience including Teaching and Industry spans more than 12 years. He is also a fellow of the International Institute of Mechanical Engineers, South Africa. Sachin M. Jagdale is presently working as the Assistant Professor at Sharad Institute of Technology, Polytechnic. Yadrav Dist. Kolhapur, India. He obtained his B.E. in Computer Science and engineering from Shivaji University and currently pursuing his M.Tech. in Computer Science and Engineering. His total teaching experience is more than 2 years.
144
Vol. 3, Issue 1, pp. 137-144
AUDIO STEGANALYSIS OF LSB AUDIO USING MOMENTS AND MULTIPLE REGRESSION MODEL
Souvik Bhattacharyya1 and Gautam Sanyal2
Department of CSE, University Institute of Technology, The University of Burdwan, Burdwan, India 2 Department of CSE, National Institute of Technology, Durgapur, India
1
ABSTRACT
Steganography is the art and science of communicating in a way which hides the existence of the communication. Important information is firstly hidden in a host data, such as digital image, text, video or audio, etc, and then transmitted secretly to the receiver. Steganalysis is another important topic in information hiding which is the art of detecting the presence of steganography. In this paper an effective steganalysis method based on statistical moment as well as invariant moments of the audio signals is used to detect the presence of hidden messages has been presented. Multiple Regression analysis technique has been carried out to detect the presence of the hidden messages, as well as to estimate the relative length of the embedded messages. The design of audio steganalyzer depends upon the choice of the audio feature selection and the design of a two-class classifier. Experimental results demonstrate the effectiveness and accuracy of the proposed technique.
KEYWORDS: Audio Steganalysis; Statistical Moments; Invariant Moments.
I.
INTRODUCTION
Steganography is the art and science of hiding information by embedding messages with in other seemingly harmless messages. As the goal of steganography is to hide the presence of a message it can be seen as the complement of cryptography, whose goal is to hide the content of a message. Although steganography is an ancient subject, the modern formulation of it comes from the prisoners problem proposed by Simmons [1].An assumption can be made based on this model is that if both the sender and receiver share some common secret information then the corresponding steganography protocol is known as then the secret key steganography where as pure steganography means that there is none prior information shared by sender and receiver. If the public key of the receiver is known to the sender, the steganographic protocol is called public key steganography [4, 8]. For a more thorough knowledge of steganography methodology the reader is advised to see [1], [2].Some Steganographic model with high security features has been presented in [3-6].Almost all digital file formats can be used for steganography, but the image and audio files are more suitable because of their high degree of redundancy [2]. Fig. 1 below shows the different categories of file formats that can be used for steganography techniques.
Figure 1: Types of Steganography
145
Vol. 3, Issue 1, pp. 145-160
Among them image steganography is the most popular of the lot. In this method the secret message is embedded into an image as noise to it, which is nearly impossible to differentiate by human eyes [11, 15, 17]. In video steganography, same method may be used to embed a message [18, 24]. Audio steganography embeds the message into a cover audio file as noise at a frequency out of human hearing range [19]. One major category, perhaps the most difficult kind of steganography is text steganography or linguistic steganography [3]. The text steganography is a method of using written natural language to conceal a secret message as defined by Chapman et al. [16]. Steganalysis, from an opponent's perspective, is an art of deterring covert communications while avoiding affecting the innocent ones. Its basic requirement is to determine accurately whether a secret message is hidden in the testing medium. Further requirements may include judging the type of the steganography, estimating the rough length of the message, or even extracting the hidden message. The challenge of steganalysis is that: Unlike cryptanalysis, where it is evident that intercepted encrypted data contains a message, steganalysis generally starts with several suspect information streams but uncertainty whether any of these contain hidden message.
Fig. 1. Generic Steganography and Steganalysis
Figure 2: Generic Steganography and Steganalysis
In this paper, a novel steganalysis method based on statistical moments of the audio signals as well as the multiple regression analysis based on the features of those statistical moments are proposed. The performance of the steganalysis has been demonstrated by extensive experimental investigation. This paper has been organized as following sections: Section II describes some review works of audio steganography. Section III reviews the previous work on audio steganalysis. Section IV describes the various methods for audio feature selection. Experimental Results of the method has been discussed in Section V and Section VI draws the conclusion.
II.
REVIEW OF RELATED WORKS ON AUDIO STEGANOGRAPHY
This section presents some existing techniques of audio data hiding namely Least Significant Bit Encoding, Phase Coding Echo Hiding and Spread Spectrum techniques. There are two main areas of modification in an audio for data embedding. First, the storage environment, or digital representation of the signal that will be used, and second the transmission pathway the signal might travel [4, 11]. A. Least Significant Bit Encoding The simple way of embedding the information in a digital audio file is done through Least significant bit (LSB) coding. By substituting the least significant bit of each sampling point with a binary message bit, LSB coding allows a data to be encoded and produces the stego audio. In LSB coding, the ideal data transmission rate is 1 kbps per 1 kHz. The main disadvantage of LSB coding is its low embedding capacity. In some cases an attempt has been made to overcome this situation by replacing the two least significant bits of a sample with two message bits. This increases the data embedding capacity but also increases the amount of resulting noise in the audio file as well. A novel method of LSB coding which increases the limit up to four bits is proposed by Nedeljko Cvejic Et al. [13, 16]. To extract secret message from an LSB encoded audio, the receiver needs access to the sequence of sample indices used during the embedding process. Normally, the length of the secret message to be embedded is smaller than the total number of samples done. There are other two disadvantages also
146
Vol. 3, Issue 1, pp. 145-160
associated LSB coding. The first one is that human ear is very sensitive and can often detect the presence of single bit of noise into an audio file. Second disadvantage however, is that LSB coding is not very robust. Embedded information will be lost through a little modification of the stego audio. B. Phase Coding Phase coding [11, 16] overcomes the disadvantages of noise induction method of audio steganography. Phase coding designed based on the fact that the phase components of sound are not as perceptible to the human ear as noise is. This method encodes the message bits as phase shifts in the phase spectrum of a digital signal, achieving an inaudible encoding in terms of signal-to-noise ratio. In figure 4 below original and encoded signal through phase coding method has been presented.
Original Signal
Encoded Signal
Figure 3: The original signal and encoded signal of phase coding technique.
Phase coding principles are summarized as under: The original audio signal is broken up into smaller segments whose lengths equal the size of the message to be embedded. Discrete Fourier Transform (DFT) is applied to each segment to create a matrix of the phases and Fourier transform magnitudes. Phase differences between adjacent segments are calculated next. Phase shifts between consecutive segments are easily detected. In other words, the absolute phases of the segments can be changed but the relative phase differences between adjacent segments must be preserved. Thus the secret message is only inserted in the phase vector of the first signal segment as follows:
A new phase matrix is created using the new phase of the first segment and the original phase differences. Using the new phase matrix and original magnitude matrix, the audio signal is reconstructed by applying the inverse DFT and by concatenating the audio segments. To extract the secret message from the audio file, the receiver needs to know the segment length. The receiver can extract the secret message through different reverse process. The disadvantage associated with phase coding is that it has a low data embedding rate due to the fact that the secret message is encoded in the first signal segment only. This situation can be overcome by increasing the length of the signals segment which in turn increases the change in the phase relations between each frequency component of the segment more drastically, making the encoding easier to detect. Thus, the phase coding method is useful only when a small amount of data, such as a watermark, needs to be embedded. C. Echo Hiding In echo hiding [14, 15, 16] method information is embedded into an audio file by inducing an echo into the discrete signal. Like the spread spectrum method, Echo Hiding method also has the advantage of having high embedding capacity with superior robustness compared to the noise inducing methods. If only one echo was produced from the original signal, only one bit of information could be encoded. Therefore, the original signal is broken down into blocks before the encoding process begins. Once the encoding process is completed, the blocks are concatenated back together to form the final signal.
147
Vol. 3, Issue 1, pp. 145-160
To extract the secret message from the final stego audio signal, the receiver must be able to break up the signal into the same block sequence used during the encoding process. Then the autocorrelation function of the signal's cepstrum which is the Forward Fourier Transform of the signal's frequency spectrum can be used to decode the message because it reveals a spike at each echo time offset, allowing the message to be reconstructed.
Figure 4: Echo Hiding Methodology.
D. Spread Spectrum Spread Spectrum (SS) [16] methodology attempts to spread the secret information across the audio signal's frequency spectrum as much as possible. This is equivalent to a system using the LSB coding method which randomly spreads the message bits over the entire audio file. The difference is that unlike LSB coding, the SS method spreads the secret message over the audio file's frequency spectrum, using a code which is independent of the actual signal. As a result, the final signal occupies a more bandwidth than actual requirement for embedding. Two versions of SS can be used for audio steganography one is the direct sequence where the secret message is spread out by a constant called the chip rate and then modulated with a pseudo random signal where as in the second method frequency-hopping SS, the audio file's frequency then interleaved with the cover-signal spectrum is altered so that it hops rapidly between frequencies. The Spread Spectrum method has a more embedding capacity compared to LSB coding and phase coding techniques with maintaining a high level of robustness. However, the SS method shares a disadvantage common with LSB and parity coding that it can introduce noise into the audio file at the time of embedding.
III.
REVIEW OF RELATED WORKS ON AUDIO STEGANALYSIS
Audio steganalysis is very difficult due to the existence of advanced audio steganography schemes and the very nature of audio signals to be high-capacity data streams necessitates the need for scientifically challenging statistical analysis [29]. A. Phase and Echo Steganalysis Zeng et. al [17] proposed steganalysis algorithms to detect phase coding steganography based on the analysis of phase discontinuities and to detect echo steganography based on the statistical moments of peak frequency [18]. The phase steganalysis algorithm explores the fact that phase coding corrupts the extrinsic continuities of unwrapped phase in each audio segment, causing changes in the phase difference [19]. A statistical analysis of the phase difference in each audio segment can be used to monitor the change and train the classifiers to differentiate an embedded audio signal from a clean audio signal. B. Universal Steganalysis based on Recorded Speech Johnson et. al [20] proposed a generic universal steganalysis algorithm that bases it study on the statistical regularities of recorded speech. Their statistical model decomposes an audio signal (i.e., recorded speech) using basis functions localized in both time and frequency domains in the form of Short Time Fourier Transform (STFT). The spectrograms collected from this decomposition are analyzed using non-linear support vector machines to differentiate between cover and stego audio signals. This approach is likely to work only for high-bit rate audio steganography and will not be effective for detecting low bit-rate embeddings.
148
Vol. 3, Issue 1, pp. 145-160
C. Use of Statistical Distance Measures for Audio Steganalysis H. Ozer et. al [21] calculated the distribution of various statistical distance measures on cover audio signals and stego audio signals vis--vis their versions without noise and observed them to be statistically different. The authors employed audio quality metrics to capture the anomalies in the signal introduced by the embedded data. They designed an audio steganalyzer that relied on the choice of audio quality measures, which were tested depending on their perceptual or non-perceptual nature. The selection of the proper features and quality measures was conducted using the (i) ANOVA test [22] to determine whether there are any statistically significant differences between available conditions and the (ii) SFS (Sequential Floating Search) algorithm that considers the inter-correlation between the test features in ensemble [23]. Subsequently, two classifiers, one based on linear regression and another based on support vector machines were used and also simultaneously evaluated for their capability to detect stego messages embedded in the audio signals. D. Audio Steganalysis based on Hausdorff Distance The audio steganalysis algorithm proposed by Liu et. al [24] uses the Hausdorff distance measure [25] to measure the distortion between a cover audio signal and a stego audio signal. The algorithm takes as input a potentially stego audio signal x and its de-noised version x as an estimate of the cover signal. Both x and x are then subjected to appropriate segmentation and wavelet decomposition to generate wavelet coefficients [26] at different levels of resolution. The Hausdorff distance values between the wavelet coefficients of the audio signals and their de-noised versions are measured. The statistical moments of the Hausdorff distance measures are used to train a classifier on the difference between cover audio signals and stego audio signals with different content loadings. E. Audio Steganalysis for High Complexity Audio Signals More recently, Liu et. al [27] propose the use of stream data mining for steganalysis of audio signals of high complexity. Their approach extracts the second order derivative based Markov transition probabilities and high frequency spectrum statistics as the features of the audio streams. The variations in the second order derivative based features are explored to distinguish between the cover and stego audio signals. This approach also uses the Mel-frequency cepstral coefficients [28], widely used in speech recognition, for audio steganalysis. Recently two new methods of audio steganalysis of spread spectrum information hiding has been proposed in [31-32].
IV.
AUDIO FEATURE SELECTION
In this section audio quality measures in terms of moments up to 7th order both statistical and invariants has been investigated for the purpose of audio steganalysis. Various moments of the audio signals are sensitive to the presence of a steganographic message embedding. Moments based features have been extracted for steganalytic measure in such a way that reflect the quality of distorted or degraded audio signal vis--vis its original in an accurate, consistent and monotonic way. Such a measure, in the context of steganalysis, should respond to the presence of hidden message with minimum error, should work for a large variety of embedding methods, and its reaction should be proportional to the embedding strength. A. Moments based Audio feature To construct the features of both cover and stego or suspicious audios moments of the audio series has been computed. In mathematics, a moment is, loosely speaking, a quantitative measure of the shape of a set of points. The "second moment", for example, is widely used and measures the "width" (in a particular sense) of a set of points in one dimension or in higher dimensions measures the shape of a cloud of points as it could be fit by an ellipsoid. Other moments describe other aspects of a distribution such as how the distribution is skewed from its mean, or peaked. There are two ways of viewing moments [30], one based on statistics and one based on arbitrary functions such as f(x) or f(x, y). As a result moments can be defined in more than one way. Statistical view Moments are the statistical expectation of certain power functions of a random variable. The most common moment is the mean which is just the expected value of a random variable as given in equation 1.
149
Vol. 3, Issue 1, pp. 145-160
= E[ X ] =
x f ( x ) dx
(1)
Where f(x) is the probability density function of continuous random variable X. More generally, moments of order p = 0, 1, 2, can be calculated as mp = E [Xp].These are sometimes referred to as the raw moments. There are other kinds of moments that are often useful. One of these is the central moments p = E [(X)p].The best known central moment is the second, which is known as the variance given in equation 2.
2 = ( x ) 2 f ( x) dx = m2 12
(2)
Two less common statistical measures, skewness and kurtosis, are based on the third and fourth central moments. The use of expectation assumes that the pdf is known. Moments are easily extended to two or more dimensions as shown in equation 3.
m pq = E[X pY q ] = x p y q f ( x, y ) dx dy
(3)
Here f(x, y) is the joint pdf. Estimation However, moments are easy to estimate from a set of measurements, xi. The p-th moment is estimated as given in equation 4 and 5.
mp =
1 N
x
i =1
p i
(4)
(Often 1/N is left out of the definition) and the p-th central moment is estimated as
p =
1 N
(x x )
i i
(5)
is the average of the measurements, which is the usual estimate of the mean. The second central moment gives the variance of a set of data s2 = 2.For multidimensional distributions, the first and second order moments give estimates of the mean vector and covariance matrix. The order of moments in two dimensions is given by p+q, so for moments above 0, there is more than one of a given order. For example, m20, m11, and m02 are the three moments of order 2. Non-statistical view This view is not based on probability and expected values, but most of the same ideas still hold. For any arbitrary function f(x), one may compute moments using the equation 6 or for a 2-D function using the equation 7.
mp =
f ( x) dx
(6) (7)
m pq = x p y q f ( x, y ) dx dy
Notice now that to find the mean value of f(x), one must use m1/m0, since f(x) is not normalized to area 1 like the pdf. Likewise, for higher order moments it is common to normalize these moments by dividing by m0 (or m00). This allows one to compute moments which depend only on the shape and not the magnitude of f(x). The result of normalizing moments gives measures which contain information about the shape or distribution (not probability dist.) of f(x).
150
Vol. 3, Issue 1, pp. 145-160
Digital approximation For digitized data (including images) we must replace the integral with a summation over the domain covered by the data. The 2-D approximation is written in equation 8.
m pq = f ( xi , y j ) xip y q j
i =1 j =1
= f (i, j ) i p j q
i =1 j =1
(8)
If f(x, y) is a binary image function of an object, the area is m00, the x and y centroids are x = m10 / m00 and y = m01 / m00 . Invariance In many applications such as shape recognition, it is useful to generate shape features which are independent of parameters which cannot be controlled in an image. Such features are called invariant features. There are several types of invariance. For example, if an object may occur in an arbitrary location in an image, then one needs the moments to be invariant to location. For binary connected components, this can be achieved simply by using the central moments, pq. If an object is not at a fixed distance from a fixed focal length camera, then the sizes of objects will not be fixed. In this case size invariance is needed. This can be achieved by normalizing the moments as given in equation 9.
pq =
pq , where = ( p + q ) + 1. 00 (9)
The third common type of invariance is rotation invariance. This is not always needed, for example if objects always have a known direction as in recognizing machine printed text in a document. The direction can be established by locating lines of text. M.K. Hu derived a transformation of the normalized central moments to make the resulting moments rotation invariant as given in equation 10.
(10)
151
Vol. 3, Issue 1, pp. 145-160
B. Regression based Analysis Method In statistics, regression analysis includes any techniques for modeling and analyzing several variables, when the focus is on the relationship between a dependent variable and one or more independent variables. More specically, regression analysis helps one understand how the typical value of the dependent variable changes when any one of the independent variables is varied, while the other independent variables are held xed. In all cases, the estimation target is a function of the independent variables called the regression function.In regression analysis, it is also of interest to characterize the variation of the dependent variable around the regression function, which can be described by a probability distribution. Regression analysis is widely used for prediction and forecasting, where its use has substantial overlap with the eld of machine learning. Regression analysis is also used to understand which among the independent variables are related to the dependent variable, and to explore the forms of these relationships. In restricted circumstances, regression analysis can be used to infer causal relationships between the independent and dependent variables. A large body of techniques for carrying out regression analysis has been developed. Familiar methods such as linear regression and ordinary least squares regression are parametric, in that the regression function is dened in terms of a nite number of unknown parameters that are estimated from the data. Nonparametric regression refers to techniques that allow the regression function to lie in a specied set of functions, which may be innite-dimensional. The performance of regression analysis methods in practice depends on the form of the data-generating process, and how it relates to the regression approach being used. Since the true form of the data-generating process is in general not known, regression analysis often depends to some extent on making assumptions about this process. These assumptions are sometimes (but not always) testable if a large amount of data is available. Regression models for prediction are often useful even when the assumptions are moderately violated, although they may not perform optimally. However, in many applications, especially with small effects or questions of causality based on observational data, regression methods give misleading results. Regression models: Regression models involve the following variables The unknown parameters denoted as ; this may be a scalar or a vector. The independent variables X. The dependent variable, Y. In various fields of application, different terminologies are used in place of dependent and independent variables. A regression model relates Y to a function of X and as given in equation 11. The approximation is usually formalized as E(Y | X) = f(X, ). To carry out regression analysis, the form of the function f must be specified. Sometimes the form of this function is based on knowledge about the relationship between Y and X that does not rely on the data. If no such knowledge is available, a flexible or convenient form for f is chosen. Linear regression: In linear regression, the model specification is that the dependent variable, yi is a linear combination of the parameters (but need not be linear in the independent variables). For example, in simple linear regression for modeling n data points there is one independent variable: xi, and two parameters, 0 and 1: Yi = 0+1xi+i , i =1,.,n.-----(12) In multiple linear regressions, there are several independent variables or functions of independent variables. For example, adding a term in xi2 to the preceding regression gives: Yi = 0+1xi+2 xi2+ i , i =1,.,n.--------(13) This is still linear regression; although the expression on the right hand side is quadratic in the independent variable xi, it is linear in the parameters 0, 1 and 2.In both cases, i is an error term and the subscript i indexes a particular observation. Given a random sample from the population, we estimate the population parameters and obtain the sample linear regression model: i= 0+1xi ------------ (14)
152
Vol. 3, Issue 1, pp. 145-160
The residual, ei = Yi - i the difference between the value of the dependent variable predicted by the model, i and the true value of the dependent variable Yi. C. Audio Steganalysis based on Moment based Multiple Regression Analysis The Steganalysis technique proposed here to test the presence of the hidden message is the combination of statistical moments and invariant moments based analysis along with multiple regression based analysis on the cover data and stego data series for the estimation of the presence of the secret message as well as the predictive size of the hidden message. Let C and S be the series modeling the distribution of the cover and stego audio signals, a multiple regression analysis can be done in order to express S (dependent variable) as a function of the cover data C (independent variable) can be modeled as multiple regression model S = I + a * C where I is the constant and a is the regression coefficient. It has been seen from the experiments that multiple regression model between two same series namely a and b can be expressed as b = (0) + (1) *a. Steganalysis approach has been designed here based on the above mentioned fact considering cover audio data as the independent data series and stego audio data as the dependent series data. From the experimental results it can be shown that with the introduction of secret message/increasing length of the secret message moments parameters also changes. Hence it can be hypothesize that regression analysis between the cover data and stego data will cluster differently for clean and stego signals. This is the basis of proposed steganalyzer that aims to classify audio signal as original and suspicious. Multiple Regression modeling is also done in order to estimate the strength of the association among the variables of a series. The steganalysis system works in four stages: Moment based feature generation, feature selection, SVM classifier selection and training, and finally, steganalysis i.e. detection of the suspicious image. The first three steps were performed using only the training set. The final step was performed on the test collection to generate the submitted results.
Figure 5: Block Diagram of the proposed steganalysis system
V.
EXPERIMENTAL RESULTS
The steganalyzer has been designed based on a training set and using various audio steganographic tools. The steganographic tools used here Steganos [15] and S-Tools [16].In the experiments 10 audio wav signals were used for training and 10 audio wav signals for testing. After applying LSB replacement algorithm for embedding secret message into the cover audio with the insertion rate of 100 char, 200 char, 400 char with increase of twice the previous one up to 51200 char various stego audios has been created. Multiple Regression analysis based on the statistical moments, invariant moments as well as the combined form of this two has been used to estimate the various regression statistics of those stego audios. This steganalyzer can predicted the approximate area of the hidden message as well as the embedding message length can also be predicted on the basis of those
153
Vol. 3, Issue 1, pp. 145-160
regression statistics. From fig 6 it can be seen that with the introduction of small message all the invariant moments values changes. Figure 7 shows the graphical change rate of invariant moments of various orders with changes in insertion rate.
Insertion Rate(in char) 0 100 200 400 800 1600 3200 6400 12800 25600 51200
1 |log|
16.8198 16.8202 16.8207 16.8189 16.8197 16.8193 16.8196 16.8201 16.8159 16.8175 16.8195
2|log|
33.0476 33.0483 33.0495 33.0458 33.0475 33.0466 33.0472 33.0483 33.0397 33.0429 33.0471
3 |log|
45.4821 45.5179 45.4880 45.4900 45.4867 45.4931 45.5013 45.4945 45.5072 45.5390 45.5394
4 |log|
45.4821 45.5179 45.4880 45.4900 45.4867 45.4931 45.5013 45.4945 45.5072 45.5390 45.5394
5 |log|
90.7469 90.8187 90.7588 90.7628 90.7561 90.7689 90.7854 90.7717 90.7973 90.8609 90.8618
6 |log|
61.8972 61.9335 61.9042 61.9043 61.9018 61.9078 61.9163 61.9100 61.9185 61.9519 61.9544
7|log|
Inf 12.4766 Inf 12.4766 Inf 12.1951 Inf Inf Inf Inf Inf
Figure 6: Invariant Moments value of chimes.wav audio signal at various embedding rate.
Figure 7: Rate of change of invariant moment of various orders at various embedding rate.
From fig 8 it can be seen that with the introduction of small message all the statistical moments value also changes and figure 9 shows the graphical change rate of statistical moments of various orders with changes in insertion rate.Finally figure 10 and figure 11 shows the changes of combined moments values of various order in tabular form and graphically respectively.
154
Vol. 3, Issue 1, pp. 145-160
Insertion Rate (in char) 0 100 200 400 800 1600 3200 6400 12800 25600 51200 M1 |log| M2 |log| M3 |log| M4 |log| M5 |log| M6 |log| M7 |log|
Inf Inf Inf Inf Inf Inf Inf Inf Inf Inf Inf
5.0144 5.0144 5.0143 5.0144 5.0144 5.0144 5.0144 5.0144 5.0144 5.0144
10.7782 10.7795 10.7790 10.7789 10.7808 10.7793 10.7844 10.7870 10.7800 10.7833
6.7931 6.7931 6.7931 6.7931 6.7931 6.7932 6.7932 6.7932 6.7931 6.7932
10.0539 10.0543 10.0543 10.0540 10.0544 10.0539 10.0570 10.0573 10.0540 10.0548
7.5902 7.5902 7.5902 7.5902 7.5902 7.5904 7.5905 7.5905 7.5902 7.5903
9.9484 9.9485 9.9486 9.9484 9.9486 9.9499 9.9505 9.9506 9.9483 9.9487
5.0144 10.7782 6.7932 10.0541 7.5903 9.9487 Figure 8: Statistical Moments value of chimes.wav audio signal at various embedding rate.
Figure 9: Rate of change of statistical moment of various orders at various embedding rate.
155
Vol. 3, Issue 1, pp. 145-160
Insertion Rate (in char) 0 100 200 400 800 1600 3200 6400 12800 25600 51200 CM1 |log| Inf Inf Inf Inf Inf Inf Inf Inf Inf Inf CM2 |log| 38.0620 38.0627 38.0638 38.0602 38.0619 38.0610 38.0616 38.0627 38.0541 38.0573 CM3 |log| 56.2603 56.2974 56.2670 56.2689 56.2675 56.2724 56.2857 56.2815 56.2872 56.3223 CM4 |log| 52.2752 52.3110 52.2811 52.2831 52.2798 52.2863 52.2945 52.2877 52.3003 52.3322 CM5 |log| CM6 |log| 69.4874 69.5237 69.4944 69.4945 69.4920 69.4982 69.5068 69.5005 69.5087 69.5422 CM7 |log| Inf 22.4251 Inf 22.4250 Inf 22.1450 Inf Inf Inf Inf CAVGM
100.8008 100.8730 100.8131 100.8168 100.8105 100.8228 100.8424 100.8290 100.8513 100.9157
39.7357 39.7585 39.7399 39.7404 39.7390 39.7426 39.7489 39.7452 39.7502 39.7712
Inf 38.0615 56.3176 52.3326 100.9159 69.5447 Inf 39.7715 Figure 10: Combined Moments value of chimes audio signal at various embedding rate.
Figure 11: Rate of change of combined moments of various orders at various embedding rate.
156
Vol. 3, Issue 1, pp. 145-160
Figure 12: Rate of Change of Average Combined Moments value at various insertion rates
To build the classes of Cover Audio signals Combined Moment values of the original 10 audio signals (chimes.wav, heartbeat.wav etc) are inserted in the database. Next step is to produce stego audio signals with those original audio signals with random bits and different rates in another database. Original audio signals database with various moments values are used for training set from which the classifiers are tuned. With help of this an incoming audio signal can be classified as stego or cover audio signal. Figure 11 shows the rate of changes of average combined moment values at various insertion rate where as fig 12 shows the combined moments value at the various segments of the original audio signal chimes.wav and also show the changes of moments value after embedding 100 char in segment1. From the figure it can be seen that moments value has been changed in segment1 with the introduction of data where as segment2, segment3 and segment 4 parameters are remain unchanged.
Figure 13: Combined Moments value at various audio segment of the original and stego audio
Thus it can be concluded that secret data has been embedded in segment1 portion of the cover audio signal. Next step is to predict the length of the hidden message. The classifier is trained with the values of Combined Moments of order 2, 3, 4, 5 and 6 of the original audio signal and stego signals with different embedding rate in order to form a relation between the embedding capacity and the combined moment values order 2, 3, 4, 5 and 6. Insertion rate can be computed from the various
157
Vol. 3, Issue 1, pp. 145-160
moments values using Multiple Regression analysis. Insertion Rate (IR) can be calculated from eqn (a) or eqn (b) given below Insertion Rate (in Char) = -41895724 + 12170436 * CM2 |log| - 2204991 * CM3 |log| + 17073509 *CM4 |log| + 5327565 *CM5 |log| - 24850928 *CM6 |log| -------------- (a) Or Insertion Rate (in %) = -117.1 + 19.32 * CM2 |log| - 28.06 * CM3 |log| - 40.7 * CM4 |log| + 51.82 * CM5 |log| - 30.73 * CM6 |log|----------------- (b) [Considering Maximum Embedding Capacity of chimes.wav is 169984 character] Embedding capacity can be predicted based on the insertion rate given in fig 13 which establishes the accuracy level of prediction is 70-75% with this steganalyzer.
Figure 14: Error rate in Prediction
From the above table in figure 13 it can be concluded that insertion rate can be predicted with an average error rate of 30-35 %.Thus we can say embedding length of the secret message can be successfully predicted 65-70% with this steganalyzer.
VI.
CONCLUSION
In this paper a new approach of LSB audio steganalysis is proposed. In this method moments based multiple regression analysis technique has been used for steganalysis. This algorithm not only detects the presence of the hidden message but can also be able to predict the length of the secret message. This method is also capable of finding the approximate hidden area of the secret message which is one of the superiority factors of the proposed method in comparison with the other existing methods.
REFERENCES
[1] [2] [3]
[4]
[5]
[6]
N.F.Johnson. and S. Jajodia. Steganography: seeing the unseen. IEEE Computer, 16:2634, 1998. JHP Eloff. T Mrkel. and MS Olivier. An overview of image steganography. In Proceedings of the fifth annual Information Security South Africa Conference., 2005. Study of Secure Steganography model by Souvik Bhattacharyya and Gautam Sanyal at the proceedings of International Conference on Advanced Computing & Communication Technologies (ICACCT-2008),Nov, 2008, Panipat, India An Image based Steganography model for promoting Global Cyber Security by Souvik Bhattacharyya and Gautam Sanyal at the proceedings of International Conference on Systemics, Cybernetics and Informatics (ICSCI-2009),Jan, 09,Hyderabad,India. Implementation and Design of an Image based Steganographic model by Souvik Bhattacharyya and Gautam Sanyal at the proceedings of IEEE International Advance Computing Conference (IACC2009) A Novel Approach to Develop a Secure Image based Steganographic Model using Integer Wavelet Transform at the proceedings of International Conference on Recent Trends in Information, Telecommunication and Computing (ITC 2010) by Souvik Bhattacharyya, Avinash Prasad Kshitij and Gautam Sanyal. (Indexed by IEEE Computer Society).
158
Vol. 3, Issue 1, pp. 145-160
[7] [8]
[9]
[10]
[11]
[12]
[13]
[14]
[15] [16] [17]
[18]
[19]
[20]
[21]
[22] [23] [24]
[25]
[26]
[27] [28]
C. Kraetzer and J. Dittmann, Pros and Cons of Mel cepstrum based Audio Steganalysis using SVM Classification, Lecture Notes in Computer Science, vol. 4567, pp. 359 377, January 2008. W. Zeng, H. Ai and R. Hu, A Novel Steganalysis Algorithm of Phase Coding in Audio Signal, Proceedings of the 6th International Conference on Advanced Language Processing and Web Information Technology, pp. 261 264, August 2007. W. Zeng, H. Ai and R. Hu, An Algorithm of Echo Steganalysis based on Power Cepstrum and Pattern Classification, Proceedings of the International Conference on Information and Automation, pp. 1667 1670, June 2008. M. K. Johnson, S. Lyu, H. Farid, Steganalysis of Recorded Speech, Proceedings of Conference on Security, Steganography and Watermarking of Multimedia, Contents VII, vol. 5681, SPIE, pp. 664 672, May 2005. H. Ozer, I. Avcibas, B. Sankur and N. D. Memon, Steganalysis of Audio based on Audio Quality Metrics, Proceedings of the Conference on Security, Steganography and Watermarking of Multimedia, Contents V, vol. 5020, SPIE, pp. 55 66, January 2003. Y. Liu, K. Chiang, C. Corbett, R. Archibald, B. Mukherjee and D. Ghosal, A Novel Audio Steganalysis based on Higher-Order Statistics of a Distortion Measure with Hausdorff Distance, Lecture Notes in Computer Science, vol. 5222, pp. 487 -501, September 2008. D. P. Huttenlocher, G. A. Klanderman and W. J. Rucklidge, Comparing Images using Hausdorff Distance, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, no. 9, pp. 850 863, September 1993. T. Holotyak, J. Fridrich and S. Voloshynovskiy, Blind Statistical Steganalysis of Additive Steganography using Wavelet Higher Order Statistics, Lecture Notes in Computer Science, vol. 3677, pp. 273 274, September 2005. N. Memon I. Avcibas and B. Sankur. Steganalysis using image quality metrics. Signal Processing., IEEE Transactions on Image Processing,12(2):221-229, 2003. M. Goljan J. Fridrich and D. Hogea. Attacking the outguess. In Proceedings of 2002 ACM Workshop on Multimedia and Security,ACM Press., 2002. W. Zeng, H. Ai and R. Hu, A Novel Steganalysis Algorithm of Phase Coding in Audio Signal, Proceedings of the 6th International Conference on Advanced Language Processing and Web Information Technology, pp. 261 264, August 2007. W. Zeng, H. Ai and R. Hu, An Algorithm of Echo Steganalysis based on Power Cepstrum and Pattern Classification, Proceedings of the International Conference on Information and Automation, pp. 1667 1670, June 2008. Paraskevas and E. Chilton, Combination of Magnitude and Phase Statistical Features for Audio Classification, Acoustical Research Letters Online, Acoustical Society of America, vol. 5, no. 3, pp. 111 117, July 2004. M. K. Johnson, S. Lyu, H. Farid, Steganalysis of Recorded Speech, Proceedings of Conference on Security, Steganography and Watermarking of Multimedia, Contents VII, vol. 5681, SPIE, pp. 664 672, May 2005. H. Ozer, I. Avcibas, B. Sankur and N. D. Memon, Steganalysis of Audio based on Audio Quality Metrics, Proceedings of the Conference on Security, Steganography and Watermarking of Multimedia, Contents V, vol. 5020, SPIE, pp. 55 66, January 2003. A.C. Rencher, Methods of Multivariate Data Analysis, 2nd Edition, John Wiley, New York, NY, March 2002. P. Pudil, J. Novovicova and J. Kittler, Floating Search Methods in Feature Selection, Pattern Recognition Letters, vol. 15, no. 11, pp. 1119 1125, November 1994. Y. Liu, K. Chiang, C. Corbett, R. Archibald, B. Mukherjee and D. Ghosal, A Novel Audio Steganalysis based on Higher-Order Statistics of a Distortion Measure with Hausdorff Distance, Lecture Notes in Computer Science, vol. 5222, pp. 487 -501, September 2008. P. Huttenlocher, G. A. Klanderman and W. J. Rucklidge, Comparing Images using Hausdorff Distance, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, no. 9, pp. 850 863, September 1993. T. Holotyak, J. Fridrich and S. Voloshynovskiy, Blind Statistical Steganalysis of Additive Steganography using Wavelet Higher Order Statistics, Lecture Notes in Computer Science, vol. 3677, pp. 273 274, September 2005. I.Avcibas, Audio Steganalysis with Content-independent Distortion Measures, IEEE Signal Processing Letters, vol. 13, no. 2, pp. 92 95, February 2006. Q. Liu, A. H. Sung and M. Qiao, Novel Stream Mining for Audio Steganalysis, Proceedings of the 17th ACM International Conference on Multimedia, pp. 95 104, Beijing, China, October 2009.
159
Vol. 3, Issue 1, pp. 145-160
C. Kraetzer and J. Dittmann, Pros and Cons of Mel cepstrum based Audio Steganalysis using SVM Classification, Lecture Notes in Computer Science, vol. 4567, pp. 359 377, January 2008. [30] MOMENTS IN IMAGE PROCESSING Bob Bailey Nov. 2002 [31] Audio steganalysis of spread spectrum hiding based on statistical moment by Zhang Kexin at the proceedings of 2nd International Conference on Signal Processing Systems (ICSPS), 2010. [32] Audio steganalysis of spread spectrum information hiding based on statistical moment and distance metric by Wei Zeng, Ruimin Hu and Haojun Ai at Multimedia Tools and Applications Volume 55, Number 3, 525-556.
[29]
AUTHORS
Souvik Bhattacharyya received his B.E. degree in Computer Science and Technology from B.E. College, Shibpur, India, presently known as Bengal Engineering and Science University (BESU) and M.Tech degree in Computer Science and Engineering from National Institute of Technology, Durgapur, India. Currently he is working as an Assistant Professor in Computer Science and Engineering Department at University Institute of Technology, The University of Burdwan. He has more than 30 research publication in his credit. His areas of interest are Natural Language Processing, Network Security and Image Processing.
Gautam Sanyal has received his B.E and M.Tech degree National Institute of Technology (NIT), Durgapur, India. He has received Ph.D (Engg.) from Jadavpur University, Kolkata, India, in the area of Robot Vision. He possesses an experience of more than 25 years in the field of teaching and research. He has published nearly 72 papers in International and National Journals / Conferences. Three Ph.Ds (Engg) have already been awarded under his guidance. At present he is guiding six Ph.Ds scholars in the field of Steganography, Cellular Network, High Performance Computing and Computer Vision. He has guided over 10 PG and 100 UG thesis. His research interests include Natural Language Processing, Stochastic modeling of network traffic, High Performance Computing, Computer Vision. He is presently working as a Professor in the department of Computer Science and Engineering and also holding the post of Dean (Students Welfare) at National Institute of Technology, Durgapur, India. He is a member of IEEE also.
160
Vol. 3, Issue 1, pp. 145-160
ANALYSIS AND COMPARISON OF COMBINATIONAL CIRCUITS BY USING LOW POWER TECHNIQUES

Suparshya Babu Sukhavasi1, Susrutha Babu Sukhavasi1, Vijaya Bhaskar M2, B Rajesh Kumar3
Assistant Professor, Department of ECE, K L University, Guntur, AP, India. M.Tech -VLSI Student, Department of ECE, K L University, Guntur, AP, India. 3 M.Tech-Embedded VLSI Student, Department of ECE, SITE, T P Gudam, AP, India.
2 1
ABSTRACT
Power dissipation is the major aspect which is effecting the digital circuits. By implementing the self resetting logic to the digital circuit, the power dissipation is drastically reduced. In the VLSI Design this low power technique is very advanced for DSP applications. The dynamic circuits are becoming increasingly popular because of the speed advantage over static CMOS logic circuits; hence they are widely used today in high performance and low power circuits. Self-resetting logic is a commonly used piece of circuitry that can be found in use with memory arrays as word line drivers. Self resetting logic implemented in dynamic logic families have been proposed as viable clock less alternatives. The combinational logic is a type of digital logic which is implemented by Boolean circuits, where the output is a pure function of the present input only. This is in contrast to sequential logic, in which the output depends not only on the present input but also on the history of the input. In this paper mainly the self resetting logic is applied for the different combinational circuits and the analysis is done very clearly. By implementing this low power technique for different logic circuits and adders, by comparison with DYNAMIC and SRCMOS logics power dissipation is drastically reduced up to 35% compared with CMOS logic circuits and observations are tabulated.
KEYWORDS: High speed, VLSI, Self-resetting logic (SRL), topologies, power dissipation
I.
INTRODUCTION
Combinational logic is used in computer circuits to do Boolean algebra on input signals and on stored data. Practical computer circuits normally contain a mixture of combinational and sequential logic. The part of an arithmetic logic unit, or ALU, that does mathematical calculations is constructed using combinational logic. Other circuits used in computers, such as half adders, full adders, half subtractors, full subtractors, multiplexers, demultiplexers, encoders and decoders are also made by using combinational logic. In todays fast processing environment, the use of dynamic circuits is becoming increasingly popular [5]. Dynamic CMOS circuits are defined as those circuits which have an additional clock signal inputs along with the default combinational circuit inputs of the static systems. Dynamic systems are faster and efficient than the static systems. A fundamental difficulty with dynamic circuits is the monotonicity requirement. In the design of dynamic logic circuits numerous difficulties may arise like charge sharing, feed through, charge leakage, single-event upsets, etc. In this paper novel energyefficient self-resetting primitive gates followed by the design of adder logic circuits are proposed. Addition is a fundamental arithmetic operation that is generally used in many VLSI systems, such as application-specific digital signal processing (DSP) architectures and microprocessors [2]. This module is the core of many arithmetic operations such as addition/subtraction, multiplication, division and address generation. Such high performance devices need low power and area efficient adder
161
Vol. 3, Issue 1, pp. 161-174
circuits. So this paper presents a design construction for primitive gates and adder circuits which reduce delay and clock skew when compared to the dynamic logic adder implementation [9]. VDD VSGR Delay chain
Inputs
Nfet logic
Cx Vx Cout Vout
Fig.1. Basic Structure of A Self Resetting Logic Circuit
Another category of dynamic circuits, called self-resetting CMOS (SRCMOS), represents signals as short-duration pulses rather than as voltage levels. When a set of pulses are sent to the inputs to a logic gate, they must arrive at essentially the same time and they must overlap with one another for a minimum duration. After a logic gate has processed a set of input pulses, a reset signal is activated that restores the logic gate to a state in which it can receive another set of input pulses. The reset operation is timed to occur after the input pulses have returned to zero [11]. Thus, there is no need for an evaluate or foot transistor since the pull-down network will be off during the reset operation, and this is one of the factors that leads to high-speed operation. Moreover, since the reset occurs immediately after each gate has evaluated, there is no need for a separate precharge phase. Since short-duration pulses are hard to debug and test, special additional test-mode features are sometimes added for these purposes. Two types of reset structures have been proposed for use in SRCMOS. In globally self-resetting CMOS [4], the reset signal for each stage is generated by a separate timing chain which provides a parallel worst-case delay path. Individual reset signals are obtained at various tap points along this timing chain in such a way that the reset pulse arrives at each stage only after the stage has completed its evaluation. Very careful device sizing based on extensive simulations over process-voltage-temperature corners are required in order to ensure correct operation. Moreover, any extra delay margin that is designed into the timing chain simply reduces the throughput by a corresponding amount. On the other hand, in locally self-resetting CMOS [5], the reset signal for each stage is generated by a mechanism local to that stage. Previous implementations of this technique have been based on single-rail domino stages in which the reset signal is obtained by sending the stages own output signal through a short delay chain. Again, this technique requires very careful simulations and device sizing in order to ensure that the reset signals do not arrive too early. As with the other technique, any timing margin that is built in will directly limit the achievable performance.
II.
SELF RESETTING CMOS DYNAMIC LOGIC
Self-resetting logic is a commonly used piece of circuitry that automatically precharge they (i.e., reset themselves) after a prescribed delay. They find applications where a small percentage of gates switch in a cycle, such as memory decoder circuits. It is a form of logic in which the signal being propagated is buffered and used as the precharge or reset signal. By using a buffered form of the input, the input loading is kept almost as low as in normal dynamic logic while local generation of the reset assures that it is properly timed and only occurs when needed [6]. A generic view of a self-reset logic is shown in Fig.1. In the domino case, the clock is used to operate the circuit. In the self-resetting case, the output is fed back to the precharge control input and, after a specified time delay, the pull-up is reactivated. There is an NMOS sub block where the logic function performed by the gate is implemented which is represented as NMOS_LF through which the input
162
Vol. 3, Issue 1, pp. 161-174
datas are loaded. The output of the gate F provides a pulse if the logic function becomes true. This output is buffered and it is connected to PMOS structure to precharge [7]. The delay line is implemented as a series of inverters. The signals that propagate through these circuits are pulses. VDD VSGR=VDD 0v Delay chain
Inputs
Nfet logic
Cx Vx=0 VDD Cout Vout =0
Fig.2. Precharge
Dynamic logic circuits are widely used in modern low power VLSI circuits. These dynamic circuits are becoming increasingly popular because of the speed advantage over static CMOS logic circuits; hence they are widely used today in high performance and low power circuits. Normally in the design of flip-flops and registers, the clock distribution grid and routing to dynamic gates presents a problem to CAD tools and introduces issues of delay and skew into the circuit design process [1]. There are situations that permit the use of circuits that can be automatically precharge themselves (i.e., reset themselves) after a prescribed delays [7]. These circuits are called post charge or self-resetting logic which are widely used in memory decoders. VDD VSGR= 0v VDD Delay chain
Inputs
Nfet logic
Cx Vx=VDD 0v Cout Vout
Fig.3. Discharge
The width of the pulses must be controlled carefully or else there may be contention between NMOS and PMOS devices, or even worst, oscillations may occur. Self-resetting logic (SRL) can be classified as a variant of domino logic that allows for asynchronous operation [11]. A basic SRL circuit is shown in Figure. A careful inspection of the schematic shows that the primary differences between this gate and the standard domino circuit are
163
Vol. 3, Issue 1, pp. 161-174
(a) The addition of the inverter chain that provides feedback from the output voltage Vout (t) to the gate of the reset pFET MR, and (b) The elimination of the evaluation nFET. Note that an odd number of inverters are used in the feedback. As discussed below, the feedback loop has a significant effect on both the internal operation of the circuit and the characteristics of the output voltage. Precharging of Cx occurs when the clock is at a value =0 and the circuit conditions are shown in Figure 2. During this time, and which is identical to the event in a standard domino circuit. As we will see, the timing of the input signals precludes the possibility of a DC discharge path to ground by insuring that the inputs to all logic nFET are 0 during precharge. The voltage on the gate of MR is at a value of so that insures that MR is in cutoff during this time. The distinct features of SRL arise when a discharge occurs. The circuit conditions are shown in Fig.3
III.
ANALYSIS OF LOGIC GATES AND ADDERS
This section presents the basic construction and simulation of primitive gates and adders in all the specified logics above they are DYNAMIC LOGIC, SRCMOS LOGIC, and SELF RESETTING LOGIC. The cells shown in Fig. is a self-resetting implementation of 2-input primitive gates. The logical functions are implemented by the NMOS stack with two input signals A and B. The delay path in this circuit is implemented with single inverter. The mechanism of self-resetting in this circuit is achieved through the PMOS transistors. The implementation of AND/OR can be obtained by placing the NMOS stack in series and parallel connection, whereas the gates NAND/NOR can be implemented with De Morgans law, an OR gate with inverted input signals behaves as a NAND gate. Similarly an AND gate with inverted input signals behaves as a NOR gate [8].
3.1 Dynamic logic implementation in logic gates and adders

Dynamic logic is one of the low power technique which is applicable for digital circuits, in which the area will be reduced where as the power reduction is less compared to Self resetting logic. Analysis of basic logic gates and their applications is done. 3.1.1 And gate implementation
Fig.4. schematic of and gate by using dynamic logic
Fig.5. simulation results of and gate by using dynamic logic
164
Vol. 3, Issue 1, pp. 161-174
3.1.2 Xor gate
Fig.6. schematic of xor gate by using dynamic logic
Fig.7. simulation results of xor gate by using dynamic logic
3.1.3
Half adder
Fig.8. schematic of half adder by using dynamic logic
Fig.9. simulation results of half adder by using dynamic logic
165
Vol. 3, Issue 1, pp. 161-174
3.1.4 Full adder
Fig.10. schematic of full adder by using dynamic logic
Fig.11. simulation results of full adder by using dynamic logic
3.2 SRCMOS logic implementation in logic gates and adders Self resetting CMOS is one of the low power technique which is applicable for digital circuits, it yields high performance but having complexity circuitry, in which the area will be reduced where as the power reduction is less compared to Self resetting logic. Analysis of basic logic gates and their applications is done. 3.2.1 And gate
Fig.12. schematic of and gate by using SRCMOS logic
166
Vol. 3, Issue 1, pp. 161-174
Fig.13. simulation results of or gate by using SRCMOS logic
3.2.2
Xor gate
Fig.14. schematic of xor gate by using SRCMOS logic
Fig.15. simulation results of xor gate by using SRCMOS logic
167
Vol. 3, Issue 1, pp. 161-174
3.2.3 Half adder
Fig.16. schematic of half adder by using SRCMOS logic
Fig.17. simulation results of half adder by using SRCMOS logic
3.2.4
Full adder
Fig.18. schematic of full adder by using SRCMOS logic
168
Vol. 3, Issue 1, pp. 161-174
Fig.19. simulation results of full adder by using SRCMOS logic
3.3 Self Resetting logic implementation in logic gates and adders

Self resetting logic is a novel low power technique which is applicable for digital circuits, in which the power reduction is less compared to other mentioned low power techniques. Analysis of basic logic gates and their applications is done. 3.3.1 And gate
Fig.20. schematic of and gate by using Self Resetting Logic
Fig.21. simulation results of and gate by using self resetting logic
169
Vol. 3, Issue 1, pp. 161-174
3.3.2 Xor gate
Fig.22. schematic of xor gate by using Self Resetting Logic
Fig.23. simulation results of xor gate by using self resetting logic
3.3.3
Half adder
Fig.24. schematic of half adder by using Self Resetting Logic
170
Vol. 3, Issue 1, pp. 161-174
Fig.25. simulation results of half adder by using self resetting logic
3.3.4
Full adder
Fig.26. schematic of full adder by using Self Resetting Logic
Fig.27. simulation results of full adder by using self resetting logic
171
Vol. 3, Issue 1, pp. 161-174
IV.
COMPARATIVE ANALYSIS OF COMBINATIONAL LOGIC CIRCUITS
In half adder circuit the sum and carry is observed for this dynamic half adder the power dissipation is calculated and it is reduced up to 15% compared to CMOS logic. The total power dissipation for this circuit is 39.05 and in dynamic full adder circuit the power dissipation is calculated and it is reduced up to 15% compared to CMOS logic. The total power dissipation for this circuit is 56.90.
Table 1 Power dissipation and delays of dynamic logic circuit Topology AND OR XOR HALF ADDER FULL ADDER Rise delay ns(output node) 0.004 0.002 0.003 0.007 0.012 Fall delay ns(output node) 0.002 0.001 0.002 0.005 0.01 Power Dissipation (mW) 26.45 26.45 47.89 39.05 56.90
In half adder circuit the sum and carry is observed for this SRCMOS half adder circuit the power dissipation is calculated and it is reduced up to 20% compared to CMOS logic[12]. The total power dissipation for this circuit is 35.19 and in this SRCMOS full adder circuit the power dissipation is calculated and it is reduced up to 20% compared to CMOS logic. The total power dissipation for this circuit is 50.21.
Table 2 Power dissipation and delays of SRCMOS LOGIC Topology AND OR XOR HALF ADDER FULL ADDER Rise delay ns(output node) 0.003 0.002 0.005 0.012 0.025 Fall delay ns(output node) 0.005 0.005 0.004 0.033 0.014 Power Dissipation (mW) 24.21 24.21 43.35 35.19 50.21
In half adder circuit the sum and carry is observed for this SELF RESETTING LOGIC half adder circuit the power dissipation is calculated and it is reduced up to 35% compared to CMOS logic. The total power dissipation for this circuit is 30.17 and in this SELF RESETTING LOGIC full adder circuit the power dissipation is calculated and it is reduced up to 35% compared to CMOS logic. The total power dissipation for this circuit is 45.12
Table 3 Power dissipation and delays of SRL circuit Topology AND OR XOR HALF ADDER FULL ADDER Rise delay ns(output node) 0.005 0.004 0.015 0.024 0.050 Fall delay ns(output node) 0.001 0.001 0.002 0.022 0.043 Power Dissipation(mW) 23.12 23.12 36.67 30.17 45.12
V.
ANALYSIS OF POWER DISSIPATION

Table 4 Comparative Analysis of Power Dissipation Power dissipation DYNAMIC LOIC SRCMOS AND 26.45 24.21 OR 26.45 24.21 XOR 47.89 43.35 HALF ADDER 39.05 35.19 FULL ADDER 56.90 50.21
172
Vol. 3, Issue 1, pp. 161-174
LOGIC SRL 23.12 23.12 36.67 30.17 45.12
Chart 1 analysis of power dissipation 60 50 40 30 20 10 0 AND OR XOR HALF ADDER FULL ADDER DYNAMIC LOIC SRCMOS LOGIC SRL
IV.
CONCLUSION
In this paper, an exhaustive analysis for commonly used high-speed primitive gates and adder circuits using self-resetting logic is implemented in 45-nm CMOS technologies. The goal was to obtain a family of gates that could simplify the implementation of fast processing circuit which overcomes the restriction due the pulses being elongated and shortened as signal traverse the logic stages. In this paper exhaustive comparison between conventional DYNAMIC LOGIC, SRCMOS and SRL were carried in terms of its parasitic value, delays and power dissipation. It is observed that the proposed circuits have offered an improved performance in power dissipation, charge leakage and clock skew when compared to DYNAMIC LOGIC and SRCMOS logic with additional burden of silicon area. Hence, it is concluded that the proposed designs will provide a platform for designing high performance and low power digital circuits and high noise immune digital circuits such as, digital signal processors and multiplexers.
REFERENCES
[1] Woo Jin Kim, Yong-Bin Kim, A Localized Self-Resetting Gate Design Methodology for Low Power IEEE 2001. [2] M. E. Litvin and S. Mourad, Self-reset logic for fast arithmetic applications, IEEE Transactions on Very Large Scale Integration Systems, vol. 13, no. 4, pp. 462475, 2005. [3] L. Wentai, C. T. Gray, D. Fan, W. J. Farlow, T. A. Hughes, and R. K. Cavin, 250-MHz wave pipelined adder in 2-m CMOS, IEEE Journal of Solid-State Circuits, vol. 29, no. 9, pp. 11171128, 1994. [4] D. Patel, P. G. Parate, P. S. Patil, and S. Subbaraman, ASIC implementation of 1-bit full adder, in Proc. 1st Int. Conf. Emerging TrendsEng. Technol., Jul. 2008, pp. 463467. [5] M. Lehman and N. Burla, Skip techniques for high-speed carry Propagation in binary arithmetic units, IRE Trans. Electron.comput., vol. EC-10, pp. 691698, Dec. 1962. [6] R. A. Haring, M.S. Milshtein, T.I. Chappell, S. H. Dong and B.A. Chapell, "Self resseting logic and incrementer" in Proc. IEEE Int. Symp. VLSI Circuits, 1996 pp. 18-19. [7] G. Yee and C. Sechen, " Clock-delayed domino for adder and combinational logic design" in proc.IEEE/ACM Int. Conf. Computer Design, Oct., 1996, pp. 332-337. [8] P. Ng, P. T. Balsara, and D. Steiss, Performance of CMOS Differential Circuits, IEEE J. of Solid-State Circuits, vol. 31, no. 6, pp. 841-846, June 1996. [9] P. Srivastava, A. Pua, and L. Welch, .Issues in the Design of Domino Logic Circuits, Proceedings of the IEEE Great Lakes Symposium on VLSI, pp. 108-112, February 1998. [10] W. Zhao and Y. Cao. New generation of predictive technology model for sub-45nm design exploration, In IEEE Intl. Symp. On Quality Electronics Design, 2006 [11] CMOS Logic Circuit Design 2002 John P Uyemura [12] Prabhakara C. Balla and Andreas Antoniou Low Power Dissipation MOS Ternary Logic Family IEE journal of solid state circuits, vol sc-19 no.5 October 1984
173
Vol. 3, Issue 1, pp. 161-174
Authors
Suparshya Babu Sukhavasi was born in A.P, India. He received the B.Tech degree from JNTU, A.P, and M.Tech degree from SRM University, Chennai, Tamil Nadu, India in 2008 and 2010 respectively. He worked as Assistant Professor in Electronics Engineering in Bapatla Engineering College for academic year 2010-2011 and from 2011 to till date working in K L University . His research interests include antennas, FPGA Implementation, Low Power Design and wireless communications.
Susrutha Babu Sukhavasi was born in A.P, India. He received the B.Tech degree from JNTU, A.P, and M.Tech degree from SRM University, Chennai, Tamil Nadu, India in 2008 and 2010 respectively. He worked as Assistant Professor in Electronics Engineering in Bapatla Engineering College for academic year 2010-2011 and from 2011 to till date working in K L University . His research interests include antennas, FPGA Implementation, Low Power Design and wireless communications.
Vijaya Bhaskar Madivada was born in A.P, India. He received the B.Tech degree in Electronics & Communications Engineering from Jawaharlal Nehru Technological University in 2010. Presently he is pursuing M.Tech VLSI Design in KL University; His research interests include FPGA Implementation, Low Power Design.
B Rajesh Kumar was born in Gudivada, Krishna (Dist.,), AP, India. He received B.Tech. in Electronics & Communications Engineering from Jawaharlal Nehru Technological University in 2010 , A.P, India. Presently he is pursuing M.Tech. in VLSI from SITE. His research interest includes Low Power and design of VLSI circuits.
174
Vol. 3, Issue 1, pp. 161-174
HIGH QUALITY DESIGN AND METHODOLOGY ASPECTS TO ENHANCE LARGE SCALE WEB SERVICES
Suryakant B Patil1, Sachin Chavan2, Preeti Patil3
1 2&3
Department of Information Technology, MPSTME, NMIMS, Shirpur, India. Department of Computer Engineering, MPSTME, NMIMS, Shirpur, India.
ABSTRACT
Network traffic has increased tremendously due to the rise in the number of Internet users. This has affected several aspects and characteristics of large scale networks such as reduced network bandwidth, increased latency, and higher response time for users who require large scale web services. This paper proposes a novel design and methodology to address these issues. It does this by optimizing network bandwidth and reduces latency and response time for large scale Web applications. The methodology works by analysing content in the proxy cache, identifying content aliasing, duplicate suppression and by the creation of the respective soft links. Here it is necessary to understand the characteristics of the web traffic before one would propose any design to manage it better. For instance, an expanding network results into reduced network bandwidth, increased latency, and higher response time for users. The present solution makes intelligent use of the proxy cache server to overcome these problems. In this study proxies were designed to enable network administrators to control internet access from within intranet. But when proxy cache is used, there develops the problem of Aliasing. Aliasing in proxy server caches occurs when the same content is stored in the cache several times. The present methodology improves performance by avoiding storing the same content in cache multiple times those results in wastage of storage space. The investigation proposed here analysed the cache content of the proxy and checked the replication of content. This work thus proved the way to increase the efficiency of large scale web services. This high quality design approach focused on the content of the access logs and user habits as users would access large scale Web services.
KEYWORDS: Large Scale Web, Cache, Web Proxy, Mirroring, and Duplicate Suppression.
I.
INTRODUCTION
Web caching consists of storing most frequently access objects to the local server at our place instead of original server. Due to this reason web server can make better use of network bandwidth, reduced workload on web servers, improves the response time for users. The local servers used to store frequently referred objects are called as web proxy servers. A proxy server acts as a mediator between the original server and the clients. The proxy server setup can be seen in Figure 1. The proxy cache also stores all of the images and sub files for the visited pages, so if the user jumps to a new page within the same site that can be possible. Aliasing in proxy servers caches occurs due to same content is stored in cache multiple times. On the World Wide Web, aliasing commonly occurs when a client makes two requests, and both the requests have the same payload [3]. The major problem associated with using the web cache is storage space requires storing the visited pages with their objects [16]. This Paper is divided into 4 sections; section 1 is an introduction of proxy web cache system which briefs about the role of the Proxy server and setup of the proxy cache server between the client side browser and web servers. Further it list out the advantages and disadvantages of using Proxy cache. Section 2 covers the related work, the concepts of the web traffic, web cache, mirroring, MD5 Hashing Algorithm followed by static caching and dynamic caching are discussed. Section 3 elaborates the survey done for Data collection and reduction with experimental analysis. The data is gathered from various systems based on the user habits and requirement. It reflects the amount of the
175
Vol. 3, Issue 1, pp. 175-185
cache space saved for respective systems after reduction based on the category of the files. Section 4 consists of results and discussions based on experimental analysis. It shows the enhancement in the large scale web services by using the proxy cache for duplication avoidance with the high quality design and methodology aspects for various categories of the web data like image, HTML, formatted, Style and other files.
1. Client request web page from proxy Proxy caching server 2. Proxy request web page from web server 3. Web server sends the page to the proxy 4. Proxy caches the pages and return to client
Web server
Figure 1. Proxy server setup
1.1 Advantages of Caching

1. Web caching reduces the workload of the remote Web server 2. If the remote server is not available due to the remote server's crash or network partitioning, the client can obtain a cached copy at the proxy. 3. It provides us a chance to analyze an organization usage patterns.
1.2 Disadvantages of using a caching:

1. The main disadvantage is that a client might be looking at stale data due to the lack of proper proxy updating. 2. The access latency may increase in the case of a cache miss due to the extra proxy processing. 3. A single proxy cache is always a bottleneck. 4. A single proxy is a single point of failure.
II.
RELATED WORK
2.1 Mirroring
Shivkumar and Garcia-Molina investigated mirroring in a large crawler data set and reported that in the WebTV client trace far more aliasing happens than expected [11]. Similarly, Bharat et al. surveyed techniques for identifying mirrors on the Internet [17]. Bharat and Broder investigated mirroring in a large crawler data set and reported that roughly 10% of popular hosts are mirrored to some extent [17]. Broder et al. considered approximate mirroring or syntactic similarity [17]. Mirroring is defined as keeping multiple copies of the content of a Web site or Web pages on different servers using different domain names. A mirrored site is nothing but an exact replica of the original site and is updated frequently to ensure that it reflects the updates made to the content of the original site. The main purpose of mirroring is to build in redundancy and ensure high availability of web
176
Vol. 3, Issue 1, pp. 175-185
documents or objects. Mirrored sited also help make access faster when the original site is geographical distant.
2.2 Web Traffic

The amount of data sent and received by visitors to a website is web traffic. It is analysis to see the popularity of web sites and individual pages or sections within a site. Web traffic can be analyzed by viewing the traffic statistics found in the web server log file, an automatically generated list of all the pages served. Traffic analysis is conducted using access logs from web proxy server. Each entry in access logs records the URL of document being requested, date and time of the request, the name of the client host making the request, number of bytes returns to requesting client, and information that describe how the clients request was treated as proxy [1]. Processing these log entries can produce useful summary statistics about workload volume, document type and sizes, popularity of document and proxy cache performance [7].
2.3 Web cache

Web caching can play a valuable role in improving service quality for a large range of Internet users. A web cache is a mechanism for the temporary storage (caching) of web documents, such as HTML pages and images, to reduce bandwidth usage, server load, and perceived lag. A web cache stores copies of documents passing through it; subsequent requests may be satisfied from the cache if certain conditions are met. There are two types of Web caches a browser cache and a proxy cache [9]. A proxy cache is a shared network device that can undertake Web transactions on behalf of a client, and, like the browser, the proxy cache stores the content. Subsequent requests for this content, by this or any other client of the cache will trigger the cache to deliver the locally stored copy of the content, avoiding a repeat of the download from the original content source [6]. A client, such as a web browser, can store web content for reuse. For example, if the back button is pressed, the local cached version of a page may be displayed instead of a new request being sent to the web server. A web proxy sitting between the client and the server can evaluate HTTP headers and choose to store web content. A content delivery network can retain copies of web content at various points throughout a network. Performance of web cache system is measured in terms of utilization, bandwidth, efficiency, hits, throughput latency, availability and response time [14].
2.4 Static Caching

It is a new approach of web caching which uses yesterdays log to predict the todays user request. Static caching is performed only once in a day. It is simpler approach to use and requires very low CPU overhead. It improves cache performance by using compression techniques. Stable web pages can be compressed to reduce their size and frees up the cache space. Cache server performance typically relies on two major factors first is byte hit ratio and second one is hit ratio. Hit ratio is nothing but the percentage of all accesses that are fulfilled by data in cache; while, byte hit ratio is the hit rate with respect to the total number of bytes in the cache [8]. The static caching algorithm defines a fixed set of URLs by analyzing the logs of previous periods. It then calculates the value of the unique URL. Depending on the value, URLs are arranged in the descending order, and the URL with the highest value is selected. This set of URLs is known as the working set. When a user requests a document and the document is present in the working set, the request is fulfilled from the cache. Otherwise, the user request is fulfilled from the origin server [8].
2.5 Dynamic Caching

In contrast to dynamic caching where cached documents are updated more than once a day. As mentioned before, the benefit of current Web caching schemes is limited by the fact that only a fraction of web data is cacheable. Non-cacheable data is of a significant percentage of the total data. For example, measurement results show that 30% of user requests contain cookies. Dynamic caching is more complex than static caching and requires detailed knowledge of the application. One must consider the candidates for dynamic caching carefully since, by its very nature,
177
Vol. 3, Issue 1, pp. 175-185
dynamically generated content can be different based on the state of the application. Therefore, it is important to consider under what conditions dynamically generated content can be cached returning the correct response. This requires knowledge of the application, its possible states, and other data, such as parameters that ensure the dynamic data is generated in a deterministic manner [5].
III.
DATA COLLECTION, REDUCTION AND EXPERIMENTAL ANALYSIS
3.1 Document Types

While downloading files and viewing websites, youll meet up with many file formats. Most are common, and encountered frequently, others are rarer and require specialist programs to open or use [2]. Here, the categorized list of the documents and files are given below HTML: .html, .shtml, .htm, and .map . Image: .gif, .jpeg, gif89, .jpg, .png, .bmp, .jpg, .pcx, .rgb, .tif, and .ras. Audio: .au, .aiff, .aif, .aifc, .mid, .snd, .wav, and .lha. Video: .mov, .qt, .avi, .mpe, .movie, .mpeg, .mpg, .mp2, and .mp3. Compressed: .z, .gz, .zip, and .zoo. Formatted: .ps, .pdf, dvi, .ppt, .tex, .rtf, .src, .doc and .wsrc. Dynamic: .cgi, .pl, and .perl. When cache is use to store the object then the URL of web page is going to be cache in the cache memory, as the URL is stored the objects associated with the Page such as images, Audio, Video files are also stored at the same place for quick retrieval of that page for next time [10].
3.2 Changing of proxy server

In most of the organizations or institution server does not support the proxy cache, so it is difficult to use main server as cache server so we have to change the proxy server from main server to other server [4]. Following are the steps to switch machine to other proxy: 1. Open the browser for ex. Internet Explorer 2. In internet explorer pull down the Tools menu and click Internet Options... 3. Click the Connections tab: 4. click the LAN Settings... button: 5. In the Address: box change "proxy1 Address" to "proxy2 Address" or vice versa and click OK. 6. Click OK on the Internet Options dialogue box to get back to the browser screen and you will now be able to get external sites.
3.3 Duplication of Data

Duplication of data means storing the multiple copies of same data object. In case of cache when we cache the object or the webpage that web page is stored at cache memory but when the different users request the same page then the multiple copies of that object or web page is stored at cache memory which results in the wastage of storage space as we all know the maintenance of cache is an expensive task so such wastage is not affordable. To avoid the problem of duplication of the data objects or web page duplicate suppression mechanism is to be used [9]. If the duplicate copy of data is saved at proxy cache then it acquires more space of storage in the analysis part given in work shows that the effect of duplication in the cache space [6].
3.4 Duplicate Suppression

You can reduce storage space requirements by avoiding duplicating copies of the same data. Content Engine provides the option to suppress storage of duplicate content elements. Duplicate suppression applies to any kind of content. Incoming content is not added to the storage area if identical content exists in the storage area; only unique content is added [14]. Many web pages having different URLs are duplicates of each other. If cache finds the duplicate copy of requested page, it will reduce the network cost when the page is served from cache.
178
Vol. 3, Issue 1, pp. 175-185
Technique like data compression, caching and content simplification can be used to avoid unnecessary use of network bandwidth [6]. Due to large network size there are many pages on web, most of those pages will not be referenced multiple times by any one cache, means the probability with which the Kth page will be referenced is 1/K. re-referenced follow a distribution similar to Zipfs law [12]. Utility of web pages can be extended using several methods such as delta encoding, perfecting and partial transfer. Simple hit ratio is always higher than weighted hit ratio because the no of references to the smaller resource is more than those larger resources.
3.5 Experimental Analysis

The experimentation carried out at the lab of our institute. Five systems from the lab are under the observation for 3 months. Those systems are use to analyse for access of internet to check the user interest and habits. File types such as .XML, .TXT, .PL, .GIF, .HTML, .CH, .CSS, .PNG, .PHP, and .BMP considered for analysis [2]. Size of the file given in table is in kb. The result of this analysis for five systems are summarised in table 1.
Table 1. Size of content According to type. Type of files System 1 2 3 4 5 AVG XML 586 0 604 0 0 1190 TXT 35 26 142 17 0 220 Pl 202 0 0 0 107 309 JPEG 1492 981 151 77 516 3217 GIF 464 312 126 44 196 1142 HTML 423 772 608 174 326 2303 CH 259 0 248 0 0 507 .CSS 1169 776 387 315 423 3070 .PNG 495 163 252 487 180 1577 PHP 131 0 0 0 50 181 BMP 0 361 49 0 0 410
Table 1 show that Style and image documents account for close to 60 - 65 percent. Unlike the web server workloads, however Table 1 show that image files consistently more requested document type (40-43 percent), followed by Style documents (about 20-22 percent). It has been observed that some of the file systems are not present on all the systems. Most of the cache space is acquired by the JPEG files.
Table 2. Reduced size of content on Disk. Type of files System 1 2 3 4 5 AVG XML 566 0 594 0 0 1160 TXT 1 6 3 1 0 11 Pl 192 0 0 0 97 289 JPEG 757 626 118 20 281 1802 GIF 215 177 18 32 45 487 HTML 300 558 539 150 93 1640 CH 216 0 248 0 0 464 .CSS 1065 744 205 300 311 2625 .PNG 385 131 135 418 161 1230 PHP 69 0 0 0 44 113 BMP 0 361 13 0 0 374
Table 2 show effect of reduction on the content stored at cache. We get much reduction effect in case of image files. If we discard the duplicate copies of the image files then we can save 50 percent of space required to store that image files, where the reduction in XML, PL and CSS files is very less. Reduction in number of copies of same files decrease the access latency time for the proxy cache [15].
179
Vol. 3, Issue 1, pp. 175-185
1600 1400 1200 1000 800 600 400 200 0
System 1
Before
After
Size (KB)
Type of files
Figure.2 Type of files present on system 1
Figure 2 shows the type of data present at the cache of first system, from Figure 2 we can say that the maximum size occupied of first system is by image files. And the files like TXT and BMP is completely absent. That shows the user interest and habits. Here first bar shows the actual size of files with replication where second bar shows the size if we discard replicated data. Aging daemon which generates dummy references to all documents to age their reference rate estimates [18]. From Figure 2 it is clear that in image file of type jpeg we get 50 percent of space is saved if we discard the replicated data. Wherein the space saved in gif file is also same i.e. near about 50 percent. In case of HTML and png it is less about 10-20 percent. Wherein case of XML, CSS, and Channel it is negligible. So after reduction of image files we can save the 50 percent storage space occupied by image files. Kai Cheng, Kambayashi, Y., Mohania M., shows that more than 60% of total cache contents are never used and the low utilization rate is largely due to users' unawareness of the cache contents [19].
1200 1000
Size(KB)
System 2
Before
After
800 600 400 200 0
Types of files
Figure 3. Type of files present on system 2
Figure 3 shows the type of data present at the cache of second system, from Figure 3 we can say that the cache occupies less space as compare to system 1. And the files like TXT, XML, PL, CHANNEL and PHP is completely absent. That shows the user interest and habits. Here first bar shows the actual size of files with replication where second bar shows the size if we discard replicated data. Storing of dynamic content in cache is not useful because that content does not used again for second time [5]. From Figure 3 it is clear that in image file of type jpeg and gif we get 40 percent of space is saved if we discard the replicated data. In case of HTML file we get 20 percent. Wherein case of PNG, CSS,
180
Vol. 3, Issue 1, pp. 175-185
and BMP it is negligible. So after reduction of files we can save 20-30 percent storage space occupied by files. At the second system most of the files are not present so the actual size occupied by cache is less but we can observe that there is replication of data. Dynamic content cache stores new copy from web server each time [5].
700 600 500

Size(KB)
System 3
Before
After
400 300 200 100 0
Type of files
Figure 4 shows the type of data present at the cache of third system, from Figure 4 we can say that the maximum size occupied is by XML and HTML files. And the files like PL and PHP is completely absent. That shows the user interest and habits. Here first bar shows the actual size of files with replication where second bar shows the size if we discard replicated data [13]. From Figure 4 it is clear that the space is saved by discarding the replicated data is negligible in case of XML and HTML files. We save maximum size by discarding duplicates of GIF and TXT files about 80-90 percent. Wherein the space saved in JPEG, HTML and BMP file is also near about 10-20 percent. Wherein case of PNG, CSS, and BMP it is 30-40 percent. So after reduction of files there is no much effect in storage space requirement.
600 500
Size (KB)
System 4
Before
After
400 300 200 100 0
Type of Files
System 4 also has the less overhead on cache size because some of the files are not present in system 4 cache such as XML, PL, CHANNEL, PHP and BMP. And the files which are present on system 4 have no much effect of reduction. In case of jpeg there is maximum effect of reduction i.e. 70 percent. Wherein case of GIF, HTML, CSS and PNG it is only 10-20 percent. From the access log of system 4
181
Vol. 3, Issue 1, pp. 175-185
we predict that there is variety of content is surf thats why there is no much effect of reduction on cache storage [6].
600 500
System 5
Berore
After
Size (KB)
400 300 200 100 0
Type of files
At system 5 we observe that XML, TXT, CHANNEL and BMP files are not present. Maximum size of cache is occupied by JPEG and CSS files. If we avoid the duplication of JPEG object then we can save 40 percent of storage space occupied by JPEG files [13]. In case of GIF and HTML the effect of reduction is better i.e. 70-80 percent. In case of PNG, PHP and Pl files the effect of reduction is very less nearly 10 percent. If avoid duplication of CSS content then we save 20 percent of storage space required to store CSS files.
IV.
Figure 7 shows the overall cache space occupied by all the files present on all the five systems as well as the effect of reduction on cache space. From Figure 7 we can say that by reducing the XML content we can save 40 percent of cache size. In case of TXT files it is 50 percent. As we observe the graph for PL, PHP and BMP files then there is no effect of reduction. JPEG and CSS files requires maximum of cache space and the effect of reduction is 50 percent in case of JPEG files and 40 percent in case of CSS files. Maximum effect of reduction is seen in HTML and PNG files i.e. 60 percent. Cache space occupied by GIF files is less but the effect of reduction is much i.e. 50 percent.
Figure.7 Average size of files on all Systems
Figure 7 shows the effect of reduction on the files present in cache of all systems from the statistics shown in figure 7 we get the effect of duplicate suppression. Duplicate suppression saves the cache space as well as the access latency is decreases for cache described in [15].
182
Vol. 3, Issue 1, pp. 175-185
9000 8000 7000 6000 5000 4000 3000 2000 1000 0 HTML IMAGE STYLE Type of files FORMATTED OTHER
Size (KB)
Figure.8 Average size of files on all Systems
Figure 8 shows the comparative graph of type of files when it is grouped category wise. Files such as GIF, JPEG, BMP and PNG is categorised into image files. CSS and PL files categorised as style file. And the remaining files in the category of other. From Figure 8 we can predict that the image files are most requested content in all. Image files occupy 43 percent of total storage space of cache. And also the effect of reduction has greater significance in saving of storage space. Major portion of storage space is occupied by the image, HTML and Style files 80-90 percent. It is clear that maximum size of cache is occupied by image file and also the effect of duplicate suppression is higher in case of image files. So the reduction in image files helps us to save the cache space as well as the access latency. By using the concept of duplicate suppression we save the space required by the files along with that the access latency at proxy is also decrease. If we store multiple copies of same object that increase the search time for lookup of specific files that increase the response time for user.
V.
CONCLUSION
The analysis based experimental results proves the need for methodology that improve the web access performance to enhance bandwidth utilization and greater connectivity speed. Here the suggested Design aspects improves the web performance in terms of reduced traffic, reduced latency, improved user response time, and optimal use of the existing bandwidth by using web caching. Content aliasing successfully detected using a web-based application, database queries and files system calls even in the large scale web applications. A considerable amount of duplicate storage can be avoided through the suggested methodology. This methodology is useful to avoid content aliasing at proxy cache that saves the unnecessary cache space required to store multiple copies of same content. It is also helpful in reducing the access latency and increase the response time of proxy cache. This methodology helps to web proxy caches to improve the performance during large web services. This work can be further optimize by the Daemon Process, which can be design and run periodically to check the consistency of the data cached and the data at the web server. This can be scheduled during the slack time with the less traffic which will not add any additional toll on the bandwidth as well as it updates the TTL Time to Live Period of the cached data; which results in the more cache hits with fresh data.
ACKNOWLEDGEMENTS
This paper would not be possible without the continuous guidance and valuable inputs of Dr. Tapan Bagchi, Director SVKMs NMIMS Shirpur Campus. The additional support taken from the following corporate professionals, for required data and design aspects based on their experience which creates innovative methodologies and data management solutions to help accelerate business breakthroughs and deliver outstanding cost efficiency: Mr. Sachin Punadikar, Senior Staff Software Engineer, NAS Team, India Software Lab, IBM India. Ozone2, Pune; Mr. Pramod Ghorpade, Dev Manager at NetApp, Bengaluru; Mr. Vinod Ghorpade, Senior Practice Manager at Wipro Technologies, Reading, United Kingdom (UK), Mr. Sujay Ghorpade, SAP PS/c Projects Consultant, Robert Bosch Engineering & Business Solutions, Bangalore.
183
Vol. 3, Issue 1, pp. 175-185
REFERENCES
Kartik Bommepally, Glisa T. K., Jeena J. Prakash, Sanasam Ranbir Singh and Hema A Murthy Internet Activity Analysis through Proxy Log IEEE, 2010. [2] Jun Wu; Ravindran, K., "Optimization algorithms for proxy server placement in content distribution networks," Integrated Network Management-Workshops, 2009. [3] Ngamsuriyaroj, S. ; Rattidham, P. ; Rassameeroj, I. ; Wongbuchasin, P. ; Aramkul, N. ; Rungmano, S. Performance Evaluation of Load Balanced Web Proxies IEEE, 2011. [4] E-Services Team, Changing Proxy Server by the Robert Gordon University, School hill, Aberdeen, Scotland-2006. [5] Chen, W.; Martin, P.; Hassanein, H.S., "Caching dynamic content on the Web," Canadian Conference on Electrical and Computer Engineering, 2003, vol.2, no., pp. 947- 950 vol.2, 4-7 May 2003. [6] Sadhna Ahuja, Tao Wu and Sudhir Dixit On the Effects of Content Compression on Web Cache Performance, Proceedings of the International Conference on Information Technology: Computers and Communications, 2003. [7] Mark S. Squillante, David D. Yaot and Li Zhang Web Traffic Modeling and Web Server Performance Analysis Proceedings of the 38' Conference on Decision & Control Phoenix, Arizona USA December 1999. [8] C. E. Wills and M. Mikhailov, Studying the Impact of More Complete Server Information on Web Caching, Computer Communications, vol. 24, no. 2, pp. 184.190, May 2000. [9] J Wang A Survey of Web Caching Schemes for the Internet - Cornell Network Research Group (C/NRG), Department of Computer Science, Cornell University 1999. [10] A. Mahanti, C. Williamson, and D. Eager, Traffic Analysis of a Web Proxy Caching Hierarchy, IEEE Network Magazine, May 2000. [11] N. Shivakumar and H. Garcia-Molina, Finding near Replicas of Documents on the Web Proc. Workshop on Web Databases, Mar. 1998. [12] L. Breslau, P. Cao, L. Fan, G. Phillips, and S. Shenker. Web caching and Zipf like Distributions: Evidence and Implications. In Proc. Infocom 99. New York, NY, March, 1999. [13] Jeffrey C. mogul A trace-based analysis of duplicate suppression in HTTP, Compaq Computer Corporation Western Research Laboratory, Nov. 1999. [14] Guerrero, C.; Juiz, C.; Puigjaner, R.; "Web Performance and Behavior Ontology," Complex, Intelligent and Software Intensive Systems, 2008. CISIS 2008. International Conference on, vol., no., pp.219-225, 4-7 March 2008. [15] Wei-Kuo Liao, Chung-Ta King Proxy Pre-fetch and Prefix Caching Parallel Processing, International Conference on, 2001. , vol., no., pp. 95- 102, 3-7 Sept. 2001. [16] Triantafillou, P.; Aekaterinidis, L.; "Web proxy cache replacement: do's, don'ts, and expectations," Network Computing and Applications, 2003. NCA 2003. Second IEEE International Symposium on, vol., no., pp. 5966, 16-18 April 2003. [17] K. Bharat, A. Broder, J. Dean, and M. R. Henzinger, A Comparison of Techniques to Find Mirrored Hosts on the WWW, IEEE Data Eng. Bull., vol. 23, no. 4, 2000. [18] Shim, J.; Scheuermann, P.; Vingralek, R.; "Proxy cache algorithms: design, implementation, and performance," Knowledge and Data Engineering, IEEE Transactions on , vol.11, no.4, pp.549-562, Jul 1999. [19] Kai Cheng; Kambayashi, Y.; Mohania, M.; , "Efficient management of data in proxy cache," Database and Expert Systems Applications, 2001. Proceedings. 12th International Workshop on , vol., no., pp.479-483, 2001. Authors Suryakant B Patil was born in Maharashtra, India, in 1975. He received the Bachelor and the Masters both in Computer Engineering, both from the Walchand College of Engineering, Sangli. He is currently pursuing the Ph.D. degree with the Department of Computer Engineering, NMIMS Mumbai and he has submitted his final thesis of the same. His research interests include Software Engineering, Project Management, MIS, Web Engineering, Mobile Communication and information Security. Sachin D Chavan was born in Maharashtra, India, in 1987. He received the Bachelor in Computer Engineering from North Maharashtra University, Maharashtra and pursuing his Masters in Computer Engineering from the NMIMS Mumbai. His research interests include Web Engineering, Design and Analysis of Algorithms, Computer Networking, Operating [1]
184
Vol. 3, Issue 1, pp. 175-185
Systems and Software Engineering. Preeti S Patil was born in Karnataka, India, in 1977. She received the Bachelor in Computer Engineering from VTU University and the Masters in Computer Engineering from Mumbai University. She is currently pursuing the Ph.D. degree with the Department of Computer Engineering, NMIMS Mumbai and she has submitted her final thesis of the same. Her research interests include Web Engineering, Mobile Communication, information Security, Databases, Data Warehousing and Mining.
185
Vol. 3, Issue 1, pp. 175-185
ENERGY EFFICIENT CLUSTER BASED KEY MANAGEMENT TECHNIQUE FOR WIRELESS SENSOR NETWORKS
T. Lalitha1 and R. Umarani2
2
Research Scholar, Bharatiar University, Coimbatore, Tamilnadu, India Research Supervisor, Bharatiar University, Coimbatore, Tamilnadu, India
A BSTRACT
Wireless Sensor Networks (WSN) is vulnerable to node capture attacks in which an attacker can capture one or more sensor nodes and reveal all stored security information which enables him to compromise a part of the WSN communications. Due to large number of sensor nodes and lack of information about deployment and hardware capabilities of sensor node, key management in wireless sensor networks has become a complex task. Limited memory resources and energy constraints are the other issues of key management in WSN. Hence an efficient key management scheme is necessary which reduces the impact of node capture attacks and consume less energy. In this paper, we develop a cluster based technique for key management in wireless sensor network. Initially, clusters are formed in the network and the cluster heads are selected based on the energy cost, coverage and processing capacity. The sink assigns cluster key to every cluster and an EBS key set to every cluster head. The EBS key set contains the pairwise keys for intra-cluster and inter-cluster communication. During data transmission towards the sink, the data is made to pass through two phases of encryption thus ensuring security in the network. By simulation results, we show that our proposed technique efficiently increases packet delivery ratio with reduced energy consumption.
K EYW ORDS: Wireless Sensor Networks, Key Management, Data Transmission, Attacks, Cluster
I.
1.1
INTRODUCTION
Wireless Sensor Network
A network comprising of several minute wireless sensor nodes which are organized in a dense manner is called as a Wireless Sensor Network (WSN). Every node estimates the state of its surroundings in this network. The estimated results are then converted into the signal form in order to determine the features related to this technique after the processing of the signals. Based on the multi hop technique, the entire data that is accumulated is directed towards the special nodes which are considered as the sink nodes or the Base Station (BS). The user at the destination receives the data through the internet or the satellite via gateway. The use of the gateway is not very necessary as it is reliant on the distance between the user at the destination and the network [1]. For supervising the physical world, the wireless sensor networks are the promising technology. In order to collect the data from the surrounding in a sensor network application, several minute sensor nodes are organized and collaborated. Sensing modals like image sensors are placed in every node and this possess the ability to communicate in the wireless environment [2].
II. ENERGY EFFICIENT CLUSTER BASED KEY MANAGEMENT TECHNIQUE

2.1 Cluster Formation
In the wireless sensor network, after the nodes are deployed in the physical environment, they first report to the base station their physical locations, and then the network starts to select cluster heads. According to the cluster head selection algorithm, each node decides if it is capable of serving as a cluster head based on the following selection criteria: a) High Energy Resources b) Wide Communication Range c) High Processing Capacity
186
Vol. 3, Issue 1, pp. 186-190
For the authentication process, the encryption mechanism is carried on. When the selection criteria are satisfied by a particular node, it is capable of being the cluster head. So, this node, Ni broadcasts a Cluster head beacon (CH_BEACON) packet. The CH_BEACON packet is encrypted with a key called as the primary key, Kpri. Ni
Kpri(CH_BEACON)
broadcast
When the neighboring nodes Si receive this message, a cluster head reply (CH_REPLY) message is sent to the node, Ni by the nodes which intend to join the cluster. The reply message contains the ID and the response content Ack.
CH_REPLY
Ni
Kpri( ID{Si}|| Ack)
Si
If the number of reply messages received by Ni is greater than a threshold Rth, then Ni can be selected as the cluster head, CH. If the number of reply messages received by Ni is greater than a threshold Rth, then Ni can be selected as the cluster head, CH. Finally, the cluster head assigns IDs to all its member nodes that intend to join the cluster.
III. EBS CONSTRUCTION

An EBS consists of several subsets of the member set collection. In the EBS, every subset is analogous to a particular key and the nodes which possess the key become the element of the subset. The dimension of the EBS is represented by (N, K, M) and it depicts a condition of a N membered secure group with numbering from 1 to N and a separate key is maintained for every subset by the key server. In EBS, if there exists a subset Ai, then every member of this subset will have knowledge about the key Ki. In EMS, there are M elements for every t [1, N] and its union is equal to [1, N] {t}. Hence, any member t can be ejected by the key server. Then re-keying is performed to enable every member to know the replacement keys for the K keys. To perform this, the M messages are multicast after encrypting them with the keys which correspond to the M elements, which has a union equal to [1, N] {t}. To restrict decipherability to selected members, encryption of every key is performed by its predecessor. A canonical enumeration technique is made use of, for the construction of EBS subsets. In the formation of subset of K objects out of K + M object set, every feasible method is taken into consideration. Matrix A is formed in order to develop a bit string sequence in its canonical (K, M), in which the K and M are already known, C (K + M, K) columns indicate the successive bit strings of which has a length of K+M objects, where K ones are present in each. For EBS (N, K, M), A is known as the canonical matrix. For instance, the canonical matrix A for EBS(8, 3, 2) enclose the enumeration of all C(5, 3) ways to form a subset of 3 keys from 5 keys, as shown in Table 1. Enumeration matrix for EBS(8,3,2)
187
Vol. 3, Issue 1, pp. 186-190
Every row in the table corresponds to a subset Ti after the construction of the matrix A, where an entry 1 in the row indicates that the corresponding node is present in the subset. Since N = 8, M9 and M10 are not useful, in Table 1, T1 = [5, 6, 7, 8], T2 = [2, 3, 4, 8], T3 = [1, 3, 4, 6, 7], T4 = [1, 2, 4, 5, 7], and T5 = [1, 2, 3, 5, 6, 8]. It is easy to prove: [1,8] [1] = T1 U T2, [1,8] [2] = T1 U T3, [1,8] [3] = T1 U T4, Hence, on the exit of any node in the network information about the keys will be updated only by two node subsets. In this protocol, only five management keys are necessary whereas 15 keys are necessary in case of LKH. This in turn minimizes the key computation and also saves space for storage. During the construction of the EBS(N, K, M) model in this protocol, the values of the parameters N, K and M are raised in order to facilitate the production of larger number of management keys. Later on, the spare keys are used for the new nodes of the cluster
IV. SIMULATION RESULTS

The proposed Energy Efficient Cluster Based Key Management (EECBKM) technique is evaluated through NS2 simulation. We consider a random network of 100 sensor nodes deployed in an area of 500 X 500m. Two sink nodes are assumed to be situated 100 meters away from the above specified area. In the simulation, the channel capacity of mobile hosts is set to the same value: 2 Mbps. The simulated traffic is CBR with UDP. The number of clusters formed is 9. Out of which, we transmit data from 4 cluster heads to the sink. 3 sensor nodes in each cluster are sending data to their cluster head. The attacker nodes are varied from 2 to 10.
Table 1 summarizes the simulation parameters used No. of Nodes 100 Area Size 500 X 500 Mac 802.11 Routing protocol EECBKM Simulation Time 50 sec Traffic Source CBR Packet Size 512 bytes Rate 250kb Transmission Range 250m No of clusters 1,2,3 and 4 sending data No. of nodes per 3 cluster sending data Transmit Power 0.395 w Receiving power 0.660 w Idle power 0.035 w Initial Energy 17.1 Joules No. of Attackers 2,4,6,8 and 10
4.1 Performance Metrics

The performance of EECBKM technique is compared with the SecLEACH scheme. The performance is evaluated mainly, according to the following metrics. Average Packet Drop: The number of packets dropped due to various attacks is averaged over all surviving data packets at the destination. Average Packet Delivery Ratio: It is the ratio of the number .of packets received successfully and the total number of packets transmitted. Energy: It is the average energy consumed for the data transmission.
188
Vol. 3, Issue 1, pp. 186-190
V.
RESULTS
5.1 Based on Attackers

In our initial experiment, we vary the number of attackers as 2,4,6,8 and 10 from various clusters performing node capture attacks.
Attackers Vs Delivery Ratio 0.45 0.445 0.44 0.435 0.43 0.425 0.42 2 4 6 8 10
Delivery Ratio
EECBKM SecLEACH
Attackers
Figure 3: Attackers Vs Delivery Ratio

Attackers Vs Drop 24000 23000 22000 21000 20000 19000 2 4 6 8 10
Packets
EECBKM SecLEACH
Attackers
Figure 4: Attackers Vs Packet Drop

Attackers Vs Energy 20 Energy(J) 15 10 5 0 2 4 6 8 10 SecLEACH EECBKM
Attackers
Figure 5: Attackers Vs Energy
When the number of attackers is increased, naturally the packet drop will increase there by reducing the packet delivery ratio. Since EECBKM reduces node capture attacks, the amount of packet drop is less, when compared with the existing schemes. Figure 3 and 4 give the packets drop and packet delivery ratio when the attackers are increased.
VI. CONCLUSION
In this paper, we have developed an efficient technique for key management in the wireless sensor network. During the formation of a cluster, initially a cluster head is selected based on eligibility criteria such as energy cost, coverage and processing capacity. After the cluster head selection, the
189
Vol. 3, Issue 1, pp. 186-190
information about all the members of the cluster is sent to the sink by the cluster head. The sink then provides the cluster head with the cluster key and the EBS key set required for the communication between the nodes. These keys are distributed to the nodes by the cluster head prior communication. After the key distribution, secure channel is established between the nodes and the cluster head. During the data transmission from the cluster members to the sink, the data passes two phases. In the first phase the data is encrypted and transmitted to the cluster head. In the second phase, the data is encrypted by another key by the cluster head and then transmitted to the sink. Thus this technique allows inter cluster as well as intra cluster communication in a very efficient manner with high security. By simulation results, we have shown that our proposed technique efficiently increases packet delivery ratio with reduced energy consumption.
REFERENCES
[1]. Lina M. Pestana Leo de Brito and Laura M. Rodrguez Peralta, An Analysis of Localization Problems and Solutions in Wireless Sensor Networks, Polytechnical Studies Review, 2008, Vol VI, ISSN: 1645-9911. [2]. Huang Lee and Hamid Aghajan, Collaborative Self-Localization Techniques for Wireless Image Sensor Networks, In Proc. of Asilomar Conf. on Signals, Systems and Computers, Oct. 2005. [3]. D.Saravanan , D.Rajalakshmi and D.Maheswari DYCRASEN: A Dynamic Cryptographic Asymmetric Key Management for Sensor Network using Hash Function, International Journal of Computer Applications (0975 8887) Volume 18 No.8, March 2011 [4]. Yoon-Su Jeong, and Sang-Ho Lee Secure Key Management Protocol in the Wireless Sensor Network, International Journal of Information Processing Systems, Vol.2, No.1, March 2006. [5]. Mohammed A. Abuhelaleh and Khaled M. Elleithy SECURITY IN WIRELESS SENSOR NETWORKS: KEY MANAGEMENT MODULE IN SOOAWSN, International Journal of Network Security & Its Applications (IJNSA), Vol.2, No.4, October 2010.
Authors:
Lalitha. T received a Masters Degree in Computer Applications in 2000 in Vysya College, Salem and received a M.Phil(CS) in 2004 Bharathidasan University, Trichy. She now doing Ph.d Part-Time in Bharatiar University, Coimbatore. She is also working as a Senior Assistant Professor in Department of MCA in Sona College of Technology, Salem. Her research interests include network security, network Simulation as well as validation and verification techniques. She has Published 14 Papers in National and International Journals and published a Book Problem Solving Techniques.
UmaRani. R. has completed her M.C.A., from NIT, Trichy in 1989. She did her M.Phil from Mother Teresa University, Kodaikanal. She received her Ph.D., from Periyar University, Salem in 2006. Her topic for research was information security. Her area of interest includes information security, data mining and mobile communications. She has published about 50 papers in national and international conferences. She has produced 20 M.Phil., (Computer Science) candidates. She is currently guiding 5 Ph.D., (Computer Science) research scholars. She is also working as Associate Professor in the Department of computer Science, Sri Sarada College for women, Salem. She has Published 35 Papers in International and Ntional Journals and 55 Papers in National and International Conferences.
190
Vol. 3, Issue 1, pp. 186-190
ON PRODUCT SUMMABILITY OF FOURIER SERIES USING MATRIX EULER METHOD

1
1
B.P.Padhy, 2Banitamani Mallik, 3U.K.Misra and 4Mahendra Misra
Department of Mathematics, Roland Institute of Technology, Berhampur, Odisha 2 Department of Mathematics, JITM, Paralakhemundi, 3 P.G.Department of Mathematics, Berhampur University, Odisha 4 Principal, N.C. College (Autonomous), Jajpur, Odisha
ABSTRACT
In this paper, a theorem on product summability of Fourier series using Matrix-Euler method is proved.
K EY W ORDS: A - mean, A(E , z ) - product mean, Fourier series.

MATHEMATICS SUBJECT CLASSIFICATION: 42B05, 42B08.
I.
Let
INTRODUCTION
be a given infinite series with the sequence of partial sums
{s n } . Let
A = (a mn ) be a
matrix .Then the sequence to-sequence transformation (1.1)
t m = a m s , m = 1,2,L
=0
defines the sequence {t m } of the A -mean of the sequence (1.2) then the series t m s , as m ,
{s n } . If
a
n =0
is said to be A summable to s .
The conditions for regularity of A -summability are easily seen to be[3] (i) sup a mn < H where H is an absolute constant.
m
(ii) lim a mn = 0
m
(iii) lim
If (1.5) then the series Let
a
n =0
mn
=1
n
(E , z ) = Enz
n n z s s , as n . =0 is said to be summable (E, z ) to a definite number s . = 1 (1 + z )n
191
(1.6) then the series
Tn =
k =0
z (1 + z )
k =0
a nk
s s as n ,
is said to be summable to s by the A(E , z ) method .
It is known [1] that (E, z ) is regular. It is supposed that the method A(E , z ) is regular through out this paper. Let f (t ) be a periodic function with period 2, integrable in the sense of Lebesgue over (-,) then
a0 + (a n cos nt + bn sin nt ) = An (t ) 2 n =1 n =0 is the Fourier series associated with f . We use the following notation through out this paper (t ) = f ( x + t ) + f ( x t ) 2 f ( x),
(1.7)
f (t ) =
K n (t ) =
1 2
(1 + z )
k k =0 =0
a nk
k k z
1 sin + t 2 . t sin 2
II.
KNOWN THEOREM
Dealing with ( N , p n )(E , z ) method of a Fourier series, Nigam,et.al[2] proved the following theorem: THEOREM-2.1: Let {p n } be a positive, monotonic, non-increasing sequence of real constants such that
Pn = p as n .
=0
If
(2.1)
t (t ) = (u ) du = O , as t +0 1 0 t
t
and (2.2)
( n) as n
where ( t ) be a positive, non-increasing function of t , then the Fourier series summable ( N , p n )(E , z ) to f (x) at the point t = x . In this paper, we have generalized it to A(E , z ) summability of Fourier series.
A (t )
n n =0
is
III.
MAIN THEOREM
Let A = (a mn ) be a regular matrix and
THEOREM -3.1:
192
Vol. 3, Issue 1, pp. 191-196
t (t ) = (u ) du = O , as 0 1 t
t
(3.1)
t +0
where ( t ) is positive ,non-increasing function of t and (3.2) ( n) as n , then the Fourier series
A (t ) is summable A(E, z )
n n =0
at the point t .
IV.
REQUIRED LEMMAS
1 . n +1
We require the following Lemmas to prove the theorem. LEMMA -4.1:
K n (t ) = O ( n )
PROOF: For 0 t
,0 t
1 , we have sinnt nsint then n +1

n
K n (t )
1 2
1 2
(1 + z )
k k =0 =0
n
a nk
k k z
(2 + 1)sin t
t sin 2
(2k + 1) k k k (1 + z )k z =0 k =0 1 n a nk (2k + 1) k = (1 + z )k (1 + z ) 2 k =0 (2n + 1) n a =

a nk 2 = O(n) .
k =0
nk
LEMMA-4.2:
1 1 K n (t ) = O , for t . n t
PROOF: For
1 t t t , we have by Jordans lemma, sin , sin nt 1 . Then n 2

1 2
n
K n (t )
(1 + z )
k k =0 =0
a nk
k k z
1 sin + t 2 t sin 2
1 2
(1 + z )
k k =0 =0
a nk
k k z t
193
Vol. 3, Issue 1, pp. 191-196
1 = 2t
a (1 + z ) (1 + z )
nk k k =0
1 = O . t
V.
PROOF OF THE THEOREM 3.1
If s n is the n-th partial sum of the Fourier series Lebesgue theorem, following Titchmarch [4] we have
A (t ) of
n n =0
f (t ) , then by using Riemann-
1 sin n + t 1 2 s n - f (x) = dt (t ) 2 0 t sin 2 z Thus, the (E, z ) transform E n of s n is given by
E nz f ( x) =
t z 2 (1 + z ) sin
n 0 k =0
(t )
2
n k
1 sin k + t dt . t
If Tn denote the A(E , z ) transform of s n , we then have
1 Tn f ( x) = 2
a nk (t ) k k k 1 (1 + z) k t =0 z sin + 2 t dt k =0 0 sin 2
n
= (t ) K n (t ) dt
0
In order to prove the theorem, under an assumption, it is sufficient to show that
(t ) K
0
(t ) dt = 0(1) as
For
0 < < ,we have Tn f ( x) = (t ) K n (t ) dt

0
n = + + (t ) K n (t ) dt . 0 1/ n = I 1 + I 2 + I 3 , say
1
Now
1/ n
I1
(t ) K
0 1/ n 0
1/ n n
(t ) dt
(t )
0
K n (t ) dt.
O(n) (t ) dt , Using Lemma -1 1 , using (3.1). = O(n)O n (n)
194
Vol. 3, Issue 1, pp. 191-196
1 = O (n) , as n . = O(1) , as n , using (3.2).
Next
I2
1 n
(t ) K n (t ) dt
(t ) = O dt , using lemma -2 t 1 / n (t ) (t ) = O + 2 dt . t 1 / n 1 / n t n 1 1 + O = O O u (u ) du , where u=1/t 1 1/ t 1/ n and 0 < < 1 .
1 1 n = O (n) + O n (n) du , 1 /
(using second mean-value (n) is monotonic) theorem for the integral in the 2nd term as
= O(1) + O(1) , as n , using (3.2) = O(1) , as n .

Finally
I 3 (t ) K n (t ) dt = O(1) as n .
by using Riemann Lebesgue theorem and the regularity condition of the method of summability. Thus, Tn f (x) = O(1) , as n . This completes the proof of the theorem.
VI.
CONCLUSION
Thus, Product Summability of Fourier series by Matrix-Euler method generalizes the ( N , p n )(E , z ) Product Summability of Fourier series and ( N , p n , q n )(E , z ) - Product Summability of Fourier series.
REFERENCES
[1]. Hardy, G.H.: Divergent series clarendon press , Oxford university press, (1949). [2]. Nigam,H.K and Kusum sharma : A study on ( N , p n )(E , q ) product summability of Fourier series , ultra scientist ,vol. 22(3) ,m 927-932, (2010). [3]. Teoplitz, O.: Uber allge-meire Lineare mittelbildungen, Prace Matimatyczno-Fizyczne, 113-119, 22(1911). [4]. Titechmalch, E. C: The theory of functions, oxford university press, 402- 403(1939).
195
Vol. 3, Issue 1, pp. 191-196
Authors Biography
B. P. Padhy has got his Ph.D from Berhampur University in 2012.He has 14 years of teaching experience. He has published around 15 research articles in various International and National journals of repute. He has also published a book Entitled Summability Methods and Applications under Lap Lambert Academic Publishing GmbH & Co.KG, Germany. Presently he is working as an Assistant Professor in mathematics in Roland Institute of Technology, Berhampur, Odisha. Banitamani Mallik is currently working as Asst. Prof. in Mathematics at JITM, , Paralakhemundi, Gajapti, Odisha which is under Centurion University of Technology & Management. She has teaching experience of more than 15yrs. She has obtained her M.Phil in Mathematics from Utkal University and doing research (Ph. D.) work in the field of Summability Theory since 2007 under Berhampur University.
U. K. Misra is working as a faculty in the P.G.Department of Mathematics, Berhampur University since last 28 years. To his credit he has guided 13 Ph.Ds and 1 D.Sc .He has published around 75 papers in various International and National journal of repute. The field of Prof. Misras research is Summability theory, sequence space, Fourier series, Inventory control, Mathematical Modeling. He is the reviewer of Mathematical review published by American Mathematical Society.
Mahendra Misra has got his Ph.D in the year 1997 from Berhampur University. He has published around 30 research articles in various International and National journals of repute. To his credit he has produced three Ph.Ds under his guidance. Presently he is working as H.O.D of P.G. Department of Mathematics in N.C. College (Autonomous), Jajpur ,Odisha.
196
Vol. 3, Issue 1, pp. 191-196
MATERIAL HANDLING AND SUPPLY CHAIN MANAGEMENT IN FERTILIZER PRODUCTION A CASE STUDY
T. K. Jack
Dept. of Mechanical Engineering, University of Port Harcourt, Nigeria
ABSTRACT
The fertilizer bagging line operation for packaging and conveying to the final delivery trucks was a source of problem for one fertilizer producer referred to here as Company N. This resulted in frequent maintenance activities from constant machine breakdown, loss of production, and the attendant loss in plant revenue. To overcome this, a technical study was conducted to address the problem. Plant site visit and technical inspection revealed the breakdowns were primarily due to fertilizer granules being trapped on the packaging/bagging conveyors, and as a result of fast acidic corrosion causing rust and eventual seizures of the conveyor rollers and lines. The resistances to free roller movement induce a drag load on the drive motors, with the additional loading causing failures of the electronic control sensors and stoppages. The unreliability of the bagging operations in the unplanned, repeated start-stop sequence in the packaging operation also led to safety concerns, with a few incidences of injuries being recorded. The engineered solution is presented in this report.
I.
INTRODUCTION
Company N was set up to meet the fertilizer demands of local farmers, and also to utilize the excess gas resources of the country. The products of the Company are Ammonia, Urea and NPK fertilizers. Ammonia is the major raw material in the production of Urea and NPK fertilizers. Expected total daily production of Urea and NPK type fertilizers without machine breakdown and no loss of time on the part of the bagging personnel, for three shifts (8 working hours per shift), in the bagging/material handling section of the Company N, is about 1500 tons for its eight bagging lines (four each for Urea and NPK type bagging).[1] This is operationally managed into six production/packaging line units (three each for Urea and NPK); with one stand-by each. However, due to frequent maintenance activities from bagging machine breakdowns, and improper orientation (in form of training) on the part of the packaging line/bagging personnel (often hourly paid contract staff), the bagging target was never achieved. This was a huge loss in revenue to the company and source of worry to the government Agricultural Ministry in meeting annual farming plans and targets. The bagging lines had teething installation challenges before final commissioning. [2] Bagging personnel were often recruited on contract basis and immediately engaged in active packaging of fertilizers, without proper orientation and training. Plant maintenance crew gangs often had other responsibilities in other plant sections and unable to give full attention to the increasing bagging machine line breakdowns. In the sections that follow is a brief description of the fertilizer packaging line operation - Fig. (1); the identified machine problem source points, and the preventive and corrective maintenance actions required.
II.
THE BAGGING/PACKAGING OPERATION OF COMPANY N
A typical bagging line arrangement consists of: (1) an in-feeder chute (provided with bag clamp and sensor, and pneumatic operation), where the bag is fed with the material (either Urea or NPK) to be bagged. (2) The bags when loaded with the product (50 kg quantity of product), then drops on to a slate/wooden conveyor, which conveys the bag to, (3) the sealing machine unit for the sealing of the
197
Vol. 3, Issue 1, pp. 197-199
inner bag (polythene protection against moisture), and (4) then conveyed on to the sewing machine for sewing of the jute type bag, from where it is, (5) conveyed on to the bag turner; it is (6) further conveyed on to the compressing unit, and finally on to, (7) the loading plate which is in a pallet form. A pallet contains twenty (20), 50 kilogram (kg) bags of product (Urea or NPK Fertilizers). (8) The Pallet is finally carried off the bagging machine by a forklift to the storage, awaiting the delivery trucks.
Feeder
Wooden Conveyor
Sealing Machine
Sewing Machine
Bag Turner
Compressing Unit
Pallet Loading Plate
To Storage and Delivery Trucks
Fig.(1): Original Installed Fertilizer Bagging Line-to-Storage/Delivery Trucks
III.
SOURCE OF MAINTENANCE PROBLEMS
Identified are two major problems. Machine associated problems, and the issues that affect worker productivity. The greatest maintenance activities occur in the sealing machine unit, pallet pusher chain, sewing machine. A source of problem is fertilizer product granules being trapped at sections of the bagging machine unit. The acidic nature of fertilizer also has a deteriorating corrosive effect particularly on the pusher chain-links. The conveyor chain links slowly rust away, leading to complete fracture of chain parts, thus resulting in difficulty in conveying the bags since the broken ends of the chain link, hook on to the conveyor rail. The conveyor rollers are also affected. Issues affecting worker productivity result from sluggish operational activities like work permit issuance delays, supervisory instructions time lag and prioritization in repair personnel allocation, delays in request for replacement spares from materials warehouse, resulting in the actual production time not exceeding 4 hours per shift. This implies an actual total production time of 12 hours for the three shifts. The question then is, can these production delays and lapses not be minimized to the barest or if possible, done away with completely. A careful study of the bagging techniques was conducted to find solutions.
IV.
FINDING SOLUTIONS
The technical services engineering consultants in collaboration with the plant maintenance engineers and the final product distribution unit, made some modifications to the production-to-distribution supply chain in which they scrapped the sealing unit and entire palletizing units for six of the eight bagging lines (three each for Urea and NPK bagging lines.), and adopted a just-in-time [3] loading method since the breakdowns often result in the delivery trucks queuing for hours and in certain instances, days. The reduced bagging/production-to-distribution chain activities carried out are: 1) Feeding the bag with the product; 2) Loading it on the wooden/slate conveyor; 3) Sewing the bag, conveying it on, and loading it straight on delivery trucks, or send for storage. The modifications made to the original bagging line are shown in fig. (2).
198
Vol. 3, Issue 1, pp. 197-199
Feeder
Wooden Conveyor
Sewing Machine
To Storage or Delivery Truck
To Customers
Fig. (2): Modified Fertilizer Bagging Line-to-Storage/Delivery Line
A preventive maintenance approach with an in house cleaning/routine maintenance to handle the corrosion problems in the conveyor chain links was adopted. Also, a general weekly maintenance on the entire lines is carried out on Sundays, since there is no bagging activity for that day. These weekly maintenance plans include, putting back to service out-of-use lines with new chain links which are often machined at the general service machine shop, to avoid delays in order delivery of out-of-stock Overseas re-order spare chain replacements. Old and worn conveyor rollers are also replaced during this period.
V.
These slight modifications, allowed for better just-in-time material flow [3] and increased the total production time to between six to seven hours per shift, with some lines recording no loss time due to breakdown, in a 24-hour continuous production period. The truck queues were completely reduced with slight queues only experienced in the weekend preceding the restart of production on Monday. It was observed that, the increased emphasis placed on safety, and proper training for bagging personnel also added to the improved productivity.
VI.
CLOSING
Significant increase in the life of the machines and improvements in packaging line/bagging activity was achieved. Life here, referring to machine time spent in continuous packaging/bagging with fewer breakdowns. The changes to the original installed design made the packaging/bagging operation more manual but reliable and stable. The study was unable to deduce if the problems were as a result of poor installation from onset, but it was suggested that in future expansion of the plant operation, designs for new bagging line units take into considerations the observations made, and modifications effected.
REFERENCES
[1] Jack, T. K., Statistic Process Control in the Fertilizer Industry, M.E.M. Technical Report, University of Port Harcourt, (Unpublished), 1989 [2] Maintenance Manager, Company N, private communications [3] Maynard, H. B., Industrial Engineering Handbook, 5th ed., McGraw-Hill, 2001 [4] Kuye, O. A., Sampling Inspection and Quality Control, Lecture Notes, University of Port Harcourt, 1988
Author
Tonye K. Jack is a Registered Engineer, and ASME member. He worked on plant maintenance and rotating equipment in the Chemical Fertilizer industry, and on gas turbines in the oil and gas industry. He has Bachelors degree in Mechanical Engineering from the University of Nigeria, and Masters Degrees in Engineering Management from the University of Port Harcourt, and in Rotating Machines Design from the Cranfield University in England. He is a University Teacher in Port Harcourt, Rivers State, Nigeria, teaching undergraduate classes in mechanical engineering. His research interests are on rotating equipment engineering, maintenance, engineering management, engineering computer programs, and applied mechanics.
199
Vol. 3, Issue 1, pp. 197-199
VLSI IMPLEMENTATION OF ERROR TOLERANCE ANALYSIS FOR PIPELINE BASED DWT IN JPEG 2000 ENCODER
Rajamanickam. G & Jayamani. S
Deptt. of ECE, KSR College of Tech., Anna University, Coimbatore, Tiruchengode
A BSTRACT
The JPEG 2000 image compression standard is designed for a broad range of data compression applications. The Discrete Wavelet Transformation (DWT) is central to the signal analysis and is important in JPEG 2000 and is quite susceptible to computer-induced errors. These errors get spread to many output transform coefficients if the DWT is implemented by using lifting scheme. This paper proposes an efficient Error Tolerance Scheme (ETS) to detect errors occurring in DWT. A pipeline-based DWT structure is also developed in this paper to speed up the error detection process. Some standard images are used as test samples to verify the feasibility of the proposed ETS design. Experimental results and comparisons show that the proposed ETS achieves better performance in error detection time and error tolerance capability.
K EYW ORDS: JPEG 2000, DWT, error detection, error tolerance.
I.
INTRODUCTION
JPEG 2000, a still-image compression standard is being designed to address many applications, e.g., internet, printing, and medical imagery [1] & [2]. Besides, it achieves better compression performance over the JPEG, JPEG 2000 also provides rich features. For example, it allows efficient lossy and lossless compressions within a single unified coding framework, provides superior image quality at low bit rates, supports additional features such as region of interest coding, and has a more flexible file format, avoiding excessive computational and memory complexity. The DWT is central to the JPEG 2000 image compression standard which includes lifting configurations for implementing the forward and inverse transforms. The main properties of DWT are the space-frequency localization and inherent multiresolution structure. In other words, wavelets allow efficient representation of a signal with a small number of nonzero coefficients. Also, wavelets take advantage of data correlation in space and frequency. DWT is implemented with computer hardware ultimately, the processing operations are susceptible to transient failures, and primarily single-event upsets, alternately termed as soft errors. These factors will increase the influences as VLSI feature sizes shrink [3]. Error tolerance is a new design and test paradigm, which takes into consideration whether erroneous outputs of defective circuits still produce acceptable results [5] & [6]. Error tolerance classifies a system as being acceptable/ unacceptable by estimating the performance degradation due to errors, rather than relying solely on the conventional perfect/imperfect classification. Error tolerance analyzes the system level effects of errors, and accepts circuits if the performance degradation can meet the application-specific or range of acceptability. This paper proposes an ETS design that targeted for detecting errors of the DWT subsystem in JPEG 2000, as DWT is one of the most important subsystems in terms of both computation and memory requirements. The remainder of this paper is organized as follows: Section II reviews the basic principles and key features of DWT by lifting scheme. In section III presents the proposed ETS structure, pipeline-based DWT design, error model definition, error detection, and tolerance strategy. Section IV shows the experimental results and comparisons for performance evaluation and discussion. Finally, Section V provides the final conclusions.
200
Vol. 3, Issue 1, pp. 200-208
II.
BACKGROUND
The DWT is usually computed through convolution and sub-sampling with a couple of filters to produce an approximation signal L (low pass filter result) and a detail signal H (high pass filter result). The multi-resolution decomposition is obtained by iterating the convolution and sub-sampling of these two filters over the approximation components. For 2-D signals, there exist separable wavelets for which the computation can be decomposed into horizontal processing (on the rows) followed by vertical processing (on the columns), using the same 1-D filters. Figure 1 shows a one level decomposition. Subsequent levels are obtained by iterating on the low pass signal LL [3] & [4].
Figure 1. A 1-Level 2-D seperable wavelet decomposition
The general block diagram of the lifting technique is illustrated in Fig. 2, which consists of four steps: 1. Split step: The original samples are separated into two disjoint sets, named even part and odd part. 2. Predict step: The even samples are multiplied by the time domain equivalent of s(z) are added to the odd samples. 3. Update step: The updated odd samples are multiplied by the time domain equivalent of t(z) are added to the even samples. 4. Scaling step: The even and odd samples are multiplied by k-1 and k, respectively.
Figure.2. Block diagram of the lifting scheme.
III.
PROPOSED ERROR TOLERANCE SCHEME
The Proposed ETS design which consists of an Input Parity Procedure (IPP), an Output Parity Procedure (OPP) and a parity analyzer. The main objective of the ETS is to compare the differences between Cin and Cout values to find the errors that occurred in the DWT. Then the parity analyzer will further analyze, whether the errors can be tolerated or not. Each row pixels of an n n image will be divided into even and odd number of data samples input for DWT operation. Thus, the size of inputdata vector X shown in Fig.3 is 1 n. A is n n matrix of wavelet transform via lifting. Additionally a
201
Vol. 3, Issue 1, pp. 200-208
1n tolerance weighting matrix W has to be developed to establish the IPP and OPP structures for error detection.[1]
Figure.3. Block diagram of the ETS design
3.1 Modeling Errors

The proposed ETS design mainly focuses on the effects of computer-induced errors, which will be modeled through transfer matrices related to the lifting sections. Forward 2D DWT, the numerical errors are caused by an underlying computer-induced error which will propagate their corrupting influence to the output. The error mechanisms can be defined by two types such as Intensive Error Model (IEM) and Distributed Error Model (DEM), as shown in figure.4. If the numerical errors influence the pixels in contiguous rows of an image, then the error model is called IEM. On the other hand, the DEM is presented if many single row pixels of an image will be influenced by the numerical errors. According to the error models, the proposed ETS design will be demonstrated as an effective method to explore the error impact in the DWT and further analyze the tolerance of errors for JPEG 2000 encoder applications [11].
Figure. 4. Error models. (a) IEM. (b) DEM
3.2 Tolerance Weighting

In order to achieve the error tolerance of human visual system and to increase the flexibility of the proposed ETS design, a tolerance weighting matrix, W has to be developed. The weighting factors of W are very important for error tolerance analysis, since the most significant data are generally centralized in the central parts of an image. The error influence in the central parts is more serious than that in the boundaries of an image. Thus, a weighting matrix has to be built for supporting evaluation of error influence when the proposed ETS is active. Fig 5 shows an image divided into some blocks to set the different parity weighting factors. Generally, good parity weighting factors should have gain factors whose ranges span three to four orders of pixel magnitude. Based on the parity weighting factor, we built a 1n tolerance weighting matrix for error detection and error tolerance evaluation purpose [1].
202
Vol. 3, Issue 1, pp. 200-208
Figure.5. Schematic representations of parity weighting of an image
3.3 Pipeline-Based DWT Design

Based on the conventional structure for one level 9/7 DWT via lifting. Initially input data samples are split into odd and even subsequences. For an nxn image, the input-data vector is represented by, X= [x0 x1 x2 x3.xn-2 xn-1] s0= [x0 x2.xn-2] = [s00 s10 ..s0(n/2)-1] d0=[x1 x3.xn-1]= [d0 d1 ..d Where sli
l th 0 0 0 (n/2)-1] th
(1) (2) (3)
and d i represent the i even and odd samples in the l stage of lifting step.
Split step
The input sequences xi are split into even and odd parts, s0i and d0i di0 = x2i+1 si0= x2i (4) (5)
Lifting step
The two splitting sequences (s0i and d0i) are performed by two lifting steps Step1 di1= di0 + (si0+s0i+1), si1= si0 +2 ( di1) Step2 di2= di1 +(si1+s1i+1) si2= si1 +2 ( di2) (8) (9) (6) (7)
Scaling step
Through the normalization factors k_1and k, the low-pass and high-pass wavelet coefficients si and di can be obtained. di= k di2 si= k-1 si2 (10) (11)
203
Vol. 3, Issue 1, pp. 200-208
The final output of the pipeline based DWT is expressed in the following matrix form
(12) where Ma, Mb, Mc, and Md are the nxn matrices and the parameters , , , and k represent floating point arithmetic for a 9/7 wavelet transform via lifting[12]. The direct mapping of 1-Dimensional lifting based architecture for the 9/7 filter is shown in the figure 6.
Figure.6. Direct mapping of 1-D lifting-based architecture for the 9/7 filter
3.4 Parity Analyzer

The parity analyzer plays the role of a comparator to check a syndrome (the difference between Cin and Cout) and determine whether a syndrome is tolerant with a chosen threshold or not. The thresholds of IEM and DEM are defined as THIEM and THDEM, respectively, to detect the Intensive and Distributed errors in an image as shown in equation (13).[9]
(13) The human visual system is more sensitive about the brightness variations than the changes in chrominance. Thus, the brightness variation detection method which plays an important role for error detection in an image is adopted here to redefine the thresholds of error detection. For the computational efficiency consideration, the criterion of brightness variation is determined by using the histogram- based method, which usually shows sensitivity to the image changes within a similar brightness condition. The histogram difference is given by equation (14). DHIS = Hj - Hi (14) where Hi and Hj signify the histograms in the ith and jth rows of an image. By setting threshold of the histogram difference DHIS, the image holding brightness variations can be detected [1].
204
Vol. 3, Issue 1, pp. 200-208
In order to accurately evaluate the quality impact of an image by brightness variations, the BVRs are normalized as fractional variations. The Brightness Variations for Intensive Error Model and Distributed Error Model is given by the equation(15) and (16) respectively. 100% (15)
100%
(16)
3.5 Acceptable Error Rate

The objective evaluation of a specific image presented here depends on the relation between BVR and acceptable error rate (AER). Error rate (ER) is the percentage of vectors for which values at a set of outputs deviate from error free response, during normal operation. AER is a percentage of acceptable errors for all injection errors; AER represents the capability of error tolerance with injection errors in an image [15]. 100% (17)
IV.
EXPERIMENT RESULTS
The 9/7 lifting DWT of the JPEG 2000 image compression standard is used as a Circuit Under Test (CUT) to demonstrate the good performance in error detection time and error tolerance capability of the proposed ETS design. Consider the lifting DWT operating with error injection values modeled by a Gaussian noise source. The error detection performance is strongly dependent on the noise variation and the selected detection threshold. Cin - Cout, from the parity analyzer of the proposed ETS is on the order of 10-10 with round off errors. Thus, the necessary thresholds have to be chosen well above this level for error detection. To examine the effectiveness of the proposed ETS design in different experimental conditions, six benchmark sequences (Lena, Peppers, Baboon, Barbara, Gold hill, and Cameraman) are selected in the experiments. The comparisons between the proposed ETS design and the work are presented in this section to demonstrate that the proposed ETS design has good performance in error detection time. Additionally, quality observation and objective analysis are discussed to evaluate the capability of error tolerance.
(a)
(b)
205
Vol. 3, Issue 1, pp. 200-208
(c)
(d) Figure.7.Quality observation for the benchmark image Baboon. (a) Error free. (b) BVRDEM = 6.2%. (c) BVRDEM=12.1%. (d)BVRDEM=18.7%.
4.1 VLSI Simulation output for pipeline based DWT
Figure.8. Simulation output for pipeline based DWT
206
Vol. 3, Issue 1, pp. 200-208
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 4.2 Table 1: Acceptable Error Rate (AER) for Test Images
Threshold 2% BVR IEM(THIEM=10-4) 4% 6% 8% 2% BVR IEM(THDEM=10-7) 4% 6% 8%
Test images Lena Peppers Baboon Cameraman
6.5 5.1 7.0 5.1
AER% 12.1 18.9 9.8 13.1 10.7 14.9 19.8 16.2
24.9 19.9 27.1 22.1
4.9 4.2 6.2 2.1
AER% 10.9 17.2 9.1 12.1 5.2 14.1 18.7 8.6
22.3 18.7 25.1 11.8
4.3 Comparisons
The proposed ETS design shown in Fig. 4 can be implemented by using a comparator and some multipliers, adders, buffers, and registers. Based on the circuit design shown in [8], the number of logic gates of the proposed ETS design is about 6,972. However, about 180 k logic gates are needed for a VLSI architecture design of JPEG 2000 encoder [3]. Thus, the area overhead of the proposed ETS design is only about 3.9 percent, which is a reasonable design for circuit testing. The comparisons between the proposed method and previous work [9] are shown in Table 2, which clearly indicates that the proposed ETS design has good performance in computational complexity (error detection time) and error tolerance capability with little area overhead.
Table2: Comparison Results DWT Design Computational Area complexity overhead Proposed work Pipelined based structure Conventional structure O(n2) O(n3) 3.9% Tolerance analysis Error Tolerance Fault Tolerance
Existing work
V.
CONCLUSION
An effectively ETS design for lifting DWT error detection and error tolerance evaluation in JPEG 2000 encoder is presented in this paper. The paper first developed a pipeline based DWT structure to support the proposed ETS design for speeding up the error detection time. Then, an IPP, an OPP, and a parity analyzer based on the weighting sum technique are built of the proposed ETS design to detect the errors. The error detection performance depends on the detection thresholds, which are determined by the brightness variations. Experimental results show that the proposed ETS with pipeline-based DWT design can significantly improve the error detection time compared with the previous work with conventional DWT structure. Additionally, according the quality observation and objective evaluation for the test images, the proposed ETS design also demonstrates the good performance in error tolerance capability.
ACKNOWLEDGEMENT
The authors would like to thank K.S. Rangasamy College of Technology, Anna UniversityCoimbatore, Tiruchengode for supporting this project for the Masters degree at Department of Electronics and Communication Engineering. Authors also would like to thank anonymous reviewers for theirs useful comments.
207
Vol. 3, Issue 1, pp. 200-208
REFERENCES
[1] Chun-Lung Hsu, Member, Yu-Sheng Huang, Ming-Da Chang, and Hung-Yen Huang,Design of an Error-Tolerance Scheme for Discrete Wavelet Transform in JPEG 2000 Encoder, IEEE Trans. May 2011. D. Shin and S.K. Gupta, A Re-Design Technique for Datapath Modules in Error Tolerant Applications, Proc. Asian Test Symp., pp. 431-437, Nov. 2008. Y.H. Seo and D.W. Kim, VLSI Architecture of Line-Based Lifting Wavelet Transform for Motion JPEG 2000, IEEE J. Solid-State Circuits, vol. 42, no. 2, pp. 431-440, Feb. 2007. T. Acharya and C. Chakrabarti, A Survey on Lifting-Based Discrete Wavelet Transform Architectures, J. VLSI Signal Processing Systems, vol. 42, no. 3, pp. 321-339, Mar. 2006. H. Chung and A. Ortega, Analysis and Testing for Error Tolerant Motion Estimation, Proc. 20th IEEE Intl Symp. Defect and Fault Tolerance in VLSI Systems, pp. 514-522, 2005. M. Breuer, S. Gupta, and T. Mark, Defect and Error Tolerance in the Presence of Massive Numbers of Defects, IEEE Trans. Design Test Computers, vol. 21, no. 3, pp. 216-227, May/June 2004. K.J. Cho, K.C. Lee, J.G. Chung, and K.K. Parhi, Design of Low-Error Fixed-Width Modified Booth Multiplier, IEEE Trans. Very Large Scale Integration Systems, vol. 12, no. 5, pp. 522-531, May2004. L. Liu, N. Chen, H. Meng, L. Zhang, Z. Wang, and H. Chen, A VLSI Architecture of JPEG 2000 Encoder, IEEE J. Solid-State Circuits, vol. 39, no. 11, pp. 2032-2040, Nov. 2004. G.R. Redinbo and C. Nguyen, Concurrent Error Detection in Wavelet Lifting Transforms, IEEE Trans. Computers, vol. 53, no. 10, pp. 1291-1302, Oct. 2004.. C. Constantinescu, Trends and Challenges in VLSI Circuit Reliability, IEEE Micro, vol. 23, no. 4, pp. 14-19, July/Aug. 2003. J. Seifert, Z. Xiaowei, and L.W. Massengill, Impact of Scaling on Soft-Error Rates in Commercial Microprocessors, IEEE Trans. Nuclear Science, vol. 49, no. 6, pp. 3100-3106, Dec. 2002. C.J. Lian, K.F. Chen, H.H. Chen, and L.G. Chen, Analysis and Architecture Design of Lifting Based DWT and EBCOT for JPEG 2000, Proc. Intl Symp. VLSI Technology, Systems, and Applications, pp. 180-183, 2001. ITU-T Recommend. T.800-ISO/IEC FCD15444-1: JPEG 2000 Image Coding System, Intl Organization for Standardization, ISO/IEC JTC1 SC29/WG1, 2000. C. Christopoulos, A. Skodras, and T. Ebrahimi, The JPEG2000 Still Image Coding System: An Overview, IEEE Trans. Consumer Electronics Design, vol. 46, no. 4, pp. 1103-1127, Nov. 2000. H.W. Haussecker, Simulation Estimation of Optical Flow and Heat Transport in Infrared Image Sequences, Proc. IEEE Workshop Computer Vision, pp. 85-93, June 2000.
[2] [3] [4] [5] [6]
[7]
[8]
[9] [10] [11] [12]
[13] [14] [15]
Authors
G. Rajamanickam, received the B.E degree in Electronics and Communication Engineering from affiliated institution of Anna University, Chennai in 2009. He is currently doing M.E degree in VLSI Design at K.S.Rangasamy College of technology, Namakkal, India. His area of interest is Image processing and VLSI.
S. Jayamani, is Assistant professor in the Department of Electronics and Communication Engineering, K.S.Rangasamy college of technology, Namakkal, India. He has 2.6 years of teaching experience. His interesting subjects are Embedded system and VLSI.
208
Vol. 3, Issue 1, pp. 200-208
LOCATION BASED SERVICES IN ANDROID

Ch. Radhika Rani1, A. Praveen Kumar2, D. Adarsh2, K. Krishna Mohan2, K.V.Kiran2
2
Asst. Professor, Dept. of C.S.E., K L University, A.P, India IV/IV B.Tech Students, Dept. of C.S.E, K L University, A.P, India
ABSTRACT
Location Based Services provide the users a set of services which originate from the geographic location of the users mobile device. Using these services it is possible for the users to find and locate other persons, vehicles, resources and also to provide location-sensitive services, in addition to tracking their own location. The request for location can originate in the mobile device or in another entity like application provider or network operator. It is possible to automatically trigger Location Based Services when the mobile device is at a particular location. These services can also originate in the users mobile device itself in order to satisfy location-based requests like finding areas of interest, checking traffic conditions, finding our friends, vehicles, resources, machines and emergency requests. In this paper we will discuss how to implement these location based services in Android after giving a brief introduction to Android and its constituents.
KEYWORDS: Android, Location based service, Location Manager, Location Provider, GPS, Geocoding,
Overlays.
I.
INTRODUCTION
Android is a combination of three components: Free and open source operating system for mobiles. Open source development platform for creating mobile applications. Devices, particularly mobile phones that run Android operating system and the applications created for it. Android SDK was released by Open Handset Alliance in the month of November of the year 2007. Android is actually developed using the kernel of Linux 2.6 and the highlighting features of Android include the following [7]: No fees for licensing, distribution and release approval GSM,3G EDGE networks for telephony IPC message passing Background processes and applications Shared data stores Complete multimedia hardware control APIs for location based services such as GPS. The Location Based Service (LBS) applications can help user to find hospitals, school, gas filling station or any other facility of interest indicated by user within certain range [2]. Just like a GPS device its location will also be updated as soon as user changes his/her position. Android can be considered as a unified software package. This software package includes an operating system, middleware and core applications. Android SDK provides some tools and APIs which are required to develop Android applications using the programming language of Java. Android platform provides open system architecture along with a powerful debugging environment. It is also characterized by optimized graphics systems, rich media support and a embeddable web browser. It uses a Dalvik virtual machine (DVM) which gives the same priority to all processes and hence optimises the execution of application.GPS, camera, compass and 3d-accelerometer are also supported by Android and they provides some useful APIs for the functions of map and location. Anyone can access, control
209
Vol. 3, Issue 1, pp. 209-220
and process the Google map which is also free and implement location based services in his mobile device [3]. The skeleton of Android framework and the jargon of Android application development are discussed in section II and section III. The design of location based services in Android is discussed in section IV, the discussion regarding the development of an application which can find route between two geographical locations is mentioned in section V and its corresponding results are mentioned in section VI. After this, section VII gives the conclusion and the section VIII gives the future scope.
II.
SKELETON OF ANDROID
The skeleton of Android framework [9] and its constituents are shown in the following figure:
Figure 1. Skeleton of Android
2.1. Applications Layer

Android ships with a set of core applications including an email client, SMS program, calendar, maps, browser, contacts, and others. All applications are built using the Java. Each of the application aims at performing a specific task that it is actually intended to do.
2.2. Application Framework Layer

The next layer is the application framework. This includes the programs that manage the phone's basic functions like resource allocation, telephone applications, switching between processes or programs and keeping track of the phone's physical location. Application developers have full access to Android's application framework. This allows them to take advantage of Android's processing capabilities and support features when building an Android application. We can think of the application framework as a set of basic tools with which a developer can build much more complex tools.
2.3. Libraries Layer

The next layer contains the native libraries of Android. These shared libraries are all written in C or C++, compiled for the particular hardware architecture used by the phone and preinstalled by the phone vendor. Some of the core libraries are listed in Fig.1.
2.4. Android Runtime Layer

Android Runtime layer includes Dalvik Virtual Machine (DVM) and a set of core java libraries. Every Android app gets its own instance of DVM. Dalvik has been written so that a device can run multiple virtual machines efficiently and it executes files with .dex (dalvik executable format) extension optimised for minimum memory.
210
Vol. 3, Issue 1, pp. 209-220
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 2.5. Linux Kernel
This layer includes Androids memory management programs, security settings, power management software and several drivers for hardware, file-system access, networking and inter process communication. The kernel also acts as an abstraction layer between hardware and rest of the software stack.
III.
JARGON OF ANDROID APP DEVELOPMENT
The basic components of an Android application include Activity, Broadcast Receiver, Service, and Content Provider. Each of these which when used for any application has to be declared in the AndroidManifest.xml. The user interface of the component is determined by the Views. For the communication among these basic components we use Intents and Intent filters which play crucial role during app development [1].
3.1. Activity
An Activity is, fundamentally, an object that has a lifecycle. An Activity is a chunk of code that does some work; if necessary, that work can include displaying a UI to the user. It doesn't have to, though some Activities never display UIs. Typically, we will designate one of our application's Activities as the entry point to our application [4].
3.2. Broadcast Receiver

Broadcast Receiver is yet another type of component that can receive and respond to any broadcast announcements.
3.3. Service
A Service is a body of code that runs in the background. It can run in its own process, or in the context of another application's process, depending on its needs. Other components "bind" to a Service and invoke methods on it via remote procedure calls. An example of a Service is a media player; even when the user quits the media-selection UI, she probably still intends for her music to keep playing. A Service keeps the music going even when the UI has completed.
3.4. Content Provider

Content Provider is a data storehouse that provides access to data on the device; the classic example is the Content Provider that's used to access the user's list of contacts. Our application can access data that other applications have exposed via a Content Provider, and we can also define our own Content Providers to expose data of our own.
3.5. Intents
Intent is a simple message object that represents an "intention" to do something. For example, if our application wants to display a web page, it expresses its "Intent" to view the URI by creating an Intent instance and handing it off to the system. The system locates some other piece of code (in this case, the Browser) that knows how to handle that Intent, and runs it. Intents can also be used to broadcast interesting events (such as a notification) system-wide. There are two types of intents namely implicit and explicit intents. Implicit intents has no specified component where as Explicit intents do specify the component.
3.6. AndroidManifest.xml
The AndroidManifest.xml file is the control file that tells the system what to do with all the top-level components (specifically activities, services, intent receivers, and content providers described below) we have created. For instance, this is the "glue" that actually specifies which Intents our Activities receive.
3.7. Views
A View is an object that knows how to draw itself to the screen. Android user interfaces are comprised of trees of Views. If we want to perform some custom graphical technique (as we might if we're writing a game, or building some unusual new user interface widget) then we would create a View.
3.8. Notification
A Notification is a small icon that appears in the status bar. Users can interact with this icon to receive information. The most well-known notifications are SMS messages, call history, and voicemail, but applications can create their own. Notifications are the strongly-preferred mechanism for alerting the
211
Vol. 3, Issue 1, pp. 209-220
user of something that needs their attention.
IV.
DESIGN OF LOCATION BASED SERVICES (LBS)
Location based services is an umbrella term is used to describe the different technologies used to find the current location of a device. The main LBS elements include: nt Location Manager Location Provider
Figure 2. Address search interface
Location Manager acts as a hook to LBS. Each of the Location providers represent a different location finding technology used to find the current location of a device . Using Location Manger we can device[7]. Perform following tasks: Findind current location Tracking the movement Setting proximity alerts for specified locations Checking all the available location providers We can use the Location Controls in DDMS perspective in Eclipse to perform location changes directly into emulators GPS Location Provider either manually or by using KML (or GPX) tabs in either Location Controls. In the Manual tab, we can send geo coordinates (latitude and longitude) manually. and GPX stands for GPS Exchange format. Most GPS systems record tracking files using GPX. KML stands for Keyhole Markup Language. It is used extensively online to define geographic information. We can write KML file by our hand or generate it using Google Earth to find directions between two generate locations. The most common location providers are GPS provider and Network provider. They can be accessed by using the static string constants mentioned below that can return the corresponding prov provider name: LocationManager.GPS_PROVIDER LocationManager.NETWORK_PROVIDER To access a specific provider we can use getProvider() method as shown in the following code snippet: String providerName = LocationManager.GPS_PROVIDER; LocationProvider gpsProvider; gpsProvider = locationManager.getProvider(providerName);
212
Vol. 3, Issue 1, pp. 209-220
To see all the available location providers we use getProviders( ) method as shown in the following code snippet: boolean e = true; List<String> providers = locationManager.getProviders(e); To select best provider we can set criteria regarding cost, power consumption, altitude, speed, accuracy, direction, we can use get Best Provider( ) method as shown in following code snippet : Criteria criteria = new Criteria(); criteria.setAccuracy(Criteria.ACCURACY_COARSE); criteria.setPowerRequirement(Criteria.POWER_LOW); criteria.setAltitudeRequired(false); criteria.setBearingRequired(false); criteria.setSpeedRequired(false); criteria.setCostAllowed(true); String bp = locationManager.getBestProvider(criteria, true);
4.1. Finding our Location

First we need to get a debugging certificate from google using MD5 fingerprint which can be generated using key tool. MD5 fingerprint is unique for any user. The Map API key thus generated is used for display of Google map in the emulator. The purpose of location-based services is to find the physical location of the device. Access to the location-based services is handled by the Location Manager system Service. To access the Location Manager, request an instance of the LOCATION_SERVICE using the get System Service() method, as shown in the following code snippet: String ss = Context.LOCATION_SERVICE; LocationManager lm; lm = (LocationManager)getSystemService(ss); Before we can use the Location Manager we need to add one or more uses-permission tags to our manifest to support access to the LBS hardware. We can have fine and coarse permissions for GPS and Network providers respectively. Permission will have coarse permission granted implicitly. We can find the last location fix determined by a particular Location Provider using the getLastKnownLocation method, passing in the name of the Location Provider. The following examplefinds the last location fix taken by the GPS provider: String provider = LocationManager.GPS_PROVIDER; Location location = locationManager.getLastKnownLocation(provider); Before using this method , we need to request updates using location manager at least once. We use request Updates method for this. It takes the location provider, minimum time, minimum distance and Location Listener object as arguments. lm.requestUpdates( provider, time,distance,listener); We can also stop receiving updates using the remove Updates method which takes the location provider as argument. lm.removeUpdates(provider);
213
Vol. 3, Issue 1, pp. 209-220
Figure 3. Finding our location
4.2. Proximity Alerts

Its often useful to have our applications react when a user moves toward, or away from, a specific location. Proximity alerts let our applications set triggers that are fired when a user moves within or beyond a set distance from a geographic location. To set a proximity alert for a given coverage area, select the center point (using longitude and latitude values), a radius around that point, and an expiry time-out for the alert. The alert will fire if the device crosses over that boundary, both when it moves out from outside to within the radius, and when it moves from inside to beyond it. When triggered, proximity alerts fire Intents, most commonly broadcast triggered, Intents. To specify the Intent to fire, we use a Pending Intent, a class that wraps an Intent in a kind of , method pointer, as shown in the following code snippet: Intent in = new Intent(MY_ACTION) Intent(MY_ACTION); PendingIntent pi = PendingIntent.getBroadcast(this, -1, in, 0);
Figure 4. Proximity Alert
If TREASURE_PROXIMITY_ALERT is static string then to start listening for proximity alerts, we can register receiver as follows: IntentFilter filter = new IntentFilter(TREASURE_PROXIMITY_ALERT); registerReceiver(new ProximityIntentReceiver(), filter);
214
Vol. 3, Issue 1, pp. 209-220
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 4.3. Geocoding and Reverse Geocoding
Geocoding lets us translate between street addresses and longitude/latitude map coordinates. This cangive us a recognizable context for the locations and coordinates used in location-based services and map-based activities. The geocoding lookups are done on the server, so our applications will require us to include an Internet uses-permission in our manifest. The Geocoder class provides access to two geocoding functions [5]: Forward Geocoding Reverse Geocoding Forward Geocoding converts the address into latitude and longitude. Reverse Geocoding converts latitude and longitude to corresponding address. First we create a Geocoder object using which we need to perform geocoding and reverse geocoding Geocoder geocoder = new Geocoder(getApplicationContext(),Locale.getDefault()); For geocoding we use getFromLocationName method of Geocoder class. The following is the sample code snippet : List<Address> locations = null; Geocoder gc = new Geocoder(this, Locale.getDefault()); try { locations = gc.getFromLocationName(streetAddress, 10); } catch (IOException e) {} For reverse geocoding we use getFromLocation method of Geocoder class. The following is the sample code snippet: List<Address> addresses = null; Geocoder gc = new Geocoder(this, Locale.getDefault()); try { addresses = gc.getFromLocation(latitude, longitude, 10); } catch (IOException e) {}
4.4. Map based Activities

To use maps in our applications we need to extend MapActivity. The layout for the new class must then include a MapView to display a Google Maps interface element. The Android maps library is not a standard Android package; as an optional API, it must be explicitly included in the application manifest before it can be used. Add the library to our manifest using a uses-library tag within the application node, as shown in the following XML snippet [8]: <uses-library android:name="com.google.android.maps"/> To see map tiles in our Map View we need to add a <uses-permission> tag to our application manifest for INTERNET. MapView controls can be used only within an Activity that extends MapActivity. Override the onCreate method to lay out the screen that includes a MapView, and override isRouteDisplayed to return true if the Activity will be displaying routing information (such as traffic directions).By default the Map View will show the standard street map. In addition, we can choose to display a satellite view, StreetView, and expected traffic, as shown in the following code snippet: mapView.setSatellite(true); mapView.setStreetView(true); mapView.setTraffic(true); We can also query the Map View to find the current and maximum available zoom levels, as well as the center point and currently visible longitude and latitude span (in decimal degrees).
215
Vol. 3, Issue 1, pp. 209-220
int m= mapView.getMaxZoomLevel(); GeoPoint center = mapView.getMapCenter(); int latS = mapView.getLatitudeSpan(); int longS = mapView.getLongitudeSpan(); We can also optionally display the standard map zoom controls using the setBuiltInZoomControls method. mapView.setBuiltInZoomControls(true); We can use the Map Controller to pan and zoom a MapView. We can get a reference to a MapViews controller using getController method . MapController mapController = myMapView.getController(); We can re-center and zoom the Map View using the setCenter and setZoom methods available on the Map ViewsMapController. mapController.setCenter(geopointobject); mapController.setZoom(1);
4.5. Creating and using Overlays

Overlays enable us to add annotations and click handling to MapViews. Each Overlay lets us draw 2D primitives, including text, lines, images, and shapes, directly onto a canvas, which is then overlaid onto a Map View [6]. We can add several Overlays onto a single map. All the Overlays assigned to a Map View are added as layers, with newer layers potentially obscuring older ones. User clicks are passed through the stack until they are either handled by an Overlay or registered as clicks on the Map View itself. Each Overlay is a canvas with a transparent background that is layered onto a Map View and used to handle map touch events. To add a new Overlay, we create a new class that extends Overlay. We override the draw method to draw the annotations we want to add, and override onTap to react to user clicks (generally made when the usertaps an annotation added by this Overlay).
Figure 5. Overlays on Maps
V.
DISCUSSION
In Android, a wide variety of applications can be developed in the field of Location Based Services. One such application is Finding the route between two locations. For preparing this application we first we need three activities namely
216
Vol. 3, Issue 1, pp. 209-220
Activity requesting source address from user. Activity requesting destination address from user. Map Activity which should display the actual route between the two locations specified in ty above activities.
For preparing the user interface of the activities requesting source and destination addresses, we need to use the corresponding XML files of the activities. Once the user interface of these two activities is ready we need to design the user interface of the Map Activity. As this activity should hold the Google Map, the procedure illustrated in the above section should be followed for the access to Google Map API. Then we should create two explicit intents (which are generated when an event is PI. performed by user like clicking a button) one for communication between the source (passing the source address to destination activity) and destination and other for commu communication between destination and Map Activity (passing the source and destination addresses to the Map Activity).The overall stepwise procedure for developing this application is mentioned below Preparing map resource and internet access for it Acquiring KML route file from google ML Drawing the Path using the following small procedure i) Building URL from source to destination ii) Connecting to URL and creating DocumentBuilder for parsing the KML file iii) Splitting each point in the path and drawing each line on the map Drawing points and lines on the Map
VI.
RESULTS
The application thus developed using the above procedure helps to find the route between the two locations. The first activity which takes the source address from the user is shown in the following user figure.
Figure 6. Source Activity
For instance if the source address is given as Benz Circle, Vijayawada, Andhra Pradesh, India, after clicking the Go button explicit intent loads the second activity which requests the destination address to be entered by the user. The second activity is shown in the following figure
217
Vol. 3, Issue 1, pp. 209-220
Figure 7. Destination Activity
For instance if the destination address is given as K L University, Vaddeswaram Andhra Pradesh, Vaddeswaram, India, after clicking the Go button explicit intent loads the MapActivity which shows the route between these two locations. The MapActivity thus loaded is shown in the following figure figure.
Figure 8. MapActivity displaying route
In the above figure the blue circle indicates the source and the green circle indicates the destination and the green curved line indicates the route between these locations.
VII.
CONCLUSION
Location Based Services are those services which provide both information and entertainment and are accessible with mobile devices through the mobile network. They utilize the ability to make use of the s geographical position of the mobile device. They can utilize multiple technologies such as the GPS technologies satellite network, cellular networks, Wi networks and other technologies. Context awareness is the Wi-Fi excellent feature for the LBS. Location Based Services can be used in a variety of aspects like vehicle tracking, monitoring driving habits, locating our employees, finding the route between two places or g to find the route to a specified location from current position etc. A of these need the use of GPS All along with some tracking programs Android provides a very nice platform for dev programs. developing LBS applications. It provides separate methods and classes for each and every separate entity involved in
218
Vol. 3, Issue 1, pp. 209-220
the development of the application. Location Based Services thus help the users in a variety of aspects and thus has a greater scope of development in the near future.
VIII.
FUTURE SCOPE
After performing a detailed survey it is observed that obtaining user location from a mobile device can be complicated. There are several reasons why a location reading (regardless of the source) can contain errors and be inaccurate. Some sources of error in the user location include:
8.1. Multitude of Location Sources

GPS, Cell-ID, and Wi-Fi can each provide a clue to users location. Determining which to use and trust is a matter of trade-offs in accuracy, speed, and battery-efficiency.
8.2. User movement

Because the user location changes, you must account for movement by re-estimating user location every so often.
8.3. Varying accuracy

Location estimates coming from each location source are not consistent in their accuracy. A location obtained 10 seconds ago from one source might be more accurate than the newest location from another or same source. These problems can make it difficult to obtain a reliable user location reading. This document provides information to help you meet these challenges to obtain a reliable location reading. It also provides ideas that you can use in your application to provide the user with an accurate and responsive geo-location experience. The future work is to eliminate these problems and provide an efficient way to find out the location of user accurately.
ACKNOWLEDGEMENTS
We thank our principal Prof. K. Rajsekhar Rao for providing guidance in understanding the concepts. We also want to thank our beloved head of the computer science department, Prof. S. Venkateswarulu for providing valuable information regarding the various categories of the applications that can be developed in Android.
REFERENCES
[1] Chris Haseman, Android Essentials, firstPress, 9_17, 2008 [2] Cludio Bettini,Sushil Jajodia,Pierangela Samarati, X.Sean Wang, Privacy in Location-Based applications, Springer,1_2,2009 [3] Georg Gartner, Felix Ortag, Advances in Location-Based Services, Springer, 4_5, 2011 [4] James Steele, Nelson To,Shan Conder,Lauren Darcey,The Android Developer's Collection, AddisonWesley Professional, 30A_40A, 2012. [5] Jerome(J. F.) DiMarzio, Android - A Programmers guide, Tata McGraw-Hill Education, 203_237,2010 [6] Mark L. Murphy, The busy coder's guide to Android development, CommonsWare, 287_292,2008 [7] Reto Meier, Professional Android 2 Application Development,Wrox, 247_264, 2010 [8] [Online] chengalva.com [9] [Online] developer.android.com Authors Ch. Radhika Rani is a Lecturer of CSE Dept. in K L University. She is very good at programming and is well versed in subjects like Microprocessors, Embedded Systems, Digital Logic Design
A. Praveen Kumar is a student studying final year of B. tech in K L University with CSE as his specialization. He has built some interesting apps in Android.
219
Vol. 3, Issue 1, pp. 209-220
D. Adarsh is a student studying final year of B. tech in K L University with CSE as his specialization. He has keen interest in the area of Android app development.
K. Krishna Mohan is a student studying final year of B .tech in K L University with CSE as his specialization. He developed many Android applications.
K. Venkata Kiran is a student studying final year of B. tech in K L University with CSE as his specialization. He has built some interesting apps in Android.
220
Vol. 3, Issue 1, pp. 209-220
A ROBUST DESIGN AND SIMULATION OF EFFICIENT MICROMECHATRONIC SYSTEM USING FSV CONTROLLED AUXILIARY DAMPED PMLSM
Sarin CR1, Ajai M2, Santhosh Krishnan3, Arul Gandhi 4
1, 3 &4 2
Department of Mechatronics Engineering, Karpagam College of Engineering, Coimbatore, India. Department of Industrial Refrigeration & Cryogenics, TKM College of Engineering, Kollam, India.
ABSTRACT
A Micromechatronic arm is best suited for many of the low range material handling systems where the power can be small but force in linear movement cases might be relatively high. Because of its simple structure and improved characteristics the PM stepper motor is an interesting alternative for low power applications where pneumatic or hydraulic linear drives are to be avoided .This paper studies and analyses a robust design and simulation of efficient Micromechatronic system using a Finite set value controlled auxiliary damped system to provide better operation of a permanent magnet linear stepper motor. The auxiliary damping control and current control circuit can improve the efficiency by reducing the vibration of forcer and adjusting better mechanical design. The processor based FSV program implementation can provide a smooth damping to achieve steady state in least time.
KEYWORDS: Auxiliary winding, damping control, FSV control, Linear Motor, Micromechatronic actuator,
PMLSM
I.
INTRODUCTION
Over the last few years the development of linear actuators has led to their increasing use in linear motion of mechatronic systems. Processor-controlled motors are one of the most versatile forms of positioning systems. Industrial applications of a Micromechatronic linear motor can be found in high speed pick and place equipment and multi-axis machine CNC machines [1]. Hydraulic or pneumatic system would formerly have used widely which lack higher degree of automation and require routine maintenance to avoid safety hazards and messy oil leaks [2]. Electric actuator systems are quiet, clean, non-toxic and energy efficient and can be integrated into sophisticated control systems even uses data bus communication [3]. The major disadvantages with electric actuators are increase in component costs about 40% higher and relatively low degree of force generated [4]. The micro-mechatronic system consists of an articulated arm with sensors and hydraulic actuators that are activated electronically. Externally controlled micro- actuation benefits from the ability of operation on arbitrary surfaces, high power at small velocities, more power per weight, increased reliability, reduced noise and longer lifetime [5][6]. The paper gives preliminary work devoted to the design and control the system with Finite Setting Value algorithm for a flexible Permanent magnet linear stepper motor associated to a hydraulic system which is more efficient by its novel design for friction compensation and momentum control by means of a force-current feedback loop control of auxiliary damp winding and a feed forward controller [7]. Linear stepping motors are an excellent solution for positioning applications that require rapid acceleration and high-speed moves with low mass payloads. This interface is intended to be used for grasping tasks, where unilateral constraints are usually present and very high power is needed which is obtained by the hydraulic block controlled
221
Vol. 3, Issue 1, pp. 221-232
International Journal of Advances in Engineering & Technology, March 2012. Technology, IJAET ISSN: 2231-1963
by PMLSM. Dramatic reductions in voltage and power requirements can be obtained by making microelectric motors and drive systems and by interfacing them with mechanical systems. mechanical The projected system of design bestow a robust design and simulation of efficient Micromechatronic nd system for the actuation of various linear mechanisms using Finite Set Value controlled auxiliary winding for the damping and vibration control of a Permanent magnet linear stepper motor which has many advantages over the conventional system design and control.
II.
MICRO-ELECTRIC PM LINEAR STEPPER MOTOR
The proposed linear motor is a brushless permanent magnet electric motor that can converts digital control pulses into axial shaft motion by dividing a full motion into a large number of steps which had its stator "unrolled" so that instead of producing a rotational torque, it produces a linear force along its length. One of the most common mode of operation is as a Lorentz Lorentz-type actuator, in which the applied force is linearly proportional to the current and the magnetic field as described in Figure 1 [7][8]. F = qv B (1)
Figure 1 : Representations of force distribution due to Lorentz force
A permanent magnet linear m micro-stepper motors as shown in Figure 2. , on the other hand, effectively have multiple "toothed" permanent magnets arranged in periphery of a central shaft of iron and linearly distributed alternative electromagnets at the inner periphery of stator The electromagnets stator. are energized step by step with an external control circuit like a microcontroller. To make the motor shaft move, first one electromagnet is given power, which makes the nearest gear's teeth magnetically , attracted to the electromagnet's teeth. When the gear's teeth are thus aligned to the first electromagnet, they are slightly offset from the next electromagnet.
Figure 2 : Schematic Design of Linear PMSM ure
A PMLSM has two parts, a stator and a forcer as shown in Figure 3. The stator hosts a three three-phase winding which generates a travelling magnetic field. The mover has a number of permanent magnets. A PMLSM is capable of generating two forces, longitudinal and normal, which can be more or less independently assigned by an a appropriate choice of the three-phase currents. So, when the next electromagnet is turned on and the first is turned off, the forcer moves slightly to align with the next one and from there the process is repeated Each of those slight movements is called a "step," with an repeated. integral number of steps making a full motion. In that way, the motor can be turned by a precise angle [9].
222
Vol. 3, Issue 1, pp. 221-232
Figure 3 : Basic Design of Linear PMSM Fig
Many advantages are achieved using this kind of motors, such as simplicity since no brushes or contacts are present, low cost, high reliability, high linear force at low speeds, and high accuracy of motion. The linear stepping motor is not subject to the same linear velocity and acceleration The acceleration. block diagram representation of a Mechatronic actuation composed of mechanical system actuated by ntation controlled electrical linear motor shown in Figure 4.
Figure 4 : Mechatronic actuation system Fig
III.
RELATED WORKS AND STRATEGIES
S. Seshagiri [10] demonstrated that stepper system could be fast and accurate with advanced feedback- position control . Bodson m., Chiasson j [11] used the technique of exact feedback linearization using full state-feedback, with extensions to the partial state feedback an experimental feedback, state-feedback and validation of the controller. Crnosija P., Kuzmanvoic B., and Ajdukovic S [12] improved performance by field control using a micro computer model based feedback controller with a programmable model-based procedure which is more users friendly. Chen et al. [13] proposed a new improved model for profile al. tracking using an advanced feedback controller with a least least-squares-based identification procedure based which was later by learning based control for precision control. These gave new way of intelligent controlling of the stepper motor to facilitate with new intelligent methods like knowledge based ing learning and artificial intelligent control. B. K. Bose [14] put forward a system for speed control of motor using intelligent neural network algorithms which give an initiation for implying such methods in control process. However, many of the above systems have shown much complication in vibration and damping control of a linear actuator basically about a permanent magnet linear stepper motor. Many systems with stepper motors need to control the acceleration / deceleration when changing the er speed and motor is not subject to the same linear velocity and acceleration as in rotary one. The PM linear motor has the highest force density among micromotors. But it had several disadvantage In any l disadvantage. of the mover's position one of its poles was generating a significant breaking force as shown in Figure 5, diminishing the total tangential force produced by the motor and thus reducing its efficiency. Beside this the magnetic flux passing through both the mover and the platen gave rise to a very strong normal force of attraction between the two armatures. This was over 10 times the peak holding tangential force of the motor. Due to this high attractive force produced of all the pole sophisticated poles linear bearing systems are required to maintain the precise air air-gap between the mover and platen [15].
223
Vol. 3, Issue 1, pp. 221-232
Figure 5 : Alignment of forcer position
Moreover, as the forcer speed increases, linear force decreases and the force curve may be extended by using current limiting drivers and increasing the driving voltage. Attempting a high-speed move with a low mass payload results in the majority of the motor linear force applied to overcoming the friction. Linear motors are not compact force generators compared to a rotary motor with a transmission offering mechanical advantage. For example to produce even 6.5 N (1.5 lb) of continuous force, a linear motors cross section is approximately 50 mm x 40 mm (2x 1.5) [16]. One of the major disadvantages of steppers is that it exhibits more vibration, as the discrete step tends to snap the shaft from one position to another due to the effect of momentum which is more at some speeds and can cause the motor to lose linear force. Thus it may affect effective force distribution, step angle resolution and accurate stepping position. So satisfactory damping of oscillations is an important issue to be addressed when dealing with the stability of shaft systems. The effect can be mitigated by accelerating quickly through the speed range, physically damping the system, or using a auxiliary flux distribution driver [17]. Tai-Sik , Seok [18] and Dong-Hun developed an active control scheme to damp the vibration of LHSM using a reluctance network based on the finite-element. But it suffers difficulties from a complex design and size of the control system. Barhoumi and Ben salah [19] designed a new model for positioning control of stepper motor using BP Neural Networks, showed many advantages over the conventional systems which make use of adaptive neural networking and knowledge base algorithms for the control. Kenneth Wang-, Norbert Chow [20] reported the satisfactory result of electronic damping based on the force constant and advanced algorithms for excitation and damping control, which is the motivation of this paper. Still the Kenneth - Norbert system suffer from some draw backs other the derived algorithms are more time consuming and the control of not much smoother. The proposed study is designed to achieve same with the use of an auxiliary winding at definite intervals in between stator poles which is exited by digitally controlled current from a processor. Finite set Value based algorithm is implemented in this system which proved a fast iterative mode of current control by probalistic selection of Finite error value coefficient.
IV.
MECHANICAL DESIGN CRITERIA
An oscillating mass will always have momentum, which is a function of its shape, mass distribution, and rate of spin. Notice, for example, that a compact object with all the mass concentrated near the center of mass spins much easier than an object that has a lot of mass located far from the center of mass. Euler's first law states that the linear momentum of a body is equal to the product of the mass of the body and the velocity of its center of mass [21] [22] [23]. The distribution of mass describes an objects mass moment (I). By knowing the mass (m), Linear Moment (I), and the objects velocity (v), momentum ( ) can be found.
I = mv (2)
Using the continuity condition,

=AV, (3)
The momentum equation simplifies to

F = Vi = AVi2 (4)
224
Vol. 3, Issue 1, pp. 221-232
The change of momentum will have two parts, momentum inside the control volume, and momentum passing through the surface. The mass, velocity and dimensions of the shaft cannot be varied to The reduce the momentum as it may affect the optimum design of shaft dimensions to obtain a sufficient imensions linear Lorentz force. As the case of the micro stepper motor the variation in dimensions make poor makes changes in momentum control Moreover the density of internal forces at every point is not control. necessarily equal, i.e. distribution of stress is at shaft pole and momentum is maximum shaft axis. This variation of internal forces throughout the body is governed by law of motion of conservation of linear momentum, which normally are applied to a mass particle but are extended in continuum mechanics to a body of continuously distributed mass. The most convenient way of control of the momentum is to use damping systems. The mechanical damping systems cannot be used in micromotors due to its huge size. The fluid damping control luid system cannot be used in electrical motors. Thus most favored mode of operation is to use electrical damping.
V.
AUXILIARY ELECTRICAL SYSTEM
To avoid the instability of the shaft after the step movement, a constant position contr system is o control designed and used in winding mechanism. Auxiliary-coil actuators are a special form of linear motor, coil capable of moving an inertial load at extremely high accelerations and relocating it to an accuracy of millionths of an inch over a limited ra range of travel [25].
Figure 6 : Auxiliary winding slots portioning
A method of reducing vibration in shaft systems by using a distortion-damper system with a two damper auxiliary winding is proposed where additional distortion source adopted [26] as described in Figure 6. Here eliminates the distortion within one sampling period by using a sampled ere sampled-data control system which makes use of the finite finite-settling-value method [27] [28]. The system is demonstrated by . applying the additional field which is opposite in polarity to the main field flux at both end to eliminate the distortion produced There must be initialization of to switch on the auxiliary windings produced. so that the winding facing the motion is switched on only after the forcer passed over the one and the other should be exited instantaneously with the pole excitation. The switching of winding is done by processor control which create a oppos opposing flux so that the forcer movement is limited within main field flux area and the auxiliary supply is eliminated after the damping control by current decaying to save power and the smooth effective distortion control [29]. The value of the excitation current in the auxiliary winding is controlled using the processing system which makes the current to decay makes gradually until the system is acquired stability which is represented using a block diagram in Figure 7.
Figure 7 : Auxiliary winding slots portioning
225
Vol. 3, Issue 1, pp. 221-232
VI.
MODELING OF THE AUXILIARY DAMPED PMLSM
Characteristics of auxiliary winding control are based on solution of electromagnetic field equations and control mode of processor based switches and control for solution of traveling magnetic field equations [3][23][31]. Basic problems of field design revolve around estimating the distribution of magnetic flux in circuit, which may include permanent magnets, air gaps, high permeability conduction elements, and electrical currents. Exact solutions of magnetic fields require complex analysis of many factors, although approximate solutions are possible based on certain simplifying assumptions [30]. Obtaining an optimum magnet design often involves experience and tradeoffs. There are two systems to be designed a system to produce an auxiliary field and current limiting circuit [31].
6.1. Auxiliary
Damping Control
Momentum equations for a motor design a feed-back controller, which can follow a desired momentum even in the case of large acceleration [32]. The mechanical part of the motor equations F =I +kfw(t)+kmi(t) Fi= (Vap(t ( +Ri(t)+Kbi(t) (
(5)
The field excitation required in auxiliary windings to eliminate distortion for unit time period
(6)
The field excitation in main winding Vapp(t) = L ( (7)
This sequence of equations leads to current to be applied on the auxiliary winding initially, =+ Vapp (t)
(8)
Thus the resulting force developed in auxiliary winding will be = - Kf w(t) + Km i(t)
(9)
The equations (6) to (9) describes the magnitude force F to be applied in order to obtain the resultant magnetic field force that can with hold the forcer head for a given time period dt.
6.2. Auxiliary
Current Control
Once the forcer reaches the main magnetic field the auxiliary supply will be initiated which will control the forcer position within the limit so that the distortion and vibration of the forcer head may reduce and the damping is within the limit [33]. Once the stability is achieved the system current must be switched off which may help to save energy. A gradually decaying current system have a better performance over the forcer than instant switching which may be attained using current limiting system. Chopping can limit the current in the winding without generating excess heat. It is quite elegant and efficient [34]. The essence of chopping is to switch the operating voltage on and off at a frequency higher than the operating range, and allow the winding itself to act as a filter, where Pulse-width modulation, in which the 'duty cycle' determines the behavior of the current.
226
Vol. 3, Issue 1, pp. 221-232
Figure 8 : Conceptual model of Chopper circuit
The most common approach is to automatically adjusting the duty cycle of the switches. The conceptual model of a Chopper circuit as shown in Figure 8, define when an incoming activate signal is received, the coil is switched on. Current through the coil is developed as a voltage across the position sensor. This voltage is proportional to the amount of current in the coil and forms an important part of our feedback loop. The value of the resistor is very low, typically 1 ohm, which equates to 1 amp, when the voltage at the sense resistor is 1 volt.) If R is large, the current through the motor windings will decay quickly when the higher level control system turns off this motor winding, but when the winding is turned on, the current ripple will be large and the power lost in R will be significant. If R is small, this circuit will be very energy efficient but the current through the motor winding will decay only slowly when this winding is turned off, and this will reduce the cutoff speed for the motor [25]. The peak power dissipated in R will be I2R during Toff and zero during Ton; thus, the average power dissipated in R when the motor winding is on will be: P = I2R (10) The sensed voltage is compared with a reference voltage, and when the sensed voltage becomes greater than the reference the logic it reduces the current value. When the voltage drops below the reference voltage the coil will give a signal to processor for further action unless of course the incoming activate signal is removed, in which case the coil is always off. Thus the feedback logic flips the switch on and off when the current is too high, maintaining a smooth current decaying. The reference voltage is typically adjustable which allows matching the current in the circuit to the motors rated current.
6.3. Finite-Settling-Value Microprocessor Control
An algorithm model of the microcontroller architecture is designed and desired operation is formulated using Finite-Settling-Value [27][28], and a software simulator implementing the auxiliary-controlled actuator model including all the nonlinear elements is formulated to predict the real system's behavior and operation quite well [35][36] . 1. START 2. Initial values of main field flux, current position of shaft, shaft dimensional characteristics, proportion momentum, and frictional loss are determined. 3. FSV error co-efficient is initialized which is based on the current decay constant and is maximum at excitation level. 4. The direction of forcer movement is determined and corresponding switches are turned on and along with auxiliary winding supply at moving side. 5. Once the forcer is moved the Ton,, Toff, I, Fi is determined and the auxiliary coil at stationary side is energized. 6. From the inertia force equation the value of the auxiliary winding current is found and the reference voltage is set. 7. Once the forcer attains stability over 60% which is identified from position sensor current, the current control circuit is activated. 8. The current control circuitry is kept on till the value reaches zero which is determined by
227
Vol. 3, Issue 1, pp. 221-232
Toff = L 9. The conditions are reset to new values when it reaches the next step procedure.
(11)
VII. RESULTS AND DISCUSSIONS

The experimental results of both the Modeling and Simulink procedures and the control system testing circuit are shown in Fig. 9 to Fig. 14. The influence of various parameters over the working conditions and characteristics is described. In the full step sequence, only one coil sector is energized at the same time and forcer moves to the same sector. Still such mode of steps reduces the step resolution. The change in the forcer position with respect to the time for an auxiliary damped system is a s shown in Figure 9.
Figure 9 : Auxiliary Damped Full Step position Control
The microcontroller based continuously varying mode of excitation as shown in Figure 9 instead of instant switching makes the system to enable a smooth forcer movement with less vibration, improved damping control and higher accuracy. The isolated auxiliary current distribution for a complete full step motion for both au auxiliary winding related to corresponding poles is described in Figure 10,
Figure 10 : Distribution for auxiliary winding current between two isolated auxiliary windings
The table of content shown below gives the current distribution of main and auxiliary winding in Half Step Sequence. Here, the step angle reduces to half the angle in full mode so that the angular resolution is also increased i.e. it becomes double the angular resolution in full mode.
228
Vol. 3, Issue 1, pp. 221-232
Current at Coil A (%) Current at Coil B (%) Current at Coil C (%) Current at Aux Coil (A1) (%) 0 100 79.5 52.7 28.2 14.9 0 0 0 0 0 0 0 Current at Aux Coil A2 (%) 100 100 79.5 52.7 28.2 14.9 0 0 0 0 0 0 0 Current at Aux Coil A3 (%) 0 0 0 0 0 0 0 0 0 6.2 16.4 21.3 31.9 Aux Current FSV Error Value (%) 100 100 80 60 40 20 0 0 0 0 0 0 100
Sl . No
Step
Time (S)
1 2 3 4 5 6 7 8 9 10 11 12 13
1 1 1 1 1 1 1 1 1 1 1 1 1
0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36 0.40 0.44 0.48 0.52
100 100 100 100 100 100 100 100 100 100 100 100 100
0 0 0 0 0 0 0 19.0 38.1 56.9 78.7 96.3 100
0 0 0 0 0 0 0 0 0 0 0 0 0
The auxiliary current distribution for a half step motion for corresponding auxiliary winding s is described in Figure 11. which shows that at each step angle the auxiliary winding current related to moving position (A2) may vary from its predistributed level gradually and the counterpart (A1) from zero to maximum, enables a smooth current decaying. For the next half step before full excitation is achieved A 3 may increase gradually to a level depend on full excitation value to obtain even varying nd of main flux and reduced vibration.
Figure 12 : Current variation in auxiliary windings (A1,A2,A3) for a half step
The velocity distribution of the forcer with respect to the duty cycle is describes that the acceleration due field excitation is varying as in a quadratic form, but later deforms due to the initial excitation in auxiliary winding as shown in Figure 13.
229
Vol. 3, Issue 1, pp. 221-232
Figure 13 : Auxiliary current Switching Characteristics
The Force Speed curve plotted in Figure14. shows that as the linear velocity is increased the output power decreases gradually a in inverse promotion do minimum . As the speed force value cannot be compromised; it is favorable to use a electro mechanical system for effective mode of operations.
Figure 14 : Linear Velocity Vs Force
VIII. CONCLUSION
The paper proposes new design criteria over the conventional system which gives a smooth energy efficient operation of the linear actuation system. The limits imposed by a auxiliary current control and the selected decay mode give a better performance and reduced distortion of the forcer specifically it's ability to follow a smooth operation. So long as these limits allow the designer to pecifically se achieve the desired resolution in the micro stepping application the devices provide a cost effective implementation. The novel algorithm is feasible and can produce significant energy savings with minimal performance degradation. Future work will involve developing alternative on online intelligent algorithms, including approaches for effective development in reduced friction characteristics and improved damping control.
REFERENCES
[1]. V. F. Cardoso, Graa Minas1and S Lanceros-Mndez, Design and fabrication of piezoelectric Micro S. actuators for micro fluidic applications Semana de Engenharia 2010 Guimares, 11 a 15 de Outubro applications, [2]. M. T. Mason, Mechanics of Robotic Manipulation. Cambridge, MA: MIT Press, 2001. [3]. Boldea, I., "Linear Electromagnetic Actuators and their Control: A Review," Proceedings of the 10th International Power Electronics and Motion Control Conference (EPE PEMC '2002), Cavtat & Dubrovnik (EPE-PEMC (Croatia), 2002. [4]. Cassat, A., Corsi, N., Moser, R., Wavre, N., "Direct Linear Drives: Market and Performance Status," Wavre, Proceedings of the 4th International Symposium on Linear Drives for Industry Applications LDIA '2003, Birmingham (UK), 2003
230
Vol. 3, Issue 1, pp. 221-232
[5]. D. Popa and H. E. Stephanou, Micro and meso scale robotic assembly, SME J. Manuf. Processes, vol. 6, no. 1, pp. 5271, 2004. [6]. P. Dario, R. Valleggi, M. C. Carrozza, M. C. Montesi, and M. Cocco, Microactuators for microrobots: A critical survey, J. Micromech. Microeng., vol. 2, no. 3, pp. 141157, Sep. 1992. [7]. Viorel I.A., Szab L., "Hybrid Linear Stepper Motors," Mediamira, Cluj (Romania), 1998. [8]. I. Boldea and S. A. Nasar. Permanent magnet linear alternators part I: Fundamental equations. IEEE Transactions on Aerospace and Electronic Systems, AES-23(1), 1987. [9]. S.A. Nasar and I. Boldea, Linear Electric Motors: Theory, Design and Practical Applications. Englewood Cliffs, NJ: Prentice-Hall, 2004. [10]. Perriard, Y., Scaglione, O. and Markovic, M. Self-Sensing Methods for PM Motors in Mechatronics Applications. Proceedings of the IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM 2010). [11]. Bodson m., Chiasson j.n., Novotnak r.t., Rekowski r.b,High Performance nonlinear feedback control of a permanent Magnet stepper motor, I trans. Control syst. Technol., 1993, 1, (1), pp. 514 [12]. Crnosija P., Kuzmanvoic B., and Ajdukovic S., Microcomputer implementation of optimal algorithms for closed-loop control of hybrid stepper motor drives, IEEE Transactions on Industrial Electronics, vol. 47, pp. 1319-1325, 2000. [13]. W. D. Chen, K. L. Yung, and K. W. Cheng, Profile tracking performance of a low ripple hybrid stepping motor servo drive, Proc. Inst. Elect. Eng.Control Theory Appl., vol. 150, no. 1, pp. 6976, Jan. 2003. [14]. B. K. Bose, Neural network applications in power electronics and motor drivesAn introduction and perspective, IEEE Trans. Ind. Electron., vol. 54, no. 1, pp. 1433, Feb. 2007. [15]. E. Fitan, F. Messine, and B. Nogarde. An electromagnetic actuator design problem: A general and rational approach. IEEE Transactions on Magnetics, 40(3), 2004. [16]. D. Zarko. A Systematic Approach to Optimized Design of Permanent Magnet Motors. PhD thesis, University of Wisconsin-Madison, 2004. [17]. B. Murphy, Design and Construction of a Precision Linear Motor and Controller , Masters Thesis, Dept. of Mechanical Engineering, Texas A&M University, May 2003. [18]. T.-S. Hwang, J.-K. Seok, and D.-H. Kim, Active damping control of linear hybrid stepping motor for cogging force compensation, IEEE Trans. Magn., vol. 42, no. 2, pt. 2, pp. 329334, Feb. 2006. [19] Barhoumi EM, Ben salah B , New Positioning Control of Stepper Motor using BP Neural Networks ,Journal of Emerging Trends in Computing and Information Sciences,,Volume 2 No.6, June 2011 [20]. Kenneth Wang-Hay Tsui, Norbert Chow Cheung, Kadett Chi-Wah Yuen, Novel Modeling and Damping Technique for Hybrid Stepper Motor, IEEE Transactions On Industrial Electronics, Vol. 56, No. 1, January 2009 [21]. Tai-Sik Hwang; Jul-Ki Seok; Dong-Hun Kim; Active damping control of linear hybrid stepping motor for cogging force compensation , Volume: 42 Issue:2,pp 329 - 334, Feb. 2006 [22]. Kaunas: Technologija, 2006. P. 125128. 9. Budig Peter-Klaus. Simplification of the mechanical design of drives with the application of direct drives ,Publication of the 4th International Symposium Topical Problems of Education in the Field of Electrical and Power Engineering. Tallinn Technical University, Kuressaare. 2007. [23]. Basak, A., Flores Filho, A. F., Nakata, T., Takahashi, N. Three dimensional computation of force in a novel brushless DC linear motor . IEEE Transactions on Magnetics. 1997. Vol. 33, No. 2, 3 p. [24]. Bobrow, J. E., Desai, J. 1990. Modeling and Analysis of a High-Torque, Hydrostatic Actuator for Robotic Applications. Experimental Robotics I, the First Int. Syrup. V. Hayward, O. Khatib (Eds.) SpringerVerlag, pp. 215-228. [25]. K.-F. Bhringer, B. R. Donald, R. R. Mihailovich, and N. C. MacDonald, A theory of manipulation and control for microfabricated actuator arrays, in Proc. IEEE Workshop MEMS, Jan. 1994, pp. 102107 [26]. Liu, C., Kuo, J. Experimental investigation and modeling of linear variable reluctance machine with magnetic-flux decoupled windings. IEEE Transactions on Magnetics. 1994. Vol. 30, No. 6, 3 p. [27]. S. Choura, Design of finite time settling regulators for linear systems, Trans. ASME, vol. 116, pp. 602608, 1994.
231
Vol. 3, Issue 1, pp. 221-232
[28]. J. M. Coron, On the stabilization in finite time of locally controllable systems by means of continuous time-varying feedback law, SIAM J. Contr. Optim., vol. 33, pp. 804833, 1995. [29]. Pfister, P.-D. and Perriard, Y. (2010) Very High Speed Slotless Permanent Magnet Motors: Analytical Modeling, Optimization, Design and Torque Measurement Methods. IEEE Transactions on Industrial Electronics, 57(1) pp. 296-303 [30]. S. F. Bart, M. Mehregany, L. S. Tavrow, J. H. Lang, and S. D. Senturia, Electric micromotor dynamics, IEEE Trans. Electron Devices, vol. 39, no. 3, pp. 566575, Mar. 1992 [31]. K.J. Binns, P.J. Lawrenson, and C.W. Trowbridge. The Analytical and Numerical Solution of Electric and Magnetic Fields. John Wiley and Sons. [32]. J Lidenholm Power System Stabilizer Performance ISSN 1401-5757, UPTEC F07 109 [33]. M. Wahln, Transfer function for excitation system and automatic voltage regulator,VG Power AB, 2004. [34]. AN1495, Application Note, Microstepping Stepper Motor Drive Using Peak Detecting Current Control, Thomas Hopkins,(2010),pp 4-9. [35]. S. Brckl, Feed-drive system with a permanent magnet linear motor for ultra precision machine tools, Proc. of IEEE International Conference on Power Electronics and Drive Systems, pp.821-826, July 1999. [36]. Ziqing Ye; Min Chen; ,DSP based field oriented hybrid stepper motor control, IEEE International Conference on E-Product E-Service and E-Entertainment (ICEEE), 2010, pp 1 - 4
Authors
Sarin CR received B.E in Electrical and Electronics Engineering from Dhanalakshmi Srinivasan Engineering College, Perambalur, India in 2010 and currently he is perusing his M.Tech in currently he is perusing his M.E in Mechatronics Engineering from Karpagam College of Engineering, Coimbatore, India He has published one paper on Intelligent Non Destructive testing and Evaluation and two papers are expected to be published in shorter. He has special interest on the fields of Robotics, Artificial intelligence, Micro mechatronic systems, intelligent control of Electrical Drives. Ajai M received B.E in Mechanical Engineering from Dhanalakshmi Srinivasan Engineering College, Perambalur,India in 2010 and currently he is perusing his M.Tech in Industrial Refrigeration and Cryogenics Engineering from Thangal Kunju Musaliar College of Engineering,Kollam,India. He has special interest on the fields of Mechanical Machine design, Quality control, Fluid Systems control, and Machine Dynamics.. Santhosh Krishnan received B.E in Electronics and Instrumentation Engineering from MAM College of Engineering, Trichy, India in 2010 and currently he is perusing his M.E in Mechatronics Engineering from Karpagam College of Engineering, Coimbatore, India. He has special interest on the fields of Robotics; Sensor based Automation and control of machines, Non Destructive testing and Evaluation.
Arul Gandhi received B.E in Electrical and Electronics Engineering from Karpagam College of Engineering, Coimbatore, India in 2007 and currently he is perusing his M.E in Mechatronics Engineering from Karpagam College of Engineering, Coimbatore, India. He has wide industrial experience in Embedded based control of machines and systems. He has special interest on the fields of Robotics, Embedded Systems, Navigation and Wireless communication, Knowledge based algorithm design.
232
Vol. 3, Issue 1, pp. 221-232
INFLUENCE OF FILLER MATERIAL ON GLASS FIBER/ EPOXY COMPOSITE LAMINATES DURING DRILLING
M.C.Murugesh1 and K.Sadashivappa2
1 2
Senior Lecturer, Dept. of Mechanical Engg., GM Institute of Technology, Davangere Professor & Head, Department of IP Engineering, Bapuji Institute of Engineering and Technology, Davangere
ABSTRACT
The use of polymeric composite materials has increased considerably over the last decade. Drilling is a frequently practiced machining process in industry due to the need for component assembly in mechanical pieces and structures. Machining processes are generally used to cut; drill or contour composite laminates for building products. In fact, drilling is one of the most commonly used manufacturing processes to install fasteners for assembly of laminate composites. The material anisotropy resulting from fiber reinforcement heavily influences the machinability during machining. Machining of fiber reinforced plastic (FRP) components is often needed in spite of the fact that most FRP structures can be made to near-net shape and drilling is the most frequently employed secondary machining process for fiber reinforced materials. The use of filler material like TiO2 and Graphite have shown that better bonding of the fiber matrix has got its effect on thrust and delamination factor values.
KEYWORDS: Drilling; Polymer-matrix composites; Thrust force; Delamination
I. INTRODUCTION
Fiber reinforced plastics have been widely used for manufacturing aircraft and spacecraft structural parts because of their particular mechanical and physical properties such as high specific strength and high specific stiffness. Another relevant application for fiber reinforced polymeric composites (especially glass fiber reinforced plastics) is in the electronic industry, in which they are employed for producing printed wiring boards. Drilling of these composite materials, irrespective of the application area, can be considered as a critical operation owing to their tendency to delaminate when subjected to mechanical stresses. With regard to the quality of machined component, the principal drawbacks are related to surface delamination, fiber/resin pullout and inadequate surface smoothness of the hole wall. Among the defects caused by drilling, delamination appears to be the most critical [1] Figure 1 shows that the factors such as cutting parameters and tool geometry/material must be carefully selected aiming to obtain best performance on the drilling operation, i.e., to obtain best hole quality, which represents minimal damage to the machined component and satisfactory machined surface.
Figure 1. Principal aspects to be considered when drilling fibre reinforced plastics
233
Vol. 3, Issue 1, pp. 233-239
Composite materials are constituted of two phases: the matrix, which is continuous and surrounds the other phase, often called as reinforcing phase [3]. Epoxy resins are widely used as matrix in many fibre reinforced composites; they are a class of thermoset materials of particular interest to structural engineers owing to the fact that they provide a unique balance of chemical and mechanical properties combined with wide processing versatility [4]. Within reinforcing materials, glass fibres are the most frequently used in structural constructions because of their specific strength properties [3]. The present study focuses on mach inability study of GFRP laminated composites with filler material Tio2 and graphite and evaluation of thrust force, delamination factor for two drill diameters at different speeds
1.1.
Machining of composite materials
The machining of GFRP is quite different from that of metals and results in many undesirable effects such as rapid tool wear, rough surface finish and defective subsurface layers caused by cracks and delaminations. At the beginning of drilling operation, the thickness of the laminated composite material is able to withstand the cutting force and as the tool approaches the exit plane, the stiffness provided by the remaining plies may not be enough to bear the cutting force, causing the laminate to separate resulting in delamination. The delaminations that occur during drilling severely influence the mechanical characteristics of the material around the hole. In order to avoid these problems, it is necessary to determine the optimum conditions for a particular machining operation. Drilling is a particularly critical operation for fiber reinforced plastics (FRP) laminates because the great concentrated forces generated can lead to widespread damage. The major damage is certainly the delamination that can occur both on the entrance and exit sides of the work piece [4]. The delamination on the exit surface, generally referred to as push down delamination, is more extensive, and is considered more severe. Hocheng and Tsao have beautifully explained the causes and mechanisms of formation of these push down delaminations and they have also reasoned out the dependence of extent of delamiantion on the feed rate [5]. In earlier studies it has been observed that the extent of delamination is related to the thrust force, feed, material properties and speed, etc. and that there is a critical value of the thrust force (dependent on the type of material drilled) below
which the delamination is negligible [6]. 1.2. Specimen Preparation

The method that is used in the present work for manufacturing the laminated composite plates is hand lay-up as shown in Figure 2, which is the oldest method that was used to get the composite materials. The type of Glass Fiber mat selected to make specimens was, Mat II (330GSM). The matrix material used was a medium viscosity epoxy resin (LAPOX L-12) and a room temperature curing polyamine hardener (K-6), both manufactured by ATUL India Ltd, Gujarat, India. This matrix was chosen since it provides good resistance to alkalis and has good adhesive properties. Based on volume fraction the calculations were made for 60-40 (60% Glass Fiber 40% Epoxy Resin) combination showed a better result. Two filler materials TiO2 and Graphite were added to Mat II 60-40 combination by keeping Epoxy percentage constant (40%). Based on literature survey the amount of filler added was 3, 6, & 9 % of Graphite and 1, 2, & 3 % of TiO2, the details are as shown in Table 1. after preparation of the specimen the specimens were tested in impact testing machine to obtain impact toughness value.
Figure 2. Hand lay-up Technique
234
Vol. 3, Issue 1, pp. 233-239
Table 1. Filler material specimen details
1.3.
Experimental Setup
The high speed steel twist drill has an 1180 point angle. Two diameters 6.35mm and 4.76mm were selected to work on radial drilling machine which has a maximum spindle speed of 2650 rpm. There details are shown in Table 2. A piezoelectric dynamometer was used to acquire the thrust force as shown in Figure 3. The damage around the holes was measured using a tool makers microscope.
Table 2: Drill tool dia and corresponding speed
Figure 3. Schematic setup to measure thrust force
II. RESULTS AND DISCUSSIONS

2.1 Thrust force
Cutting forces are very useful for drill-wear monitoring, because these forces generally increase with tool wear. Thus, within the tool wear region, cutting forces provide good assessment of the tool condition. If the tool cannot withstand the increased cutting forces, catastrophic tool failure becomes inevitable. Consequently, tool life, which is a direct function of tool wear, is best determined by monitoring thrust force. Due to the thrust developed during drilling, many common problems exist. Some of the problem causes in drilling are fiber breakage, matrix cracking, fiber/matrix debonding, fiber pull-out, fuzzing, thermal degradation, spalling and delamination. The thrust force and torque developed in drilling operation is an important concern. Monitoring of thrust force in drilling is needed for the industry. In Figure 4 a qualitative trend of thrust force as a function of the drilling is shown. As can be seen, a pushing action is exerted by the drill on the work piece. In the first phase the thrust force continues to increase as an increasing part of the cutting lips is engaged in the material; in the second phase the thrust force remains at an almost constant value as the drill sinks into the work piece. In the third phase the thrust force rapidly decreases when the twist drill exits.
235
Vol. 3, Issue 1, pp. 233-239
Figure 4. Responses of cutting forces during drilling showing key process points
The value of thrust force was measured using a piezoelectric dynamometer. Figures 5 and 6 show the results of the thrust force for the two sets of drilling tests, as a function of the cutting parameters.
5a
5b
Figure 5. Effect of spindle speed on thrust force for 6.35mm and 4.76mm twist drill for TiO2 filler material.
6a
6b
Figure 6. Effect of spindle speed on thrust force for 6.35mm and 4.76mm twist drill for graphite filler material.
After drilling glass fiber reinforced epoxy composite laminates manufactured by hand lay-up; using two different HSS twist drill and various cutting speeds, the cutting speed is the cutting parameter that has the highest physical as well as statistical influence on the thrust force and surface roughness in GFRP material, the following conclusions can be drawn: From the Figures 5 and 6 it could be seen that as the speed increases the thrust force decreases for both the drill diameters. This is due to abrasiveness inherent property of the filler material.
236
Vol. 3, Issue 1, pp. 233-239
As the filler percentage increased from (1% to 3%) TiO2 and (3% to 6%) Graphite the thrust force values are less at all the speed intervals and has shown downward trend. When comparing among drill diameters 4.76mm have recorded higher values than 6.35 mm
2.2 Delamination factor (Fd) Delamination is caused by different drilling parameters. Several ratios were established for damage evaluation. One of them is delamination Factor (Fd), a ratio between the maximum delaminated diameter (Dmax) and drill diameter (D0). Fd = Dmax/D0. Figure 7 shows Tool Makers microscope with which delamination was measured.
Figure 7. Schematic view of delamination factor and a view of tool makers microscope
Delamination is commonly classified as peel-up delamination at the twist drill entrance and pushdown delamination at the twist drill exit as shown in Figure 8.
Figure 8. Delamination at the twist drill entrance (left) and exit (right) when drilling laminates
9a 8b
Figure 9.The Effect of Spindle Speed on the Delamination Factor (Fd) for 6.35 mm and 4.76 mm twist drill for TiO2 filler material.
10a
10b
Figure 10.The Effect of Spindle Speed on the Delamination Factor (Fd) for 6.35mm and 4.76 mm twist drill for Graphite filler material.
237
Vol. 3, Issue 1, pp. 233-239

From the Figures 9 and 10 it could be seen that as the speed increases the delamination factor decreases for both drill diameters . It is seen that decrease in spindle speed results in increase in the thrust force. This increase in thrust force causes increased delamination. Hence , thrust force is recognized as the main cause of damage As the filler percentage increased from 1% to 3% for TiO2 and 3% to 9% for Graphite the delamination factor values are less at all the speeds and has shown downward trend.
III.
CONCLUSIONS
Based on the experimental results presented, the following conclusions can be drawn: Considerable efforts have been focused on the better understanding of the phenomena associated with the cutting mechanism. Conventional high speed steel twist drills are used for drilling operation Abrasion, was the principal wear mechanism and led to the elevation of the thrust force. The increasing of thrust force as a result of increasing drill pre-wear leads to destroying the matrix and micro-cracking at the ply interfaces, which deteriorates the surface finish. The principal factors used to evaluate the performance of the process are undoubtedly the damage caused at the drill entry or exit of the hole produced. The damage decreases with cutting parameters, which means that the composite damage is smaller for higher cutting speed within the cutting range tested. Delamination decreases as the spindle speed is elevated, probably owing to the fact that the cutting temperature is elevated with spindle speed, thus promoting the softening of the matrix and inducing less delamination. The effect of addition of filler material like TiO2 and Graphite have shown that higher the percentage lesser the values of thrust and delamination factor , which insists that the better bonding of the filler material with the fiber matrix have increased the capacity of force sustainability.
REFERENCES
[1]. A.M. Abrao et.al,. Drilling of fiber reinforced plastics: A review, Journal of Materials Processing Technology 186 (2007) 17. [2]. W.D. Callister, Materials Science and Engineering: An Introduction, sixth ed., Wiley, Canada, 2002. [3]. M.A. Boyle, C.J. Martin, J.D. Neuner, Epoxy resin, in: H. Hansmann (Ed.), Composites Compendium, ASM Handbook FB MVU, Werkstofftechnologien, Kunststofftechnik, 2003. [4]. S. Jain, D.C.H. Yang, Delamination-free drilling of composite laminates, Trans. ASME J. Eng. Ind. 116 (4) (1994) 475481. [5]. H. Hocheng, C.C. Tsao, Analysis of delamination in drilling composite materials using core drill, Aust. J. Mech. Eng. 1 (1) (2003) 4953. [6]. H. Saghizadeh, C.K.H. Dharan, Delamination fracture toughness of graphite and aramid epoxy composites, Trans. ASME J. Eng. Mater. Technol. 108 (1986) 290295. [7]. Paulo Davim, J; et al: Experimental study of drilling glass fiber reinforced plastics (GFRP) manufactured by hand lay-up, Composites Science and Technology, vol. 64, 2004, 289-297. [8]. Campos Rubio J; et al: Effects of high speed in the drilling of glass fibre reinforced plastic: Evaluation of the delamination factor, 'International Journal of Machine Tools & Manufacture, vol. 48, 2008, 715720. [9]. Abrao, AM; et al: The effect of cutting tool geometry on thrust force and delamination when drilling glass fibre reinforced plastic composite, 'Materials and Design, vol. 29, 2008, 508-513. [10]. Mohan, NS; et al: Delamination analysis in drilling process of glass fiber reinforced plastic (GFRP) composite materials, Journal of Materials Processing Technology, vol. 186, 2007, 265-271. [11]. Paulo Davim, J; et al: Drilling fiber reinforced plastics (FRPs) manufactured by hand lay-up: influence of matrix (Viapal VUP 9731 and ATLAC 382-05), Journal of Materials Processing Technology, 155 156, 2004, 1828833. [12]. Okenwa, I Okoli; Ainullotfi Abdul - Latif: Failure in composite laminates: overview of an attempt at prediction,Composites: Part A, vol. 33, 2002, 315-321. [13]. Khashaba, UA; et. al: Machinability analysis in drilling woven GFR/epoxy composites: Part II Effect of drill, 'wear, Composites: Part A, vol. 41, 2010, 11301137. [14]. Hocheng, H; Tsao, CC: Effects of special drill bits on drilling-induced delamination of composite materials, International Journal of Machine Tools & Manufacture, vol. 46, 2006, 14031416. [15]. Khashaba, UA: Delamination in drilling GFR - thermoset composites, Composite Structures, vol. 63, 2004, 313327.
238
Vol. 3, Issue 1, pp. 233-239
[16]. Malhotra, SK: Some studies on drilling of fibrous composites, Journal of Materials Processing Technology, vol. 24, 1990, 291-300. [17]. Hocheng, H; Tsao, CC: The path towards delamination-free drilling of composite materials, 'Journal of Materials Processing Technology, vol. 167, 2005, 251264. [18]. Hocheng, H; Tsao, CC: Effects of exit back-up on delamination in drilling composite materials using a saw drill and a core drill, 'International Journal of Machine Tools & Manufacture, vol. 45, 2005, 12611270. [19]. Chung - Chen Tsao; Wen - Chou Chen: Prediction of the location of delamination in the drilling of composite laminates, Journal of Materials Processing Technology, vol. 70, 1997, 185-189. [20]. Caprino, G; Tagliaferri, V: Damage development in drilling glass fibre reinforeced plastics, 'International Journal of Machine tools and Manufacturing, vol. 35,1995, 817-829. [21]. S. Basavarajappa;K.V.Arun; J Paulo Davim: Effect of Filler Materials on Dry Sliding Wear Behavior of Polymer Matrix Composites A Taguchi Approach, Journal of Minerals & Materials Characterization & Engineering, Vol. 8, No.5, pp 379-391, 2009.
M.C.MURUGESH received his Diploma in Mechanical Engineering from Bapuji Polytechnic, Davangere, got his BE degree and Masters degree in Mechanical Engineering from UBDT Engineering college, Mysore University, Karnataka in 1993 and 2004.He has served as mechanical engineer in Aditya Birla Group company , Harihar for seven years. His research interest area is on composite materials, presently doing Ph.D work on the same field under Visvesvaraya Technological University, Karnataka. Currently he is a senior Faculty member from Department of Mechanical Engineering GM Institute of Technology, Davangere, Karnataka State, India. K. SADASHIVAPPA received his BE degree and Masters Degree from Mysore University in the production field in 1984 and 1990 respectively. He received the Ph.D degree in Mechanical from IIT Madras in 1997. Currently he is a Professor and Head, Department of Industrial & Production, BIET, Davangere, Karnataka. He has guided two Ph.D scholars. He has got two National awards in his name from CSIR & AICTE New Delhi. He has published number of papers in journals and conferences. He has also published four book chapters under AICTE CEP publication for Engineering teachers.
239
Vol. 3, Issue 1, pp. 233-239
LFSR TEST PATTERN FOR FAULT DETECTION AND DIAGNOSIS FOR FPGA CLB CELLS
Fazal Noorbasha, K. Harikishore, Ch. Hemanth, A. Sivasairam, V. Vijaya Raju
Department of ECE, KL University, Vaddeswaram, Guntur (Dist.), AP, India
A BSTRACT
The increasing growth of sub-micron technology has resulted in the difficulty of VLSI testing. Test and design for testability are recognized today as critical to a successful design. Field Programmable Gate Arrays (FPGAs) have been used in many areas of digital design. Because FPGAs are reprogrammable, faults can be easily tolerated once fault sites are located. In this paper, we discuss about fault detection and fault diagnosis techniques for FPGA CLBs. The most of the discussion will be made using Configurable Logic Block (CLB) instead of whole FPGA for simplicity. In order to generate testing pattern we aid Linear Feedback Shift Register (LFSR). Fault detection and location will be carried out using output response analyzer. Diagnosis will be discussed accordingly. For this analysis we have taken XC4000 FPGA. VHDL is used as HDL language.
K EY WORDS: FPGA, CLB, LFSR, XC4000, VHDL

I. INTRODUCTION
Field programmable gate arrays (FPGAs) are a popular choice among VLSI devices, any logical circuit can be implemented into the FPGA at low cost. It consists of an array of configurable logic blocks (CLBs), programmable interconnect and programmable Input/output blocks. Many methods have been proposed to test FPGAs [1]. In some works, the circuits under consideration are programmed FPGAs, in which logic circuits have been implemented. Since an FPGA can be programmed in many different ways, this method is not applicable to manufacturing time testing, as we do not know the final configuration. Testing faults in general FPGAs has been proposed by many researchers. In these methods, the FPGA under test is not mapped to a specific logic function. As a result, multiple test sessions are usually required, with each session dealing with one configuration. The BIST architecture requires the addition of three hardware blocks to a digital circuit: a pattern generator, a response analyzer and a test controller. Examples of pattern generators are a ROM with stored patterns, a counter and a linear feedback shift register. LFSR is constructed using flip-flops connected as a shift register with feedback paths that are linearly related using XOR gates. An LFSR can be used for generation of pseudo-random patterns, polynomial division, response compaction etc. [2]. LFSR is more popular for implementation of both TPG and ORA due its compact and simple structure. Figure 1 shows the generic FPGA architecture for VLSI design.
Figure 1. Generic FPGA architecture
240
Vol. 3, Issue 1, pp. 240-246
A typical response analyzer is a comparator with stored responses or an LFSR used as a signature analyzer. Traditional chip testing deals with fault detection only, while fault diagnosis is often conducted at the system level. This is because components in the chip cannot be repaired. However, faults in FPGAs can be easily tolerated by not including faulty elements in the final circuit. Therefore, FPGA chips with faults can still be used if we can identify the fault sites. In this paper, we propose a chip-level diagnosis methodology for faults in CLBs. Our method is also based on the BIST technique, which means that the testing process is conducted by the chip itself, and the requirement for external ATE support is limited.
II.
PRELIMINARIES AND FAULT DIAGNOSIS
In order to diagnose faults, there must first be a way to test modules in FPGAs. A candidate for this purpose is the built-in self-test (BIST). This structure reconfigures part of the functional circuit to be a test pattern generator (TPG), and some other to be an output response analyzer (ORA) [4]. The rest circuit consists of the circuit under test (CUT). The TPG is either a linear feedback shift register (LFSR) that generates pseudorandom test sequences, or simply a counter that provides an exhaustive test set.
Figure 2. Connections between testing module and set of CLBs under test
The test inputs are fed to the CUT, while the output responses are collected and analyzed by the ORA. The ORA can be either a signature analyzer or a comparator-based analyzer [5]. Our test architecture works as follows. An FPGA is divided into disjoint sets of CLBs, where each set can be configured into a TPG and ORA as shown in figure 2. Such a set acts as a module in the PMC model since it is able to test another module and determine whether the CUT passes or fails the given test. All the CLBs under test are programmed in the same way; therefore, they perform the same logic function and could be given the same test patterns. Thus outputs of the TPG are fed to all CLBs in the set under test, and the results are analyzed by the ORA. Since each CLB can be programmed in many ways, it is not possible to test any CLB in a single test run. As a result, a complete test of all faults in a CLB usually requires several steps, and in each step a CLB is programmed in a particular way. Our fault diagnosis process contains two sessions: 1) horizontal diagnosis, and 2) vertical diagnosis. Horizontal diagnosis is illustrated in figure 3 (a). The outputs of the rightmost cells are compared with correct results for identifying faulty rows.
Figure 3. (a) Horizontal Diagnosis (b) Vertical diagnosis
241
Vol. 3, Issue 1, pp. 240-246
The interconnection structures for vertical diagnosis are shown in figure 3(b). The outputs of the bottommost cells are compared with correct results for identifying faulty columns. By intersecting faulty columns with faulty rows, the faulty cells can be identified. The diagnosis resolution of our approach is a single CLB due to the structure of a basic cell. We arranged the CLBs shown above in an array of 88 for the experiment purpose. Since CLBs are configurable; we can inject faults into CLBs by simulating the effects of logic faults. We consider two types of faults: stuck-at faults and open faults. We simulate the effect of a stuck-at fault by connecting the faulty line to logic 0 or 1. To simulate the effect of an open fault, all we have to do is to disconnect the faulty line from its driving source [6]. Once by using the above diagnosis we can find the faulty CLBs easily. Once the location is found we can avoid those CLBs from programming and FPGA still be used efficiently.
III.
LFSR FOR TEST PATTERN GENERATION
Different structures of LFSR will generate different sequence of test pattern. It means that if the BIST time is limited, the structure of LFSR will affect the BIST time and Fault Coverage (FC) of Circuit under Test (CUT). LFSRs are more popular because of their compact and simple design. Cellular Automaton LFSR are more complex to design but provide patterns with higher randomness and perform better in detection of faults such as stuck-open or delay faults, which need two-pattern testing. A software tool is used which automatically generates built-in self-test blocks into VHDL models of digital circuits by giving the suitable values of initial seed and primitive polynomial in TPG block [3]. Figure 4 shows the LFSR for test pattern of G and F of XC4000 FPGA. It will generate 8bits, 4-bit for G and 4-bit for F.
Figure 4. LFSR test pattern signals generator of G and F inputs of XC4000 FPGA
IV.
TESTING OF FPGA CLB
A simplified diagram of the CLB in Xilinx XC4000 Family is shown in figure 5. This CLB has 13 inputs (I= 13), in which one is the clock input (K), nine signals are input to the combinational part (F1 to F4, G1 to G4, and Hl), and the other three are for the sequential part (DIN, S/R, and EC). There are four outputs in a CLB (0 = 4), in which two are combinational outputs and the other two are for sequential circuits. The combinational part consists of three look-up tables (LUTs), and three multiplexers (MUXs), whose outputs are H1, X, and Y, respectively. The sequential part is made up of the remaining components: two D flip-flops (F=2), the S/R control, and the remaining MUXs.
242
Vol. 3, Issue 1, pp. 240-246
Figure 5. Simplified diagram of a CLB in XC4000
The testing of FPGAs falls into two categories: testing of unprogrammed FPGAs (configurationindependent testing) and testing of programmed FPGA (configuration-dependent testing) [7]. Here, we focus on the testing of unprogrammed FPGAs. The LUTs consist of SRAM. To test the memory elements, each bit has to be set to both 1 and 0. Therefore, at least two phases are required to exercise all possible faults in the LUTs. There are 4-to-1 MUXs in a CLB. As result, we need at least four test phases so that each input-output connection of these MUXs can be exercised. We found that four test phases are enough to exercise all possible configurations in the CLB. These configurations are given as follows. Phase I: The LUTs are configured as exclusive-OR (XOR) of the nine inputs. F is connected to both flip flops, G is connected to Y and H is connected to X. Phase 2: The LUTs are configured as exclusive-NOR (XNOR) of the nine inputs. G is connected to both flip flops, F is connected to X, and Hs is connected to Y. Phase 3: The LUTs implement XOR, and DIN is connected to both flip-flops. Phase 4: The LUTs implement XNOR and His connected to both flip-flops.
Figure 6. Comparator based ORA
The combinational part is tested in the first two phases. Flip-flops are tested in all four phases. In order to fully test the upper MUXs, in phase 1 the four connections are (C 1, C2, C3, C4)=>(Hl, DIN, S/R, EC) (i.e., H1 is connected to C1, DIN is connected to C2, etc.). In phase 2, the connection is (C2, C3, C4, Cl) => (Hl, DIN, S/R, EC), and in phase 3 we have (C3, C4, C1, C2) => (Hl, DIN, S/R, EC). Finally, in phase 4 the connection is (C4, C1, C2, C3) => (Hl, DIN, S/R, EC). As a result, all connections are exercised in four phases. The testing pattern is generated using 8-bit LFSR whose polynomial is x8 +x6 +x5 +x +1. The output response analyzer we used here is a comparator based
243
Vol. 3, Issue 1, pp. 240-246
ORA. There are B CLBs in a node, and each CLB has four output lines. Since all C-mode CLBs are configured in the same way, they should give the same output response if all CLBs are fault-free [8]. Thus the same output signals in all CLBs are fed to one ORA for comparison as shown in figure 6(e.g., output YQs in all C-mode CLBs are sent to the same ORA). We need four ORAs as each CLB has four output lines. In each ORA, all B inputs are fed to a XOR gate. If all CLBs under test are fault-free, the output of the XOR gate should always be zero. Unless the number of faulty CLBs is even and those faults are identical, the XOR gate will generate a 1 output at least once, and that will set the FF to be 1 and the detected fault is recorded.
V.
SIMULATION RESULTS
The following results are obtained using XILINX tool. The Xilinx XC4000 model is used for fault modeling. Initially we designed front end model. The function generators are programmed with simple functions .For testing purpose we generated TPG using 8-bit LFSR. The test pattern results S/R control is shown in below figure 7. The LFSR G and F test patterns are shown in below figure 8. The outputs of LFSR are fed to the CLB function generators. LFSR used here is of external type. The LFSR generates 8-bits using this 8-bits we have taken two 4-bits for G and F to test the FPGA CLBs. Initially it was implemented in DSCH and later in Xilinx for better view of results. Faults are incorporated into the CLBs for testing purpose. Faults that are incorporated here are stuck-at fault and open fault. By using this test pattern we have tested only stuck at faults and open faults.
Figure 7. S/L control signals for XC4000 FPGA CLBs
Figure 8. G and F input test pattern signals of XC4000 FPGA generated by 8-bit LFSR
244
Vol. 3, Issue 1, pp. 240-246
VI.
CONCLUSION
In this paper we present a methodology for the diagnosis of faulty CLBs in FPGAs. The reference CLB we have used is XC4000 series. For generation of test pattern 8-bit LFSR is used. Analysis purpose we have injected some of the common faults like stuck at fault and open fault. In this method all the testing is done within the FPGA which prevents the use of external hardware. The main advantage of this method is testing time of the CLBs mainly depends on number of faults rather than the chip size. Hence it yields more advantages while diagnosing the larger chips. For testing of delay faults and stuck-open faults, test patterns which are more random in nature provide better fault coverage, so in many cases they are being looked upon as the alternatives to conventional LFSRs for test pattern generation and output response analysis in BIST. Also the resolution of the fault diagnosis algorithm in interconnect testing is greatly dependent on the structure of the original application configuration. The complexity of the CLB test configuration will increase in some worst case. All these problems need to be studied in the future research work in this area.
REFERENCES
[1]. D. Gil, J. Gracia, J.C. Baraza, P.J. Gil, (2003) Study, Comparison and Application of different VHDLBased Fault Injection Techniques for the Experimental Validation of a Fault-Tolerant System, Microelectronics Journal, vol. 34, no. 1, pp. 41-51. [2]. Guihai Yan,E, Yinhe Han, And Xiaowei Li, (2011) SVFD: A Versatile Online Fault Detection Scheme via Checking Of Stability Violation, IEEE Transactions On Very Large Scale Integration (VLSI) Systems, Vol.19, No.9, pp.1627-1640. [3]. Shikha Kakar, Balwinder Singh and Arun Khosla, (2009) Implementation of BIST Capability using LFSR Techniques in UART, International Journal of Recent Trends in Engineering ,Vol 1, No. 3, pp. 301-304. [4]. Mehdi B. Tahoori,(2011) High Resolution Application Specific Fault Diagnosis of FPGA, IEEE Transactions On Very Large Scale Integration (VLSI) Systems, Vol. 19, No. 10, PP.1775-1786. [5]. I. Pomeranz and S. Reddy,(2004) Improving the stuck-at fault coverage of functional test sequences by using limited-scan operations, IEEE Transactions On Very Large Scale Integration (VLSI) Systems, vol. 12, no. 7, pp.780788. [6]. P. Bernardi,(2006) A pattern ordering algorithm for reducing the size of fault dictionaries, in Proc. IEEE VLSI Test Symp., pp. 4551. [7]. N. Hernandez and V. Champac, (2008) Testing skew and logic faults in SOC interconnects, in Proc. IEEE Comput. Soc. Annu. Symp. VLSI, pp. 151156. [8]. A. Namazi, M. Nourani, and M. Saquib, (2007) A robust interconnect mechanism for nanometer VLSI, presented at the Int. Test Synthesis Workshop (ITSW), San Antonio, TX.
Authors Fazal Noorbasha was born on 29th April 1982. He received his, B.Sc. Degree in Electronics Sciences from BCAS College, Bapatla, Guntur, A.P., Affiliated to the Acharya Nagarjuna University, Guntur, Andhra Pradesh, India, in 2003, M.Sc. Degree in Electronics Sciences from the Dr. HariSingh Gour University, Sagar, Madhya Pradesh, India, in 2006, M.Tech. Degree in VLSI Technology, from the North Maharashtra University, Jalgaon, Maharashtra, INDIA in 2008, and Ph.D. Degree in VLSI from Department Of Physics and Electronics, Dr. HariSingh Gour Central University, Sagar, Madhya Pradesh, India, in 2011. Presently he is working as a Assistant Professor, Department of Electronics and Communication Engineering, KL University, Guntur, Andhra Pradesh, India, where he has been engaged in teaching, research and development of Low-power, High-speed CMOS VLSI SoC, Memory Processors LSIs, Fault Testing in VLSI, Embedded Systems and Nanotechnology. He is a Scientific and Technical Committee & Editorial Review Board Member in Engineering and Applied Sciences of World Academy of Science Engineering and Technology (WASET), Advisory Board Member of International Journal of Advances Engineering & Technology (IJAET), Member of International Association of Engineers (IAENG) and Senior Member of International Association of Computer Science and Information Technology (IACSIT). He has
245
Vol. 3, Issue 1, pp. 240-246
published over 20 Science and Technical papers in various International and National reputed journals and conferences. Harikishore Kakarla was born in Vijayawada, Krishna (Dist.), AP, India. He received B.Tech. in Electronics & Communication Engineering from, JNTU, Hyderabad, AP, India, M.Tech. from SKD University, Anantapur, AP, India. He is pursuing Ph.D in the area of VLSI in KL University, Vijayawada, AP, India. He has published 03 International Journals and 01 National Conference Level. Presently he is working as a Assistant Professor, Department of Electronics and Communication Engineering, KL University, Guntur, Andhra Pradesh, India, where he has been engaged in teaching, research and development of Low-power, High-speed CMOS VLSI SoC, Memory Processors LSIs, Fault Testing in VLSI, Embedded Systems and Nanotechnology.
246
Vol. 3, Issue 1, pp. 240-246
FUZZY BASED ROTOR GROUND FAULT LOCATION METHOD FOR SYNCHRONOUS MACHINES
Mohanraj. M, Rani Thottungal and Manobalan.M
Department of Electrical and Electronics Engineering, Kumaraguru College of Technology, Coimbatore, India
A BSTRACT
This paper presents a fuzzy rule based on-line rotor ground fault location method for synchronous machines with static excitation systems, which, combined with rotor ground fault protection, can detect and locate faults in the rotor. Synchronous machine field winding is fed by rectifiers through an excitation transformer. The main contribution of this technique to locate the position of the ground fault in the rotor winding online, reduce the repair time. The system is based on the analysis of the ac and dc components of the excitation voltage and the voltage measured in a grounding resistance located in the neutral terminal of the excitation transformer. This technique has been validated through computer simulations.
K EYW ORDS: AC generator excitation, fault location, power generation protection, synchronous generator
excitation.
I.
INTRODUCTION
The use of protection devices in power systems is absolutely necessary in order to safeguard them against short circuits, overloads and, in general, abnormal operations, or faults. The protection system in generating units is especially important since it must reliably guarantee the power supply as per IEEE Standard C37.102 [1]. Some of the most common malfunctions in synchronous generators, such as vibrations and unbalanced stator voltages, are caused by ground faults in the rotor. Generally, the initial ground fault does not cause any damage to the machine, because this circuit is usually ungrounded. However, the probability of a second fault increases after the first one since it establishes a reference for voltages induced in the field, thereby increasing the stress to ground at other points on the field winding. If this second fault occurs, the field winding will be partially short circuited, producing unbalanced fluxes in the machine, with the consequent vibrations and unbalance in the stator voltage.
Fig. 1. Device for rotor fault detection.
247
Vol. 3, Issue 1, pp. 247-254
Most rotor ground fault detection devices for synchronous generators are based on detecting abnormal values in certain electrical variables, such as the stator no-load voltage [2], or the air gap flux [3], so they can only detect double faults. To achieve the early detection of the initial failure of the rotor to ground before severe damage occurs in the generator, authors presented in [4] a detection technique for synchronous generators with static excitation systems, where the excitation field windings are fed by an excitation transformer through a rectifier. Fig. 1 shows a scheme of the proposed technique This technique exhibited two important advantages when compared to current fault detection systems. 1) It could discriminate between a ground fault on the ac or the dc side of the excitation circuit. 2) It did not need any additional injection source to detect the fault, because the proposed detection technique is supplied by the power network itself. This technique required a Wye secondary in the excitation transformer for the rectifier power supply. In this situation, the proposed technique requires a high-value grounding resistance. Otherwise, if the system is fed by a transformer with delta secondary winding, an artificial neutral will be required (i.e., an earthing transformer or zig-zag impedance). In any case, once detected, locating the fault can be a long and laborious process, especially with multipole synchronous generators. When the fault is on the rotor winding, the connections between poles must be opened in order to locate the ground fault. Except for some cases [6], in order to determine whether the fault is on the rotor winding or in the external buses and dc source, the machine has to be removed from service. This paper presents a fuzzy rule based rotor ground fault location method for synchronous machines applicable to the fault detection system presented in [4].The fuzzy rule is based on the value minimum and maximum variation of Vdc,Vac,Rf,Vfdc,Vfac due to the effect of fault at different position of winding. The variation of values is studied from the paper [5]. This new system eliminates the current need to disconnect all poles to locate the fault. Moreover, having available information about the position of the ground fault in the excitation winding could save time in the generator repair process. This fact is of particular interest in the case of hydro generators because of the feasibility of removing one pole without requiring a complete rotor extraction. We aim to make a contribution to the field of fault location in synchronous machines, where there are already developments in ground fault location for the stator winding [7], in fault location between the core laminations [8] and even in insulation fault location is using partial discharges [9]. The subject of rotor fault location in electrical machines has been extensively studied in the case of induction machines, where there are very significant contributions in the detection of broken bars, running the machine in transient state [10],[11] and even in magnetic saturation conditions [12], and its combination with the effects of eccentricity[13]. However, with a few exceptions [6], there are not many contributions to rotor fault location methods for synchronous machines. This is the field that will be covered in this paper. This section II presents the principles of the proposed method. Then, Section III analysis the results of the simulations obtained by applying the proposed method. Finally, section IV concludes with the main contributions of the proposed technique.
II.
PRINCIPLES OF THE ROTOR GROUND FAULT LOCATION METHOD
This procedure for rotor ground fault location was developed with the analysis of numerous laboratory tests, which followed the development of a previous rotor ground fault protection technique presented by the authors in [4]. The rotor ground fault protection technique was based on the analysis of the ac component of the voltage at the grounding resistance. But also a dc component appears in the grounding resistance in case of earth fault in the field winding. The procedure proposed in this paper will use both ac and dc components to locate the rotor ground fault. The neutral of the excitation transformer is grounded, so the dc voltage supplied by the rectifier has a ground reference. In case of normal operation, the voltage between the midpoint of the field winding and ground is zero. Moreover, the voltage between the positive brush and ground is half the voltage supplied by the rectifier (Vfdc/2), in a similar way the voltage at the negative brush has the same amplitude but negative polarity (Vfdc/2). The distribution of the voltage to ground is linear along the
248
Vol. 3, Issue 1, pp. 247-254
field winding. For a general position in the field winding X (%), the voltage between this point and ground can be obtained as presented in (1) as VT . This voltage relationship is illustrated in Fig. 2
VT = V fdc(X 50/100) (1)
Fig. 2. Excitation dc equivalent circuit.
Fig. 3. Equivalent circuit for rotor ground fault dc current calculation for any rotor position X (%).
The dc current that flows from the field winding to earth, in case of earth fault, is calculated using Thevenins theorem. Then, the equivalent circuit between any point of the field winding and earth consists of a voltage source and a resistance. The voltage source corresponds to the voltage in open circuit, according to (1). On the other hand, the resistance of this equivalent circuit be approximated by the grounding can resistance Rg, as this resistance is much higher (around 5 k , to limit the ground fault current) than the resistances of the field winding, the transformer or the rectifier. Finally, the dc equivalent circuit for any rotor position X and the fault resistance is presented in Fig. 3. Hence, the dc current depends on the position of the rotor earth fault and the fault resistance. The principle behind this method is the relationship between the dc voltage component in the grounding resistance V dc and the position of the fault along the field winding. It is important to realize that the voltage measurements available with the machine online are the ac and dc components of the field winding voltage and the voltage in the grounding resistance in the event of a fault. A. Relationship between V dc and Fault Position The dc component of the voltage in the grounding resistance is related to the rotor fault position. The amplitude of V dc is maximum, with positive polarity, when the fault occurs in the negative terminal, which will be considered as the start of the winding (0%). If, on the other hand, the fault occurs in the positive terminal, which is at the end of the winding (100%), V dc has the same amplitude, but the polarity is negative. In contrast, V dc is negligible for faults at the midpoint of the winding (50%).
Fig. 4. V dc-Fault position relationship.
Fig. 5. V dc-Fault position relationship for different Rf.
The value of V dc for different fault positions and a particular fault resistance Rf is shown in Fig. 4. The maximum value is referred to as V dc0. The maximum value of the dc component V dc0 depends on the value of the fault resistance Rf and the value of the dc component of the excitation field voltage Vfdc. B. Estimate of the Fault Resistance Rf. In order to estimate the fault resistance Rf, the ac components should be used. In a similar way, the ac current that flows from the field winding to earth depends on the ac component of the excitation
249
Vol. 3, Issue 1, pp. 247-254
voltage Vfac and the fault resistance Rf. The equivalent ac circuit is composed by a voltage source and impedance. The voltage source corresponds to the ac component of the excitation voltage Vfac, and the impedance corresponds to the ground resistance Rf. In this case, the ground fault position does not affect the value of the current.
Fig. 6. Equivalent dc circuit for V dc0 calculation.
Fig. 7. Equivalent ac circuit for Rf calculation.
After analyzing numerous tests, it was determined that the fault resistance Rg could be reasonably estimated using this ac equivalent circuit, as shown in Fig. 7. The dc and ac equivalent circuits, as shown in Figs. 6 and 7, are checked using all test measurements, where the fault resistance Rf and the rotor fault position X are known in advance. Therefore, the fault resistance Rf can be estimated according to the following expression: Rf = Rg ((V fac / V ac)-1) C. Fuzzy Rule. The fuzzy rule is based on the value minimum and maximum variation of Vdc,Vac,Rf,Vfdc,Vfac due to the effect of fault at different position of winding. If (fdc is PS) and (dc is PS) and (rf is NVB) the (output1 is 60) (1) If (fdc is PS) and (dc is PS) and (rf is NB) then (output1 is 80) (1) If (fdc is PS) and (dc is PS) and (rf is NS) then (output1 is 100) (1) If (fdc is PS) and (dc is PS) and (rf is mf4) then (output1 is 0) (1) If (fdc is PS) and (dc is PS) and (rf is PS) then (output1 is 20) (1) If (fdc is PS) and (dc is PS) and (rf is PB) then (output1 is 40) (1) If (fdc is PS) and (dc is PS) and (rf is PVB) then (output1 is 50) (1) If (fdc is NB) and (dc is NB) and (rf is mf4) then (output1 is 0) (1) If (fdc is NS) and (dc is NS) and (rf is mf4) then (output1 is 0) (1) If (fdc is NS) and (dc is NS) and (rf is PS) then (output1 is 20) (1) If (fdc is Z) and (dc is Z) and (rf is PVB) then (output1 is 50) (1) If (fdc is PS) and (dc is PS) and (rf is PB) then (output1 is 60) (1) If (fdc is PB) and (dc is PB) and (rf is PB) then (output1 is 80) (1) If (fdc is PB) and (dc is PB) and (rf is NS) then (output1 is 100) (1) If (fdc is z) and (dc is NB) and (rf is mf4) then (output1 is 0) (1) If (fdc is z) and (dc is NS) and (rf is PS) then (output1 is 20) (1) If (fdc is z) and (dc is Z) and (rf is PB) then (output1 is4 0) (1) If (fdc is z) and (dc is PS) and (rf is NVB) then (output1 is6 0) (1) If (fdc is z) and (dc is PB) and (rf is NB) then (output1 is 80) (1) If (fdc is z) and (dc is NB) and (rf is PVB) then (output1 is 50) (1) If (fdc is z) and (dc is NS) and (rf is NS) then (output1 is 100) (1) If (fdc is PS) and (dc is NB) and (rf is mf4) then (output1 is 0) (1) If (fdc is PS) and (dc is NS) and (rf is PS) then (output1 is 20) (1) If (fdc is PS) and (dc is Z) and (rf is PB) then (output1 is 4 0) (1) If (fdc is PS) and (dc is PS) and (rf is NVB) then (output1 is6 0) (1) If (fdc is PS) and (dc is PB) and (rf is NB) then (output1 is 80) (1)
250
Vol. 3, Issue 1, pp. 247-254
If (fdc is PS) and (dc is NB) and (rf is PVB) then (output1 is 50) (1) If (fdc is PS) and (dc is z) and (rf is NS) then (output1 is 100) (1) If (fdc is PB) and (dc is NB) and (rf is mf4) then (output1 is 0) (1) If (fdc is PB) and (dc is NS) and (rf is PS) then (output1 is 20) (1) If (fdc is PB) and (dc is Z) and (rf is PB) then (output1 is 4 0) (1) If (fdc is PB) and (dc is PB) and (rf is NVB) then (output1 is6 0) (1) If (fdc is PB) and (dc is PS) and (rf is NB) then (output1 is 80) (1) If (fdc is PB) and (dc is PS) and (rf is PVB) then (output1 is 50) (1) If (fdc is PB) and (dc is PS) and (rf is NS) then (output1 is 100) (1) D. Rotor ground fault location block diagram. To summarize the complete rotor fault location procedure, we include the rotor location block diagram in Fig. 8. The algorithm steps for estimating the rotor fault position are as follows. 1) The measurement of the excitation voltage Vf and grounding resistance voltage V yields their ac and dc components Vfac, Vfdc, V ac, and V dc, respectively. 2) The ground fault resistance Rf is estimated from the ac components which is already calculated. 3) Based on the values of Rf, Vfdc, Vdc specific fault position is calculated using Fuzzy logic controller.
Fig. 8. Rotor fault location device layout.
The proposed rotor ground fault location method was simulated using the MATLAB/Simulink Power Systems software. The simplified schematics of the MATLAB simulation are shown in Fig. 9, representing a 5-kVA synchronous machine and its excitation system, similar to the machine used in the experimental tests. Detailed data on the simulated devices can be found in Tables I and II. In the simulation, the excitation transformer is modelled as a three-phase voltage source; the rotor windings of the synchronous machine are modelled with ten sets of resistances, inductances, and capacitors; finally, the rectifier is modelled with a three-phase thyristor-controlled rectifier bridge. In the example, the excitation transformer neutral grounding resistance is a 5k resistance.
251
Vol. 3, Issue 1, pp. 247-254
Fig. 9. Simplified schematics of the excitation system for rotor ground fault location simulation TABLE I
CHARACTERISTICS OF SYNCHRONOUS GENERATOR USED IN THE EXPERIMENT
TABLE II
SYNCHRONOUS GENERATOR STATIC EXCITATION SYSTEM TECHNICAL DATA
III.
SIMULATION RESULT
Fig. 10. Voltage waveform measured at the grounding resistance for a fault at 0% winding position with Rf = 0.
Fig. 11. Voltage waveform measured at the grounding resistance for a fault at 50% winding position with Rf =0.
252
Vol. 3, Issue 1, pp. 247-254
Fig. 12. Voltage waveform measured at the grounding resistance for a fault at 100% winding position with Rf = 0.
Fig. 13. Voltage waveform measured at the grounding resistance for a fault at 0% winding position with Rf = 5k
Fig. 14. Voltage waveform measured at the grounding resistance for a fault at 50% winding position with Rf = 5k .
Fig. 15. Voltage waveform measured at the grounding resistance for a fault at 100% winding position with Rf = 5k .
Figs. 1012 show the voltage in the grounding resistance for a fault in the negative brush (0%), midpoint (50%), and positive brush (100%), respectively. A zero fault resistance and a field voltage of 20 V dc (corresponding to a firing angle of 48) are used for the simulations. The simulations at the same conditions, but using a fault resistance of 5 k , are included in Figs. 13 15. As expected, the values are lower than in the previous case without fault resistance.
IV.
CONCLUSION
The method is able to locate the fault with the machine online. Moreover, it offers a significant improvement over the detection technique proposed by the authors in [4]. The fuzzy controller locates the position of fault in the rotor of synchronous machines faster due to the pre programmed laboratory value. In practical application it reduces the time needed for calculation. It directly determines the fault position in rotor winding. The main advantage of this technique is that it contributes valuable information on the fault position before tripping the generator. It is possible to incorporate it easily into actual rotor ground protection systems. Having information on the position of a ground fault in the excitation winding saves time in the generator repair process.
REFERENCES
[1] IEEE Guide for AC Generator Protection, IEEE Standard C37.102, 2006. [2] M. Kiani, W-J. Lee, R. Kenarangui, and B. Fahimi, Detection of rotor faults in synchronous generators, in Proc. IEEE Int. Symp. Diagnosis Elect. Mach., Power Electron. Drives, 2007, pp. 266271. .[3] R. L. Stoll andA.Hennache, Method of detecting and modelling presence of shorted turns in DC field winding of cylindrical rotor synchronous machines using two airgap search coils, IEEE Proc. Electric Power Appl., vol. 135, no. 6, pp. 281294, Nov. 1988.
253
Vol. 3, Issue 1, pp. 247-254
[4] C. A. P. Gaona, F. Blazquez, P. Fras, and M. Redondo, A novel rotor ground fault detection technique for synchronous machines with static excitation, IEEE Trans. Energy Convers., vol. 18, no. 4, pp. 965973, Dec. 2010. [5]Carlos A. Platero, Member, IEEE, Francisco Blazquez, Member, IEEE, Pablo Fras, and Miguel Pardo, New On-Line Rotor Ground Fault Location Method for Synchronous Machines with Static Excitation IEEE Trans. Energy Convers., vol. 26, no. 2, pp. 572580, Jun. 2011. [6] I. Kerszenbaum and J. Lopetrone, Novel Hall-effect turbo-generators rotor DC ground-fault localizer, in Proc. IEEE Int. Electric Machines Drives Conf. Rec., 1997, pp. TC1/5.15.3. [7] M. Zielichowski and M. Fulczyk, Analysis of operating conditions of ground-fault protection schemes for generator stator winding, IEEE Trans. Energy Convers., vol. 18, no. 1, pp. 5762, Mar. 2008. [8] G. B. Kliman, S. B. Lee, M. R. Shah, R. M. Lusted, and N. K. Nair, A new method for synchronous generator core quality evaluation, IEEE Trans. Energy Convers., vol. 19, no. 3, pp. 576582, Sep. 2004. [9] J. Borghetto, A. Cavallini, A. Contin, G. C. Montanari, M. de Nigris, G. Pasini, and R. Passaglia, Partial discharge inference by an advanced system. Analysis of online measurements performed on hydrogenerator, IEEE Trans. Energy Convers., vol. 19, no. 2, pp. 333339, Jun. 2004. [10] M. Riera-Guasp, M. F. Cabanas, J. A. Antonino-Daviu, M. Pineda- Sanchez, and C. H. R. Garca, Influence of nonconsecutive bar breakages in motor current signature analysis for the diagnosis of rotor faults in induction motors, IEEE Trans. Energy Convers., vol. 25, no. 1, pp. 8089, Mar. 2010. [11] H. Douglas, P. Pillay, and A. K. Ziarani, Broken rotor bar detection in induction machines with transient operating speeds, IEEE Trans. Energy Convers., vol. 20, no. 1, pp. 135141, Mar. 2005. [12] J. Sprooten and J. C. Maun, Influence of saturation level on the effect of broken bars in induction motors using fundamental electromagnetic laws and finite element simulations, IEEE Trans. Energy Convers., vol. 24, no. 3, pp. 557564, Sep. 2009. [13] Z. Liu, X. Yin, Z. Zhang, D. Chen, and W. Chen, Online rotor mixed fault diagnosis way based on spectrum analysis of instantaneous power in squirrel cage induction motors, IEEE Trans. Energy Convers., vol. 19, no. 3, pp. 485490, Sep. 2004. Biographies M. Mohanraj, from Erode, finished his UG in Bharathiar University and obtained PG in Power System Engineering in Annamalai University and he is currently working as Assistant Professor in EEE department, Kumaraguru College of Technology, Coimbatore and a Life member of ISTE. His research area includes Wind Energy Conversion, Solar Energy, Machines and Power Quality issues.
Rani Thottungal, obtained her B.E and M.E degrees from Andhra University and Doctorate from Bharathiar University. She is currently working as Professor and Head in EEE department, Kumaraguru College of Technology, Coimbatore. Her research interest includes Power System, Power Inverter and Power Quality Issues.
M. Manobalan, from Neyveli, finished his UG in St.Josephs college of Engineering and pursuing his final year M.E. in Power Electronics and Drives in Kumaraguru College of Technology, Coimbatore.
254
Vol. 3, Issue 1, pp. 247-254
TURNING PARAMETER OPTIMIZATION FOR SURFACE ROUGHNESS OF ASTM A242 TYPE-1 ALLOYS STEEL BY TAGUCHI METHOD
Jitendra Verma1, Pankaj Agrawal2, Lokesh Bajpai3
1
Department of Mechanical Engineering, Samrat Ashok Technological Institute, Vidisha (M.P) 464001, India. 2&3 Professor in Department of Mechanical Engineering, Samrat Ashok Technological Institute, Vidisha (M.P.) 464001, India.
ABSTRACT
The purpose of this research paper is focused on the analysis of optimum cutting conditions to get lowest surface roughness in turning ASTM A242 Type-1 ALLOYS STEEL by Taguchi method. Experiment was designed using Taguchi method and 9 experiments were conducted by this process. The results are analyzed using analysis of variance (ANOVA) method. Taguchi method has shown that the cutting speed has significant role to play in producing lower surface roughness about 57.47% followed by feed rate about 16.27%. The Depth of Cut has lesser role on surface roughness from the tests. The results obtained by this method will be useful to other researches for similar type of study and may be eye opening for further research on tool vibrations, cutting forces etc.
KEYWORDS: ASTM A242 Type-1 alloy steel; Machining; Dry turning; Signal-to- noise ratio; Taguchi method
I.
INTRODUCTION
Increasing the productivity and the quality of the machined parts are the main challenges of metalbased industry; there has been increased interest in monitoring all aspects of the machining process. Surface finish is an important parameter in manufacturing engineering. It is a characteristic that could influence the performance of mechanical parts and the production costs. The ratio between costs and quality of products in each production stage has to be monitored and immediate corrective actions have to be taken in case of Deviation from desired trend. Surface roughness measurement presents an important task in many engineering applications. Many life attributes can be also determined by how well the surface finish is maintained. Machining operations have been the core of the manufacturing industry since the industrial revolution [1] and the existing optimization researches for Computer Numerical Controlled (CNC) turning were either simulated within particular manufacturing circumstances or achieved through numerous frequent equipment operations. These conditions or manufacturing circumstances are regarded as computing simulations and their applicability to real world industry is still uncertain and therefore, a general optimization scheme without equipment operations is deemed to be necessarily developed. Surface roughness is commonly considered as a major manufacturing goal for turning operations in many of the existing researches. The machining process on a CNC lathe is programmed [13]. Many surface roughness prediction systems were designed using a variety of sensors including dynamometers for force and torque. Taguchi and Analysis Of Variance (ANOVA) can conveniently optimize the cutting parameters with several experimental runs well designed [15]. Taguchi parameter design can optimize the performance characteristics through the settings of design parameters. This study describe how to select the control factors levels (cutting speed, feed rate, Depth of cut) that can minimize the effect of noise factors on the response (surface roughness). An experimental work will be conducted to analyse the influence of
255
Vol. 3, Issue 1, pp. 255-261
cutting parameters (control factors) on surface roughness (signal factor), then select the optimal cutting parameters which lead to optimal response by assistance of optimal signal factor [18]. Sundaram and Lambert [20, 21] considered six variables i.e. speed, feed, and depth of cut, time of cut, nose radius and type of tool to monitor surface roughness. To improve the efficiency of these turning processes, it is necessary to have a complete process understanding and model. To this end, a great deal of research has been performed in order to quantify the effect of various hard turning process parameters to surface quality. These factors can be divided into a) setup variables, b) tool variables, and c) work piece variables. In order to gain a greater understanding of the turning process it is necessary to understand the impact of the each of the variables, but also the interactions between them. It is impossible to find all the variables that impact surface roughness in turning operations. In addition, it is costly and time consuming to discern the effect of the every variable on the out put. In order to simplify the problem, one needs to eliminate or select specific variables that correspond to practical applications.
II.
TAGUCHI METHOD
Taguchi method is a powerful tool for the design of high quality systems. It provides simple, efficient and systematic approach to optimize designs for performance, quality and cost [22]. Taguchi method is efficient method for designing process that operates consistently and optimally over a variety of conditions. To determine the best design it requires the use of a strategically designed experiment [23]. Taguchi approach to design of experiments in easy to adopt and apply for users with limited knowledge of statistics, hence gained wide popularity in the engineering and scientific community [17-18]. The desired cutting parameters are determined based on experience or by hand book. Cutting parameters are reflected. Steps of Taguchi method are as follows: (1) Identification of main function, side effects and failure mode. (2) Identification of noise factor, testing condition and quality characteristics. (3) Identification of the main function to be optimized. (4) Identification the control factor and their levels. (5) Selection of orthogonal array and matrix experiment. (6) Conducting the matrix experiment. (7) Analysing the data, prediction of the optimum level and performance. (8) Performing the verification experiment and planning the future action. [4]
III.
EXPERIMENTAL SET UP AND CUTTING CONDITIONS
MATERIALS AND METHODS Experimental procedures and conditions: In this study ASTM A242 TYPE-1 ALLOY steel and 250 mm long with 50 mm diameter was used as work material for experimentation using a lathe turning machine. The chemical composition of the selected work piece is shown as
Table 1 Composition OF ASTM A242 Type-1 ALLOYS STEEL C 0.15% Mn 1.0% Si 0.4% P 0.15% Cr 0.5% Cu 0.2% S 0.05% P 0.045% Nb+V+Ti 0.15
Universal turning machine tool was used in the experiments.. Cutting speed, feed rate and depth of cut were selected as the machining parameters to analyse their effect on surface roughness. A total of 9 experiments based on Taguchis orthogonal array were carried out with different combinations of the levels of the input parameters. Among them, the settings of cutting speed include 100, 125 and 150 rpm; those of feed rate include 0.05, 0.1, 0.15 mm rev-1; the depth of cut is set at 0.5, 1.0 and 1.5 mm. The cutting parameters are shown in the Table 2. Three levels of cutting speed, three levels of feed rate and three levels of depth of cut were used and are shown in the Table 2. The different alloying elements present in a work piece are shown in the table 1.
256
Vol. 3, Issue 1, pp. 255-261
Table 2 Cutting parameters Symbol A B C Parameters/Level Cutting speed Feed rate Depth of cut Level 1 -18.30 -16.46 -16.78 Level 2 -20.28 -18.85 -18.5 Level 3 -15.78 -19.05 -19.09 Units m/min mm/rev. mm
IV.
4.1 Analysis of S/N ratio based on Taguchi Method To select an appropriate orthogonal array for experiments, the total degrees of freedom need to be computed. The degrees of freedom are defined as the number of comparisons between process parameters that need to be made to determine which level is better and specifically how much better it is. For example, a Three-level process parameter counts for four degrees of freedom. The degrees of freedom associated with interaction between two process parameters are given by the product of the degrees of freedom for the two process parameters [8]. The mean S/N ratio for each level of the other cutting parameters can be computed in the similar manner. The mean S/N ratio for each level of the cutting parameters is summarized and called the S/N response table for surface roughness. In addition, the total mean S/N ratio for the nine experiments for surface roughness, listed in Table.3 the greater S/N ratio corresponds to the smaller variance of the output characteristic around the desired value.
Table 3 Experimental Results Test. No 1 2 3 4 5 6 7 8 9 Cutting Speed Feed rate 100 100 100 125 125 125 150 150 150 0.05 0.1 0.15 0.05 0.1 0.15 0.05 0.1 0.15 DOC 0.5 1.0 1.5 1.5 0.5 1.0 1.0 1.5 0.5 Mean Surface Roughness Ra (m) 5.62 10.04 09.88 10.25 9.30 11.60 5.04 6.93 6.27 Signal to Noise Ratio (S/N) -14.99 -20.03 -19.89 -20.21 -19.36 -21.29 -14.18 -17.18 -15.99
According to Taguchi method, S/N ratio is the ratio of Signal representing desirable value, i.e. mean of output characteristics and the noise representing the undesirable value i.e., squared deviation of the output characteristics. It is denoted by and the unit is dB. The S/N ratio is used to measure quality characteristic and it is also used to measure significant welding parameters [9].
Table No. 4 Response table
Parameters/Level Cutting speed Feed rate Depth of cut Level 1 -18.30 -16.46 -16.78 Level 2 -20.28 -18.85 -18.5 Level 3 -15.78 -19.05 -19.09 Max-Min 4.5 2.59 2.31 Rank 1 2 3
257
Vol. 3, Issue 1, pp. 255-261
Level 1 -13 -14 -15 -16 -17 S/N -18 RESPO NSE -19 -20 -21 Level 2 Level 3
-15.78 -18.3 -20.28
Cutting speed
Average Cutting speed
Fig. 1 CUTTING SPEED VS S/N RESPONSE

Level 1 Level 2 Level 3
S/N RESPONSE
-13 -14 -15 -16 -17 -18 -19 -20 -21
-16.46
-18.85
-19.05
Feed Rate
Average Feed Rate
Fig. 2 FEED RATE VS S/N RESPONSE

Level 1 Level 2 Level 3
S/N RESPONSE
-13 -14 -15 -16 -17 -18 -19 -20 -21
-16.78 -18.5
-19.09
Depth of cut
Average Depth of cut
Fig. 3 DEPTH OF CUT VS S/N RATIO
4.2 Analysis of variance (ANOVA) The main aim of ANOVA is to investigate the design parameters and to indicate which parameters are significantly affecting the output parameters. In the analysis, the sum of squares and variance are calculated. F-test value at 95 % confidence level is used to decide the significant factors affecting the process and percentage contribution is calculated [19]. The ANOVA analysis for percentage calibration is shown in Table-5
258
Vol. 3, Issue 1, pp. 255-261
Table 5 Result of ANOVA for Surface Roughness contribution
Symbol A B C Error Total Parameter Cutting Speed Feed Rate Depth of Cut DOF 2 2 2 2 8 Sum of Square 30.5208 12.4602 8.6427 1.4823 53.106 Mean Square 15.2604 6.2301 4.3213 0.74115 26.553 Percentage Contribution 57.47% 23.46 % 16.27 % 2.79 %
Fig. 4 Percentage contribution by Pie Chart for surface roughness
V.
CONCLUSION
The following are conclusions drawn based on the tests conducted on turning ASTM A242 Type-1 ALLOYS STEEL and 250 mm long with 50 mm diameter. 1. From the ANOVA, Table 5 and the P value, the cutting speed is the only significant factor which contributes to the surface roughness i.e. 57.47 % contributed by the cutting speed on surface roughness. 2. The second factor which contributes to surface roughness is the feed rate having 23.46 %. 3. The third factor which contributes to surface roughness is the depth of cut having 16.27%. 4. The Validation experiment confirms that the error occurred was less than 2.79% between equation and actual value. 5. It is recommended from the above results that cutting of 18.30 to 15.78 m/min can be used to get lowest surface roughness. 6. Taguchi gives systematic simple approach and efficient method for the optimum operating conditions. So now it is found by this research how to use Taguchis parameter design to obtain optimum condition with lowest cost, minimum number of experiments and industrial engineers can use this method.
REFERENCES
[1]
[2] [3] [4]
Elias N. Malamas, Euripides G.M. Petrakis, Michalis Zervakis A SURVEY ON INDUSTRIAL VISION SYSTEMS, APPLICATIONS AND TOOLS Department of Electronic and Computer Engineering Technical University of Crete Chania Crete Greece J. Z. Zhang, J. C. Chen, E. D. Kirby, Surface roughness optimization in an end-milling operation using the Taguchi design method, Journal of Materials Processing Technology 184, pp. 233-239, 2007. P. J. Ross, Taguchi technique for quality engineering, New York: McGraw-Hill, 1988. Onkar N. panday total quality management
259
Vol. 3, Issue 1, pp. 255-261
[5]
[6] [7]
[8] [9] [10]
[11] [12]
[13]
[14] [15] [16]
[17] [18] [19]
[20] [21] [22] [23]
N. Nalbant, H. Gokkaya, G. Sur, Application of Taguchi method in the optimization of cutting parameters for surface roughness in turning, Materials and Design, date received 21.07.2006 and date accepted 06.01.2006. I.A. Choudhury, M.A. El-Baradie, Surface roughness prediction in the turning of high strength steel by factorial design of experiments, Journal of Materials Processing technology, 67 (1997) 55-67. B. Erginc, Z. Kampu, B. Sustarsic, The use of the Taguchi approach ro determine the influence of injection molding parameters on the properties of green parts, Journal of achievement in Materials and Manufacturing Engineering, 15, 2006. P.J. Ross. 2005. Taguchi Techniques for Quality Engineering. 2nd Ed. Tata McGraw Hill. Ugur Esme. 2009. Application of Taguchi method for the optimization of resistance spot welding process. The Arabian Journal for Science and Engineering. 34(28): 519-528. A. G. Thakur, T. E. Rao, M. S. Mukhedkar and V. M. Nandedkar APPLICATION OF TAGUCHI METHOD FOR RESISTANCE SPOT WELDING OF GALVANIZED STEE ARPN Journal of engineering and Applied Sciences 2006-2010 Asian Research Publishing Network (ARPN). All rights reserved S. Thamizhmanii, S. Saparudin and S. Hasan, Analyses of roughness, forces and wear in turning gray cast iron, Journal of achievement in Materials and Manufacturing Engineering, volume20, issues 1-2, 2007. Palanikumar, L. Karunamoorthy, R. Krathikeyan, Assessment of factors influencing surface roughness on the machining of glass reinforced polymer composites, Journal of Materials and Design, 27 (2006) 862871. Xue Ping, C. Richard Liu, Zhenqiang Yao, Experimental study and evaluation methodology on hard surface integrity, International Journal Advanced Manufacturing Technology, ODI 10.1007/s00170-0060576-6. T. Tamizharasan, T. Selvaraj, A. Noorul Hag, Analysis of tool wear and surface finish in hard turning, International Journal of Advanced Manufacturing Technology (2005), DOI 10/1007 /s 00170-004-2411-1. W.H. Yang, Y.S. Tang, Design optimization of cutting parameters for turning operations based on Taguchi method, Journal of Materials Processing Technology, 84 (1998) 122-129. Ersan Aslan, Necip Camuscu Burak Bingoren, Design of optimization of cutting parameters when turning hardened AISI 4140 steel (63 HRC) with Al 2 O3 + TiCN mixed ceramic tool, Materials and Design, date received 21.07.2005 and date accepted 06.01.2006. D.C. Montgometry, Design and analysis of experiments, 4th edition, New York: Wiley; 1997. N. Nalbant, H. Gokkaya, G. Sur, Application of Taguchi method in the optimization of cutting parameters for surface roughness in turning, Materials and Design, date received S. Thamizhmanii*, S. Saparudin, S. Hasan Analyses of surface roughness by turning process using Taguchi method; Journals of Achievements in Materialsand Manufacturing Engineering VOLUME 20ISSUES 1-2January-February2007 R.M. Sundaram, B.K. Lambert, Mathematical models to predict surface finish in fine turning of steel, Part I, International Journal of Production Research 19 (1981) 547556. R.M. Sundaram, B.K. Lambert, Mathematical models to predict surface finish in fine turning of steel, Part II, International Journal of Production Research 19 (1981) 557564. W.H. Yang, Y.S. Tang, Design optimization of cutting parameters for turning operations based on Taguchi method, Journal of Materials Processing Technology, 84 (1998) 122-129. Ersan Aslan, Necip Camuscu Burak Bingoren, Design of optimization of cutting parameters when turning hardened AISI 4140 steel (63 HRC) with Al 2 O3 + TiCN mixed ceramic tool, Materials and Design, date received 21.07.2005 and date accepted 06.01.2006.
ABOUT THE AUTHOR

Jitendra Verma was born in 15th Sep 1986. He received his B.Tech in Manufacturing Technology from JSS Academy of Technical Education, Noida (U.P.) in 2007. Currently, He is pursuing his M.Tech (C.I.M.) from Samrat Ashok Technological Institute, Vidisha (M.P.). His research interests are surface roughness, welding, Turning and machining.
Pankaj Agrawal was born in 18th August 1967. He is currently working as a Professor in Mechanical Engineering Department of Samrat Ashok Technological Institute, Vidisha (M.P.). He has more then 18 years experience in teaching, one year industry and 10 years of research experience. He has done his B.E. in Mechanical Engineering from Samrat Ashok Technological Institute, Vidisha (M.P.) in 1990. He has done M.Tech and then Ph.D. in 2003 from BARKATULLAH UNIVERSITY BHOPAL in 2003. He has published many papers in
260
Vol. 3, Issue 1, pp. 255-261
various journals and conferences of international repute. His main interests are hybrid manufacturing, stereo lithography, Supply Chain Management and Flexible Manufacturing Systems etc. Lokesh Bajpai was born in 19th December 1960.He is currently working as Professor in Mechanical Engineering Department of Samrat Ashok Technological Institute, Vidisha (M.P.). He has done B.E. from GEC Jablpur (M.P.) in 1984.He has done M.E., Ph.D. F.I.E. (India), MISME, M.I.S.C.A. He has more then 24 years experience in teaching and 14 years of research experience. He has published many papers in various journals and conferences of international repute. His main interests are computer integrated manufacturing, Flexible Manufacturing, computer aided process planning etc.
261
Vol. 3, Issue 1, pp. 255-261
PREDICTION OF HEAT-RELEASE PATTERNS FOR MODELING DIESEL ENGINE PERFORMANCE AND EMISSIONS
B.VenkateswaraRao1 and G. Amba Prasad Rao2
Graduate Student, 2Associate Professor Department of Mechanical Engineering, National Institute of Technology, Warangal, India
1
A BSTRACT
In the present paper phenomenological modelling has been adopted for modelling diesel engine processes considering heat losses, and variable specific heats using double-Weibe function for the heat release. High speed diesel fuel C10.8H18.7 is considered for calculations. Fuel injection timing engine speed, compression ratio, inlet charge pressure and temperature are observed to be pertinent parameters affecting diesel engine performance. Effect of exhaust gas recirculation on the formation and emission of oxides of nitrogen and soot density are also studied. Numerical experiments are performed by writing a computer code in C++ and heat release (both pre-mixed and diffusion phases), in-cylinder pressure and temperature histories and emissions are predicted. It was found that early injection timing leads to higher levels of pressure and temperature in the cylinder. It is observed that fuel injection timing and fraction of exhaust gas recirculation as critical factors in affecting engine performance as well as emissions.
K EYW ORDS: Diesel engine, double Wiebe function, EGR, Performance, Emissions
I.
INTRODUCTION
Direct injection diesel engines exhibit better performance as far as fuel economy is concerned compared to gasoline engines. Of late, due to stringent emission norms researchers and leading manufacturers are aiming for the development of clean diesel engines. In this regard, computer simulations are found to be prominent tools for arriving at the optimum designs and to make the diesel technology be more competitive. Though there are various models such thermodynamics and fluid dynamics based models available, phenomenological models are attractive in the light of less computational complexities involved. These models can be made more attractive by imposing all possible practical conditions the diesel engine experiences and to predict performance near to actual cycle simulations. Typical direct injection diesel engine combustion process comprises of four phases viz; ignition delay, pre-mixed, diffusion and late burning. Abu-Nada et al. [1-3] carried out engine simulations taking into account the effect of heat transfer, friction, and temperature dependent specific heats on the overall engine performance. Miyamoto et al. [4] the model was originally developed for spark ignition (SI) engines; they claimed that it could be extended and modified to simulate compression ignition (CI) engines as well. This results in a significant shift in the rate of heat release model from the simple Weibe function commonly used for SI engines. A double peak heat release model becomes more representative CI engines [4].Arregle et al. [6] studied the influence of injection parameters and running conditions on heat release in a Diesel engine. Galindo et al. [7] used four different Weibe functions to account for pilot injection, premixed, diffusion and late combustion in the heat release model. Chemla et al. [8] used a zero-dimensional rate of heat release model for the simulation of direct injection diesel engine. Aithal [9] studied effect of EGR fraction on diesel engine performance considering heat loss and temperature dependent properties of the working fluid. The objective of the present work is to analyze the performance of a CI engine using a phenomenological-thermodynamics based model considering double-Weibe function for diesel
262
Vol. 3, Issue 1, pp. 262-269
combustion process. The developed model predicts in-cylinder temperatures and pressures as functions of the crank angle, variation of premixed, diffusion, heat release pattern by varying the engine operating and design parameters. The novelty of present model lies in considering effects of heat losses (both convection and radiation) and temperature-dependent specific heats in addition to ignition delay and EGR fraction. These features can be very useful in providing more realistic estimations for the performance and emission parameters in comparison to the existing models for CI engines.
II.
PHENOMENOLOGICAL MODELLING
(1)
The governing equation for calculating variation of in-cylinder pressure with respect to crank angle is:
dp k 1 dQin dQloss P dV P dk = + k d V d d V d k 1 d dQloss In Eq. (1), the rate of heat loss is expressed as: d dQloss 1 = hA( )(Tg Tw ) d
The convective heat transfer coefficient is given by the Woschni model as [10]: 0.55 0.8 h = 3.26 D 0.2 P 0.8Tg w , The mean piston speed is related as
(2)
(3) (4)
Up =
2NS 60
Based on the Woschnis formula calculate the convective mode of heat losses, assume it as 70% of total losses, based on that calculate 30% of heat losses, assume this 30% of heat losses is due to radiation. On the other hand, the rate of the heat input function [5, 7 and 8]:
m p 1
dQin (heat release) can be modeled using a dual Weibe d

mp
Qp dQin m = a p d p p
exp a p
Q + a d d
m d d
md 1
exp a d
md
(5)
Where p and d refers to premixed and diffusion phases of combustion. The parameters p and d represent the duration of the premixed and diffusion combustion phases. Also, Qp and Qdrepresent the integrated energy release for premixed and diffusion phases respectively. The constants a, mp, md are selected to match experimental data. For the current study, these values are selected as 6.9, 4, and 1.5 respectively [5]. It is assumed that the total heat input to the cylinder by combustion for one cycle is: (6) Qin = m f LHV Equations for ignition delay and fraction of fuel burned in premixed combustion are taken from Heywood [10]. Eq. (1) is discretized using a first order finite difference method to solve for the pressure at each crank angle (). Once the pressure is calculated, the temperature of the gases in the cylinder can be calculated using the equation of state as:
Tg =
P ( )V ( ) mR g
(2.7) The instantaneous cylinder volume, area, and displacement are taken from literature [10]. The initial NO formation rate written as
d [ NO] 6 1016 69,090 1/ 2 = exp [O2 ]e [ N 2 ]e mol/cm3.s 1/ 2 dt T T

The formation rate of soot particles is expressed in the Arrhenius form
(8)
263
Vol. 3, Issue 1, pp. 262-269
Ef dmsoot = A f m fuel p 0.5 exp RT dt

Initial temperature T [0] =
(9)
m a Ta + megr Tegr ( m a + m egr )
(10) (11)
P[0] V [0] = (ma + megr ) Rg T[0]
The following assumptions were made in modeling the compression and power strokes: i. All thermodynamic variables were assumed to be spatially uniform throughout the engine at any instant of time (zero-dimensional model) and variables varied temporally during the engine cycle. ii. Compression and expansion were assumed to take place in a series of quasi-equilibrium processes. iii. Compression stroke began at = 0 0 and expansion stroke ended at = 360 0 .Fuel injection began at = i and ended at = f . iv. v. vi. vii. For i ig , fuel was injected but not burned due to the ignition delay. Fuel combustion took place for ig . For ig , only a prescribed fraction of the instantaneous fuel mass available in the cylinder was burned. EGR fraction was defined as the ratio of the mass of exhaust gas to the mass of fresh air m egr m a at BDC.
0
Effect of soot formation on the reduction of the flame temperature (due to radiation) was considered. viii. EGR was assumed to consist of CO2 , H 2 O, O2 , N 2 (effect of residual gases neglected). The energy equation expressing the relationship between pressure and crank angle was solved using Eulers method (First order finite difference expression), to obtain the work output and cycle efficiency. The specifications of engine and other pertinent parameters are given Table 1.
Table 1. Specifications of Engine and fuel Fuel Cetane number 45 Lower heating value (kJ/kg) Molecular weight (kg/kmol) Stoichiometric air fuel ratio Compression ratio Bore X Stroke (m X m) Connecting rod length (m) Swept volume (m3) Engine speed range (rpm) Equivalence ratio, Injection timing Duration of combustion Wall temperature (K) 44.3103 148.6 14.36:1 18:1 0.105 X 0.125 0.1 1.082103 10005000 0.6 12 to 8 70 400
III.
Numerical experiments are performed, making use of the equations and imposing relevant conditions for the chosen diesel engine configuration, for predicting engine performance and emissions.
264
Vol. 3, Issue 1, pp. 262-269
3.1. Engine performance modeling: To begin with, heat release calculations are done for different crank angles;the pattern is shown in Figure.1, taking into consideration the heat transfer in the engine and variable specific heats of the air, using a dual Weibe function. This result is treated as base line data; it can be observed that there are three dominant combustion stages, namely premixed, diffusion and late combustion phases. The predicted combustion phases are in good agreement with the Lyn and Ways model [10].
Figure.1. Heat release pattern for the diesel engine at N=2500 rpm, =0.6, and FIB 8bTDC
The predicted heat release pattern is utilized to derive the in-cylinder pressure history to obtain engine performance. In-cylinder pressure variation with crank angle is obtained as shown in Figure.2, it can be observed that a deviation of combustion curve from the motoring pressures and peak pressure reaching about 90 bar occurring very near to TDC.
Figure 2. In-cylinder pressure for the diesel engine at N=2500 rpm, =0.6, and FIB 8bTDC.
Engine heat transfer from hot gases to walls comprises both convection, radiation and affects its performance, and emissions. The predominant portion of heat is lost through convection and radiation losses are significant in diesel engine due to soot formation accounting to about 25 to 35 per cent of the total heat transfer. For a given mass of fuel within the cylinder, higher heat transfer to the combustion chamber walls will lower average combustion gas temperature and simultaneous reduction in pressure decreases the work per cycle transferred to the piston. Figure3. 3represents rate of heat loss versus crank angle for the Diesel engine running at 2500 rpm and =0.6.
Figure 3 Diesel engine heat loss history for N=2500 rpm and =0.6 and FIB 8bTDC.
265
Vol. 3, Issue 1, pp. 262-269
By advancing the fuel injection timing, ignition delay will increase which causes increase in the fraction of fuel burned in pre-mixed combustion phase and thus increases premixed combustion stage as illustrated in Figure 4. In early injection timing, most of the heat release taking place before piston reaches the TDC and for the case of late injection timing heat release still continues even after piston crosses the TDC. However, an interesting observation is that 20 degrees after TDC, the gas temperature and pressure in the cylinder is still highest for late injection timing since heat release continues even after piston crosses the TDC. Also, the early injection timing develops higher pressures and temperatures leading to higher power output compared to delayed injection timing. The effect is illustrated in Figures 5 and 6.
Figure 4 Variation of premixed combustion heat profiles for different injection timings.
Figure 5. In-cylinder pressures for different injection timings.
Figure 6.In-cylinder temperatures for different injection timings.
With increase in engine speed, the absolute time in milliseconds gets reduced. The actual time for accumulation of fuel will be reduced with lesser amount of fuel injected. This effect reduces the ignition delay in milliseconds, causing decrease in the fraction of fuel burned in pre-mixed combustion. However, the duration of pre-mixed combustion increases with increase in speed. As the pre-mixed heat release decreases with speed, the combustion phasing shifts to diffusion with relatively higher values. The speed has significant effect on heat loss as the speed decreases, the
266
Vol. 3, Issue 1, pp. 262-269
absolute time taken for heat exchange increases resulting in more heat loss to the cylinder walls. These effects are well predicted through the present numerical experiments. With the consequent reduction in heat losses, eventually the energy is utilized in developing more work. This can be observed with increase in in-cylinder pressures for the given conditions. Also, increase in engine speed decreases time for exchange of heat with cylinder walls resulting in reduced heat losses and finally yielding higher temperatures within the cylinder. The ultimate effect of speed is that it increases engine efficiency. 3.2. Engine emissions modeling: Though direct injection diesel engines are known to be fuel efficient compared gasoline engines, challenges still exist with regards to emission of higher levels of NOx and PM. Thus researchers and leading manufactures are resorting to limit these harmful emissions. Numerical experiments are performed to observe the emission formation phenomenon. Dilution of the intake air with cooled recirculated exhaust gas limits the production high temperatures with subsequent reduction of NOx due to a lowering of the adiabatic flame temperature, reduction in oxygen content of the intake mixture and reduction of heat input inside the combustion chamber. The popular technique to accomplish this task is exhaust gas recirculation (EGR). EGR also reduces the mixture-averaged ratio of specific heats (k) of the charge, pressure, temperature leading to a reduction in the thermodynamic cycle efficiency. Also, with increase in EGR fraction heat release in both premixed and diffusion combustion phase reduces, as shown in Figs.3.7 and 3.8 respectively. Utilizing Zeldovich mechanism related equations for NOx formation with in-cylinder temperature history the appropriate equations are solved for predicting NOx emissions. Apart from, EGR retarding of fuel injection beginning is also known one of the techniques for NOx reduction methods. The effect of EGR alone is plotted as Figure 9. Thus, it can be noticed that as the EGR fraction increases, there appears to substantial reduction in NOx formation and emission and also with retarding timing there is reduction in NOx emission [10].
Figure 7. Variation of pre-mixed combustion with EGR fraction
Figure 8. Variation of diffusion combustion with EGR fraction
267
Vol. 3, Issue 1, pp. 262-269
Figure 9. Variation of NOx emissions with EGR fraction
In engines, the formation mechanisms for NOx and soot (or PM) are opposite to each other. High levels of EGR can also lead to high levels of PM due to a reduction in the oxygen levels and poor combustion quality. Additionally, EGR leads to higher specific fuel consumption and engine noise while adversely affecting the lubricating oil quality and engine durability. Also, early injection timing leads to higher levels of pressure and temperature which causes drop in the net soot density. Effect of EGR is shown in Figure 10. From Figures 9 and 10, it can be observed clearly that the factors which decrease the NOx would increase soot and vice-versa. Therefore, a trade-off should be obtained without emitting either harmful pollutants or in deterioration of engine gross performance [10].
Figure 10. Variation of soot emissions with EGR fraction
IV.
CONCLUSIONS
In the present work, DI diesel engine flow processes for performance and emissions are modelled using dual Wiebe function for heat release analysis. Based on the work, the following conclusions are drawn: Advancing of fuel injection timing, ignition delay increases which causes increase of premixed combustion with a minor effect on diffusion combustion. Early injection timing leads to higher levels of pressure and temperature inside the combustion chamber. Early injection timing leads to the formation of higher levels of NOx emission and lower levels of soot emission. Increase in engine speed, ignition delay increases in crank angle degrees, but decreases in milliseconds causing drop in premixed combustion however, with increased diffusion phase. Thermal efficiency, pressure and temperature will increase by increasing the engine speed. By increasing the fraction of EGR supply to the engine, heat input, pressure, temperature, adiabatic flame temperature, and efficiency will decrease. Increase in EGR fraction reduces the formation of NOx emissions and increase the formation of soot emission.
ACKNOWLEDGEMENTS
The authors wholeheartedly thank the authorities of NIT Warangal for their cooperation in carrying out the work and permitting to publish work.
268
Vol. 3, Issue 1, pp. 262-269
REFERENCES
[1] E. Abu-Nada, I. Al-Hinti, B. Akash, Al-Sarkhi, Thermodynamic analysis of spark ignition engine using a gas mixture model for the working fluid, International Journal of Energy Research 31 (2007) 1031 1046. E. Abu-Nada, Sakhrieh ,I. Al-Hinti , A. Al-Ghandoor , B. Akash Computational thermodynamic analysis of compression ignition engine International Communications in Heat and Mass Transfer 37 (2010) 299303. E. Abu-Nada, I. Al-Hinti, A. Al-Sarkhi, B. Akash, Effect of piston friction on the performance of SI engine: a new thermodynamic approach, ASME Journal of Engineering for Gas Turbines and Power 130 (2) (2008) 022802-1. J I Ghojel Review of the development and applications of the Wiebe function: a tribute to the contribution of Ivan Wiebe to engine research Int. J. Engine Res. Vol. 11, 2010. N. Miyamoto, T. Chikahisa, T. Murayama, R. Sawyer, Description and analysis of diesel engine rate of combustion and performance using Weibe's functions, SAE paper 850107, 1985. J. Arrgle, J.M. Garcia, J.J. Lopez, C. Fenollosa, Development of a zero-dimensional Diesel combustion model. Part 1: analysis of the quasi-steady diffusion combustion phase, Applied Thermal Engineering 23 (2003) 13011317. J. Galindo, J.M. Lujan, J.R. Serano, L. Hernandez, Combustion simulation of turbocharger HSDI Diesel engines during transient operation using neural networks, Applied Thermal Engineering 25 (2005) 877898. F.G. Chemla, G.H. Pirker, A. Wimmer, Zero-dimensional ROHR simulation for DI diesel engines a generic approach, Energy Conversion and Management 48 (2007) 29422950. AithalSM .Impact of EGR fraction on diesel engine performance consideringheat loss and temperature dependent properties of the working fluid.Int J Energy Res 2008; 33:415-30. Heywood JB. Internal Combustion Engine Fundamentals, New York: McGraw Hill; 1988.
[2]
[3]
[4] [5] [6]
[7]
[8] [9] [10]
Authors Biographies B.VenkateswaraRao is the Graduate Student Dept. of Mechanical Engineering, NIT Warangal, India. His areas of interest are IC Engines, Engine simulation
G.Amba Prasad Rao is working as Associate Professor in Dept. of Mechanical Engineering, NIT Warangal-506 004, India. His total teaching experience is 22 years. His areas of interest are IC Engines, Alternate Fuels, Emissions and its control, Engine simulation.
269
Vol. 3, Issue 1, pp. 262-269
MULTI-OBJECTIVE OPTIMIZATION OF CUTTING PARAMETERS FOR SURFACE ROUGHNESS AND METAL REMOVAL RATE IN SURFACE GRINDING USING RESPONSE SURFACE METHODOLOGY
M. Janardhan1 and A. Gopala Krishna2
1
Department of Mechanical Engineering, Abdul Kalam Institute of Technological Sciences, Kothagudem- 507120, A.P, India. 2 Department of Mechanical Engineering, University College of Engineering, JNTU, Kakinada-533 003, AP, India.
ABSTRACT
Surface grinding is the most common process used in the manufacturing sector to produce smooth finish on flat surfaces. Surface quality and metal removal rate are the two important performance characteristics to be considered in the grinding process. The economics of the machining process is affected by several factors such as abrasive wheel grade, wheel speed, depth of cut, table speed and material properties. In this work, empirical models are developed for surface roughness and metal removal rate by considering wheel speed, table speed and depth of cut as control factors using response surface methodology. In this paper, Response surface methodology (RSM) has been applied to determine the optimum machining parameters leading to minimum surface roughness and maximum metal removal rate in Surface grinding process operation on EN24 steel. The second order mathematical models in terms of machining parameters were developed for metal removal rate (MRR) and Surface roughness on the basis of experimental results. The model selected for optimization has been validated with F-test. The adequacy of the models is tested on output responses have been established with analysis of variance (ANNOVA). An attempt has also been made to optimize cutting parameters using multiobjective characteristics for the developed prediction models using Response surface methodology (RSM)
K EYW ORDS: Surface grinding, MRR, Surface roughness, RSM, Optimization.
I.
INTRODUCTION
Grinding is a complex machining process with lot of interactive parameters, which depend upon the grinding type and requirements of products. The surface quality produced in surface grinding is influenced by various parameters given as follows .(i) Wheel parameters: abrasives, grain size, grade, structure, binder, shape and dimension, etc.(ii) Work piece parameters: fracture mode, mechanical properties, and chemical composition, etc.(iii) Process parameters: wheel speed, depth of cut, table speed, and dressing condition, etc.(iv) Machine parameters: static and dynamic characteristics, spindle system, and table system, etc.[1] surface roughness is an performance index to meet the technical standards and customer satisfaction this performance index depends on various machining parameters. The selection of proper combination of machining parameters yields the desired surface finish and metal removal rate [2] the proper combination of machining parameters is an important task as it determines the optimal values of surface roughness and metal removal rate. It is necessary to develop mathematical models to predicate the influence of the operating conditions [3]. In the present work mathematical models has been developed to predicate the surface roughness and metal removal rate with the help of Response surface methodology, Design of experiments [30]. The Response surface methodology (RSM) is a practical, accurate and easy for implementation. The study of most important variables effecting the quality characteristics and a plan for conducting such experiments is called design of experiments(DOE) [31].The experimental data is used to develop
270
Vol. 3, Issue 1, pp. 270-283
mathematical models for second order models using regression methods. Analysis of variance is employed to verify the validity of the model. RSM optimization procedure has been employed to optimize the output responses, surface roughness and metal removal rate subjected to grinding parameters namely wheel speed, table sped and depth of cut using multi objective function model.
II.
METHODOLOGY
In this work, experimental results were used for modeling using Response surface methodology, is a practical, accurate and easy for implementation. The experimental data was used to build first order and second order mathematical models by using regression analysis method. This developed mathematical models were optimized by using the RSM optimization procedure for the output responses by imposing lower and upper limit for the input machining parameters namely table speed, wheel speed, and depth of cut.
2.1 Design of Experiments (DOE)

The study of most important variables affecting quality characteristics and a plan for conducting such experiments is called the Design of Experiments. G.Taguchi (1959) of Japan, by developing the associated concept of linear graph, was able to device numerous variants based on the OA design, which can easily be applied by an engineer or a scientist without acquiring advanced statistical knowledge for working out the design and analysis of even complicated experiments (Ross J. Philip, 1989).These methods have the advantage of being highly flexible and readily enable allocation of different levels of factors, even when these levels are not the same in number for all the factors studied [5]. The beauty of these methods lies in cutting to the bare minimum the size of experimentation. At the same time yielding results with high precision, thus, by a mere 27experiments, we may be able to evaluate all the main effects. The Design layout in Taguchis Method explained below: 1. List down the Response, Factors and levels along with the desired interactions. 2. Find the Degrees of Freedom for each factor and for each interaction. 3. Compute the Total Degrees of Freedom (TDF). 4. The minimum number of trials (MNE) is equal to total degrees of freedom Plus one (+1). 5. Choose the nearest orthogonal array series like: L4, L8, L16 or L9, L27, etc. 6. Draw the required Linear Graph (LG). 7. Number the linear Graph by starting with the Number 1 for factor A and Number 2 for factor B. Then check whether any interaction exists. If not, proceed with the Number 3 for factor C. If there is an interaction, check with the interaction Table, which Column is to be allotted to the interaction? Then Proceed with the next number for the next factor. 8. Complete the numbering as described until the following is achieved. All the factors and interactions are numbered. There is no repetition of numbers. The interaction numbers are as per the Interaction table. The numbers used do not exceed the number of columns permitted for the orthogonal array table. 9. Write the column numbers against each factor. That is the Design Assignment. Rewrite the OA Table with only those columns represented by factors and all the rows as per the OA Table. Replace the 1, 2 &3 in the Table with the Physical value of the level from the Factors and Levels identified. This completes the Design Layout.
2.2 Response Surface Methodology (RSM)

Response Surface Methodology is combination of mathematical and statistical technique [30-31], used develop the mathematical model for analysis and optimization. By conducting experiment trails and applying the regression analysis, the output responses can be expressed in terms of input machining parameters namely table speed, depth of cut and wheel speed. The major steps in Response Surface Methodology are 1. Identification of predominate factors which influences the surface roughness, Metal removal rate. 2. Developing the experimental design matrix, conducting the experiments as per the above design matrix.
271
Vol. 3, Issue 1, pp. 270-283
3. 4. 5. 6. 7. Developing the mathematical model. Determination of constant coefficients of the developed model. Testing the significance of the coefficients. Adequacy test for the developed model by using analysis of variance (ANNOVA). Analyzing the effect of input machining parameters on output responses, surface roughness and metal removal rate
III.
MATHEMATICAL FORMULATION
The first order and second order Mathematical models were developed using multiple regression analysis for both the output responses namely surface roughness and metal removal rate. Multiple regression analysis [17-22] is a statistical technique, practical, easy to use and accurate. The aim of developing the mathematical models is to relate the output responses with the input machining parameters and there by optimization of the machining process. By using these models, optimization problem can be solved by using Taughis optimization procedure as multi objective function model. The mathematical models can be represented by Y= f (V,N,d) (1) Where Y is the output grinding response, V, N, d are the table speed, wheel speed, depth of cut respectively In this work the following mathematical models were formulated. Metal removal rate, MRR=K1Va1Nb1dc1e (2) a2 b2 c2 (3) Surface roughness, Ra =K2V N d e To determine the above constants and exponents, this mathematical model have to be linearisied by performing the logarithmic transformation which as follows ln MRR= lnk1+a1lnV+b1lnN+c1lnd (4) (5) ln Ra= lnk2+a2lnV+b2lnN+c2lnd The constant and exponents can be determined by the method of least squares. The first order and second order linear models, develop from the above functional relationship using the least square regression analysis can be represented as follows Y1= Y-e = b0x0+b1x1+b2x2+b3x3 (6) Where Y1 is first order output response of metal removal rate, Y is the measured metal removal rate, x1x2x3 are the logarithmic transformations of table speed, wheel speed, depth of cut, respectively. The second order polynomial of output response will be given as Y2=Y-e= b0x0+b1x1+b2x2+b3x3+b12x1x2+b13x1x3+b23x2x3+b11x12+b22x22+ b33x32 (7) Where Y2 is second order output response of metal removal rate Y is the measured metal removal rate,b0,b1, b2, b3, b12, b13, b23, b11, b22 b33 are estimated by the method of lest squares. The validity of this mathematical model will be tested using F test, Chi-Square test before going for optimization.
IV.
EXPERIMENTAL DETAILS
A set of experiments were conducted on surface grinding machine to determine effect of machining parameters namely table speed(m/min),wheel speed(RPM),depth of cut(mm) on output responses namely surface roughness and metal removal rate. The machining conditions were listed in table 1.Three levels and three factors used to design the orthogonal array by using design of experiments(DOE) and relevant ranges of parameters as shown in Table 2.Grinding wheel used for the present work is the aluminum oxide abrasives with vitrified bond, WA 60K5V is used. The selected L27 orthogonal array to conduct the experiments is shown in the Table 3 along with the output responses, MRR and surface roughness. MRR was calculated as the ratio of volume of material removed from the work piece to the machining time. The surface roughness, Ra was measured in perpendicular to the cutting direction using with Surface Roughness tester SJ-201 at 0.8mm cutoff value. An average of six measurements was taken at six different places to record the output response, surface roughness. These results will be further used to analyze the effect of input machining parameters on output responses with the help of RSM and design expert software.
272
Vol. 3, Issue 1, pp. 270-283
Table 1 Machining conditions (a) (b) (c) (d) Work piece material: EN 24 steel Chemical composition: Carbon 0.35-0.45/ Silicon 0.10-0.35/ Manganese 0.45-0.70/ Nickel 1.30-1.80 / Chromium 0.90-1.40/ Moly 0.20-0.35/ Sulphur 0.050 (max)/ Phosphorous 0.050(max) and balance Fe Work piece dimensions: 155mm x 38mm x 38mm Physical properties: Hardness-201BHN, Density-7.85 gm/cc, Tensile Strength-620 Mpa (e) (f) Grinding wheel: Aluminum oxide abrasives with vitrified bond wheel WA 60K5V Grinding wheel size :250 mm ODX25 mm widthx76.2 mm ID Table 2 Levels of independent control factors S.No Control Factor Symbol Levels of factors -1 0 +1 1 Wheel speed N 1250 1650 2050 2 Table speed V 7.5 10 12.5 3 Depth of cut d 5 10 15 Unit RPM m/min m
Trail no 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Table 3 Experimental observations Surface Depth Roughness Wheelspeed Table speed of cut (Ra)(m) (N)(RPM) (V)(m/min) (d)(m) 1250 1250 1250 1250 1250 1250 1250 1250 1250 1650 1650 1650 1650 1650 1650 1650 1650 1650 2050 2050 2050 2050 2050 2050 2050 7.50 7.50 7.50 10.0 10.0 10.0 12.5 12.5 12.5 7.50 7.50 7.50 10.0 10.0 10.0 12.5 12.5 12.5 7.50 7.50 7.50 10.0 10.0 10.0 12.5 5 10 15 5 10 15 5 10 15 5 10 15 5 10 15 5 10 15 5 10 15 5 10 15 5 1.034 1.440 1.624 1.324 1.591 1.721 1.38 1.679 1.940 1.180 1.56 1.684 1.490 1.641 1.716 1.501 1.697 1.826 1.361 1.582 1.703 1.460 1.632 1.805 1.513
Metal Removal rate (gm/min) 5.510 10.28 15.505 7.350 13.28 22.07 9.190 16.89 26.09 6.260 13.13 17.68 11.63 15.61 21.17 12.42 18.17 25.87 7.760 13.21 19.40 10.35 15.16 24.16 12.64
273
Vol. 3, Issue 1, pp. 270-283
26 27 2050 2050 12.5 12.5 10 15 1.734 2.072 22.44 30.44
V.
DEVELOPMENT OF EMPIRICAL MODELS
In the present study, Empirical models of second order for the output responses, Surface roughness (Ra), Metal removal rate (MRR) in terms of input machining parameters in actual factors were developed by using the RSM [23-27]. The developed models are further used for optimization of the machining process. To determine the regression coefficients of the developed model, the Statistical analysis software, MINITAB, 16V is used. The second order models were developed for output responses due to lower predictability of the first order model to the present problem. The following equations were obtained in terms of uncoded factors Ra = - 0.4485 + 0.0005N + 0.1236V + 0.0975d - 0.0022V2 - 0.0017d2 - 0.00002NV - 0.000013Nd + 0.000053Vd (8)
MRR = - 0.9022 + 0.0023N - 0.3760V - 0.20041d - 0.000001N2 + 0.0118V2 + + 0.0203d2 + 0.00036 NV + 0.00007Nd + 0.1006Vd (9)
Analysis of variance (ANNOVA) is employed to test the significance of the developed models. The multiple regression coefficients of the second order model for surface roughness and metal removal rate were found 0.9325 and 0.9781 respectively. The R2 values are very high,close to one,it indicates that the second order models were adequate to represent the machining process. The "Pred RSquared" of 0.8027 is in reasonable agreement with the "Adj R-Squared" of 0.8967 in case of surface roughness. The "Pred R-Squared" of 0.9498 is in reasonable agreement with the "Adj R-Squared" of 0.9666 in case of MRR. Similarly,The Model F-value of 26.09 for surface roughness and The Model F-value of 84.51 for metal removal rate implies the model is significant. There is only a 0.01% chance that a "Model F-Value" this large could occur due to noise. The analysis of variance (ANNOVA) of response surface quadratic model for surface roughness and metal removal rate were shown in Table 4 and Table 5 respectively. Adeq Precision" measures the signal to noise ratio. A ratio greater than 4 is desirable.S/N ratio of 18.415 &32.54 for surface roughness and MRR indicates an adequate signal. This model can be used to navigate the design space. The P value for both the models is lower than 0.05 (at 95% confidence level) indicates that the both the models were considered to be statistically significant.The normal probability plots of residuals for Ra and metal removal rate are shown in the fig 1 and fig 2 respectively. From these plots, it can be concluding that the residuals lies on a straight line which implies that the errors are distributed normally and the developed regression models are well fitted with the observed valves. The Plot of Predicted versus actual response for surface roughness and MRR are shown in the fig 3 and fig 4 respectively and show that the models are adequate without any violation of independence or constant assumption.
Table 4 ANOVA for Response Surface Quadratic Model of Ra p-value Sum of Squares df Mean Square F Value 1.18 0.071 0.26 0.82 4.332E-003 7.550E-003 5.333E-006 4.603E-007 1.157E-003 9 1 1 1 1 1 1 1 1 0.13 0.071 0.26 0.82 4.332E-003 7.550E-003 5.333E-006 4.630E-007 1.157E-003 26.09 14.09 52.24 163.66 0.86 1.50 1.061E-003 9.211E-005 0.23 Prob>F 0.0001 Significant 0.0016 < 0.0001 < 0.0001 0.3662 0.2371 0.9744 0.9925 0.6374
Source Model N V d NV Nd Vd N2 V2
274
Vol. 3, Issue 1, pp. 270-283
d2 Residual Cor.Total Std.Dev. Mean C.V PRESS 0.011 0.085 1.27 0.071 1.59 4.46 0.25 1 17 26 0.011 5.026E-003 R-Squared Adj R-Squared Pred R-Squared Adeq Precision 2.23 0.1537
0.9325 0.8967 0.8027 18.415
Table 5 ANOVA for Response Surface Quadratic Model of MRR Source Model N V d NV Nd Vd N2 V2 d2 Residual Cor.Total Std.Dev. Mean C.V PRESS Sum of df Squares 1098.50 48.00 237.73 790.36 1.52 0.22 18.99 0.086 0.033 1.55 24.55 1123.05 9 1 1 1 1 1 1 1 1 1 17 26 1.2 15.69 7.66 56.37 Mean Square 122.06 48 237.73 790.36 1.52 0.22 18.99 0.086 0.033 1.55 1.44 F Value 84.51 33.24 164.61 547.25 1.05 0.15 13.15 0.060 0.023 1.07 p-value Prob>F < 0.0001 significant < 0.0001 < 0.0001 < 0.0001 0.3189 0.6994 0.0021 0.8101 0.8825 0.3144
R-Squared Adj R-Squared Pred R-Squared Adeq Precision
0.9781 0.9666 0.9498 32.524
Figure 1. Normal probability plot of residuals for Ra
275
Vol. 3, Issue 1, pp. 270-283
Figure 2. Normal probability plot of residuals for MRR
Figure 3. Comparison of Predicted and actual values for Ra
Figure 4. Comparison of Predicted and actual values for MRR
VI.
INTERPRETATION OF DEVELOPED MODELS
The detailed main effects and interaction effects for both the outputs are discussed in the following sections. It should be noted that if a particular parameter does not influence the output during the course of evaluation, it gets eliminated.
276
Vol. 3, Issue 1, pp. 270-283
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 6.1 Effect of process parameters on surface roughness (Ra)
6.1.1 Direct effects The direct effect of process parameters on output response, surface roughness is shown in figs 5 to 7. From Fig. 5, it is observed that increase in wheel speed tends to improve the finish. With carbide tools particularly, slow speed is not at all desirable since it means wastage of time and money and tools wear out faster. Fig. 6 shows the effect of table speed on roughness. As the table speed increases, finish gets poorest because the tool marks show on the work piece. The effect of depth of cut on surface roughness is shown in Fig. 7. It is noted from Fig. 7, that the increase in depth of cut makes the finish poor. Hence smaller values of table speed and depth of cut and larger value of wheel speed must be selected in order to achieve better surface roughness during the process.
Figure 5. Direct effect of wheel speed on Ra
Figure 6. Direct effect of table speed on Ra
Figure 7. Direct effect of depth of cut on Ra
277
Vol. 3, Issue 1, pp. 270-283
6.1.2 Interaction effects The three dimensional surface plots for the surface roughness are shown in figs 8-10.In each of these graphs, two cutting parameters are varied while the third parameter is held as its mid value. From fig 8, it is observed that best surface roughness was obtained at the lowest depth of cut and low wheel speed combination. The surface roughness results also indicate a poor surface finish for a lower depth of cut at a lower wheel speed. This behavior is due to the plugging action of the tool on the work piece surface at lower depth of cut. It is seen from these graphs that there is significant amount of curvature indicating non linearity in the variation. From fig 9, it is observed that best surface roughness is obtained at low depth of cut and low table speed. The 3d surface graph of Ra at constant depth of cut of 10 microns is shown in fig 10. From these graphs it is observed that there is switching of the curvature effect. It indicates that the reversal in behavior depending on the combination of the machining parameters. It also points towards significant contribution from the interaction of the machining parameters.
Figure 8. Interaction effect of wheel speed and depth of cut on Ra
Figure 9. Interaction effect of depth of cut speed and table speed on Ra.
Figure 10. Interaction effect of wheel speed and table speed on Ra.
278
Vol. 3, Issue 1, pp. 270-283
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 6.2 Effect of Process parameters on MRR
6.2.1 Direct effects The direct effect of process parameters on output response, surface roughness is shown in Figs 11 to 13. From Fig. 11, it is observed that increase in wheel speed tends to increase the MRR; where as the other two machining parameters are kept at its mid value. It is observed from the direct effects, depth of cut plays more vital role on MRR than other two parameters. Material removal rate in machining process is an important factor because of its vital effect on the industrial economy. Increasing the table speed, wheel speed and depth of cut leads to an increase in the amount of Material removal rate. But the most influential factors are table speed, and depth of cut. The highest value of MRR is obtained at the extreme range of the input parameters in all the interaction plots. Also the MRR increases gradually with the depth of cut.
Figure 11. Direct effect of wheel speed on MRR
Figure 12. Direct effect of Table speed on MRR
Figure 13. Direct effect of depth of cut on MRR
6.2.2 Interaction effects The 3D surface graphs for metal removal rate are shown in the fig 14 to 16 and shows that the graphs are curvilinear profile as the empirical model developed is quadratic. In each of these graphs, two cutting parameters are varied while the third parameter is held as its mid value. From fig 14, it is
279
Vol. 3, Issue 1, pp. 270-283
observed that best surface roughness was obtained at the lowest depth of cut and low wheel speed combination. The surface roughness results also indicate a poor surface finish for a lower depth of cut at a lower wheel speed. This behavior is due to the plugging action of the tool on the work piece surface at lower depth of cut. It is seen from these graphs that there is significant amount of curvature indicating non linearity in the variation. From these graphs it is observed that there is switching of the curvature effect. It indicates that the reversal in behavior depending on the combination of the machining parameters. It also points towards significant contribution from the interaction of the machining parameters.
Figure 14. Interaction effect of wheel speed and table speed on MRR
Figure 15. Interaction effect of wheel speed and depth of cut on MRR
Figure 16. Interaction effect of table speed and depth of cut on MRR
VII.
FORMULATION OF THE PROBLEM
In the process of optimization, the aim is to maximize the MRR and minimize the surface roughness (Ra), which forms the multi objective optimisation problem and these two are conflicting in nature [24]. The optimization problem for MRR and surface roughness (Ra) with feasible limits of control variables are represented in the equations (9) and (10) respectively after eliminating the insignificant terms.
280
Vol. 3, Issue 1, pp. 270-283
Minimize Ra = - 0.4485 + 0.0005N + 0.1236V + 0.0975d - 0.0022V2 - 0.0017d2 - 0.00002NV 0.000013Nd + 0.000053Vd (10) Maximize MRR = - 0.9022 + 0.0023N - 0.3760V - 0.20041d - 0.000001N2 + 0.0118V2 + 0.020d2 + 0.00036 NV + 0.00007Nd + 0.1006Vd (11) Subjected to 1250 RPMN2050 RPM 7.5 m/min V12.5 m/min 5 m d 15 m Once the optimization problem is formulated, then it is solved using a Response surface optimisation.
VIII.
OPTIMIZATION OF THE PROBLEM
Optimization of machining parameters increases the utility for machining economics, a response surface optimization is attempted using Minitab software for individual machining parameters in surface grinding. Table 6 shows the RSM optimization results for the surface roughness and MRR parameters in surface grinding. It also includes the results from confirmation experiments conducted with the optimum conditions.
Table 6 RSM optimization for output responses Parameter Objective Function Min Max Optimum combination N 1250 2050 V 7.5 12.5 d 5 15 Predicted response 1.128 29.48 Expected Value 1.034 30.44 % of error
Ra MRR
8.3 3.25
IX.
RESULTS
The optimum results for the output responses namely surface roughness and Metal removal rate interms of machining parameters namely wheel speed, table speed and depth of cut on EN 24 steel on CNC surface grinding machine using Minitab software were determined and presented in Table 6.The confirmation experiments were conducted and there is an good agreement between predicted and experimental values. It is found that the error in prediction of the optimum conditions is about 3 to8 %. Thus the response optimization predicts the optimum conditions fairly well
X. Conclusions
In this study an experimental investigation performed to evaluate the surface roughness and MRR parameters of EN 24 steel in surface grinding operation has been presented. A plan of experiments has been prepared in order to test the influence of cutting speed, feed rate and depth of cut on the output parameters. The obtained data have been statistically processed using response surface method. The empirical models of output parameters are established and tested through the analysis of variance to validate the adequacy of the models. It is found that the surface roughness and MRR parameters greatly depend on work piece materials. A response surface optimization is attempted using Minitab software for output responses in surface grinding.
REFERENCES
[1] Anne Venu gopal, P.V.Rao,Selection of optimum conditions for maximum material removal rate with surface finish and damage as constraints in SiC grinding, International Journal of Machine Tools & Manufacture 43 (2003) 1327-1336. [2] Jae Seob Kwak, Application of Taguchi and response surface methodologies for geometric error in Surface grinding process, international journal of machine tools &manufacture 45 (2005) 327-334. [3] P.V.S.Suresh, P.V.Rao, etc, A genetic algorithmic approach for optimization of surface roughness prediction model, international journal of machine tools &manufacture 42 (2002) 675-680.
281
Vol. 3, Issue 1, pp. 270-283
[4] I.A.Choudhury Surface roughness prediction in the turning of high strength steel by factorial design of experiments journal of materials processing technology 67 (1997) 55-61. [5] Anne Venu gopal, P.V.Rao, The optimization of the grinding of silicon carbide with diamond wheels using genetic algorithms, international journal of advanced manufacturing technology 22(2003) 475480. [6] P.G.Benardos, Predicting surface roughness in machining international journal of machine tools &manufacture 43 (2003) 833-844. [7] P.Krajnik, J.Kopac, A.Sluga, Design of grinding factors based on response surface methodology, journal of materials processing technology 162 (2005) 629-636. [8] L.P.Khoo and C.H.Chen, Integration of response surface methodology with genetic algorithm, international journal of advanced manufacturing technology 18(2001) 483-489. [9] N. Suresh Kumar Reddy and P. V. Rao, Selection of an optimal parametric combination for achieving a better surface finish in dry milling using genetic algorithms, International Journal of Advanced Manufacturing Technology, v 28, n 5-6, March, 2006, p 463-473. [10] T.S.Lee, T.O.Ting, Y.J.Lin, Than Htay, A particle swarm approach for grinding process optimization analysis, International journal of advanced manufacturing technology DOI. [11] Radu Pavel.Anil Srivastava, An experimental investigation of temperatures during conventional and CBN grinding, International journal of advanced manufacturing technology 2006. [12] X.M.Wen, A.A.O.Tay and A.Y.C.Nee, Micro-computer-based optimization of the surface grinding process, Journal of Material Processing Technology, 29 (1992) 75-90. [13] K.Ramesh, S.H.yeo, S.Gowri&L.Zhou ,Experimental Evaluation of super high speed grinding of advanced ceramics, International journal of advanced manufacturing technology,17(2001)87-92. [14] K.Palanikumar, Modeling and analysis of surface roughness in machining glass fibre reinforced plastics using response surface methodology, journal of materials and design, 2006. [15] Yusuf Sahin,A.Riza Motorcu, Surface roughness model for machining mild steel with coated carbide tool, journal of materials and design,26,(2005) 321-326. [16] Jae Seob Kwak, An analysis of grinding power and surface roughness in external cylindrical grinding of hardened SCM 440 Steel using the response surface method., international journal of machine tools &manufacture 46 (2006) 304-312. [17] M.Y.Noordin , V.C.Venkatesh, S.Sharif, S.Elting, A.Abdullah Application of response surface methodology in describing the performance of coated carbide tools when turning AISI 1045 steel,Journal of material processing technology145 (2004) 46-58. [18] Tugrul Ozel, Yigit karpat Predictive modeling of surface roughness and tool in hard turning using regression and neural networks. International journal of machine tools & manufacture 45 (2005) 467479. [19] Sanjay Agarwal, P.Venkateswar rao , Experimental investigation of surface /subsurface damage formation and material removal mechanisms in sic grinding. International journal of machine tools &manufacture 48(2008) 698-710. [20] N.Alagumurthi, K. palaniradja and V.Soundararajan,cylindrical grinding a review on surface integrity. International journal of precision engineering and manufacturing vol. no 3. [21] Stephen malkin and changsheng guo, Model based simulation of grinding process department of mechanica l& industrial engineering, university of Massachusetts Amherst, MA 01003; united technologies research center 411 silver lanes East Hartford, ct 06108. [22] Shaji, S. and Radhakrishnan, V., 2003, Analysis of process parameters in surface grinding with graphite as lubricant based on the Taguchi method, Journal of Materials Processing Technology, 141:51-59. [23] Dhavlikar, M.N., Kulkarni, M.S. and Mariappan, V., 2003, Combined Taguchi and dual response method for optimization of a centerless grinding operation, Journal of Materials Processing Technology, 132:90-94. [24] Hecker, R.L. and Liang, S.Y., 2003, Predictive modeling of surface roughness in grinding,International Journal of Machine Tools &Manufacture, 43:755-761. [25] Zhong, Z.W., Khoo, L.P. and Han, S.T., 2006,Prediction of surface roughness of turned surfaces using neural networks, International Journal of Advanced Manufacturing Technology, 28:688-693. [26] Sun, X., Stephenson, D.J., Ohnishi, O. and Baldwin,A., 2006, An investigation into parallel and cross grinding of BK7 glass, Precision Engineering,30:145-153. [27] Atzeni, E. and Iuliano, L., 2008, Experimental study on grinding of a sintered friction material, Journal of Materials Processing Technology, 196:184-189. [28] Choi, T.J., Subrahmanya, N., Li, H. and Shin, Y.C., 2008, Generalized practical models of cylindrical plunge grinding process, International Journal of Machine Tools and Manufacture, 48:61-72.
282
Vol. 3, Issue 1, pp. 270-283
[29] Liu, C.H., Chen, A., Chen, C.A. and Wang, Y.T., 2005, Grinding force control in an automatic Surface finishing system, Journal of Materials Processing Technology, 170:367-373. [30] Montegomery. D. C., 1991, Design and Analysis of experiments, Wiley, India. [31] Response Surface Methodology, Raymonds.H.Mayres & Douglas.E.Montgomery, Second Edition,John Whiely Publishers. Authors M. Janardhan obtained his B.Tech (Mechanical Engineering) & M.Tech (Mechanical Engineering) from S.V University in 1989 and 1999 respectively. He is at the finishing stage of PhD degree in JNT University, Kakinada, India. His employment in teaching includes at the levels of lecturer, Assistant professor, Associate professor and Professor at various engineering colleges over a span of more than 22 years. Presently, He is working as a principal at Abdul Kalam Institute of Technological Sciences, Kothagudem- 507120, A.P, India. He published 6 research papers in international conferences & International Journals. He authored a textbook entitled" dynamics of machinery" published by Hi-Tech Publishers, Hyderabad. He is life member of ISTE. A. Gopalakrishana obtained his B.Tech (Mechanical Engineering) from S.K University in 1994, M.Tech (Mechanical Engineering) from S.V University in 1996 respectively. He received his doctorate degree (PhD) in Mechanical Engineering from JNT University, Kakinada, India in 2006. His teaching experience includes at the levels of lecturer, Assistant professor and Associate professor at JNT University, Kakinada over a span of more than 12 years. Presently, He is working as a Associate professor at JNT University, Kakinada A.P, India. He published 80 research papers in national /international conferences & Journals. He is life member of ISTE and Institution of Engineers.
283
Vol. 3, Issue 1, pp. 270-283
HARMONIC REDUCTION IN CASCADED MULTILEVEL INVERTER WITH REDUCED NUMBER OF SWITCHES USING GENETIC ALGORITHMS
C. Udhaya Shankar1, J.Thamizharasi1, Rani Thottungal1, N. Nithyadevi2
1
Department of EEE, Kumaraguru College of Technology, Coimbatore, India 2 Department of Mathematics, Bharathiyar University, Coimbatore, India
ABSTRACT
In this paper, a new topology of cascaded multilevel inverter using a reduced number of switches is proposed. The new topology has the advantage of reduced number devices compared to traditional configurations and can be extended to any number of levels. This topology results in reduction of installation area, cost, computational time and has simplicity of control system. This structure consists of series connected sub-multilevel inverter blocks. The GA technique finds the optimal solution set of switching angles, if it exits, for each required harmonic profile. Both simulation results and experimental verification of the proposed inverter topology for different number of levels and different harmonic profiles are presented.
KEYWORDS:
Multilevel inverter, Cascaded multilevel inverter, H bridge, Full-bridge, Sub-multilevel inverter, Selective harmonic elimination, Programmed PWM, Genetic algorithms.
I.
INTRODUCTION
A Multilevel inverter is a power electronic system that synthesizes a desired output voltage from several DC voltages as inputs. The concept of utilizing multiple small voltage levels to perform power conversion was presented by a MIT researcher [1,2]. Advantages of this approach include good power quality, good electro-magnetic compatibility, low switching losses and high voltage capability. The first introduced topology is the series H-bridge design [1]. This was followed by the diodeclamped inverter [24] which utilizes a bank of series capacitors to split the dc bus voltage. The flying-capacitor (or capacitor clamped) [5] topology uses floating capacitors to clamp the voltage levels. Another multilevel design, involves parallel connection of inverter phases through interphase reactors [6]. One particular disadvantage of multilevel inverter is the great number of power semi-conductor switches needed. So, in practical implementation, reducing the number of switches and gate driver circuits is very important. Genetic algorithms (GAs) are stochastic optimization techniques. Genetic Algorithms are applied in this to compute the switching angles in a cascaded multilevel inverter to produce the required fundamental voltage while, at the same time, 3rd and 5th harmonics are reduced. It is shown in [79] that the problem of harmonic elimination is converted into an optimization task using binary coded genetic algorithms (GA).Various components of GAs such as chromosomes, fitness function, reproduction, crossover and mutation are illustrated as applied to the present work.
II.
CONVENTIONAL CASCADED MULTILEVEL INVERTER
The cascaded multilevel inverter consists of series connections of n full bridge topology fig.1shows the configuration of cascaded multilevel inverter.
284
Vol. 3, Issue 1, pp. 284-294
Fig. 1. Configuration of cascaded multilevel inverter
III.
SUGGESTED TOPOLOGY
Fig. 2 shows the suggested basic unit for a sub-multilevel inverter. This consists of a capacitor (with dc voltage equal to Vdc) with two switches S1 and S2. Table 1 indicates the values of Vo for states of switches S1 and S2. It is clear that both switches S1 and S2 cannot be on simultaneously because a short circuit across the voltage Vdc would be produced. It is noted that two values can be achieved for Vo.
The basic unit shown in Fig. 2 can be cascaded as shown in Fig. 3.
285
Vol. 3, Issue 1, pp. 284-294
Fig. 3. Cascaded basic unit
The output voltage of conventional multilevel inverter for all times has zero or positive value. In the following, we propose a new method for determination of magnitudes of dc voltage sources which are used in the proposed multilevel inverter. The number of maximum output voltage steps of the n series basic units can be evaluated by, Nstep=n+1 The maximum output voltage is given by, Vo,max=n*Vdc
IV.
EXTENDED STRUCTURE
Fig 4. The proposed structure for generating both positive and negative voltages
286
Vol. 3, Issue 1, pp. 284-294
V.
COMPARISON OF THE SUGGESTED STRUCTURE WITH THE CONVENTIONAL MULTILEVEL INVERTER
Tables 4 and 5 compare the power component requirements among the conventional and suggested multilevel inverters for the same number of the output voltage steps, respectively.
287
Vol. 3, Issue 1, pp. 284-294
VI.
SIMULATION DIAGRAM OF 11-LEVEL (CONVENTIONAL) CASCADED MULTILEVEL INVERTER
Fig 5. 11-level Cascaded Multilevel inverter(Cascaded) Table 5: Switching sequence for Conventional topology
VII.
OUTPUT WAVEFORMS (CONVENTIONAL)
Fig 6.Output Voltage Waveform
288
Vol. 3, Issue 1, pp. 284-294
Fig 7.Output Current waveform
Fig 8.THD Analysis(Conventional)
VIII.
SIMULATION DIAGRAM OF 11-LEVEL (PROPOSED) CASCADED MULTILEVEL INVERTER
Fig 9. 11-level Cascaded Multilevel inverter(proposed)
289
Vol. 3, Issue 1, pp. 284-294
Table 6: Switching sequence for proposed topology
IX.
OUTPUT WAVEFORMS (PROPOSED)
Fig 10.Output Voltage Waveform
Fig 11.Output Current waveform
290
Vol. 3, Issue 1, pp. 284-294
Fig 12.THD Analysis (Proposed)
X.
MATHEMATICAL METHOD OF SWITCHING
The control of the new family of multilevel inverters is to choose a series of switching angles to synthesize a desired sinusoidal voltage waveform synthesized by a 2s + 1 level inverter, where s is the number of switching angles, which also equals the number of dc sources. To reduce 3rd and 5th order harmonics in the 11-level inverter, three nonlinear equations can be set up as follows: Cos((1))+Cos((2))+Cos((3))+Cos((4))+Cos((5))=M Cos (3(1))+Cos(3(2))+Cos(3(3))+Cos (3 (4))+Cos(3(5))=0 Cos(5(1))+Cos(5(2))+Cos(5(3))+Cos(5(4))+Cos(5(5))=0 Where, Modulation index, M = Vm/5Vdc
XI.
SOLUTION USING GENETIC ALGORITHMS
A GA for optimization is different from classical optimization methods in several ways: random versus deterministic operation, population versus single best solution and selecting solutions via survival of the fittest.The solution to the harmonic elimination problem is five switching angles 1, 2, 3, 4, 5. Each switching angle is called a gene. A chromosome consists of all the genes and in this case there are five genes in one chromosome. Thus, each chromosome represents a possible solution to the problem
A. Encoding of a Chromosome
The population size remains constant throughout the whole process.. The most used way of encoding is a binary string. Indeed, there are many other ways of encoding. The encoding depends mainly on the problem considered. In this study, a binary coding system is used. A string then could look like this (in the binary case): String 1 1101100100110110 String 2 1101111000011110 A string in GAs may be divided into a number of substrings. The number of sub-strings, usually, equals the number of problem variables.
B. Fitness Function
291
Vol. 3, Issue 1, pp. 284-294
The fitness function is the function responsible for evaluation of the solution at each step. The objective here is determining the switching angles such that some selected harmonics are minimized or put equal to zero.
C. Selection
The GA performs a selection process in which the most fit members of the population survive, and the least fit members are eliminated.
D. Crossover
Crossover operates on selected genes from the parent chromosomes and creates new offspring. Crossover can be illustrated as follows :(j is the crossover point): Chromosome 1 11011j00100110110 Chromosome 2 11011j11000011110 Offspring 1 11011j11000011110 Offspring 2 11011j00100110110
E. Mutation
After performing crossover, mutation takes place. Mutation is used to prevent all the solutions in the population falling into a local optimum of the solved problem.. In case of binary encoding, we can switch a few randomly chosen bits from 1 to 0 or from 0 to 1. Mutation can be illustrated as follows: Original offspring 1 1101111000011110 Original offspring 2 1101100100110110 Mutated offspring 1 1100111000011110 Mutated offspring 2 1101101100110110 The technique of mutation (as well as crossover) depends mainly on the encoding of the chromosomes.
XII.
SIMULATION DIAGRAM WITH GA CONTROLLER
Fig .13 11-level Cascaded Multilevel inverter with GA Controller
292
Vol. 3, Issue 1, pp. 284-294
Fig 14.THD Analysis (with GA)
XIII.
RESULTS OBTAINED
Table 7: THD Comparison
TYPE CONVENTIONAL PROPOSED PROPOSED WITH GA THD 18.75% 13.27% 7.86% 3rd HARMONIC 13.57% 11.8% 0.6% 5th HARMONIC 4.9% 2.8% Eliminated
In Conventional Cascaded multilevel inverter, the observed THD is 17.2 %.In our proposed topology of multilevel inverter with reduced number of switches ,the observed THD value is 13.2% and then with GA control techniques, the THD is further reduced to 7.86% and the 3rd and 5th Order harmonics has been eliminated.
XIV.
CONCLUSION
The selective harmonic elimination of a new family of multilevel inverters using GA has been presented. The new configuration has the advantage of a reduced number of switching devices compared to traditional configurations of the same number of levels. The GA technique usually produces more than one possible solution set for each harmonic profile and a given specific modulation index. For multiple solutions, the solution that gives the lowest THD is selected. Both simulation and experimental results show that the algorithm can be effectively used for selective harmonic elimination of the new family of multilevel inverters and results in a dramatic decrease in the output voltage THD.
REFERENCES
[1] Baker RH. Electric power converter. US Patent 03-867-643; February 1975. [2] Baker RH. High-voltage converter circuit. US Patent 04-203-151; May 1980. [3] Nabae A, Takahashi I, Akagi H. A new neutral-point clamped PWM inverter. In: Proceeding of the industry application society conference; 1980.
293
Vol. 3, Issue 1, pp. 284-294
[4] Fracchia M, Ghiara T, Marchesini M, Mazzucchelli M. Optimized modulation techniques for the generalized N-level converter. In: Proceeding of the IEEE power electronics specialist conference, vol. 2; 1992. p. 120513. [5] Meynard TA, Foch H. Multi-level conversion: high voltage choppers and voltage source inverters. In: Proceedings of the IEEE power electronics specialist conference, vol. 1; 1992. p. 397403. [6] Ogasawara S, Takagali J, Akagi H, Nabae A. A novel control scheme of a parallel current-controlled PWM inverter. IEEE Trans Ind Appl 1992;28(5):102330. [7] B. Ozpineci, L.M. Tolbert, J.N. Chaisson, Harmonic optimization of multilevel converters using genetic algorithms, IEEE Power Electronics Lett. (2005) 14. [8] B. Ozpineci, L.M. Tolbert, J.N. Chaisson, Harmonic optimization of multilevel converters using genetic algorithms, in: Proceeding of IEEE Power Electronics Specialist Conference Letters, 2004, pp. 39113916. [9] K. Sundareswaran, A.P.Kumar, Voltage harmonic elimination in PWMA.C. chopper using genetic algorithm, IEE Pro-Electr. Power Appl. 151 (1) (2004) 2631.
Authors Biography
C. UDHAYA SHANKAR MIEEE. received his B.E. degree in Electrical and Electronics Engineering from Bharathiyar University, Coimbatore India in 2001 and ME degree in Power electronics and drives from Vellore Institute of Technology, India in 2002. Recently he is a PhD candidate at Anna university of Technology, Coimbatore, India. He is currently working as Senior grade Assistant professor in Kumaraguru college of Technology, Coimbatore, India. His main interest research is optimization techniques and its application to Power Electronics, Power quality, FACTS devices and their control.
J.THAMIZHARASI was born in salem India, in 1989. She received the B.E (Electronics and Instrumentation Engineering) degree in Vivekanandha College of Engineering for women, Thiruchengode, India in 2010. Now she is doing M.E (Power Electronics and Drives) in Kumaraguru College of Technology, Coimbatore, India. Her areas of interest are Power Electronics and Drives, Power Quality, and Renewable Energy. N. NITHYADEVI received her U.G degree in Mathematics, Kongunadu Arts and Science College, Coimbatore, India in 2000 and M.Sc degree in Mathematics from Kongunadu Arts and Science College, Coimbatore in 2002 and M.Phil from Kongunadu Arts and Science College, Coimbatore in 2003. She received her PhD for Applied Mathematics from Bharathiyar University in 2006 and she got the Post Doctoral Fellowship from National Cheng Kung University Taiwan, ROC in 2008. She is having Research and Teaching Experience of Five Years. She has presented many papers in different International Journals. She is currently working as Assistant professor in Bharathiyar University, Coimbatore, India.
294
Vol. 3, Issue 1, pp. 284-294
FINGERPRINT BASED GENDER IDENTIFICATION USING FREQUENCY DOMAIN ANALYSIS

Ritu Kaur1 and Susmita Ghosh Mazumdar2
2
M. Tech Student, RCET Bhilai, India Reader, Department of Electronics & Telecom, RCET Bhilai, India
ABSTRACT
Although Fingerprints are one of the most mature biometric technologies and are considered legitimate proofs of evidence in courts of law all over the world, relatively little machine vision method has been proposed for gender identification. Few researchers addressed the use of fingerprint for gender identification which will be more helpful in short listing the suspects. In this paper, a novel method is proposed to estimate gender by analysing fingerprints using fast Fourier transform (FFT), discrete cosine transform (DCT) and power spectral density (PSD). A dataset of 220 persons of different age and gender is collected as internal database. Initially the fingerprints of the subject were tested and after the manual analysis threshold is specified. Frequency domain calculations are compared with predetermined threshold and gender is determined. Of the samples tested, 99 samples identified exactly out of 110female samples and 87 samples identified exactly out of 110 male samples.
KEYWORDS:
Discrete cosine transform, frequency domain, fast Fourier transform, gender identification, and power spectral density.
I.
INTRODUCTION
Within todays environment of increased importance of security and organization, identification and authentication methods have developed into a key technology. Such requirement for reliable personal identification in computerized access control has resulted in the increased interest in biometrics. Fingerprints are one of the most mature biometric technologies and are considered legitimate proofs of evidence in courts of law all over the world. Based on the varieties of the information available from the fingerprint we are able to process its identity along with gender, age and ethnicity. Fingerprint is an impression of friction ridges, from the surface of the finger-tip. Fingerprints have been used for personal identification for many decades; more recently becoming automated due to advancements in the computing capabilities Fingerprints have some important characteristics that make them invaluable evidence in crime scene investigations: 1. A fingerprint is unique to a particular individual, and no two fingerprints possess exactly the same set of characteristics. 2. Fingerprints do not change over the course of persons lifetime (even after superficial injury to the fingers). 3. Fingerprint patterns can be classified, and those classifications then used to narrow the range of suspects. In this paper, we proposed a method that detects the gender of a person using fingerprints by frequency domain analysis. Here we obtain the fundamental frequency of various transforms and use them for gender Classification. This application is helpful in short listing the suspects and victims from crime and to boost the performance of a system which is used for person recognition and human computer interfaces. The remainder of this paper is organized as follows: brief literature of various gender recognition algorithms using fingerprint is discussed in section 2. Frequency domain analysis is discussed in section 3. The proposed system is discussed in section 4. The dataset and experimental results are described in section 5.
295
Vol. 3, Issue 1, pp. 295-299
II.
RELATED WORKS
Although the fingerprint plays vital role in the identification and verification, relatively few machine vision method has been proposed for gender identification. In this section we briefly review and summarize the prior researches in gender classification. M.D. Nithin et al [1] has applied baye's theorem on the rolled fingerprint images belonging to south Indian population and found fingerprint possessing ridge density < 13 ridges/25 mm2 is most likely to be of male origin and ridge count > 14 ridges/25mm2 are most likely to be of female. Similar results is obtained by Dr. Sudesh Gungadin MBBS [2] using Ridge density by counting the ridges in the upper portion of all fingers which shows that a finger print ridge of < 13 ridges/25 mm2 is more likely of male origin and finger print ridge of > 14 ridges/25mm2is more likely of female origin. Acree MA [6] used Ridge density in a particular space to classify gender using fingerprint. He showed that female have a higher ridge density compared with male. Kralik M and Vladimir Novoiny [5] showed that the males have higher ridge breadth defined as the distance between the centers of two adjacent valleys, than females. Ahmed Badawi et al [7], used ridge thickness to valley thickness ratio (RTVTR), and white lines count features for the classification. According to them, the female's fingerprint is characterized by a high RTVTR; while the male's fingerprint is characterized by low RTVTR Dr. A. Bharadwaja et al [8] correlated relation between the fingerprint pattern and blood groups between male and female. In this paper, instead of traditional ridge related analysis, we proposed a frequency domain analysis of fingerprint using FFT, 2D-DCT and PSD.
III.
FREQUENCY DOMAIN ANALYSIS
The gender identification is made through the frequency domain instead of the traditional spatial domain. The transforms FFT, DCT and PSD are chosen for the fingerprint analysis. Fourier transform plays a vital role in image processing applications. It contains most of the information of the spatial domain image. DCT transforms an image from the spatial domain to the frequency domain and provide better approximation of image. DCT transforms a set of data which is sampled at a given sampling rate to its frequency components. The fundamental frequencies of these transforms are used for gender identification. The 2D FFT pair is given by
, , = = 1 1 , ,
Where, 0 m, k M-1, 0 n, l N-1
The discrete cosine transform (DCT) is a Fourier-related transform similar to the discrete Fourier transform (DFT), but using only real numbers. The equation for the two-dimensional DCT is
, = 2 , cos 2 +1 2 cos 2 +1 2
Where C(m),C(n)= 1/2 for m,n=0 and C(m),C(n)=1 otherwise The 2D DCT is computed by applying 1D DCT (vertically) to columns and the resulting vertical DCT is applied with 1D DCT (horizontally).Let F(signal) is Fourier transform of the signal and the PSD is found by , PSD=|abs(F(signal))|^2/N where N is Normalization factor.
IV.
PROPOSED SYSTEM LEVEL DESIGN
A fingerprint based gender identification system constitutes of digital images of fingerprint as its input which is then transformed into frequency domain, compared with the predetermined thresholds
296
Vol. 3, Issue 1, pp. 295-299
and finally, gender is declared. The figure below shows the block diagram of the proposed gender identification system by frequency domain analysis of fingerprints.
Figure-1: Block Diagram of the proposed Gender identification scheme. The proposed gender identification system follows the following steps: 1. Input from the database is given to the gender identification system. 2. FFT transforms the given input and generates the output. Threshold is set to TH1. Rule is set in such a way that if the fundamental frequency (FF) is greater than TH1 the decision is female and if the FF is less than TH1 the decision will be male. 3. DCT transforms the given input and generates the output. Threshold is set to TH2. Rule is set in such a way that if the fundamental frequency (FF) is greater than TH2 the decision is female and if the FF is less than TH2 the decision will be male. 4. PSD transforms the given input and generates the output. Threshold is set to TH3. Rule is set in such a way that if the fundamental frequency (FF) is less than TH3 the decision is female and if the FF is greater than TH3 the decision will be male. 5. Comparing the decisions by all the transforms, if two decisions are male, the result is announced as male and if two decisions are female, the result is announced as female.
V.
5.1 Data Set

The database is basic requirement for any research work An internal dataset of fingerprints (left and right forefinger) for 220 persons of different ages and gender (110 males, and 110 females) were obtained from different colleges that used biometric fingerprint sensor for marking the attendance and were analysed using frequency domain analysis. The internal database is of 8 bit gray level with a size of 109 x 108 each. The developed algorithm has been tested using the MAT LAB 7.1 image processing tool.
Figure- 2: Sample Fingerprint images from our database
297
Vol. 3, Issue 1, pp. 295-299
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 5.2 Threshold Setting Setting the threshold for each transform is an important part of the gender identification process. Initially 50 fingerprints of both male and female are examined with FFT, DCT and PSD and the fundamental frequencies were obtained for each case. After the manual analysis, proper threshold was set for the gender classification. Table below gives the details of the transform, threshold and the threshold condition for the identification of gender.
Table-1: Threshold setting of FFT, DCT and PSD Transforms and Threshold FFT DCT PSD FF<1500000 FF<17000 FF>9000000000 FF>1500000 FF>17000 FF<9000000000
Gender Male Female
5.3 Results Table-2 shows the FFT, DCT and PSD transform results of fingerprints of 10 female subjects. The proposed system performs well and successfully classifies the subjects as females. Few readings were found to have deviations from the desired results.
Table 2: Results of FFT, DCT and PSD for female samples
Fingerprint Sample FFT Threshold> 150000 DCT Threshold> 17000 PSD Threshold<9000000000
1 2 3 4 5 6 7 8 9 10
2279661 1570108 1454314 2330334 2258674 1580512 2284175 2000917 2182986 1848986
21010.92356 14471.19513 13403.958 21477.96078 20817.49294 14567.08555 21052.52769 18441.82716 20119.90027 17041.52657
16050830074 7648114668 6569115014 16606900769 15758103976 7749077120 16114143245 12248922996 14578942394 10583366169
Table-3 shows the FFT, DCT and PSD transform results of fingerprints of 10 male subjects. The proposed system performs well and successfully classifies the subjects as males. Few readings were found to have deviations from the desired results.
Table 3: Results of FFT, DCT and PSD for male samples
Fingerprint Sample FFT Threshold<150 000 DCT Threshold< 17000 PSD Threshold>9000000000
1 2 3 4 5 6 7 8 9 10
1720734 1010376 1817095 1447936 2249503 1736023 1666175 1246556 1369293 1777807
15859.46794 9312.320079 16747.59719 13345.17396 20732.96669 16000.38188 15356.61467 11489.11739 12620.34599 16385.49196
9059365136 3124565177 10097352413 6512066008 15480756894 9337090299 8489722112 4838737388 5737451302 9665436481
Of the total samples tested, the performance efficiency of the proposed system was found to be 90% for female (99 samples identified exactly out of 110female samples) and 79.09% (87 samples identified exactly out of 110male samples) for male.
298
Vol. 3, Issue 1, pp. 295-299
VI.
CONCLUSIONS
In this paper, instead of traditional ridge related analysis, we proposed a frequency domain analysis of fingerprint to identify the gender. From the internal database, 110 male samples and 110 female samples were tested. Optimal threshold for each transform is chosen for better results. It is found that the proposed algorithm produces an accurate decision of 90% for female and 79.07% for male. The performance can be enhanced further by using good quality fingerprint images. In future, more work can be done in frequency domain to find different parameters and different transforms that can be applied in gender identification which will more accurate and suitable for all types of applications. Gender identification results using frequency domain analysis showed that this method could be considered as a prime candidate for use in forensic anthropology in order to minimize the suspects search list and give a likelihood probability value of the gender of a suspect.
REFERENCES
[1.] M.D. Nithin, B. Manjunatha, D.S. Preethi and B.M. Balaraj, "Gender differentiation by finger ridge count among South Indian population," Journal of Forensic and Legal Medicine, vol 18 (2), pp. 79-81, Feb 2011. [2.] Dr. SudeshGungadin MBBS, MD "Sex Determination from Fingerprint Ridge Density," Internet Journal of Medical Update, Vol. 2, No. 2, Jul-Dec 2007. [3.] G. G. Reddy, Finger dermatoglyphics of the Bagathas of Araku Valley (A.P.), American Journal of Physical Anthropology Volume 42, Issue 2, pages 225-228, March 1997. [4.]Yvonne K. Dillon, Julie Hayne S and Macie J Hennerberg [8]," The relationship of the number of Meissner's corpuscles to dermatoglyphic characters and finger size," journal of anatomy, vol 199, pp. 577-584, November 2001. [5.] KRALIK M., NOVOTNY V. 2003. Epidermal ridge breadth: an indicator of age and sex in paleodermatoglyphics. Variability and Evolution, Vol. 11: 5-30, Adam Mickiewicz University, Faculty of Biology, Institute of Anthropology, Poznan. [6.] Acree, M. Is there a gender difference in fingerprint ridge density? Forensic Science International 1999 May; 102 (1): 35-44. [7.] A. Badawi, M. Mahfouz, R. Tadross, and R. Jantz.Fingerprint-based gender classification.The International Conference on Image Processing, Computer Vision, and Pattern Recognition, June 2006. [8.] Dr. A.Bharadwaja , Dr.P.K.Saraswat, Dr.S.K.Aggarwal, Dr.P.Banerji, and Dr.S.Bharadwaja, "Pattern of finger prints in different ABO blood groups." Journal of Indian Academy of forensic medicine, vol 26(1), pp 69, March 2004. [9]. Anil K. Jain, Sarat C. Dass, and KarthikNandakumar Soft Biometric Traits for Personal Recognition Systems Proceedings of International Conference on Biometric Authentication, LNCS 3072, pp. 731-738, Hong Kong, July 2004 [10]. S.M.E. Hossain , G. Chetty Next Generation Identity Verification Based on Face-Gait Biometrics 2011 International Conference on Biomedical Engineering and Technology IPCBEE vol.11 (2011) (2011) IACSIT Press, Singapore [11]. Michael D. Frick, Shimon K. Modi, Stephen J. Elliott, Ph.D., and Eric P. Kukula, Member IEEE Impact of Gender on Fingerprint Recognition Systems 5th International Conference on Information Technology and Applications ICITA 2008 ISBN: 978-0-9803267-2-7
Authors
Ritu Kaur is currently pursuing Masters Degree program in Digital Electronics in Chhattisgarh Swami Vivekananda Technical University, India.
Susmita Ghosh Mazumdar is Reader in Rungta College of Engineering and Technology in Chhattisgarh Swami Vivekananda Technical University, India.
299
Vol. 3, Issue 1, pp. 295-299
APPLYING THE GENETIC ALGORITHMS OF SORTING THE ELITIST NON-DECISIVE SOLUTIONS IN THE CASE STUDY "RESOURCE ALLOCATION"
Khashayar Teimoori
Department of Mechanical and Aerospace Engineering, Science and Research Branch, Islamic Azad, University, Tehran, Iran
ABSTRACT
To achieve the purposes of the permanent development, a good energy Policy should take into account the various range of the political, economical, social and environmental considerations. On the other hand, the energy resources and technologies at hand are limited. Due to the complexity of the elements affecting the decision-making process, the usage of modelling methods in the energy sector has major importance. Unfortunately, the old methodologies of the energy systems Modelling are not suitable for multi-objective purposes, and they dont have the capability to lead the systems to the desirable optimum point. Given the development of multi-objective evolutionary algorithms and the increasing usage of them in engineering problem solving, the motivation for this research in optimizing energy systems based on multi-objective purposes, have been inconsistent and not co-scaled. In this paper, using genetic algorithms of sorting the elitist non-decisive solutions, the resource allocation was considered as a sample, which was modulated and analyzes. The present study shows that applying post-research decision-making methods based on multi-objective optimization, besides developing the possibility of analyzing the balance multi-objective, inconsistent and nonco-scaled purposes in energy systems, will also increase the power of policy-maker and politicians in predicting and improving the results from different decisions and support them in making better decisions. Applying other new optimizing methods in analyzing energy systems is recommended.
KEYWORDS:
Allocation.
Energy Policy, Multi-Objective Purpose Optimization, Genetic Algorithms, Resource
I.
INTRODUCTION
According to the recent studies in the framework of a general equilibrium model can be calculated 1 CGT; works carrying energy prices on economic sectors and inputs to put the experience and analysis. Developed general equilibrium pattern contains 6 cantons: 1-production, 2-energy, 3commercial sector, 4-production factors, 5-block prices, 6-market clearing block. The result shows soaring energy prices caused less relative price deviation and decrease indiscriminate energy utilization in terms of productivity and household's energy. On the other hand increasing productivity prices, inflation in economy, economic prosperity decreased in society by soaring energy prices, increase every countys incomes by growth energy aspects. Hence they can compensate for reduced welfare and developing atonement infrastructure; by astronomical prices. In short period of time by putting the people with exiguous incomes and directly paying subsides, compensated the low level of convenience.
CGT: Compated General Equilibrium 1 CES: Constant Elasticity Substitution 2
300
Vol. 3, Issue 1, pp. 300-305
Hence they can compensate for reduced welfare and developing atonement infrastructure; by astronomical prices. In short period of time by putting the people with exiguous incomes and directly paying subsides, compensated the low level of convenience. At the time, they have to put the efficient investment as an infrastructure for transportation and remarkably increasing public transportation, according to the recent studies, has got the high test fee, production expenditure and welfare caused it to scale down.
I.1
Calculated Relative Equilibrium Pattern Framework (CGE):
Developed general equilibrium pattern contains 6 blocks which every individuals of them covering several blocks; such as: 1-production, 2-energy, 3-commercial, 4- Production factors, 5- price and 6- market clearing.
I.1.1
Production block:
Manufacturing block in aimed pattern is well known as the three nested production function. Assuming an imperfect replacement of production factors from the production function of Constant Elasticity Substitution. [First Diagram] At the first level, production (ADI) that each part of the function is the two relative inputs (QINTI) and (QVAEI) is the total value added. [1] A D i = a i [ i Q V A E i i + (1 i ) Q IN T a i ] 1 / i (1)
Production Experiences
Goods (n) Internal bazaar
Rudimentary Productions
Goods (1)
Internal bazaar Finance Work Force
Exports
Rudimentary product Set one
Export
Synthetic Set of Energy
Energy (1)
Import Internal Energy (n) Import Internal

Fig.1 Structure of the production section
Standard reduced form for maximum benefit is the fixed proportion of inputs:
Q VAEi i P IN T i 1 + i ( 2) =[ * ] Q IN T i (1 i ) P IN T i
Wherein (PVAEI) and (PINTI) is the total value added price energy inputs and relative input prices for I section. It is worth to mention that the second formula (2) titled optimal condition CGS inputs function, and CGS followed the Oiler's laws: [2]
( P A D i ) A D i = ( P V A E i ) Q V A E i + ( P IN T i ) Q IN T i (3)
Which PDAI is the total production value, Therefore this is the demand for input function: [3]
301
Vol. 3, Issue 1, pp. 300-305
j = c1 .....cn
Q IN T j , i = a in, it Q IN Ti (4) j
Wherein AJI is the technical coefficient data and output, QINTJI: inputs produced by sector J and consume by sector I. Total value added-energy input(QVAE) assuming an imperfect replacement between initial factors & energy inputs in subdominant CES figure, function of the total value added (QVAI) and the whole energy inputs (QVEI).[4]
Q V A E i = a iva e [ iva e Q V Ai i
vac
+ (1 iva e ) Q V E i i ] 1 / i
va c
vae
( 5)
According to the first production function, optimal established input and following the function of CES 2 from the Oiler's law shall specify the demand and price: [5]
Q V Ai iva e P V E i 1 + iva e =[ ] ( 6) Q VEi (1 iva c ) P V Ai
( P V A E i )( Q V A E i ) = ( P V Ai )( Q V Ai ) + ( P V E i )( Q V E i ) (7)
Also, the total value added (QVAI), in subdominant form CES the workforce and capital (HSJ). [6]
2
from (QFFI) inputs which contains
Q V A i = a iv a
iv a e Q F f
va f ,i
va
va i
( 8)
And optimal input Equation (the first law), is obtained from the final factor prices and final production benefits Equality:
W F f , i = P V Ai a iva Q V Ai [
f
Q F f, i i ] 1 fvai Q F f, i i ,
va
( 9)
Wherein (WFFI), is the input prices for F which defined section I. [7, 8]
I.1.2
Energy section:
Six energy carries petrol, kerosene, fuel oil, LPG and electricity. Considering that the aim of this study is about investigating on mentioned carrier prices on various economic sections. Thus the CGE pattern has been developed to determine the prices. For this purpose the whole energy input function (QVEI), is a CES function from 6 inputs energy. [9]
Q VEi = a
ve i
iv e ve [ i Q F E i , e ] e
iv e
(10)
Wherein (QFEei) is the energy carriers and e=1, 2.........6 indices are for every individual carrier. The first step for choosing a beneficial energy carrier: Q V Ei (11) P D E i ,e = P E E i Q F E i,e Where in (PDEie), the price of every carrier and (PEEi) the whole energy input prices.
Q VEi = Q F E i, e Q F E i ,e
Q VEi = Q F E i, e Q F E i ,e
Ve i
Ve a iv e iv e iv e Q F E i, e i e
i V e
1 1
(12)
( 1)
By replacing above equation in equation number 10 we would have:

Ve i Q V E i iv e Q F E i, e i e And as a result of the first step we would have: Ve i
ve
(13)
302
Vol. 3, Issue 1, pp. 300-305
Q VEi (14) i Q VEi P D E i ,e = P D E i Q F E Ve ai And by calculating equation number 13 which is by the (QFEie), demand for carriers will be determined: [10, 11, and 12]
Ve 1 i ,e ve
Ve P D E i a i i ( iV e + 1 ) Q FEi Ve P E E i i And the whole energy carrier prices are defined as follows: P E E i Q V E i = P D E i ,e Q F E i ,e
iV e
(15)
(16)
I.2
Foreign Trade Block:
I.2.1
Export section:
In this section by equation of internal production (Xdi) to the internal market (XXDi) and export Qui, utilize the Constant Elasticity of transformation of equation [13]. One of the crucial properties of this function is the conversion between the production for foreign and internal market.
xd i = B i i q e i
I.2.1.1
iT
+ (1 i ) xxd
iT
(17)
Export Supply Equation:
Indeed, equation for export is implying, each section can produce two kinds of goods for the domestic market and foreign market. Thus the combination of manufactured goods and for domestic market and foreign for maximum sale:
p x i xd i = p e i q e i + pd i xxd i
To limit the export function (4-16 equation) will become:
(18)
pe 1 i qe i = xxd i i pd i i
1 1 iT
(19)
Wherein (Pxi) is the whole manufactured goods prices, Xdi is the amount of domestic production, (Quei) amount of export, (Pdi) prices of goods supplied have been manufactured and sold, XXDi is the domestic sales. [14, 15, 16]
I.2.1.2
Export Prices:
A unit for Export prices according to the definition of foreign export prices PWEI and TEI is for the rates and taxes for exports: [18, 19]
p e i = p w e i e x r (1 + te i )
(20)
This is very important to mention this point about the last equation that is assumed that the price of the export and the finance rate (related to the financial rules in every country) in flexible and the prices of the external export is constant. The main of the non-flexibility of the recent prices is related to be small economy compare with the world Economy that it has very far reaching. According to the absorption equation, the price of composite commodities equal to the sum of production:
pi xi = p d
xxd
+ pm
qm
(21)
Wherein PI is price of composite commodities, XI is the amount of domestic manufacturing production and export, QMI is the total amount of export. Imperfect replacement between internal and exporting production is one of the aspects that modern theories considered them. KURGMAN and HELPMANs Researches is in the frame work of various production, therefore implies on a specified manufacturing section which can export or import the production as they produce. [20]
303
Vol. 3, Issue 1, pp. 300-305
II.
ANALYSIS OF RESULTS AND DISCUSSION
The present study attempted to show in the general form of equating (CGE), the effects of the increasing the carrier energy prices into the sections and securitize this important factors in every sections of each country that is demonstrated in Fig.2. In this figure, 12 sections are showed to the effects of the accounting matrix in every year and we can see how the sections of each country can influence in the accounting of one year. [21]
Fig.2 Effects of each section into the accounting of financial economy
III.
CONCLUSIONS AND FUTURE SCOPE
Many are faced with staggering capital requirements for new plants, significant fluctuations in demand and energy growth rates, declining financial performance and political or regulatory and consumer concern about rising prices. Results shows that increasing of the energy carriers price with prepare of decreasing in the aberrance of relating prices, decreased the waste consume of the households. Until the beginning of the industrial revolution several centuries ago, this was the major environmental impact of human activities. Today, we are approaching the limit of available land for agricultural purposes, and only more intensive use of it can provide food for future increases of world population.
REFERENCES
[1]. Bhattacharyya, S.C. (1996). Applied General Equilibrium Models for Energy Studies: A Survey .Energy Economics 18: 145-164. [2]. Blitzer, C. R. (1986). Analyzing Energy-Economy Interactions in Developing Countries, Energy Journal pp.471-501. [3]. Boyd, R and N.D. Uri,(1993), The economic impact of taxes on refined petroleum products in the Philippines , Energy : The International Journal ,18,31-47. [4]. Boyd, R. and N.D. Uri, (1999). A note on the Economic Impact of Higher Gasoline and Electricity Prices in Mexico, Journal of Policy Modeling , Vol 21 , no.4,pp527-534.
304
Vol. 3, Issue 1, pp. 300-305
[5]. Boyd, Roy and Noel D. Uri (1991). The Impact of a Broad Based Energy Tax on the U.S. Economy. Energy Economics,13(4),258-273. [6]. Clarete, R.L. and J.A. Roumasset, (1986), CGE models and development policy analysis: problems, pitfalls, and challenges, American Journals of Agricultural Economics, 68, 1212-1216.[5] [7]. De Melo, j. (1988). Computable general equilibrium models for trade policy analysis in developing countries: A surve .Journal of Policy Modeling. 10:4.469-503. [8]. Dervis, IC,J.De melo and S. Robinson, (1982),General Equilibrium Models for Development Policy, A World Bank research publication, Cambridge University Press. [9]. Dixon,P.B.,Parmenter B. R,.Sutton,J. and Vincent,D.P.(1997),ORANI:A Multisectoral Model of the Australian Economy . North Holland press. [10]. Hazilla, M.and R.J. Kopp (1990), Social Coast of Environmental Quality Regulations: AgeneralEquilibrium Analysis , Journal of Political Economy, vol.98,no.4,853-873. [11]. Hudson, E, A. and Jorgenson D .W.(1974). U.S energy policy and economic growth, 1975-2000.Bell Journal of Economics and Management Science, 5:2, pp.461-514. [12]. Jorgenson, D.W. (1984). Economrtric Methods for Applied General Equilibrium Analysis, in H. Scarf and J. Shoven (eds.), Applied General Equilibrium Analysis, in H. Scarf and J. Shoven (eds), Applied General Equilbrium Analysis, Cambridge: Cambridge University press, 139-207. [13]. Jorgenson,D.W. and P.J. Wilcoxen, (1993a), Energy Prices, Productive and Economic Growth ,Annual Review of Energy and Environmen,18,343-395. [14]. Jorgenson. D.W. and P.J. Wilcoxen, (1993b),Reducing US carbon emissions: An econometric general equilibrium assessment, Resource and Energy Economics, 15,7-25. [15]. Lenjosek, G. and j. Whalley, (1985),A small open Economy Model Applied to an Evaluation of Canadian Energy Policies Using 1980 Data, Journal of Policy Modeling 8(1):89-110. [16]. Washington,D.C.:International Food Policy Research Institute. Manne, A.S. and R.G. Richels, (1977), ETA-MACRO: A model of energy economy [17] Interaction, in J. Hitch, (ed), Modeling Energy-Economy Interactions; Five Approaches, Research paper. No.5. Resources for the Future, Washington, DC. [18] Naqvi, Farzana(1998). A computable General Equilibrium Model of Energy , Economy and Equity Interaction in Pakistan. Energy Economics,20(4),347-373.National University. Mimeo. [19] Nornam, V.D. and J. Haaland (1987),VEMOD- a Ricardo Viner-Hechscher-Ohlin-Jones Model of Factor Price Determination, Scandinavian Journal of Economics, 89(3),251-70. [20] Resosudarmo, B. (2003). Computable General Equilibrium model on air pollution abatement policies with Indonesia as a case study. Economic Record 79 (0),63-73. [21] Robinson, Sherman. (1989). Multisecotral models in H.B. Chenery and T.N. Srinivasan, eds., Handbook of Development Economics Amsterdam, North Holland. Savings. Oxford University Press, London. Author Khashayar Teimoori was born in Tehran, Iran in 1992. Currently he is pursuing B.Tech-final year degree in the field of mechanical engineering from Islamic Azad University, Science and Research Branch of Tehran-Iran. Currently he is a manager of the Backstretch-team (In the fields of Robotics) in the address of www.backstretch-team.info, and he is a manager of webdesigning in the Rheosociety Group researches that it can be see in the address of www.rheosociety.com . Now he is a member in the technical society as ASME, ISME (Iranian society of mechanical Engineering), and IMS (Iranian Mathematical Society). His special interests are Computational Mechanics, Rheology, Robotics and special field of energy considerations.
305
Vol. 3, Issue 1, pp. 300-305
POWER CONDITIONING IN BATTERY CHARGERS USING SHUNT ACTIVE POWER FILTER THROUGH NEURAL NETWORK
1
P.Thirumoorthi, M IEEE, 2Jyothis Francis and 3 N.Yadaiah, SM IEEE

1&2
Dept. of Electrical and Electronics Engg., Kumaraguru College of Technology, Coimbatore, India 3 Dept. of Electrical and Electronics Engg., JNTUH College of Engg., Hyderabad, India
ABSTRACT
In this paper, single layer perceptron control algorithm is presented for a single phase shunt active power filter to improve the power factor and to reduce the source harmonics. This control scheme is based on a neural harmonic estimator. The reference current signal generated by this neural harmonic estimator is used to generate gating control pulses for active power filter switches. The control scheme have two control loops namely dc voltage regulation loop and current control loop. The performance of the proposed neural harmonic estimator is evaluated and compared with a linear reference generator based control scheme, incorporating the same voltage and current control loops as in the proposed controller. The proposed APF controller forces the supply current to be sinusoidal, with low current harmonics, and to be in phase with the voltage. Simulations are carried out using Matlab Simulink and the results show that the proposed system is capable of compensating the harmonic current to acceptable level.
KEYWORDS: Active power filter, Harmonics, single layer Perceptron, feedfoward Neural network, Selective
compensation.
I.
INTRODUCTION
Harmonic compensation have become increasingly important due to the intensive use of power converters and other nonlinear loads which results in the deterioration of power system voltages and current waveforms. Thus the current wave form can become quite complex depending up on the type of load and its interaction with other components in the system. One of the major effects of power system harmonics is to increase the current in the system. It also causes other problems like greater power losses in distribution and operation failure of protection devices. Due to these problems the quality of electrical power is an object of great concern. Power line conditioner like Active Power filter (APF) can be used to minimize the harmonic distortion current [1]-[3]. The main purpose of shunt active power filter is to supply the harmonic current absorbed by the system. To this end the control of APF systems has great importance. For the reference-signal generation, by means of direct method consists of sensing the load current and extracting the harmonic content [4]- [5]. As an alternative, the indirect method generates a sinusoidal reference signal by means of grid-voltage sensing. In that case, the grid current is forced to follow this sinusoidal signal, and thus, the load harmonics are indirectly given by the APF inductor current [6][8]. The linear reference generator based control strategy consists of two control loops and a resonant selective harmonic compensator. Reference signal is generated by indirect method [6]. That is by means of sensing the grid voltage. Resonant selective harmonic compensator is used for generating
306
Vol. 3, Issue 1, pp. 306-314
reference signal. It consists of several generalized integrators like second order band pass filter with high gain and low band width. Thus it will not affect the dynamics of control loop. The reference signal tracking is performed by the inner current loop. Outer voltage loop is accountable to regulate the capacitor voltage. The main advantages of this method are that good dynamic and transient performance obtained. By this approach most harmful harmonics from load current can be eliminated. In this method harmonics are processed individually. In order to improve the performance of the inner current loop, optimal, neural, and model reference adaptive controls have been used recently [13][15]. Other approaches utilize nonlinear regulators, such as sliding-mode control and hysteretic control [16], [17]. All the previously mentioned controls attenuate the current harmonics only to a certain level. In this paper, the design of shunt active power filter based on single layer feedforward neural network is presented. The proposed ANN control scheme consists of three control blocks namely voltage control, current control and a neural reference generator. The reference current signal generated by this neural reference generator contains the harmonic components that will be eliminated. This reference signal is controlled by means of a PI controller, which in turn, controls the pulse width modulation (PWM) switching pattern generator [5]-[8]. The output of the PWM generator controls the power switches. The neural reference generator and the other two control block play an important role in the dynamic response of the system. These blocks determine the accuracy and order of the harmonics to be injected. Neural reference generator has inherent learning capability that can give improved precision by interpolation unlike the standard look up table method and space vector modulation. The reference current can be determined using the distorted source current of the system [16]. This paper is organized as follows. Section II describes the active power filter topology. Section III presents the estimation of compensating current. Section IV explains the control of active filter. Section V verifies the expected features of both controllers by means of matlab simulation results. Section VI is conclusion.
II.
ACTIVE POWER FILTER FOR BATTERY CHARGER
Battery charger is used as the nonlinear load in the system shown in fig. 1. It draws a non sinusoidal current from supply. In this work, the load consists of a four diodes full bridge rectifier with a capacitor in parallel with a resistor in the dc side. Active power filters are the best known tool for the current harmonic compensation. Fig.1 shows the schematic diagram of a single phase active power filter in a closed loop manner. Diode bridge rectifier with resistor and capacitor connected in parallel act as a nonlinear load The APF can be controlled in a proper way to attain high resistance against higher order harmonics produced by the load. The active power filter through its control mechanism shapes the grid current to sinusoidal. The compensation principle of shunt active power can be explained as follows. Under normal condition the supply voltage can be represented as vs(t)= vmsinwt (1) But when the non linear load is connected to the supply, it will draw non sinusoidal current. Thus the load current will contain fundamental component and all other higher order harmonics. It can be represented as iL(t)=
n sin(nwt+n)
(2)
The shunt connected active power filter will generate a harmonic current iF(t) which compensate the harmonics present in the source current and make source current purely sinusoidal in nature.
s(
)=
F(
) +
L(
(3)
The compensation current iF is exactly equals to the harmonic content of the load current iL. Hence APF needs to calculate iF accurately and instantaneously.
307
Vol. 3, Issue 1, pp. 306-314
Fig. 1. Active power filter circuit
Fig. 2. Linear Reference generator
III.
ESTIMATION OF COMPENSATING CURRENT
This section deals with the reference current generating schemes, including the linear resonant selective harmonic compensator and proposed neural harmonic estimator.
3.1 Linear Reference Generator

In the basic approach of reference generation grid voltage vs is multiplied with k(t) [6]-[8]
Isref = vs.k(t). (4)
In this method reference signal will follow the sensed grid voltage. It will result in the reflection of distortions in the grid voltage. In order to overcome this drawback, a controller with linear harmonic compensator can be used which is as shown in Fig 2. This generator uses a band pass filter G1(s) and a harmonic compensator G2(s). Grid voltage vs and supply current is is processed through these filters and produces the reference signal. Band pass filters of G2(s) have closed loop operation G1(s) =
( )
(5)
G2(s) =
(6)
Where is the damping factor, = 2f, and k is the gain at the fundamental frequency f. n can take the values of 3, 5, . . . N, where N is the highest current-harmonic component to be attenuated, and kn is the band pass gain of each filter.
3.2 Neural Harmonic Estimator

In this method, the linear reference generator scheme is replaced by artificial neural network (ANN) made up of single layer perceptron. The ANN is trained offline, using a set of training data generated
308
Vol. 3, Issue 1, pp. 306-314
by Fourier analysis of the source current. In neural networks, there are two main processes involvedtraining and testing. In the training process, the network is trained with suitable input and output patterns which is called data set, so that the outputs of the neural network approximate the target values for various input training patterns in the training set. In the testing process, the performance of the network is verified by using the data outside the training data set. Neural network for a harmonic component detection is consisted of 2-layers network which input layer =49 units, and a single output layer. Before feeding data to ANN, the source current signals are sampled at a uniform rate in a half cycle of voltage source as shown in Fig 3. So time values are discrete, k t with k=0, 1, 2.. and then given to the ANN for its training, together with the expected output. The input is given as a continuous variable. The input signal flows through a gain or weight. The weights can be positive or negative corresponding to acceleration or inhibition of the flow of signals. The summing node produces a weighted sum of signals with a bias value which is initially taken as zero. It then passes to the output through the transfer function which is usually nonlinear, such as sigmoid, inverse-tan, hyperbolic or Gaussian type. When a set of input values are presented to the ANN, step by step calculations are made in the forward direction to drive the output pattern. A cost functional given by the squared difference between the net output and the desired net output for the set of input patterns is generated and this is minimized by gradient descent method altering the weights one at a time starting from the output layer After training next is the testing phase. In this period we have value of weighing vectors and best output. Using these data we can find out the best value of reference current generated. Current regulation is performed using a PI controller Fig.3 shows the block diagram of APF control scheme based on neural reference generator. The neural reference generator will perform the task of current harmonic computation and generate reference current signal. This current will send to current control block which can be realized using a PI controller. The output of PI controller is a control voltage signal which produces gating pulses for inverter control through PWM generator.
Fig.3. APF control scheme based on neural harmonic estimator
IV.
CONTROL OF ACTIVE FILTER
Indirect control method includes an outer voltage control loop and inner current control loop. In the indirect control the reference signal is generated by sensing the grid voltage and grid current is forced to follow this sinusoidal signal. This will reduce the harmonics present in the grid current.[7]-[8].. Indirect current control scheme attenuate the current harmonics to a high level, while maintaining the stability. By this method more harmful harmonics can be easily attenuated. The reference current
309
Vol. 3, Issue 1, pp. 306-314
generated by this method contain information regarding dominating harmonic content that need to be eliminated.
4.1 DC Voltage Regulation Loop

Outer voltage loop is responsible for the capacitor voltage regulation. Fig 4 shows the block diagram of outer voltage loop. In the outer voltage loop, square of the capacitor voltage is compared with squared reference voltage. The output is regulated with the help of PI compensator [13]-[14]. Squared values are used in order to make the design of loop simpler. In order to reduce the ripple at the output of PI compensator, a LPF can be added. The cutoff frequency of LPF can be made less than twice the grid frequency
4.2 Current Control Loop

Inner current control loop will track the reference signal generated by the reference generator. Fig.5 shows the block diagram of current control loop. Through this current control, grid current is forced to follow the reference current created. The error signal is regulated using a PI compensator and control signal is generated. This control signal is used to generate gating pulses to shunt active power filter through PWM technique
Fig.4. Outer voltage loop
Fig.5 Current control loop. System Parameters are specified in Table I. Table I Symbol Vs f Rs R C L C1 System Parameter Grid Voltage Grid frequency Nonlinear load series resistance Nonlinear load resistance Nonlinear load capacitance Active filter inductance Active filter capacitance Value 230V 50Hz 4 ohm 50 ohm 65 uF 15mH 1mF
V.
SIMULATION RESULTS
The active power filter circuit using neural reference generator and linear reference generator are established in Matlab Simulink environment. Simulink model using neural network is shown in Fig. 6. The system parameters used in these simulations are provided in Table 1. A system with supply voltage 230 V, 50 Hz is used. The DC reference voltage is set at 350V. The filter inductor is 15mH
310
Vol. 3, Issue 1, pp. 306-314
and the DC bus capacitor is 1mF. This control system is able to detect and eliminate most of the harmonics present in the source current
Fig. 6 simulation model of APF based on neural harmonic estimator
. Fig. 7 grid current waveform and FFT analysis without filter
311
Vol. 3, Issue 1, pp. 306-314
Fig. 8 Grid current waveform and FFT analysis with filter using linear reference generating control technique.
Fig.9 Grid current waveform and FFT analysis with filter using neural reference generator.
The nonlinear load consists of a series resistor Rs with an uncontrolled bridge rectifier connected to a capacitor C and a resistive load R. A proportional proportional-integral (PI) voltage loop has been used to set the proper magnitude of the line current. The proposed neural network based control strategy and linear controller have been tested in Matlab Simulink and results are compared. These results reveal, with the use of neural controller, the magnitude of the harmonic components is considerably reduced in the grid current. The proposed neural control system is able to detect the largest load harmonics and to compensate them properly. and Fig. 7 shows the distorted current waveform when a nonlinear load is connected across the grid. Through the analysis of the waveforms given it is clear that THD of grid current 26.28% when load is connected. It is reduced to 9.88% by the use of APF with linear control scheme. It i shown in Fig 8. % is Fig 9 shows that the neural controller is able to reduce this harmonics further, to 7.82%. Compared to the linear controller proposed neural controller has the advantage of increased overall efficiency with its high learning rate. It can adapt itself to compensate for variations in nonlinear current or nonlinear rning loads. Higher order harmonics are attenuated very efficiently using neural controller controller.
VI.
CONCLUSIONS
In this paper, an active power filter has been designed and the control methods have been presented. Proposed controller has been designed to mitigate the selective harmonic currents The shunt active currents. filter has been used to compensate a nonlinear load harmonic current. The filters parameters are made adaptive versus the grid current fluctuations. The inverter of the shunt active filter is current controlled by a PI controller whose reference is given by a neural filter. Compared to linear current control scheme, neural controller has high attenuation against high order harmonics. In linear current
312
Vol. 3, Issue 1, pp. 306-314
control conventional PI control loop that regulates the average level of the filter capacitor voltage and a resonant selective harmonic compensator are used. This controller is compared with neural network which shows the better performance in compensation of current harmonics.
ACKNOWLEDGEMENTS
The authors would like to thank the management of Kumaraguru College of Technology and JNTUH College of Engineering for providing infrastructure and support.
REFERENCES
[1] M. El-Habrouk, M. K. Darwish, and P. Mehta, Active power filters:A review, Proc. Inst. Elect. Eng. Elect. Power Appl., vol. 147, no. 5,pp. 403413, Sep. 2000. [2] T. C. Green and J. H.Marks, Control techniques for active power filters,Proc. Inst. Elect. Eng.Elect. Power Appl., vol. 152, no. 2, pp. 369381,Mar. 2005. [3] J. C.Wu and H. L. Jou, Simplified control method for single-phase active power filter, Proc. Inst. Elect. Eng.Elect. Power Appl., vol. 143, no. 3,pp. 219224, May 1996. [4] C. Y. Hsu and H. Y. Wu, A new single-phase active power filter with reduced energy-storage capacity, Proc. Inst. Elect. Eng.Elect. Power Appl., vol. 143, no. 1, pp. 2530, Jan. 1996. [5] H. Komurcugil and O. Kukrer, A new control strategy for single-phase shunt active power filters using a Lyapunov function, IEEE Trans. Ind. Electron., vol. 53, no. 1, pp. 305312, Feb. 2006 [6] B. N. Singh, Sliding mode control technique for indirect current controlled active filter, in Proc. IEEE Region 5 Annu. Tech. Conf., New Orleans, LA, Apr. 2003, pp. 5158. [7] D. A. Torrey and A. M. A. M. Al-Zamel, Single-phase active powerfilters for multiple nonlinear loads, IEEE Trans. Power Electron., vol. 10,no. 3, pp. 263272, May 1995. [8] V. F. Corasaniti, M. B. Barbieri, P. L. Arnera, and M. I. Valla, Hybrid active filter for reactive and harmonics compensation in a distribution network, IEEE Trans. Ind. Electron., vol. 56, no. 3, pp. 670 677,Mar. 2009 [9] P. Kumar and A. Mahajan, Soft computing techniques for the control of an active power filter, IEEE Trans. Power Del., vol. 24, no. 1,pp. 452461, Jan. 2009. [10] L. Asiminoaei, F. Blaabjerg, S. Hansen, and P. Thogersen, Adaptive compensation of reactive power with shunt active power filters, IEEETrans. Ind. Appl., vol. 44, no. 3, pp. 867877, May/Jun. 2008 [11] M. Cirrincione, M. Pucci, G. Vitale, and S. Scordato, A single-phase shunt active power filter for current harmonic compensation by adaptive neural filtering, in Proc. 12th IEEE EPE-PEMC, Portoroz, Slovenia,Aug. 30Sep. 1, 2006, pp. 18301835. [12] M. Cirrincione, M. Pucci, and G. Vitale, A single-phase DG generation unit with shunt power filter capability by adaptive neural filtering, IEEE Trans. Ind. Electron., vol. 55, no. 5, pp. 20932110, May 2008 [13] G. Bhuvaneswari and M. G. Nair, Design, simulation, and analog circuit implementation of a three-phase shunt active filter using the I cosphi algorithm, IEEE Trans. Power Del., vol. 23, no. 2, pp. 12221235,Apr. 2008. [14] D. Casadei, G. Grandi, R. K. Jardan, and F. Profumo, Control strategy of a power line conditioner for cogeneration plants, in Proc. IEEE PES C,Charleston, SC, Jun. 1999, pp. 607612 [15]. Niebur D, Germond AJ. Unsupervised Neural Net Classification of Power System Static Security States. Int. Journal of Electrical Power and Energy System.1992; 14(2-3): 233-242. [16]M. Rukonuzzaman and M. Nakaoka, Single-phase shunt active power filter with adaptive neural network method for determining compensating current, in Proc. 27th Annu. IEEE IECON, Nov. 29Dec. 2 2001, vol. 3, pp. 20322037. [17] D. Gao and X. Sun, A shunt active power filter with control method based on neural network, [18] W.M. Grady, M.J. Samotyj, A.H. Noyola, Survey of Active Power Line Conditioning Methodologies, IEEE Trans. on Power Delivery, vol. 5, no. 3, 1990, pp. 1536-1542. [19] M. El-Habrouk, M.K. Darwish, Design and Implementation of Modified Fourier Analysis Harmonic Current Computation Technique for Power Active Filters Using DSPs, IEE Proc. Electr. Power Appl., vol. 148, no. 1, 2001, pp. 21-28. [20]Selective Harmonic-Compensation Control for Single-Phase Active Power Filter With High Harmonic Rejection Jaume Miret, Member, IEEE, Miguel Castilla, Jos Matas, Josep M. Guerrero, Senior Member, IEEE, and Juan C. Vasquez, IEEE Transactions On Industrial Electronics, Vol. 56, No. 8, August 2009. [21] Mohd Amran Mohd Radzi, Member, IEEE, and Nasrudin Abd. Rahim, Senior Member,IEEE Neural Network and Bandless Hysteresis Approach to Control Switched Capacitor Active Power Filter for Reduction of Harmonics ieee Transactions On Industrial Electronics, Vol. 56, No. 5, May 2009
313
Vol. 3, Issue 1, pp. 306-314
[22] Avik Bhattacharya, Student Member, IEEE, and Chandan Chakraborty, Senior Member, IEEE A shunt active power filter with enhanced performance using ann-based predictive and adaptive controllers ieee transactions on industrial electronics, vol. 58, no. 2, february 2011
Authors
P. Thirumoorthi received B.E. in Electrical and Electronics Engineering from Madras University, India, 1993, M. Tech. in Power Electronics and Drives from SASTRA University, India, 2002. He is working as Associate Professor in Kumaraguru college of Technology, India and currently he is pursuing Ph.D. in Electrical Engineering at J. N. T. University, Hyderabad, India. His research interest includes Power Quality and Power Filters.
Jyothis Francis received B.Tech. Degree in Electrical and Electronic Engineering from Calicut University, Kerala, India in 2010. She is currently pursuing M.E. (Power electronics and drives) in Kumaraguru College of technology, Coimbatore, Tamil Nadu, India. Her area of interest includes Power Electronics, Industrial drives and power quality improvement. N. Yadaiah received B.E. in Electrical Engineering from College of Engineering, Osmania University, Hyderabad, India, 1988, M. Tech. in Control systems from I.I.T. Kharagpur, India, 1991 and Ph.D. in Electrical Engineering from J. N. T. University, Hyderabad, India,2000. He received Young Scientist Fellowship (YSF) of Andhra Pradesh State Council for Science and Technology, in 1999. Currently he is Professor and Head in Electrical & Electronics Engineering at JNTUH College of Engineering, Hyderabad and holding two research projects. He has 75 publications to his credit in International journals/conferences. He has visited as Visiting Professor to University of Alberta, from May to July 2007. He is Fellow of Institution Engineers (India), Fellow of IETE (India), Senior Member of IEEE, and Life Member of Systems Society of India and ISTE. He is editorial board member to Journal of Computer Science (India). His research interest includes: Adaptive Control, Artificial Neural Networks, Fuzzy logic, Nonlinear Systems and Process Control.
314
Vol. 3, Issue 1, pp. 306-314
DATA LEAKAGE DETECTION

Archana Vaidya, Prakash Lahange, Kiran More, Shefali Kachroo & Nivedita Pandey
1
Department of Computer Engineering, S.V.I.T., Nashik, M.H., India
A BSTRACT
Modern business activities rely on extensive email exchange. Email leakages have become widespread, and the severe damage caused by such leakages constitutes a disturbing problem for organizations. We study the following problem: A data distributor has given sensitive data to a set of supposedly trusted agents (third parties). If the data distributed to third parties is found in a public/private domain then finding the guilty party is a nontrivial task to distributor. Traditionally, this leakage of data is handled by water marking technique which requires modification of data. If the watermarked copy is found at some unauthorized site then distributor can claim his ownership. To overcome the disadvantages of using watermark [2], data allocation strategies are used to improve the probability of identifying guilty third parties. The distributor must assess the likelihood that the leaked came from one or more agents, as opposed to having been independently gathered by other means. In this project, we implement and analyze a guilt model that detects the agents using allocation strategies without modifying the original data. The guilty agent is one who leaks a portion of distributed data. We propose data allocation strategies that improve the probability of identifying leakages. In some cases we can also inject realistic but fake data record to further improve our changes of detecting leakage and identifying the guilty party. The algorithms implemented using fake objects will improve the distributor chance of detecting guilty agents. It is observed that by minimizing the sum objective the chance of detecting guilty agents will increase. We also developed a framework for generating fake objects.
K EYW ORDS: Sensitive Data, Fake Objects, Data Allocation Strategies
I.
INTRODUCTION
Demanding market conditions encourage many companies to outsource certain business processes (e.g. marketing, human resources) and associated activities to a third party. This model is referred as Business Process Outsourcing (BPO) and it allows companies to focus on their core competency by subcontracting other activities to specialists, resulting in reduced operational costs and increased productivity. Security and business assurance are essential for BPO. In most cases, the service providers need access to a company's intellectual property and other confidential information to carry out their services. For example a human resources BPO vendor may need access to employee databases with sensitive information (e.g. social security numbers), a patenting law firm to some research results, a marketing service vendor to the contact information for customers or a payment service provider may need access to the credit card numbers or bank account numbers of customers. The main security problem in BPO is that the service provider may not be fully trusted or may not be securely administered. Business agreements for BPO try to regulate how the data will be handled by service providers, but it is almost impossible to truly enforce or verify such policies across different administrative domains. Due to their digital nature, relational databases are easy to duplicate and in many cases a service provider may have financial incentives to redistribute commercially valuable data or may simply fail to handle it properly. Hence, we need powerful techniques that can detect and deter such dishonest. We study unobtrusive techniques for detecting leakage of a set of objects or records. Specifically, we study the following scenario: After giving a set of objects to agents, the distributor discovers some of
315
Vol. 3, Issue 1, pp. 315-321
those same objects in an unauthorized place. (For example, the data may be found on a web site, or may be obtained through a legal discovery process.) At this point the distributor can assess the likelihood that the leaked data came from one or more agents, as opposed to having been independently gathered by other means. We develop a model for assessing the guilt of agents. We also present algorithms for distributing objects to agents, in a way that improves our chances of identifying a leaker. Finally, we also consider the option of adding fake objects to the distributed set.
II.
PROBLEM DEFINITION
Suppose a distributor owns a set T = {t1 ,tm } of valuable data objects. The distributor wants to share some of the objects with a set of agents U1,U2 , ,Un but does wish the objects be leaked to other third parties. An agent Ui receives a subset of objects Ri which belongs to T, determined either by a sample request or an explicit request, Sample Request Ri = SAMPLE ( T,mi ) : Any subset of mi records from T can be given to Ui. Explicit Request Ri = EXPLICIT ( T,condi ) : Agent Ui receives all the T objects that satisfy condi . The objects in T could be of any type and size, e.g., they could be tuples in a relation, or relations in a database. After giving objects to agents, the distributor discovers that a set S of T has leaked. This means that some third party called the target has been caught in possession of S. For example, this target may be displaying S on its web site, or perhaps as part of a legal discovery process, the target turned over S to the distributor. Since the agents U1,U2 , ,Un, have some of the data, it is reasonable to suspect them leaking the data. However, the agents can argue that they are innocent, and that the S data was obtained by the target through other means.
2.1 Agent Guilt Model

Suppose an agent Ui is guilty if it contributes one or more objects to the target. The event that agent Ui is guilty for a given leaked set S is denoted by Gi | S. The next step is to estimate Pr {Gi | S }, i.e., the probability that agent Gi is guilty given evidence S. To compute the Pr {Gi | S}, estimate the probability that values in S can be guessed by the target. For instance, say some of the objects in t are emails of individuals. Conduct an experiment and ask a person to find the email of say 100 individuals, the person may only discover say 20, leading to an estimate of 0.2. Call this estimate as Pt, the probability that object t can be guessed by the target. The two assumptions regarding the relationship among the various leakage events. Assumption 1: For all t, t S such that t the provenance of t is independent of the provenance of . The term provenance in this assumption statement refers to the source of a value t that appears in the leaked set. The source can be any of the agents who have t in their sets or the target itself. Assumption 2: An object t S can only be obtained by the target in one of two ways. A single agent Ui leaked t from its own Ri set, or The target guessed (or obtained through other means) t without the help of any of the n agents. To find the probability that an agent Ui is guilty given a set S, consider the target guessed t1 with probability p and that agent leaks t1 to S with the probability 1-p. First compute the probability that he leaks a single object t to S. To compute this, define the set of agents Vt = { Ui | t Ri } that have t in their data sets. Then using Assumption 2 and known probability p, we have Pr {some agent leaked t to S} = 1- p ..1.1
Assuming that all agents that belong to Vt can leak t to S with equal probability and using Assumption 2 obtain, , Pr { leaked to } = | | 1.2 0, Given that agent Ui is guilty if he leaks at least one value to S, with Assumption 1 and Equation 1.2 compute the probability Pr { Gi | S}, agent Ui is guilty,
316
Vol. 3, Issue 1, pp. 315-321
..1.3
2.2 Data Allocation Problem

The distributor intelligently gives data to agents in order to improve the chances of detecting a guilty agent. There are four instances of this problem, depending on the type of data requests made by agents and whether fake objects [4] are allowed. Agent makes two types of requests, called sample and explicit. Based on the requests the fakes objects are added to data list. Fake objects are objects generated by the distributor that are not in set T. The objects are designed to look like real objects, and are distributed to agents together with the T objects, in order to increase the chances of detecting agents that leak data.
The Fig. 1 represents four problem instances with the names EF, E , SF an S , where E stands for explicit requests, S for sample requests, F for the use of fake objects, and for the case where fake objects are not allowed. The distributor may be able to add fake objects to the distributed data in order to improve his effectiveness in detecting guilty agents. Since, fake objects may impact the correctness of what agents do, so they may not always be allowable. Use of fake objects is inspired by the use of trace records in mailing lists. The distributor creates and adds fake objects to the data that he distributes to agents. In many cases, the distributor may be limited in how many fake objects he can create. In EF problems, objective values are initialized by agents data requests. Say, for example, that T = {t1,t2 } and there are two agents with explicit data requests such that R1 = {t1 ,t2 } and R2 = {t1}. The distributor cannot remove or alter the R1 or R2 data to decrease the overlap R1 \ R2 . However, say the distributor can create one fake object (B = 1) and both agents can receive one fake object (b1 =b2 = 1). If the distributor is able to create more fake objects, he could further improve the objective.
2.3 Optimization Problem

The distributors data allocation to agents has one constraint and one objective. The distributors constraint is to satisfy agents requests, by providing them with the number of objects they request or with all available objects that satisfy their conditions. His objective is to be able to detect an agent who leaks any portion of his data. We consider the constraint as strict. The distributor may not deny serving an agent request and may not provide agents with different perturbed versions of the same objects. The fake object distribution as the only possible constraint relaxation The objective is to maximize the chances of detecting a guilty agent that leaks all his data objects. The Pr {Gj |S =Ri } or simply Pr { Gj | Ri} is the probability that Uj agent is guilty if the distributor discovers a leaked table S that contains all Ri objects. The difference functions , is defined as: , = Pr {Gi|Ri}- Pr {Gj|Ri} ..1.4 1) Problem definition: Let the distributor have data requests from n agents. The distributor wants to give tables R1, ..Rn to agents U1 , . . . ,Un respectively, so that
317
Vol. 3, Issue 1, pp. 315-321
Distribution satisfies agents requests; and Maximizes the guilt probability differences (i, j ) for all i, j = 1. . . n and i = j. Assuming that the Rt sets satisfy the agents requests, we can express the problem as a multi-criterion 2) Optimization problem: Maximize (. . . , ( i, j ), . . . ) i ! = j . 1.5 (over R1 , . . . ,Rn ) The approximation [3] of objective of the above equation does not depend on agents probabilities and therefore minimize the relative overlap among the agents as | | , . . . ) i != j ....1.6 Minimize (. . . , (over R1 , . . . ,Rn ) | | This approximation is valid if minimizing the relative overlap, maximizes ( i, j ).
2.4 Objective Approximation

In case of sample request, all requests are of fixed size. Therefore, maximizing the chance of detecting | | is equivalent to minimizing (|Ri Rj|). The a guilty agent that leaks all his data by minimizing, minimum value of (|Ri Rj|) maximizes (|Ri Rj|) and ( i, j ) , since (|Ri |) is fixed. If agents have explicit data requests, then overlaps |Ri Rj| are defined by their own requests and |RiRj| are fixed. Therefore, minimizing | Ri | j is equivalent to maximizing | Ri | (with the addition of fake objects). The maximum value of | Ri | minimizes (Ri ) and maximizes ( i, j ), since ( Ri Rj) is fixed. Our paper focuses on identifying the leaker. So we propose to trace the ip address of the leaker. The file is send to the agents in the form of email attachments which need a secret key to download it. This secret key is generated using a random function and send to the agent either on the mobile number used at registration or to the other global email service account such as gmail. Whenever the secret key mismatch takes place the fake file gets downloaded. To further enhance our objective approximation ip address tracking [5][6][7] is done of the system where fake object is downloaded. Various commands are available for getting ip address information. Ping, tracert, nslookup, etc anyone may be used to get it. The ip address is traced with the time so as to overcome problem of dynamic ip addressing. But as we are doing this for an organisation there is no problem of dynamic ip. Or else if looking for the ip address universally it is unique for that period of time therefore it can be traced to unique system of the leaker.
III.
ALLOCATION STRATEGIES
In this section the allocation strategies [1] solve exactly or approximately the scalar versions of Equation 1.7 for the different instances presented in Fig. 1. In Section A deals with problems with explicit data requests and in Section B with problems with sample data requests.
3.1 Explicit Data Request

In case of explicit data request with fake not allowed, the distributor is not allowed to add fake objects to the distributed data. So Data allocation is fully defined by the agents data request. In case of explicit data request with fake allowed, the distributor cannot remove or alter the requests R from the agent. However distributor can add the fake object. In algorithm for data allocation for explicit request, the input to this is a set of request R1,R2,,Rn from n agents and different conditions for requests. The e-optimal algorithm finds the agents that are eligible to receiving fake objects. Then create one fake object in iteration and allocate it to the agent selected. The e-optimal algorithm minimizes every term of the objective summation by adding maximum number bi of fake objects to every set Ri yielding optimal solution. Step 1: Calculate total fake records as sum of fake records allowed. Step 2: While total fake objects > 0 Step 3: Select agent that will yield the greatest improvement in the sum objective i.e. i =
318
Vol. 3, Issue 1, pp. 315-321
Step 4: Create fake record Step 5: Add this fake record to the agent and also to fake record set. Step 6: Decrement fake record from total fake record set. Algorithm makes a greedy choice by selecting the agent that will yield the greatest improvement in the sum-objective.
3.2 Sample Data Request

With sample data requests, each agent Ui may receive any T from a subset out of different ones.
Hence, there are different allocations. In every allocation, the distributor can permute T objects and keep the same chances of guilty agent detection. The reason is that the guilt probability depends only on which agents have received the leaked objects and not on the identity of the leaked objects. Therefore, from the distributors perspective there / |T| are different allocations. An object allocation that satisfies requests and ignores the distributors objective is to give each agent a unique subset of T of size m. The s-max algorithm allocates to an agent the data record that yields the minimum increase of the maximum relative overlap among any pair of agents. The s-max algorithm is as follows. Step 1: Initialize Min_overlap 1, the minimum out of the maximum relative overlaps that the allocations of different objects to Ui Step 2: for k {k |tk Ri} do Initialize max_rel_ov 0, the maximum relative overlap between Ri and any set Rj that the allocation of tk toUi Step 3: for all j = 1,..., n : j = i and tk Rj do Calculate absolute overlap as abs_ov |Ri Rj | + 1 Calculate relative overlap as rel_ov abs_ov / min (mi ,mj ) Step 4: Find maximum relative as max_rel_ov MAX (max_rel_ov, rel_ov) If max_rel_ov min_overlap then min_overlap max_rel_ov ret_k k Return ret_k It can be shown that algorithm s-max is optimal for the sum-objective and the max-objective in problems where M |T| and n < |T|. It is also optimal for the max-objective if |T| M 2 |T| or all agents request data of the same size. It is observed that the relative performance of algorithm and main conclusion do not change. If p approaches to 0, it becomes easier to find guilty agents and algorithm performance converges. On the other hand, if p approaches 1, the relative differences among algorithms grow since more evidence is needed to find an agent guilty. The algorithm presented implements a variety of data distribution strategies that can improve the distributors chances of identifying a leaker. It is shown that distributing objects judiciously can make a significant difference in identifying guilty agents, especially in cases where there is large overlap in the data that agents must receive.
IV.
RELATED WORK
The presented guilt detection approach is related to the data provenance problem [8]: tracing the lineage of S objects implies essentially the detection of the guilty agents. Suggested solutions are domain specific, such as lineage tracing for data warehouses [9], and assume some prior knowledge on the way a data view is created out of data sources. Our problem formulation with objects and sets is more general and simplifies lineage tracing, since we do not consider any data transformation from Ri sets to S. As far as the allocation strategies are concerned, our work is mostly relevant to watermarking that is used as a means of establishing original ownership of distributed objects. Watermarks were initially
319
Vol. 3, Issue 1, pp. 315-321
used in images, video and audio data [2] whose digital representation includes considerable redundancy. Our approach and watermarking are similar in the sense of providing agents with some kind of receiver-identifying information. However, by its very nature, a watermark modifies the item being watermarked. If the object to be watermarked cannot be modified then a watermark cannot be inserted. In such cases methods that attach watermarks to the distributed data are not applicable. Finally, there are also lots of other works on mechanisms that allow only authorized users to access sensitive data. Such approaches prevent in some sense data leakage by sharing information only with trusted parties. However, these policies are restrictive and may make it impossible to satisfy agents requests.
V.
CONCLUSION
In doing a business there would be no need to hand over sensitive data to agents that may unknowingly or maliciously leak it. And even if we had to hand over sensitive data, in a perfect world we could watermark each object so that we could trace its origins with absolute certainty. However, in many cases we must indeed work with agents that may not be 100% trusted, and we may not be certain if a leaked object came from an agent or from some other source. In spite of these difficulties, we have shown it is possible to assess the likelihood that an agent is responsible for a leak, based on the overlap of his data with the leaked data and the data of other agents, and based on the probability that objects can be guessed by other means. Our model is relatively simple, but we believe it captures the essential trade-offs. The algorithms we have presented implement a variety of data distribution strategies that can improve the distributors chances of identifying a leaker. We have shown that distributing objects judiciously can make a significant difference in identifying guilty agents, especially in cases where there is large overlap in the data that agents must receive. Our future work includes the investigation of agent guilt models that capture leakage scenarios that are not studied in this paper. For example, what is the appropriate model for cases where agents can collude and identify fake tuples? A preliminary discussion of such a model is available in. Another open problem is the extension of our allocation strategies so that they can handle agent requests in an online fashion (the presented strategies assume that there is a fixed set of agents with requests known in advance).
ACKNOWLEDGMENTS
We would like to thank Prof. S. M. Rokade, Head of Department, Computer Engineering Department, SVIT, Nashik for his encouragement and motivation to write this paper. Last but not least, we would like to thank all our friends and classmates for giving us timely suggestions.
REFERENCES
[1]. Rudragouda G Patil, Development of Data leakage Detection Using Data Allocation Strategies International Journal of Computer Applications in Engineering Sciences [VOL I, ISSUE II, JUNE 2011, [ISSN: 2231-4946]. [2]. S. Czerwinski, R. Fromm, and T. Hodes. Digital music distribution and audio watermarking. [3]. L. Sweeney. Achieving k-anonymity privacy protection using generalization and suppression, 2002. [4]. S. U. Nabar, B. Marthi, K. Kenthapadi, N. Mishra, and R. Motwani. Towards robustness in query auditing. In VLDB 06. [5]. Stevens Le Blond, Chao Zhang Arnaud Legout, Keith Ross, Walid Dabbous, Exploiting P2P Communications to Invade Users Privacy [6]. The best tools and methods to track down suspect IP addresses and URLs http://www.techrepublic.com/blog/networking/the-best-tools-and-methods-to-track-down-suspect-ipaddresses-and-urls/3456 [7]. How to Trace an IP Address http://www.wikihow.com/Trace-an-IP-Address [8]. P. Buneman, S. Khanna, and W. C. Tan. Why and where: A characterization of data provenance. In J. V. den Bussche and V. Vianu, editors, Database Theory - ICDT 2001, 8th International Conference, London, UK, January 4-6, 2001, Proceedings, volume 1973 of Lecture Notes in Computer Science, pages 316-330. Springer, 2001
320
Vol. 3, Issue 1, pp. 315-321
[9]. Y. Cui and J. Widom. Lineage tracing for general data warehouse transformations. In The VLDB Journal, pages 471480, 2001. Authors Archana S. Vaidya is currently working as an Assistant Professor in the Computer Engineering Department of Sir Visvesvaraya Institute of Technology (S.V. I.T), Chincholi, Nasik (INDIA). She received her Masters Degree in Computer Engineering from V. J. T. I., Mumbai University (INDIA) in 2010 & Bachelors Degree in Computer Engineering from Walchand College Of Engineering Sangli Shivaji University (INDIA) in 2002 .She has an academic experience of 10 years (since 2002). She has taught Computer related subjects at undergraduate level. Her areas of interest are Data Structures ,Database Management Systems ,Operating Systems & Distributed Systems . She is life member of ISTE (Indian Society Of Technical Education) She has published 06 papers in National Conferences & 4 papers in International Conferences & Journals. She has guided 15 B. E. projects. Kiran More was born in Ahmednagar, Maharashtra, India in year 1989. He is currently pursuing degree in Computer Engineering from Sir Visvesvaraya Institute of Technology, Nashik
Shefali Kachroo was born in Jammu & Kashmir, India in year 1990. She is currently pursuing degree in Computer Engineering from Sir Visvesvaraya Institute of Technology, Nashik
Nivedita Pandey was born in Mumbai, Maharashtra, India in year 1990. She is currently pursuing degree in Computer Engineering from Sir Visvesvaraya Institute of Technology, Nashik
Prakash Lahange was born in Nashik, Maharashtra, India in year 1990. He is currently pursuing degree in Computer Engineering from Sir Visvesvaraya Institute of Technology, Nashik
321
Vol. 3, Issue 1, pp. 315-321
EXTRACTION OF VISUAL AND ACOUSTIC FEATURES OF THE DRIVER FOR REAL-TIME DRIVER MONITORING SYSTEM
Sandeep Kotte
Department of Computer Science & Engineering, Dhanekula Institute of Engineering & Technology, Vijayawada, India.
ABSTRACT
Driving is one of the most dangerous tasks in our everyday lives. Statistics show that over the past couple of decades the majority of the accidents are not only due to the poor vehicle technical conditions but also due to the driver's inattentiveness. The major cause for the inattentiveness includes aspects such as drowsiness (sleepy), fatigue (lack of energy) and emotions/stress (for example sad, angry, joy, pleasure, despair and irritation). In order to improve attentiveness of the driver, the Indian Government has introduced regulations amongst others concerning driving time and rest periods for the drivers. Even though these regulations helped partially in improving the driver attentiveness, they largely ignore the real-time dynamic ergonomics which is influenced by diverse factors such as the traffic control strategies, road geometry, vehicle characteristics, changing traffic scenarios, weather, etc. Different approaches have been proposed for monitoring the driver states, especially drowsiness and fatigue. Fatigue is traditionally measured by observing the eyelid movements. The drowsiness is generally measured by analyzing either head movements patterns or eyelid movements or face expressions or all the lasts together. Concerning emotion/stress recognition visual sensing of face expressions is helpful but generally not always sufficient. Therefore, one needs additional information that can be collected in a non-intrusive manner in order to increase the robustness of the emotion/stress measurement in the frame of a non-intrusive monitoring policy. We choose and find acoustic information emanating from the driver to be appropriate in the analysis of the emotions of driver, provided the driver generates some vocal signals by speaking, shouting, crying, etc., which is not un-common during the driving process. From these acoustic signals, this work extracts the spatial and temporal acoustic features and correlates it to the emotions of the driver. In this paper, a demonstration on how one can distinguish the emotion based on acoustic features (or combination of features) by testing them over Berlin emotion database.
KEYWORDS: Feature Extraction, Fatigue, Classifier, LDA, Berlin Database
I.
INTRODUCTION
Driver monitoring plays a major role in order to assess, control and predict the driver behavior. The research concerning driver monitoring systems was started nearly from the 1980's. In the first stages of this research, researchers developed driver monitoring systems based on inferring both driver behavior and state from the observed/measured vehicle performance. In a following generation, driver state and behavior has been directly assessed by intrusive systems measuring the physiological characteristics. But these techniques require driver cooperation since they are intrusive. Further, they may also disturb the driver behavior. And more recently, a significant research has been focusing on developing non-intrusive techniques, generally based on machine vision, directly measure in a nonintrusive manner the driver state.
1.1. Theoretical background

Due to increased number of speech driven applications in the recent years the automatic assessment of emotions from the drivers speech signal has become a research interest [1][2]. Such assessment
322
Vol. 3, Issue 1, pp. 322-333
provides information about the driver's satisfaction with the cars infotainment system and in particular, it improves efficiency and friendliness of human-machine interfaces. Moreover it reflects the driver's perception of the traffic situation and thus reveals his/her stress level. Therefore the emotional state also affects the driving capability which is of utmost importance for the safety of all occupants [4][3]. It allows for monitoring of physiological state of individuals in several demanding work environments, can be used to augment automated medical or forensic data analysis systems [5]. The performance of such a system is studied in the acoustically demanding environment of vehicular noise while driving. These systems mainly focus on (a) electro physiological data (e.g. EEG: [7][6]), and (b) behavioural expression data (gross body movement, head movement, mannerism, and facial expression; [8]) in order to characterize the user state. But these electrode-(EOG/EEG reaching 15% error rate for fatigue detection; [9]) or video based instruments still do not fulfill the demands of an everyday life measurement system. They have some shortcomings like (a) lack of robustness against environmental and individual-specific variations (e.g. bright light, wearing correction glasses, and angle of face or being of Asian race) and (b) lack of comfort and longevity due to electrode sensor application. In contrast to these electrode or video-based instruments, the utilization of voice communication as an indicator for emotional states matches the demands of everyday life measurement. Contact free measurements as voice analysis are non-obtrusive (not interfering with the primary driving task) and favorable for emotion detection, since an application of sensors would cause annoyance, additional stress and often impairs working capabilities and mobility demands. In addition, speech is easy to record even under extreme environmental conditions (temperature, high humidity and bright light), requires merely cheap, durable, and maintenance free sensors and most importantly, it utilizes already existing communication system hardware. Furthermore, speech data is omnipresent in many professional driver settings. Given these obvious advantages, the renewed interest in computational demanding analyses of vocal expressions has been enabled just recently by the advances in computer processing speed [12][13][10][11]. The first investigations to emotion detection from speech were conducted around the mid of the 1980s using statistical properties of certain acoustic features [15][14]. Later, the evolution of computer architectures introduced the detection of more complicated emotions from the speech. The research towards detecting human emotions is increasingly attracting the attention of the research community [15]. Nowadays, the research is focused on finding powerful combinations of classifiers that increase the classification efficiency in real-life speech emotion detection applications. Some of these techniques are used to recognize the frustration of a user and change their response automatically. Speech based emotion detection has lots of useful applications. Some of them are human-robotic interfaces [16], smart call-centers [18][19][17], intelligent spoken tutoring systems [20], spoken dialog research. The emotions can be observed from information about the language, what we say and how we say it. How we say something is more important than what we say. In the literature [21], the main focus is on the phonetics and acoustic properties of the affective spoken language. The emotions like anger, joy and sadness affect the driver attentiveness and are therefore relevant for the driving process. Thus we will devote particular attention in this research to those emotions that are badly/negatively affecting the driver behaviour [22]. The important voice features to consider for emotion classification are: Fundamental frequency (F0) or Pitch, Intensity (Energy), Speaking rate, Voice quality and many other features that may be extracted/calculated from the voice information are the formants, the vocal tract cross-section areas, the MFCC (Mel Frequency Cepstral Coefficient), Linear frequency cepstrum coefficients (LFCC), Linear Predictive Coding (LPC) and the teager energy operator-based features [15]. Certain features in the voice of a person can be used to infer the emotional state of the particular speaker. The real-time extracting the voice characteristics conveys emotion and attitude in a systematic manner and it is different from male and female.
323
Vol. 3, Issue 1, pp. 322-333
Table 1: Some variations of acoustic variables observed in relation to emotions [21] Emotion Anger Joy Pitch High mean Increased mean Normal or lower Decreased than normal mean Intensity Wide increased range Increased range Speaking rate Increased Increased Voice quality Breathy; blaring timbre Sometimes breathy; moderately Blaring timbre Resonant timbre
Sadness
Narrow range
Slow
Feature extraction from visual and acoustic information of the driver is an important and basic task to know the driver behavior/emotions. Generally, a feature is a set of measurements. Each measurement contains a piece of information, and specifies the property or characteristics of the object present in the given input. In the daily life, humans are able to guess and understand the state of the other persons by observing multiple features such as body action, voice information and the interpreted knowledge of understanding what he is saying and how he says it. Observing and extracting the multiple features is less complex task (in some cases it is obvious) for humans, for machines it is much more complex. This work mainly focuses on (a) identifying the useful features in the acoustic information (b) extracting the features in real time without missing the frames and (c) correlate the features to parameters/dimensions of the extended ergonomics status vector.
II.
NON-INTRUSIVE FEATURE EXTRACTION APPROACHES
Feature extraction is used to reduce the large input data into smaller data and it converts the data into small features sets or feature vectors (n-dimensional vector to store numerical features which represents an object). Feature extraction is defined as the process of extracting the feature from a source data, where the data can be embedded from high dimensional data set [23]. We calculate different feature sets for different applications. For computer vision applications edges, corners are calculated as features for images. Features like data and noise ratio, length of sound and relative power are calculated for pattern recognition applications.
2.1. Feature Extraction from the Visual Information

In 1990s, researchers introduced appearance based linear subspace techniques, statistics related techniques, to reduce the dimensionality and to extract the useful visual features. The introduction of the linear subspace techniques is a milestone in the visual feature extraction concept. The performance of appearance based techniques heavily depends upon the quality of the extracted features from image [23]. The appearance based linear subspace techniques extract the global features, as these techniques use the statistical properties like the mean and variance of the image [24]. The major difficulty in applying these techniques over large databases is that the computational load and memory requirements for calculating features increase dramatically for large databases [24]. In order to increase the performance of the feature extraction techniques, the nonlinear feature extraction techniques are introduced. In order to improve the performance of the emotion recognition systems, we have to extract both linear and nonlinear features. We have many nonlinear feature extraction techniques, such as radon transform and wavelet transform. The radon transform based nonlinear feature extraction gives the direction of the local features. This process extracts the spatial frequency components in the direction of radon projection is computed [25]. When features are extracted using radon transform, the variations in this facial frequency are also boosted [25]. The wavelet transform gives the spacial and frequency components present in an image. The performance of these feature extraction approaches are systematically evaluated in our previous work over FERET database for face recognition application as shown in Figure. 1.
324
Vol. 3, Issue 1, pp. 322-333
Figure 1: On Performance comparison of different face recognition approaches with profile right images [25].
2.2. Feature Extraction from the Acoustic Information

Speech is easy to record even under extreme environmental conditions (temperature, high humidity and bright light), requires merely cheap, durable and maintenance free sensors and most importantly, it utilizes already existing communication system hardware. Furthermore, speech data is omnipresent in many professional driver settings. Given these obvious advantages, the renewed interest in computational demanding analyses of vocal expressions has been enabled just recently by the advances in computer processing speed. The first investigations to emotion recognition from speech were conducted around the mid of the 1980s. Due to increased number of speech driven applications in the recent years the automatic assessment of emotions from the drivers speech signal has become a research interest [26][27]. Such assessment provides information about the drivers satisfaction with the cars infotainment system and in particular, it improves efficiency and friendliness of humanmachine interfaces. For the machine based state estimation, we will not focus on the voice content but rather on voicesignal features that are relevant for an emotional state inference. In this regard, to make the system more robust in predicting the driver state we will analyze the acoustic information such as pitch, intensity, speaking rate, voice quality, etc. in order to extract appropriate features [27]. Acoustic features extraction is challenging in many aspects as it highly depends on the age and gender of the person. The acoustic features are quite varying for different age groups and different gender. Angry males show higher levels of energy than angry females. It is found that males express anger with a slow speech rate as opposed to females who employ a fast speech rate under similar circumstances [27]. Acoustic information will however be precious, whenever available, to better assess and understand the effect of driving process ergonomics on the driver state and mood. The real-time extraction of acoustic characteristics from the voice signal conveys emotion and attitude in a systematic manner and it is different from male and female. Acoustic features are used to recognize the frustration of the driver (i.e. recognize vocal signals by shouting, crying, etc which are not un-common during the driving process). The performance also depends upon the given (input) acoustic information. In this work, input data is taken from audio sensors like microphone and the feature extraction algorithms are executed to extract the features in real time. Features are extracted from the real time data by performing time and frequency domains algorithms [23]. These algorithms extract temporal, spectral features and cepstral coefficients. These features are extracted based on the amplitude and spectrum analyzer of the audio data. The basic approach to the extraction of acoustic features is frame blocking such that a stream of given input audio signal is converted to a set of frames. And the time duration of each frame is about 10~30ms. If the frame size is shorter than 10ms we may miss some important information and sometimes we cannot extract valid acoustic features. If the frame size is longer than 30ms redundancy may occur and we cannot catch the time-varying characteristics of the audio signals. The important voice features to consider for emotion classification are: Fundamental frequency (F0) or Pitch, Intensity (Energy), Speaking rate, Voice quality and many other features that may be
325
Vol. 3, Issue 1, pp. 322-333
extracted/calculated from the voice information are the formants, the vocal tract cross-section areas, the MFCC (Mel Frequency Cepstral Coefficient), Linear frequency cepstrum coefficients (LFCCs), Linear Predictive Coding (LPC) and the teager energy operator-based features [27]. Pitch is the fundamental frequency of audio signals, which is equal to the reciprocal of the fundamental period. It can also be defined as the highness or lowness of a sound. Generally for pitch estimation wavelet transforms is used. The shape of the vocal track is modified depending up on the emotion [28]. The MFCC is spectrum of the spectrum used to find the number of voices in the speech. It has been proven beneficial in speech emotion detection, and speech detection tasks. The teager energy operator is used to find the number of harmonics due to nonlinear air flow in the vocal track [29][30]. The LPC provides an accurate and economical representation of the envelope of the short-time power spectrum. One of the most powerful speech coding analysis techniques providing very accurate estimates of speech parameters and is known as being relatively efficient for computation at the same time. The LFCC is similar to MFCC but without the perceptually oriented transformation into the Mel frequency scale; emphasize changes or periodicity in the spectrum, while being relatively robust against noise. These features are measured from the mean, range, variance and transmission duration between utterances [26][27]. To calculate these voice features, different techniques are used. The variation in the feature set for Happy is shown in Figure. 2 and Figure. 3 respectively.
Figure 2: Extracted features from the joy emotional audio file.
Figure 3: Detected emotions in 58 seconds representing in a pie.
326
Vol. 3, Issue 1, pp. 322-333
III.
EMOTION CLASSIFICATION
Emotion detection has been implemented on a variety of classifiers including Fishers linear discriminant analysis (FLDA), maximum likelihood classifier (MLC), neural network (NN), k-nearest neighbor (k-NN), and Gaussian mixture model (GMM) [27]. The main criterion in evaluating the effectiveness of the classification algorithm in this work is the scalability of the classifier. By considering the importance of scalability of the classification algorithm, the algorithms are evaluated over the open source Berlin database. For validation purposes, we create the test and training set databases using Berlin Database [31]. A training set database contains 535 utterances speech recordings 10 actors: 5 males, 5 females 10 utterances (in German) by each voice Seven different emotional speeches (anger, joy, neutral, boredom, fear, sadness and disgust)[31]. The test database contains different emotions of the persons present in the training database. The backend classifier is working with a fixed length feature vector and is used for classification as shown in Figure. 4.
Figure 4: The overall feature extraction approach from acoustic information.
Linear discriminant analysis (LDA) classifier is used for feature selection (i.e. converts the variable large dimensional feature vectors into the fixed small dimensional feature vector). LDA uses information about the class [23]. A class contains one person with different emotions. LDA tries to maximize the between class variance and minimize the within class variance. In other words, it decreases the distance between same class files and increases the distance between different class files [23]. Because of that LDA easily recognizes the emotions among large databases. The success rate of the different popular classifiers is compared in Table 2.
Table 2: Major Classification techniques and their Success rates Classification techniques Success rate Linear discriminant analysis (LDA) k-nearest neighbor (k-NN) neural network (NN) maximum likelihood classifier (MLC) Gaussian mixture model (GMM) 67.22% 63.33% 57.78% 55% 53.33%
3.1. Proposed Classifier

The LDA performs considerably better when compared to above classifiers for Berlin database. But the performance of LDA is also not sufficient for real world applications.
327
Vol. 3, Issue 1, pp. 322-333
The scalability of the real world emotion recognition system is limited, as the computational load and memory requirements increase dramatically with the large data sets. So scalability is major issue here. Up to now, in LDA we are using linear metric (Euclidean distance). Linear metric cannot compare different dimensions of the acoustic feature vectors accurately. To improve the performance (success rate and process speed), we propose the nonlinear Hausdorff metric based LDA. A special nonlinear metric Hausdorff distance which is able to compute the distance between different sized matrices having a single common dimension, like the acoustic matrices representing our acoustic feature vectors [32][33].
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Gaussian mixture model (GMM) maximum likelihood classifier (MLC) neural k-nearest Linear network (NN) neighbor (k- discriminant NN) analysis (LDA) LDA using Hausdorff distance Success rate
Figure 5: The graphical representation of the success rate for different classifiers.
Hausdorff distance measure is often used in content-based retrieval applications [32][33]. Hausdorff distance is meant as a measure between two point collection A and B in a metric space S (whose distance is d), it can be viewed as dissimilarity measure between two feature vectors A and B. By considering the hausdorff distance measure instead of the linear eculidean distance measure, the success rate of the LDA algorithm is increased by around 20 % as shown in Figure. 5.
3.2. Summary
In this chapter, Linear Discriminant Analysis (LDA) is explained in detailed. LDA only uses a second order statistics. LDA is using any information about the class. It tries to maximize the between class variance and minimize the within class variance.
IV.
SYSTEM ARCHITECTURE & IMPLEMENTATION RESULTS
The overall architecture of the system is shown in below Fig: 6. The architecture depicts the process of transforming given input speech signals to driver emotions.
328
Vol. 3, Issue 1, pp. 322-333
Figure 6: The overall architecture of the emotion detection system. (a)Training
4.1. Signal Preprocessing

As the input data is recorded using audio sensors like microphone, the recorded data may be affected by noise due to the weather conditions or any other disturbances. To reduce the noise affect, we performed filter operations which also optimize the class separability of features. This filter operation is performed with pre-emphasis high pass filter. The main goal of pre-emphasis is to boost the amount of energy in the high-frequencies. Mainly boosting is used to get more information from the higher formants available to the acoustic model and to improve the phone recognition performance. Lower frequencies have more energy in voiced segments compared to higher frequencies and this is called spectral tilt [34]. The Figure.7 is a plot of the frequency response of before & after filtering.
Figure 7: Frequency Response of Signal Preprocessing High Pass Filter (a) original
4.2. Frame Blocking and Windowing

In the window operation, the large input data is divided into small data sets and stored in sequence of frames. While dividing, some of the input data may be discontinuous. So to achieve the continuity the sequences of frames are overlapped. This window operation is performed using hamming window method to reduce the spectral leakage in the input data.
329
Vol. 3, Issue 1, pp. 322-333
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 4.3. Zero Crossing Rate and Short Time Energy
In time domain, the input data is recorded for every instance of time. The audio features in time domain are calculated based on the amplitude of the audio data and the variation of amplitude. These audio features are also known as temporal features. The rate at which zero crossings occur is a simple measure of a signal. Zero-crossing rate is measure of number of times in a given time interval/frame that the amplitude of the speech signals passes through a value of zero. The amplitude of the speech signal varies with time. Generally, the amplitude of unvoiced speech segments is much lower than the amplitude of voiced segments. The energy of the speech signal provides a representation that reflects these amplitude variations [35][36].
4.4. Mel frequency cepstral coefficient (MFCC)

MFCC is based on human hearing perceptions which cannot perceive frequencies over 1Khz. In other words, in MFCC is based on known variation of the human ear's critical bandwidth with frequency. MFCC has two types of filter which are spaced linearly at low frequency below 1000 Hz and logarithmic spacing above 1000Hz. A subjective pitch is present on Mel Frequency Scale to capture important characteristic of phonetic in speech. It turns out that humans perceive sound in a highly nonlinear way. Basic parameters like pitch and loudness highly depend on the frequency, adding weight to components at lower frequencies. In Fig: 8 this behaviour is illustrated, relating the perceived pitch to the physical frequency. The pitch associate with a tone is measured on the so-called mel-scale (By definition 1,000 mels correspond to the perception of a sinusoidal tone at 1000kHz, 40dB above the hearing threshold). The graph clearly shows that the perceived pitch increases all the more slower as we go to higher frequencies. Essentially we observe a logarithmic increase that is illustrated by the almost linear curve (with respect to a logarithmic scale) in Figure: 8 at high frequencies. MFCCs extensively use this property and add weight to lower frequencies, because more discriminative information can be found there.
Figure 8: Relation between the perceived pitch and frequency [38]
4.5. Pitch Extraction

Pitch is the fundamental frequency of audio signals, which is equal to the reciprocal of the fundamental period. This is mainly explained in terms of highness or lowness of a sound. Pitch in reality can be defined as the repeat rate of a complex signal, i.e., the rate at which peaks in the autocorrelation function occur. The three main difficulties in pitch extraction arise due to following factors. Vocal cord vibration does not necessarily have complete periodicity especially at the beginning and end of the voiced sounds. From speech wave, vocal cord source signal can be extracted but its extraction is difficult if has to be extracted separately from the vocal tract effects.
330
Vol. 3, Issue 1, pp. 322-333
The fundamental frequency possesses very large dynamic range.
4.6. Berlin Database of Emotional Speech (BDES)

It is developed by the Technical University, Institute for Speech and Communication, Department of Communication Science, Berlin [37]. It has become one of the most popular databases used by researchers on speech emotion recognition, thus facilitating performance comparisons with other studies. 5 actors and 5 actresses have contributed speech samples for this database it mainly has 10 German utterances, 5 short utterances and 5 longer ones and recorded with 7 kinds of emotions: happiness, neutral, boredom, disgust, fear, sadness and anger. The sentences are chosen to be semantically neutral and hence can be readily interpreted in all of the seven emotions simulated. Speech is recorded with 16 bit precision and 48 kHz sampling rate (later down-sampled to 16 kHz) in an anechoic chamber.
4.7. Summary
In this chapter, pitch, zero-crossing rate, short time energy and MFCC is explained in detailed. Zerocrossing rate and short time energy are calculated for the voiced and unvoiced signals. After adding energy, delta, and double delta features to the 12 cepstral features, totally 39 MFCC features are extracted. Again, one of the most useful facts about MFCC features is that the cepstral coefficients tend to be uncorrelated, which turns out to make our acoustic model much simpler.
V.
CONCLUSION
Acoustic information will be precious whenever available, to better assess and understand the effect of driving process ergonomics on the driver state and mood, since these are drawn in an non-intrusive manner. The acoustic features are quite varying for different age groups and different gender. For the machine based state estimation, we will not focus on the voice content but rather on voicesignal features that are relevant for an emotional state inference. In this regard, to make the system more robust in predicting the driver state we analyzed the acoustic information such as pitch, short time energy, zero crossing rate and MFCC etc. in order to extract appropriate features. From the literature emotion recognition based on acoustic information has been implemented on a variety of classifiers including Fisher's linear discriminant analysis (FLDA), maximum likelihood classifier (MLC), neural network (NN), k-nearest neighbor (k-NN), Bayes classifier, support vector classifier, artificial neural network (ANN) classifier and Gaussian mixture model (GMM) etc. An experimental result shows that the LDA classifier with linear metric produces 67.22% recognition rate. To improve the recognition rate later we used a special non linear metric called Hausdorff distance measure. The recognition rate is improved to 85% approximately. The overall system does determine driver's emotions such as anger, despair, pleasure, sadness, irritation and joy.
VI.
FUTURE WORK
The present system already chosen some of the features, but we can try with remaining features like voice quality and speaking rate etc. Although the current MFCC already given very good emotion recognition performance, a further exploitation may contribute to the development of even more powerful features. Moreover, the features will be tested under more complex real-world conditions (e.g. reverberant and noisy speech). To simulate emotion detection in cars, direct interfacing between MATLAB and the TMS board is required. TMS board is a micro-controller which is embedded with a pipelined digital signal processor for real-time signal processing.
REFERENCES
[1]. Christian Martyn Jones and Ing-Marie Jonsson. Automatic recognition of affective cues in the speech of car drivers to allow appropriate responses. In OZCHI '05: Proceedings of the 17th Australia conference on Computer-Human Interaction, pages 1-10, Narrabundah, Australia, Australia, 2005. Computer-Human Interaction Special Interest Group (CHISIG) of Australia.
331
Vol. 3, Issue 1, pp. 322-333
[2]. B. Schuller et al. Effects of in-car noise-conditions on the recognition of emotion within speech. In Proc. DAGA, 2007. [3]. Clifford Nass, Ing-Marie Jonsson, Helen Harris, and Ben Reaves, Jack Endo, Scott Brave, and Leila Takayama. Improving automotive safety by pairing driver emotion and car voice emotion. In CHI '05: CHI '05 extended abstracts on Human factors in computing systems, pages 1973-1976, New York, NY, USA, 2005. ACM. [4]. J. Healey and R. Picard. Smartcar: detecting driver stress. Volume 4, pages 218 -221 vol.4, 2000. [5]. J. G. Taylor, K. Scherer, and R. Cowie. Introduction: 'emotion and brain:Understanding emotions and modelling their recognition'. Neural Netw., 18(4):313-316, 2005. [6]. Sommer D. Holzbrecher M. Golz, M. and T. Schnupp. Automatic recognition of affective cues in the speech of car drivers to allow appropriate responses. In RS4C (Eds.), Proceedings 14th International Conference Road Safety on Four Continents, Bangkok, Thailand, 2007. [7]. Batliner A. Hnig, F. and E. Nth. Fast recursive data-driven multi-resolution feature extraction for physiological signal classification. In In J. Hornegger, et al. (Eds.): 3rd Russian-Bavarian Conference on Biomedical Engineering, pages 47-52, Erlangen, 2007. [8]. Robert Horlings, Dragos Datcu, and Leon J. M. Rothkrantz. Emotion recognition using brain activity. In CompSysTech '08: Proceedings of the 9th International Conference on Computer Systems and Technologies and Workshop for PhD Students in Computing, pages II.1-1, New York, NY, USA, 2008. ACM. [9]. Martin Golz, David Sommer, Mo Chen, Udo Trutschel, and Danilo Mandic. Fusion of state space and frequency domain features for improved microsleep detection. In W. Dutch et al. (Eds.), Proceedings International Conference Artificial Neural Networks (ICANN 2005), pages 753-7592, 2005. [10]. Anton Batliner, Stefan Steidl, Bjrn Schuller, Dino Seppi, Kornel Laskowski, Thurid Vogt, Laurence Devillers, Laurence Vidrascu, Noam Amir, Loic Kessous, and Vered Aharonson. Combining efforts for improving automatic classification of emotional user states. In Proc. IS-LTC 2006, Ljubliana, pages 240-245, 2006. [11]. P Juslin and K Scherer. Vocal expression of affect. The New Handbook of Methods in Nonverbal Behavior Research, January 2005. [12]. M. J. Owren and J.-A Bachorowski. Measuring emotion-related vocal acoustics. In J. Coan and J. Allen (Eds.). Handbook of emotion elicitation and assessment, pages 239-266. New York: Oxford University Press, 2007. [13]. Seppi D. Steidl S. Vogt T. Wagner J. Devillers L. Vidrascu L. Amir N. Kessous L. Schuller B., Batliner A. and Aharonson V. The relevance of feature type for the automatic classification of emotional user states: Low level descriptors and functionals. In Proceedings of Interspeech, pages 2253-2256, 2007. [14]. R. Van Bezooyen. Characteristics and Recognizability of Vocal Expressions of Emotion. Foris Pubns, USA, June 1984. [15]. Dimitrios Ververidis and Constantine Kotropoulos. Emotional speech recognition: Resources, features, and methods. Speech Communication, 48(9):1162 -1181, 2006. [16]. T. Kanda, K. Iwase, M. Shiomi, and H. Ishiguro. A tension-moderating mechanism for promoting speech-based human-robot interaction. Pages 511-516, aug. 2005. [17]. Englert R. Stegmann J. Burleson W Burkhardt F., Ajmera J. Detecting anger in automated voice portal dialogs. In Proceedings of Interspeech, Pittsburgh, 2006. [18]. Van Ballegooy M. Englet R. Huber R Burkhart, F. An emotion aware voice portal. In Proc. Electronic Speech Signal Processing ESSP, 2005. [19]. Laurence Devillers and Laurence Vidrascu. Real-life emotion detection with lexical and paralinguistic cues on Human-Human call center dialogs. Proc. INTERSPEECH' 06. Pittsburgh, 2006. [20]. Hua Ai, Diane J. Litman, Kate Forbes-riley, Mihai Rotaru, Joel Tetreault, and Amruta Pur. Using system and user performance features to improve emotion detection in spoken tutoring dialogs. In Proceedings of Interspeech, pages 797-800, 2006. [21]. J I. Murray, Arnott. Toward the simulation of emotion in synthetic speech: A review of the literature on human vocal emotion. The Journal of the Acoustical Society of America, 93(2):10971108, 1993. [22]. Thurid Vogt, Elisabeth Andre, and Johannes Wagner. Automatic recognition of emotions from speech: A review of the literature and recommendations for practical realisation. Pages 75-91, 2008. [23]. O. M. E. Wahlstrom, N. Papanikolopoulos, "Vision-based methods for driver monitoring," in Proceedings of the 6th IEEE International Conference on Intelligent Transportation Systems, Shanghai, China, 2003, pp. 903-908.
332
Vol. 3, Issue 1, pp. 322-333
[24]. A. E. M. C. Esra Vural, Gwen Littlewort, Marian Bartlett and Javier Movellan, "Drowsy Driver Detection through Facial Movement Analysis Springer Berlin / Heidelberg, vol. 4796, pp. 618, 2007. [25]. H. D. Vankayalapati and K. Kyamakya, "Nonlinear Feature Extraction Approaches for Scalable Face Recognition Applications," ISAST transactions on computers and intelligent systems, vol. 2, 2009. [26]. R. V. Bezooijen, "The Characteristics and Recognisability of Vocal Expression of Emotions," Foris, Drodrecht, the Netherlands, 1984. [27]. J. W. E. A. e. Thurid Vogt, "Automatic Recognition of Emotions from Speech: a Review of the Literature and Recommendations for Practical Realisation," in Affect and Emotion in HumanComputer Interaction: From Theory to Applications Berlin, 2008. [28]. J. A. I. Murray, "Toward the simulation of emotion in synthetic speech: A review of the literature on human vocal emotion," Journal of the Acoustical Society of America, vol. 93 (2), pp. 10971108, 1993. [29]. C. K. D. Ververidis, "Emotional speech recognition: Resources, features, and methods," presented at the Speech Communication, 2006. [30]. L. G. Yongjin Wang, "An investigation of speech-based human emotion recognition," pp. 15 18, 2004. [31]. M. R. A. Paeschke, W. Sendlmeier, B. Weiss, A Database of German Emotional Speech," Proc. Interspeech, 2005. [32]. T. Barbu, "Discrete speech recognition using a hausdorff based metric," in Proceedings of the 1st Int. Conference of E-Business and Telecommunication Networks, ICETE, Setubal, Portugal, 2004, pp. 363-368. [33]. T. Barbu, "Speech-dependent voice recognition system using a nonlinear metric," International Journal of Applied Mathematics, vol. 18, pp. 501-514, 2005. [34]. James H. Jurafsky, Daniel Martin. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition. Prentice Hall Ser (2ND), May 2008. [35]. Tobias Andersson. Audio classification and content description. Master's thesis, Lulea University of Technology, Multimedia Technology, Ericsson Research, Corporate unit, Lulea, Sweden, March 2004. [36]. Abdillahi Hussein Omar. Audio segmentation and classification. Master's thesis, Informatics and Mathematical Modelling, Technical University of Denmark, DTU, Richard Petersens Plads, Building 321, DK-2800 Kgs. Lyngby, February 2005. [37]. M. Rolfes W. Sendlmeier F. Burkhardt, A. Paeschke and B. Weiss. Berlin database of emotional speech on-line. In Interspeech: http://pascal.kgw.tu-berlin.de/emodb/index-1024.html, pages 1517-1520, 2005. Author Sandeep Kotte Graduated in Information technology from JNTU in 2007 and M.S in Information technology from the University of Klagenfurt, Austria in 2010 specialized in Intelligent Transportation System, pervasive computing and Business Informatics. Currently working as Assistant professor in Dhanekula Institute of Engineering & Technology, India.
333
Vol. 3, Issue 1, pp. 322-333
REVIEW ON THEORY OF CONSTRAINTS

CH. Lakshmi Tulasi1, A. Ramakrishna Rao2
Research Scholar, Dept of Mech. Engg, S.V.U.College of Engineering, Tirupati, A.P., India. 2 Professor, Dept of Mech. Engg, S.V.U.College of Engineering, Tirupati, A.P., India.
1
A BSTRACT
Theory of constraints (TOC) is a philosophy of management put forth by Eliyahu M. Goldratt, which claims that each system has at least one constraint. Although initially in manufacturing method, TOC has now developed into a theory of Management: a powerful systemic problem structuring and problem solving methodology which can be used to develop solutions with both intuitive power and analytical rigour. This philosophy is applied in many functional areas of companies, ranging from production flow management, marketing, services and project management to being a tool of logical reasoning. The present work aims at highlighting the extant literature with respect to the application of TOC and summaries important findings on the theory and practice of TOC.
K EYW ORDS:
management.
Theory of constraints, Production flow management, marketing, services, project
I.
INTRODUCTION
The theory of constraints (TOC) was invented by Dr. Eliyahu M. Goldratt in 1984 in his book The Goal[22], Israeli physicist turned business guru, developed a revolutionary method for production scheduling [21] which was in stark contrast to accepted methods available at the time, such as MRP. TOC is evolved from the OPT (Optimized Production Timetables) system and was later known under the commercial name of Optimized Production Technology (OPT) . Central to the TOC philosophy was that any organization or system has a constraint or a small number of constraints which dominate the entire system. The theory of constraints (TOC) adopts the common idiom "A chain is no stronger than its weakest link" as a new management paradigm. This means that processes, organizations, etc., are vulnerable because the weakest person or part can always damage or break them or at least adversely affect the outcome. The analytic approach with TOC comes from the contention that any manageable system is limited in achieving more of its goals by a very small number of constraints, and that there is always at least one constraint. Hence the TOC process seeks to identify the constraint and restructure the rest of the organization around it, through the use of five focusing steps. The secret to success lies in managing these constraints and the system as it interacts with these constraints, to get the best out of the whole system. The main motivation for the research review reported in this paper was the realization that TOC is growing very rapidly, and we simply did not know what was out here; i.e. what had already been tackled. This review paper will first outline the background to TOC, and then report on the application of paradigms of TOC in various fields and the findings. The theory of constraints (TOC) is a management philosophy that has been effectively applied to Manufacturing processes and procedures to improve Organizational effectiveness [25]. In its brief 25-years history, TOC has developed rapidly in terms of both methodology ([9], [12]) and area of applications ([33], [46]).
334
Vol. 3, Issue 1, pp. 334-344
Most of the work is carried out in the application of Logistics Paradigm in the field of Project Management. The literature concerning about the feeding buffers is well illustrated in ([24], [44], [36], [1], and [50]). The methodology used in the project management is critical chain project management to find the critical chain and to find the project buffers and the feeding buffers. Three TOC Paradigms that have evolved over the last twenty-five years Logistics, Global Performance Measures and thinking process ([2], [14]). More recently [4] have referred to these paradigms as decision making, Performance Measurement system and thinking Process. The review of literature on TOC is spanned from the Logistics Paradigm to the Thinking process Paradigm. In section 2, we reviewed the Logistics Paradigm and application of logistics paradigm in the various fields. Where as in section 3, we reviewed the Global performance Measure Paradigm and application of the paradigm in the services. Section 4, deals with the Thinking process Paradigm and the various fact finding tools in that paradigm and finally in section 5, this paper concludes with some directions for further research on TOC.
II.
LOGISTICS PARADIGM
Originally the Logistics Paradigms are looking for system constraints in order to increase throughput. The Logistics Paradigm of the TOC has evolved from the scheduling software called Optimized Production Technology (OPT) which in turn based on the following nine rules [24]. This included using Drum-Buffer-Rope (DBR) scheduling technique and the Five Focusing steps of TOC. The DBR methodology synchronizes resources and material utilization in an organization. Resources and materials are used only at a level that contributes to the organizations ability to achieve throughput. Because random disruption is inevitable in any organization, DBR methodology provides a mechanism for protecting total throughput of the system by the use of Time-Buffers (T-Bs).
2.1 Application of Logistics Paradigm

The logistics paradigm was applicable in Manufacturing, supply Chain, Project Management and in services [6]. In this Section we reviewed the application of Logistics Paradigm in Manufacturing, Supply Chain, and Project Management and in services (Table 1).
Table 1: application area of logistics paradigm and references Main area Sub area Manufacturing Logistics Paradigm Supply Chain Project Management Services Reference [34], [15], [10], [11], [18], [53], [62], [49], [19], [39], [55], [70]. [28], [48], [71]. [24], [64], [38], [40], [69],[44], [36],[1],[50], [26], [65], [47], [66], [17], [72], [61], [35], [51], [27],[52]. [31], [6], [67], [63], [30], [46], [16], [54], [57].
Literature concerning TOC in Manufacturing can be broken down into four categories. TOC, Materials requirement Planning (MRP), Manufacturing Resources Planning (MRP II) and JIT Manufacturing. The buffer stock allocation in serial and assembly type of production lines, the key concept in this is the Long Pull (Constant Work-In-Process inventory) Kanban System [34]. A comparison is made between the TOC and the JIT manufacturing in [15]. The Drum-buffer-Rope scheduling technique can be used in services as well as in manufacturing ([10], [11], [53], and [19]). The comparison of Traditional JIT and TOC Manufacturing in a Flow shop in [18]. A survey based comparison is made between performance and in change of performance of firms using JIT and TOC and suggested that greatest performance and improvement in performance occurred to adopters of TOC [55]. JIT and TOC buffering philosophies are compared and suggested that improved system performance stems from the strategic placement of buffers in DBR, which maximizes protection of the constraint from variation rather than attempting to protect each individual station is given by [70].The DBR technique can be used to manage supply chains ([28], [48], and [71]) and to improve current production operations [62]
335
Vol. 3, Issue 1, pp. 334-344
TOC project management process is a pull process that promotes low WIP level, time buffers to protect the critical chain of tasks and resources, and no multitasking- which is defined as splitting a workers time between two or more priority projects, to the detriment of the overall schedules. Literature concerning TOC in Project Management in Logistics Paradigm is that it was first popularized by the novel, The Goal [22] that applied the principles to operations management. Since 1997 it has found applications in two areas within project management. The first application is scheduling of a single project to reduce project duration and simply project control. This is the main theme of the novel Critical Chain [24]. A practical approach for implementing critical chain Management is suggested by [38]. The application of TOC in the scheduling of single project is investigated [64]. The Theory of Constraints Management system is investigated [40]. Planning and controlling multiple, simultaneous, independent projects in the resource constrained environment [69]. The TOC Project Management approach is to schedule all non-critical activities as late as possible, but with buffers. The objective of these buffers (called feeding buffers because they are placed where non-critical paths feed into the critical chain) is to prevent delay of the execution of work on the critical chain when work on a non-critical path is delayed. The illustration of the literature of these feeding buffers is given by ([24], [44], [36], [1], and [50]).A detail information of how these five steps provide the framework for coordinating and controlling activities in both manufacturing and project environment [26]. The literature review on application of TOC in Project Management is given by [65]. Critical chain scheduling is not only a technique for the development and tracking of project schedules it is coherence and comprehensive approach to project management that encompasses and affect other processes and practices associated with project management as well [47]. The TOC approach can also be applied to the other areas of Project Management such as project cost management and project risk management [66]. The application of theory of Constraints Multi-Project Management at the Boeing Commercial Airline Company, Highlighting project end date predictability, modified scheduling algorithm and balancing resources across multiple projects [17]. The comparison and contrast of the advantages and disadvantages of traditional project management and TOC project management [72]. Several examples of how placing buffers or blocks of unscheduled time to account for delays allows project manager to minimize their risks and achieve project success [61]. A number of sources of bias in performance of project to schedule and cost estimates and provides recommendations to size buffers that ensure your projects come in under your baseline schedule and budget [35]. Although Critical Chain Project Management has a number of valuable concepts, it does not provide a complete solution to project management needs and the organizations should be very careful about the exclusion of conventional project management techniques [51]. A development of five-step innovative approach for the application of the management philosophy Theory of Constraints is suggested by [27]. An algorithm to determine critical sets and critical clouds and applied to a sample project and the results are presented in a condensed, project manager friendly, graphical format [52]. The Logistics Paradigm in services is that the five step focusing process has been applied to processes and procedures within services. The focusing steps have been used to improve logistics functions with the military [67]. The five focusing steps can be used in re-engineering of administrative functions [63]. The five step process has been used to improve information flow ([31], [6], and [16]. It has been used to improve service times [46]. It has been used to improve sales [30]. It has been used in medical settings [54]. The drum- buffer-rope scheduling technique can be used in services as well. While manufacturing DBR uses to schedule machinery, services may use DBR to schedule people within the organization, to set appointments for customers, or to predict lead times for customers. The buffer management can be used to identify pr4oblems and weaknesses that will cause disruption to the system [57].
III.
GLOBAL PERFORMANCE MEASURES PARADIGM
Under Toc all company performance measures are driven by the global goal of making money now and in the future. Throughout the methodology/ philosophy, the three measurements of throughput, inventory/ investment and operating expense serve to focus the improvement activity so as to achieve a global optimum. Throughput is defined as the rate at which the system generates money through
336
Vol. 3, Issue 1, pp. 334-344
sales. Inventory/ investment are defined as all the money the system invests in purchasing things the system intends to sell. Operating expense is defined as all the money the system spends in turning inventory into throughput [24]. These changes in definition provide more realistic measurements and targets for the organizations operations. This is accomplished by increasing throughput, reducing inventory and reducing operational expense. The measurements are also related back to general accounting methods for tax and reporting purpose [61]. All measurements and activity are linked to increasing throughput (T) first, reducing inventory/ investment (I) second and lowering operating expense (OE) third. Net Profit= T OE (1) Return on Investment= (T OE)/I (2) Cash Flow= T OE+/- I (3) The equations govern all activity and show the direct link to organizational profit ([13], [23], [60],]. Net profit is an absolute measure reflecting the companys ability to make money now and in the future, Return on Investment is a relative measure and finally Cash Flow is a survival Measurement [37]. TOC does not consider value added costs as part of inventory valuation. Similarly in operating expenses, no distinction is made between direct or indirect, long and short- term expenses.
Table 2: application area of global performance measure and references Main Area Sub area References Global Performance Measure Services ([58],[59]), [67], [43], [41],[42], [29], [5], [69], [7],
Services share this goal, defining throughput based on sales. However services may have to be a little more creative when it comes to defining inventory and operating expenses (OE). Application of Global performance measure and its applications and the references are given in Table2. Under TOC the more effective statement of corporate goals are those that lead to more effective measures of inventory and operating expenses within the service is suggested by [41]. The most services will have limited amount of traditional inventory ([7], [42], and [43]), the service is often produced at the time of sale and cannot be carried in inventory, and inventory will be a smaller fraction of the service firms assets than it would be for a manufacturing firm. Just because inventory is often a smaller fraction of assets for services, these global performance measures can still be utilized ([5], [41], [69], [29], [58], [59], and [67]).
IV.
THINKING PROCESS PARADIGM
The major component of TOC that underpins all the other parts of the methodology is the TOC Thinking Processes. Goldratt states that managers make three decisions when dealing with constraints: What to change? What to change to? and How to cause the change? The TOC logical thinking process (TP) has evolved to answer these generic questions [63].The past ten to fifteen years have shown that it is often managerial policies are most often the main constraint [50] the thinking process also helps in these situations. The thinking process consists of trees or logic diagrams that provide a road map for change, by addressing the three basic questions of What to change, What to change to, and How to cause the change. In much the same way as the 5 Focusing Steps focus on the constraint, the Thinking Processes focus on the factors that are currently preventing the system from achieving its goals. As shown in Fig.1 the Thinking Process follows the cause-effect logic, necessary condition logic and verifying sound logic. The current reality tree (CRT) depicts the current state of affairs, designed to identify the system constraint, link causes and effects within the current operation to reveal root causes of problems. Future reality trees (FRT) are used to test potential solutions by diagramming cause and effect relationship for events in the future. Transition trees (TT), also called a cause-and effect trees, is a flow diagram describing the states of the system as it changes based on a prescribed action plan; it is an implementation plan that has been time sequenced. Two other tools are slightly different. The
337
Vol. 3, Issue 1, pp. 334-344
evaporating cloud (EC) and the prerequisite tree (PRT) are used to identify necessary conditions. These tools complete the sentence In order to havewe must--- and are used to identify and overcome obstacles to meeting an objective or implementing a solution. The PRT provides a bridge between the future reality tree and the transition tree. As such, the PRT is also time-sequenced. Fig 2 shows the Thinking Process procedure for solving the problem. Three of the trees (Current and Future Reality trees and the Transition Tree) use cause-and effect logic. They are built up by constructing connections between observed effects and cause on the basis of sufficient cause. Sufficiency can be of 3 types: A is sufficient to cause C or If both A and B occurs together, then they will be sufficient to cause C or A and B (separately) both contributes to C, and between them is sufficient to cause C. The Evaporating Cloud and the Prerequisite Tree both use necessary condition thinking: In order to achieve A we must have B. The logic rules are called the Categories of Legitimate Reservation ([12], [45]), and have been proposed for use in validating Systems Dynamics models [3]. An excellent straight forward explanation of these building blocks is given in [56]. A fuller description and examples of these logics were given in ([21], [45], [12], and [32]). TOC Systems Thinking Process Techniques
CauseEffect Logic (Sufficient Condition Logic)
Necessary condition Logic
Verifying Sound logic (Categories of Legitimate Reservation)
Fig 1: TOC Thinking Process
The Problem Current Reality Tree Cloud Cloud + Injection Future Reality Tree The Solution Decision Loop
Fig 2: Thinking Process Procedure
4.1. Current Reality Trees (CRT)
338
Vol. 3, Issue 1, pp. 334-344
An Existing condition is called as a reality [20]. The tools he has designed are intended to use to analyze and deal with a system condition, or reality, with which the TOC practitioner is unhappy. A Current Reality Tree as a logic structure which has been designed to depict that state of reality as it currently exists in a given system [12]. It is constructed from top-down: from observed undesirable effects, postulating likely causes for those effects, which are then tested via the CLR. One such test is to predict (and check for) other effects that would also arise if this cause did exist- hence the term Effect-Cause-Effect. The CRT is designed to achieve the following objectives [12]: Provide the basis for understanding complex systems Identify undesirable effects (UDEs) exhibited by a system Relate UDEs through a logical chain of cause and effect to root causes Identify, where possible, a core problem that eventually produces 70% or more of the systems UDEs. Determine at what points the root causes and/or core problem lie beyond ones span of control or sphere of influence Isolate those few causative factors (constraints) that must be addressed in order to realize the maximum improvement of the system Identify the one simplest change to make that will have the greatest positive impact on the system. (P.64) Dettmer described the CRT as functional rather than organizational and as such is blind to internal and external system boundaries. CRTs may also include positive feedback loops: generally there will be at least one feedback loop which constitutes a vicious cycle. The existence of a loop usually opens up more possibilities for the setting of remedial action: a change in or below a loop will have a significant effect.
4.2. Evaporating Clouds (EC)

Once the TOC practitioners have identified what to change, the second step in the process deals with the search for a plausible solution to the root cause; that is, what to change to. This task is accomplished with the aid of the Evaporating Cloud (EC) and the Future Reality Tree (FRT). Unlike the trees, the EC has a set format with 5 boxes. The practitioner identifies two opposing wants, that represent the conflict, the need that each want is trying to satisfy, and a common objective or goal that both needs are trying to fulfill. This direct conflict is often the same as that underlying the CRT. Traditionally in resolving these conflicts, managers have sought compromise solutions [20]. The EC start with an objective, which is the opposite of the core problem. The object has a minimum of two requirements listed. Each requirement has a prerequisite. From the objective, the requirements (minimum of two) are listed. Each requirement will have at least one prerequisite. It is the prerequisites that depict the tug-of-war. What is needed is a set of injections that can be used to break the validity of any one of the assumptions. This is the first step in freeing our self from the binding controversy. In constructing the EC, one injects the ideal answer, which would burst the cloud and thereby remove the problem. The EC is intended to achieve the following purposes [12]: Confirm that the conflict exists Identify the conflict perpetuating a major problem Resolve conflict Avoid compromise Create solutions in which both sides win Create new breakthrough solutions to problems Explain in depth why a problem exists Identify all assumptions underlying problems and conflicting relationships.
4.3 Future Reality Trees (FRT)

Once a solution, called an injection, has been identified via the EC method practitioners assume for the next exercise that it has been achieved and start to build the Future Reality Tree (FRT). The tree is constructed and scrutinized to test the solution, once again using an effect-cause-effect method. The
339
Vol. 3, Issue 1, pp. 334-344
FRT identifies what to change as well as considering its impact on the future of the organization. The Future Reality Tree is the thinking process that enables a person to construct a solution that, when implemented, replaces the existing undesirable effects with desirable effects without creating new ones (Goldratt, 1993) Step-by-step the solution is created, and each stem is scrutinized to logically show that once the injection are implemented, the desirable effects can be accomplished. The resulting tree originates in one or more injections and ends in desirable effects which really reflect the opposite of the UDEs in the CRT. Glodratts Categories of Legitimate Reservation (CLR) provide guidelines for communicating any reservations about the validity of the elements and connections within the trees [12, 3]. The FRT serves the following purposes: Enables effectiveness testing of new ideas befor3e committing resources to implementation Determines whether proposed system changes will produce the desired effects without creating negative side effects Reveals through negative branches, whether proposed change will create new or collateral problems as they solve old problems, and what additional actions are necessary to prevent any such negative side effects from occurring Provides a means of making beneficial effects self-sustaining through deliberate incorporation of positive reinforcing loops Provides a means of assessing the impacts of localized decisions on the entire system Provides an effective tool for persuading decision makers to support a desired course of action Serves as an initial planning tool
4.4. Prerequisite Trees (PRT)

Once the practitioners have identified what to change to, the third step in TOC deals with implementing the solution. The one of the TOC principle is that ideas are not yet solutions [20]. It cannot be called a solution until implementation is complete and the system is working as intended. The PRT is intended to identify obstacles that prevent the injection from the EC being implemented. The PRT uses a different logic from the previous trees, both of which use sufficiency logic (which basically asks Is this enough?) to establish cause and effect relationships. The PRT uses necessary logic, as does the Evaporating Cloud. Asking the following two questions to check whether a PRT is needed [12]: Is the objective a complex condition? If so, a PRT may be needed to sequence the intermediate steps to achieve it. Do I already know exactly how to achieve it? If not, then a PRT will help map out possible obstacles, the steps involved in overcoming them, and the appropriate sequence. The PRT is used to achieve the following objectives [12]: To identify obstacles preventing achievement of a desired course of action, objective, or injection (solution idea arising from the Evaporating Cloud). To identify the remedies or conditions necessary to overcome or otherwise neutralize obstacles to a desired course of action, objective, or injection. To identify the required sequence of actions needed to realize a desired course of action. To identify and depict unknown steps to a desired end when one does not know precisely how to achieve them.
4.5 Transition Trees

The last tool in the TOC Thinking Process is the Transition Tree, [34] allow practitioners to determine the actions necessary to implement the solution. Practitioners use the effect-cause-effect method to construct and scrutinize the details of the action plan, called the Transition Tree. As in the construction of the FRT, each step is scrutinized using CLRs for negative branches. The FRT as a strategic tool in which major changes can be outlined [12]. The implementation of these, however, will require complex interventions needing greater detail of action to be taken, which is the intended use for the Transition Tree. The Transition Tree as an operational or tactical tool.
340
Vol. 3, Issue 1, pp. 334-344
The purpose of a Transition Tree is to implement change [12]. He says that the Transition Tree structure started off as a four-element tree, with a fifth element being added later. The use of the four or five element tree is situational. He states that the five-element tree is the preferred methodology when constructing step by step procedures and there is a need to explain to others exactly why each step is required. [12] outlined the original four elements of the Transition Tree as: 1. A condition of existing reality 2. An unfulfilled need 3. A specific action to be taken 4. An expected effect of the integration of the preceding three Each succeeding level of the Tree is built upon the previous level, with the expected effect taking the place of the unfulfilled need. These build progressively upward to an overall objective or desired effect. The fifth element added to the Transition Tree is: 5. The rational for a need at the next higher level of the tree. The Transition Tree has nine basic purposes [12]: Provide a step by step method for action implementation Enable effective navigation through a change process Detect deviation in progress toward a limited objective Adapt or redirect effort, should plans change Communicate the reasons for action to others Execute the injections developed in the EC or FRT Attain the intermediate objectives identified in a PRT Develop tactical action plans for conceptual or strategic plans Preclude undesirable effects from arising out of implementation. (p. 284)
4.6 Summary of the Thinking Process

The five stage Theory of constraints thinking process begins with a Current Reality Tree, which diagnoses what, in the system, needs to be changed. The Evaporating Cloud is then used to gain an understanding conflict within the system environment or, as Goldratt prefers to call it, the reality that is causing the conflict. The Evaporating Cloud also provides ideas of what can be changed to break the conflict and to resolve the core problem. The Future Reality Tree takes these ideas for change and ensures the new reality created would in fact resolve the unsatisfactory systems conditions and not cause new ones. The Prerequisite Tree determines obstacles to implementation and ways to overcome them and the Transition Tree is a means by which to create a step-by-step implementation plan. All of Glodratts tools are designed to overcome resistance to change by creating a logical path which can be followed.
V.
CONCLUSIONS
In this paper we have synthesized large body of knowledge of three paradigms of TOC logistics, global performance measure and thinking process. The review shows that the vast majority of the papers have concentrated on the concept and philosophy enhancement of TOC. Several articles have been published in the production sector also. In the application category a number of articles reports the application of TOC concepts in the area of production and management accounting. Few papers have been published on the comparison of TOC with various Existing theories such as MRP and JIT. This review shows that most of the work is carried in the field of Project Management in the Logistics Paradigm. The review shows some of the applications in the field of services in the Logistics paradigm as well as in the Global Performance Measure paradigm. The Theory of Constraints provides an effective, systematic approach for identifying constraints to the overall business and developing a plan to alleviate these constraints. TOC provides a global system methodology that promotes achieving the organizational goal of making more money both now and in the future.
341
Vol. 3, Issue 1, pp. 334-344
From the analysis of these paradigms it is observed that there are some important elements which have not yet been adequately addressed. This analysis shows that some important issues in the Project Management is not adequately addressed Simulation of the case studies of organizations by identifying the bottleneck stations and developing a detail schedule for it. Validation of the network models with the simulation can be carried out in the Project Management. A comparison of the Project management network models with the CPM can be done. REFERENCES
[1]. Barber P., Tomkins C. & Graves A., (1999) Decentralized site management- a case study. International Journal of Project Management, vol 17(2), 113-120. [2]. Blackstone, J.H., (2001) theory of constraints--A Status Report, International Journal of Production Research, vol 39 (6), 1053-1080. [3]. Bladerstone, S.J.,(1999) Increasing User Confidence in systems Dynamics Models through Use of an Established Set of Logic Rules to Enhance Forrester and Senges Validation Tests, in Systems Thinking for the Next Millennium, Proceedings of the 17th International Systems Dynamics Conference and 5th Australian and New Zealand Systems Conference, Wellington, 20-23 July. [4]. Boyd. L. & Gupta, M., (2004) constraints management: what is the theory? International Journal of Production and Operations Management, vol 24 (4), 350-371. [5]. Bramorski, T., Madan, M.S. & Motwani, J., (1996) Application of the Theory of Constraints in Banks, The Bankers Magazine, vol 180(1), 53-59. [6]. Coman, A. & Ronen, B., (1994) Is Management by Constraints: Coupling IS Effort to Changes in Business Bottlenecks, Human Systems Management, vol 13(1), 65-70. [7]. Cook, D.P., Goh, C. & Chung, C.H.,(1999) Service Typologies: A State of the Art Survey, Production and Operations Management, vol 8(3), 318-338. [8]. Cox, Blackstone & Schleier,.(2003) Managing Operations: A Focus on Excellence, [9]. Cox, J. F. & M.S. Spencer, (1998)The Constraints Management Handbook, The St Lucie Press/ APICS series on Constraints Management, Boca Raton, FL, [10]. Demmy, W.S. & Petrini, A.B., (1992) The Theory of constraints: A New Weapon for Depot Maintenance Planning and Control, Air Force Journal of Logistics, vol 16(3), 6-11. [11]. Demmy, W.S. & Demmy, B.S., (1994) Drum-Buffer-Rope Scheduling and Pictures for the Yearbook, Production and Inventory Management Journal, vol 35(3), 45-47. [12]. Dettmer, H. W. (1997) Glodratts Theory of constraints: A system approach to Continuous Improvement, ASQC Quality press, Milwaukee. [13]. Dettmer, H. William, (1998), Breaking the Constraints to World-Class Performance, ASZ quality Press, Milwaukee, WI. [14]. Draman R.H., (1995) A new approach to the development of business plans: A Cross Fundamental Model Using. The theory of constraints philosophies. Ph.D dissertation, University of Georgia. [15]. Fawcett, S.E. & Pearson, J.N., (1991) Understanding and applying Constraint Management in Todays Manufacturing Environments, Production and Inventory Management Journal, vol 32(3), 46-55. [16]. Feather, J.J. & Cross, K.F.,(1988) workflow Analysis, Just-in-time Techniques Simplify Administrative Process in Paperwork Operation, Industrial Engineering, vol 20(1), 32-40. [17]. Fenbert, JA. & Fleener, NK.(2002) Implementing TOC multi-project management in a research organization, Frontiers of Project Management Research and Application: Proceedings of PMI Research Conference 2002; 2002 July 14-2002 July 17; AIPM, USA; PMI; [18]. Gardiner, S.C., Blackstone, Jr., J.H. & Gardiner, L.R., (1994) The evolution of the Theory of Constraints, Industrial Management, vol 36(3), 13-16. [19]. Gillespie, M.W., Patterson, M.C. & Harmel, B., (1999) TOC Beyond Manufacturing, Industrial Management, vol 41(6), 22-25. [20]. Goldratt, E. M., (1990b) What is this thing called Theory of Constraints and How Should it be Implemented? North River Press, New York, NY. [21]. Goldratt, E.M.(1994) Optimized Production Timetable: A Revolutionary Program for Industry, APICS 23rd annual conference Proceedings, Goldratt, E.M., Its Not Luck, North River Press Publishing corporation, Great Barrington, MA. [22]. Goldratt, E.M. & Cox, J.,(1992) second revised edition, The Goal, North River Press, Great Barrington, MA. [23]. Goldratt, E. M., & Fox, J.(1986), The Race, North River Press, New York, NY.. [24]. Goldratt, E. M., & Fox, J., (1987) Critical Chain, North River Press, Corton-on-Hudson, New York,.
342
Vol. 3, Issue 1, pp. 334-344
[25]. Goldratt, E. M., & Fox, J., and Robert E.,(1986) The Race, North River Press, great Barrington, MA. [26]. Gray, V., Felan, J., Umble, E. & Umble M., (2000) A Comparison of drum-buffer-rope (DBR) and Critical Chain (CC) buffering Techniques. Project Management Research at the Turn of the Millennium. Proceedings of PMI Research Conference 2000. 21-24 June 2000; (21-24 June 2000); AIPM Pennsylvania, USA: Project Management Institute; [27]. Gregory, Alan & Kearney Gillian, (2004) Restriction Buster. Project 16(10), 20-22. [28]. Gupta, S., (1997) Supply Chain Management in Complex Manufacturing, IIE Solutions, vol 29(3), 18-21. [29]. Hinneburg, P.A., Lynch, W. and Black, J. (1996), Lean Logistics, in 1996 APICS Constraints Management Symposium Proceedings, APICS, Detroit, Michigan, 89-94. [30]. Hodgdon, B., (1998) Identifying an Elevating the Constraint Sales skill, in 1998 APICS Constraints Management Symposium Proceedings, APICS, Seattle, Washington, 62-63. [31]. Jolley, A. & Patrick, A., (1990) The Office Factory, Management Today, July, 100-102. [32]. Kendall, G.I., (1998) Securing the Future: Strategies for exponential Growth Using the Theory of Constraints, St. Lucie Press/ APICS Series on Constraints Management: Boca Raton, FL. [33]. Klein, D. & DeBruine, M., (1995) A Thinking Process for Establishing Management Policies. Review of Business, vol 16, No.3:31-37. [34]. Lambrecht, M. & Segaert, A., (1990) Buffer Stock Allocation in Serial and Assembly Type of Production Lines, International Journal of Operations and Production Management, vol 10(2), 47-61. [35]. Leach, L., (2003) Schedule and cost buffer sizing: how to account for the bias between project performance and your model, Project Management Journal, vol 34(2), 34-47. [36]. Leach, LP. (1999) Critical Chain Project Management Improves Project Performance. Project Management Journal, vol 30(2), 39-51. [37]. Lockamy, III, A. & Spencer, M.S.,( 1998) Performance Measurement in a Theory of Constraints Environment, International Journal of Production Research, vol 36(8), 2045-2060. [38]. Lynch, W.E. & Newbold, R.,(1998) A Practical Approach for Implementing Critical Chain Management, in 1998 APICS Constraints Management Symposium Proceedings, Seattle, Washington, 6467. [39]. Mabin, V.J. & Balderstone, S.J.,( 2000) The World of the Theory of Constraints: A Review of the International Literature, St, Lucie Press, Boca Raton, FL. [40]. McMullen, Jr., T.B., (1998) Introduction to the Theory of Constraints Management Systems, St. Lucie Press/APICS Series on Constraints Management, Boca Raton, FL. [41]. Motwani, J., Klien, D. & Harowitz, R., (1996a) The Theory of constraints in Services: Part 1the basics, Managing Service Quality, vol 6(1), 53-56. [42]. Motwani, J., Klien, D. & Harowitz, R., (1996b) The Theory of constraints in Services: Part 2 examples from healthcare, Managing Service Quality, vol 6(2), 30-34. [43]. Motwani, J. & Vogelsang, K., (1996) The theory of Constraints in Practiceat quality engineering, Inc., Managing Service Quality, vol 6(6), 443-473. [44]. Newbold, RC. (1998) Project management in the fast lane--- applying the Theory of Constraints. St. Lucie press. [45]. Noreen, E., Smith, D.A. & Mackey, J.T., (1995) The Theory of Constraints and its Implications for Management Accounting, The North River Press Publishing Corporation: Great Barrington, MA. [46]. Olson, C.T., (1998) The Theory of Constraints: Application to a service firm, Production and Inventory Management Journal, vol 39(2), 55-59. [47]. Patrick FS., (2001) Buffering against risk- critical chain and risk management. PMI Seminar and Symposium Proceedings 2001; 2001 Nov 1- 2001 Nov 10; AIPM (CD Rom). USA: PMI; [48]. Perez, J.L. (1997), TOC for World Class Global Supply Chain Management, Computers and Industrial Engineering, vol 33(1-2), 289-293. [49]. Rahman, S., (1998) Theory of Constraints: A Review of the Philosophy and its Applications, International Journal of Operations & Production Management, vol 18(4), 336-355. [50]. Rand, GK. (2000) Critical Chain: the theory of Constraints applied to project management, International Journal of Project Management, vol 18(3), 173-177. [51]. Raz, T., Barnes, R. & Dvir, D. (2003), A Critical Look at Critical Chain Project Management. Project Management Journal, vol 34(4), 24-32. [52]. Rivera, F. & Duran, A. (2004), Critical Clouds and Critical sets in resource-constrained projects. International Journal of Project Management, vol 22(6), 489-497. [53]. Ronen, B., Gur, R. & Pass, S., (1994) Focused Management in Military Organizations: An Avenue for Future Industrial Engineering, Computers and Industrial End, vol 27(1-4), 543-544. [54]. Roybal. H., Baxendale, S.J. & Gupta, M. (1999), Using Activity-Based Costing and Theory of Constraints to Guide Continuous Improvements in Managed Care, Managed Care Quarterly, vol 7(1),1-10.
343
Vol. 3, Issue 1, pp. 334-344
[55]. Sale, M.L. & Inman, R.A., (2003) Survey-Based Comparison of Performance and Change in Performance of Firms Using Traditional Manufacturing, JIT and TOC, International Journal of Production Research, vol 41(4), 829-844. [56]. Scheinkopf, L., (1999) Thinking For change: Putting the TOC Thinking Processes to Use, St, Lucie Press/ APICS Series on Constraints Management: Boca Raton, FL. [57]. Schragenheim, E. & Ronen, B., (1991) Buffer Management: a Diagnostic tool for Production Control, Production and Inventory Management Journal, vol 32(2), 74-79. [58]. Simons, J.V. & Moore, Jr., R.I., (1992a) The Theory of Constraints Approach to focused Improvement, Air Force Journal of Logistics, vol 16(3), 1-5. [59]. Simons, J.V. & Moore, Jr., R.I., (1992b) Improving Logistics flow Using the Theory of Constraints, Logistics Spectrum, vol 26(3), 14-18. [60]. Smith, Debra, (2000), The Measurement Nightmare: How the Theory of Constraints Can Resolve Conflicting Strategies, Policies and Measures, The St. Lucie Press/ APICS Series on Constraints Management, Boca Raton, FL. [61]. Sood, S., (2003) Taming Uncertainty: Critical- Chain buffer management helps minimize risk in the project equation. PM Networks, vol 17(3), 56-59. [62]. Spencer, M.S. & Cox, III, J.F., (1995) Optimum Production Technology (OPT) and the Theory of Constraints (TOC): Analysis and Genealogy, International Journal of Production Research, vol 33(6), 14951504. [63]. Spencer, M.S. & Wathen, s., (1994) Applying the Theory of Constraints Process Management Technique to an Administrative Function at Stanley Furniture, National Productivity Review, vol 13(3), 379-385. [64]. Steyn, H., (1996.) An investigation into the fundamentals of critical chain project management. International Journal of Project Management, PM [65]. Steyn, H., (2001) An Investigation into the fundamentals of critical chain project scheduling. International Journal of Project Management, vol 19(6), 363-369. [66]. Steyn, H., (2002) Project Management Application of the Theory of Constraints beyond Critical Chain Scheduling. International Journal of Project Management, vol 20, 75- 80. [67]. Underwood, J.W., (1994) Applying the Theory of Constraints to army Logistics, Army Logistician, July/august (4), 13-17. [68]. Wahlers James L. & Cox James F., (1994) Competitive factors and performance measurement: applying Theory of Constraints to meet customer needs, International Journal of Production Economics, vol 37, 229-240. [69]. Walker, II, E.D. & Cox, III, J.F., (1998) Generic Airlines Reservation Sales Office, Journal of Systems Improvement, vol 2(1), 9-16. [70]. Watson, K.J. & Patti, A., (2008) A comparison of JIT and TOC buffering philosophies on system performance with unplanned M/C downtime, International Journal of Production Research, vol 46(7), 18691885. [71]. Watson, K.J. & Polito, T., (2003) Comparison of DRP and TOC Financial Performance within a Multi-Product, Multi-Echelon Physical Distribution Environment, International Journal of Production Research, vol 41(4), 741-765. [72]. Wei, C. C., Liu, P. H. & Tsai, Y. C.,(2002) Resource- constrained project management using enhanced theory of constraint, International Journal of Project Management, vol 20(7), 561-567.
Authors
She is having 7 years of teaching experience in both U.G and P.G levels. She pursued her M.Tech degree in the field of Industrial Engineering. Currently pursuing her Ph.D as a full time research scholar.
A. Ramakrishna Rao is a vice-principal and professor in the department of mechanical Engineering, S.V.U. College of engineering, S.V. University Tirupati. He has more than 31 years of experience in teaching for both U.G and P.G level and in Research. He has published more than 55 papers in referred national and international journals. He has presented more than 45 research articles in national and international conferences. His current area of research includes Supply Chain Management, Operations Management, Lean Manufacturing, Graph Theory and TOC.
344
Vol. 3, Issue 1, pp. 334-344
TURBO BRAKE ASSISTS TURBOTOR: CONCEPTUAL DEVELOPMENT AND EVALUATION

G. Ramkumar and Amrit Om Nayak
Student, Department of Mechanical Engineering, Thiagarajar College of Engineering, Madurai, India
ABSTRACT
Turbotor employs the principles of turbo machinery and automotive engineering in order to achieve effective braking assistance for the modern day disc as well as drum brakes. These brakes are intended for the purpose of braking assists during high speed motion. The major components include a turbine or rotor, pump, nozzle, enlarged brake fluid chamber, DOT 5.1 braking fluid. The functioning involves directing a jet of high velocity brake fluid (DOT 5.1 in this case) at the blades of the rotor which is coupled with the wheel. The braking circuit is used as such with the enlargement of the brake fluid chamber. A pump is added to the set up to enable increase of mass flow rate of fluid and also to ensure the continuity of fluid flow in the entire circuit. A speed sensor is integrated along with the circuit to activate the pump such that the braking assists are put to use only after a particular velocity is reached as braking assists are not necessary at lower speeds.
KEYWORDS: brake Assists, brake fluid, mass flow rate, and nozzle
I.
INTRODUCTION
The modern era has seen unprecedented technological growth and has propelled automotives on an ever rising curve of speed. Still, despite chargers, turbochargers, twin turbochargers or NOX, there are limits which cannot be surpassed by a land based vehicle in terms of speed. Brakes on the other hand are a part of automobiles that abide by only those limitations imposed to them by the ability of human body to withstand rapid decelerations. Otherwise, it would be a lot easier stopping a car than making it go insanely fast. The drum based braking system can be considered to be the forefather of the modern day brake.The man largely credited with the development of the modern day drum brake is French manufacturer Louis Renault, in 1902. This braking systems was external, a feature which soon turned into a problem. Dust, heat and even water rendered them less effective. It was time for the internal expanding shoe brake. By placing the shoes inside the drum brake, dust and water were kept out, allowing the braking process to remain effective. As the vehicles spilled out the assembly plants, they started becoming both faster and heavier. The earlier brakes were effective, but they had a tendency to ineffectively distribute heat. This feature made room for the creation of the disc braking system. The turbotor will take the development process in the field of automotive braking a step further. It is an attempt to reduce our dependence on frictional braking and thereby reduce wear and tear of brake components. The turbotorutilizes the concepts governing turbo machinery and automotive engineering in order to achieve effective braking assistance for the modern day disc as well as drum brakes. The principle involved encompasses the targeting of a jet of viscous fluid onto the blades of a turbine which is attached to the wheel. This jet of fluid retards the motion of the turbine and in the process damps or slows down the rotation of the wheels. The turbotor is a potential brake assist in high speed conditions thereby reducing the work done by the disc or drum brakes. This results in decreased braking distance as well as braking time. The major components include a turbine or rotor, pump, nozzle, enlarged brake fluid chamber, DOT 5.1 braking fluid. The functioning involves
345
Vol. 3, Issue 1, pp. 345-353
directing a jet of high velocity brake fluid (DOT 5.1 in this case) at the blades of the rotor which is coupled with the wheel. The braking circuit is used as such with the enlargement of the brake fluid chamber. A pump is added to the set up to enable increase of mass flow rate of fluid and also to ensure the continuity of fluid flow in the entire circuit. A speed sensor is integrated along with the circuit to activate the pump such that the braking assists are put to use only after a particular velocity is reached as braking assists are not necessary at lower speeds. The concept development and evaluation of turbotor is organised into six sub-sections. The preliminary section describes the Theory behind the functioning of turbotor, and is followed by the second section which discusses the Construction of Turbo Brake Assist system and the integral components that form the heart and soul of the system. The third section illuminates the Working principles and driving concepts of the system. The detailed Mathematical study of existing braking systemand turbotor is illuminated in the fourth section. The fifth section consists of Results and Discussion which throws light on the inferences from mathematical study of the brake assist system. The sixth section is the Conclusion which elucidates the practicality of the concept and performance of turbotor as an effective brake assist.
II.
THEORY
We deem the turbotor to be a success even if it caters to at least 35 per cent of the braking force from the disc as well as the drum brakes. The main purpose of this system is to function as a braking assist. If the turbotor is capable of providing excessive braking which is close to that obtained from the primary brake system, this would lead to locking of the wheels and very sudden deceleration resulting in utmost discomfort for the driver. Therefore the working of this system will be limited to a velocity range in order to accommodate for the above mentioned features. The total braking force obtained from the front disc brakes and the rear drum brakes is calculated independently. The braking force obtained from the turbotor is calculated by evaluating the pressure density or force of the jet of viscous fluid (DOT 5.1 braking fluid) on the turbine blades. The weight distribution between front and rear axles during dynamic braking is calculated. This will enable us to decide the amount of braking assist required at the front and the rear wheels of the vehicle. The use of DOT 5.1 braking fluid will ensure that heat generated during damping of turbine motion will not pose a threat as it is hygroscopic in nature. The concept is explained in detail in the following sections.
III.
CONSTRUCTION
The integral parts of the Brake Assists are depicted in the schematic diagram Fig.1and the components are explained subsequently.
Fig.1 Turbotor schematic layout
3.1.Master cylinder chambers:

The Turbotor setup consists of a master cylinder which is partitioned into two chambers in order to provide undisturbed volumes for normal braking system as well as for Turbotor. The position of the
346
Vol. 3, Issue 1, pp. 345-353
master cylinder chambers is the same as in the existing braking system. The master cylinder chamber for Turbotor system is provided with pipelines to carry brake fluid to and fro from the rotor setup. These pipelines for the Turbotor are called auxiliary brake fluid lines(inlet and outlet). The other chamber of master cylinder is provided with pipelines for disc and drums brakes and is thus left undisturbed.
3.2. Auxiliary brake fluid pipelines:

The auxiliary brake fluid pipe lines are connected to the master cylinder chamber for turbotor at one end and the other end consists of nozzle to impart high velocity to the brake fluids at the striking end to the turbotor. There are four inlet and outlet brake fluid lines attached to the master cylinder chamber for turbotor. The outlet pipe lines from the turbotor carrythe brake fluid that is accumulated after striking the turbotor to the master cylinder chamber for turbotor. The auxiliary inlet pipelines to the turbotors at the front and rear is symmetric at the point along its longitudinal length as shown in the Fig. 1. This symmetric arrangement ensures equal mass flow rate of fluid to both left and right turbotors at front and rear axles.
3.3.Hydraulic pump:
The hydraulic pump that is going to be used here is a diaphragm type positive displacement reciprocating pump. A diaphragm pump is a positive displacement pump that uses a combination of the reciprocating action of a rubber, thermoplastic or teflon diaphragm and suitable non-return check valves to pump a fluid. Sometimes this type of pump is also called a membrane pump.
Fig.2. Diaphragm pump schematic
When the volume of a chamber of either type of pump is increased (the diaphragm moving up), the pressure decreases, and fluid is drawn into the chamber. When the chamber pressure later increases from decreased volume (the diaphragm moving down), the fluid previously drawn in is forced out. Finally, the diaphragm moving up once again draws fluid into the chamber, completing the cycle. This action is similar to that of the cylinder in an internal combustion engine.
3.4. Turbotor:
Fig.3. Tubotor schematic
The turbotor schematic is shown in Fig. 3. After the fluid jet hits the turbotor blades, it is allowed to pass through the orifice holes and collect in the collecting chamber. This collected brake fluid is now sent back via the outlet turbotor outlet back to the MCC 2 as shown in Fig. 1.
347
Vol. 3, Issue 1, pp. 345-353
IV.
WORKING
The overall working of the system is quite simple. The brake fluid is pumped from MCC 2 and is sent via the auxiliary inlet lines to the nozzle. High pressure fluid is sprayed on the turbotor blades which in turn decelerates the blades. Since the turbotor is coupled to the axle, it in turn causes overall deceleration of the vehicle. The brake fluid collects in the collecting chamber of the turbotor through the orifices and is sent back to the MCC 2 via the auxiliary fluid outlet lines with the help of the pump. This entire turbotor system aids the primary braking system and thereby leads to reduction in braking time and braking distance. This also reduces the wear and tear experienced by the primary braking system.
V.
MATHEMATICAL STUDY OF EXISTING BRAKE SYSTEM
NOTE: The numerals beside the equations are intended to signify the number of equations only.
5.1. Drum brakes braking equations:

The maximum wheel torque is limited by wheel slip and is given by: = Where, = = = = (1)
The torque produced at the drum brakes caused by friction between the lining and the drum which is necessary to bring the vehicle to a standstill is given by: = Where, = = = = Both wheel and drum torques must be equal up to the point of wheel slip but they act in opposite direction to each other. Therefore they may be equated. = = Therefore Force between lining and drum is: = / (N) (5) The above mentioned equation is the braking force produced by drum brakes in order to bring the vehicle to a stop. (3) (4) (2)
5.2. Disc brakes braking equations

The normal clamping thrust on each side of the disc acting through the pistons multiplied by coefficient of friction generated between the disc and pad interfaces produces a frictional forceon both sides of the disc. = in N (6)
348
Vol. 3, Issue 1, pp. 345-353
If the resultant frictional force acts through the centre of the friction pad the mean distance between centre of the pad pressure and centre of the disc will be /2 = in m (7) Accordingly, the frictional braking torque will be dependent upon twice the frictional force (on both sides) and the distance the pad is located from the disc centre of rotation. That is, =2 i.e. , =2 (Nm) (Nm) (8)
5.3. Braking dynamics of Turbotor The maximum sustained deceleration required by the specifications is 0.65g(where g is the acceleration due to gravity and is taken as 9.81 m/s2) The braking force required to bring the car to a stop is given by, = in kN (9) The average power absorbed by the brakes is given by, = /2in kW (10) The static weight distribution of the car is given by, Weight on the front axle: = / + (11) Weight on the rear axle: = / + (12) Assuming that the car is established, under steadystate braking, application of Newtons law provides the equations to determine the dynamic weight distribution on the front and rear axles during braking. / + + / + (13) = / + / + (14) = NOTE*: All weight distributions are in kN. The following diagram indicates the parameters as noted in above formulae
Fig. 4 Vehicle force diagram
5.4. Mathematical study of proposed brake assist system

We know that the braking power obtained at the conventional brakes is the maximum braking power that can be provided at the turbotor for it to act safely as a brake assist. This power is carried to the turbotor by the brake fluid. The brake power (in kW) that needs to be imparted to the brake fluid is given as: = (15) Where = braking force in kN and
= velocity of the brake fluid in m/s
349
Vol. 3, Issue 1, pp. 345-353
This is the velocity of the brake fluid (in m/s) obtained at the nozzle outlet in the auxiliary inlet fluid line. The head developed in the nozzle is given be
=2
(16)
Where g = acceleration due to gravityin m/s And H = head developed in m
The necessary power to be given to the brake fluid is obtained with the help of a suitable reciprocating pump (as described in Fig. 1) which is similar to the fuel pumps in use today. The power rating (in kW) of the pump is given by = Where = density of the brake fluid in kg/m3 A= area of cross-section in m2 = stroke length in m N = rpm of the pump = total head= Where = suction head in m = delivery head in m The head obtained at the nozzle outlet is the delivery head of the pump (without considering the losses in pipeline). From the above equation we can obtain the value of suction head of the pump. The pump is similar to fuel pump considering Head in terms of pressure. Head in terms of pressure can be obtained using the formula, = 10.197 / (18) Where, H = head in m SG = Specific gravity of the brake fluid P = pressure of the fluid in bar Since turbotor spins at the same rpm of the wheel, Rpm of the turbotor can be obtained from: / = 0.00595 / (19) = Where, = tangential velocity of the turbotor in m/s = diameter of the turbotor in m = Rpm of the turbotor From the velocity triangle of the turbotor blades, Fig 5, we get the following equations, = Where, = Absolute velocity of the brake fluid at inlet of the blade in m/s = brake fluid whirl velocity at the inletof the blade in m/s = cos (22) (21) /60 (20) + / 60 1000 (17)
350
Vol. 3, Issue 1, pp. 345-353
Where, = whirl velocity of the brake fluid at outlet in m/s = tangential velocity of the turbotor in m/s = angle of deflection of the brake fluid from the blade surface = Where, = braking force given by the jet of brake fluid in kN And Where, = discharge of the brake fluid in m3/s = area of cross-section in m2

(23)
(24)
= velocity of the brake fluid at the nozzle outlet in m/s = + (25)
Where, = work done by the brake fluid in kNm The velocity diagram for the blade is shown below:
Fig. 5 Velocity diagram for blade
In the above velocity diagram, = absolute velocity of the brake fluid at inlet in m/s = turbotor wheel velocity in m/s = whirl velocity of the brake fluid at inlet in m/s = relative velocity of the brake fluid with respect to blade at inlet in m/s = absolute velocity of the brake fluid at outlet in m/s = turbotor wheel velocity in m/s = whirl velocity of the brake fluid at outlet in m/s = relative velocity of the brake fluid with respect to blade at outlet in m/s And = = U = turbotor wheel velocity in m/s = , neglecting the effect of friction between fluid and blade. The heat developed in turbotor during braking is given by the following set of equations: = 35% / (26) Where, = braking force in kN
351
Vol. 3, Issue 1, pp. 345-353
= mass of the vehicle in kg = Where, = time taken for deceleration in s = velocity of the vehicle before braking in m/s = velocity of vehicle after braking in m/s / (27)
= Where, = heat energy absorbed by brakes in kJ = power absorbed by the brake fluid during braking in kW = time taken for deceleration in s Considering the energy absorbed by a single turbotor at front axle, we have: = /2 The temperature change in the turbotor during braking is given by: = Where, = heat energy generated in turbotor in kJ = mass of the turbotor in kg = specific heat at constant volumeof material of the turbotor in kJ/kg.K = change in temperature of the turbotor
(28)
(29) (30)
VI.
Here Ford Fiesta hatchback edition is taken as a case-study for the Turbo Brake Assists. The velocity for actuation of turbotor is taken as 19.44 m/s. We now assume that the vehicle is moving with an initial velocity of 27.78 m/s. With mass of the car as 1110kg and deceleration of 6.3765 m/s2 and the braking force obtained from the conventional braking system is 7.0779 kN. Now considering 35 per cent of the existing braking force to be obtained by Turbo Brake Assists, the braking force is calculated as 2.4773 kN. This is the braking force necessary to be impacted by the brake fluid to the turbotor. Considering 59.7 per cent of braking force at front axle (as the weight distribution ratio for Ford Fiesta is 59.7/40.3 in percentage), the braking force at the front wheels is obtained as 1.4789 kN. At each turbotor in the front axle, the braking force is obtained as 0.7395 kN. The average power absorbed by the conventional brakes during braking is the average power absorbed by the brake fluid during braking in Turbotor.Therefore the average power to be imparted to the brake fluid at each turbotor in the front wheels is calculated with = 0.7395 and = 27.7778 m/s. and is obtained as 10.27036 kW.This average power has to be given to the brake fluid in order to perform braking efficiently.The power to be absorbed by the brake fluid during braking is equal to the power given to the brake fluid before striking the turbotor, and the velocity is calculated as 13.8889 m /s. This is the velocity obtained at the nozzle out let before striking the turbotor. The above calculations give the necessary design conditions and values for desired efficient performance of Turbo Brake Assists i.e., 35 per cent of existing braking force. And in order to achieve this, a power of 10.27036 kW is to be given to the brake fluid for obtaining a velocity of 13.8889 m/s. The power is provided by the reciprocating hydraulic pump.The Head developed at the nozzle outlet is calculated as 9.8319m. This head is the delivery head for the pump. The total head for the pump is calculated using = 1050 kg/m3(property of DOT 5.1 brake fluid), g = 9.81 m/s2, Diameter of the cylinder = 64 mm, L = 175 mm, N = 5000 rpm, P = 10.27036 kW and the total head is calculated as 20.61456 m. In terms of pressure, the total head, suction head, delivery head are calculated as 2.1227 bar, 1.11031bar and 1.01240 bar respectively. Now considering the turbotor blade part, the absolute velocity of the brake fluid at inlet of the blade is obtained as 13.88890001 m/s. The tangential velocity of the wheel is obtained as 85.426471 m/s. With radius of the wheel = 7.5 inches, gear ratio at 5th gear is 0.756, final drive ratio is 4.07:1, and vehicle speed is 62.137119 miles/hr and value of n = 4284.3909 rpm.The whirl velocity of brake fluid at the blade outlet is obtained as 18.2031434 m/s. Here angle of deflection, is taken as 20 degrees.
352
Vol. 3, Issue 1, pp. 345-353
NOTE*: Since turbotor tangential velocity is greater than brake fluid velocity, the negative sign obtained in the values are omitted and only magnitudes are considered. The force imparted by the jet of brake fluid onto the blade surface is obtained as 0.58782 kN. Work done by the brake fluid is obtained as 50.21538792 kW. The above value is obtained for a nozzle diameter of 0.04m. The time taken for deceleration is calculated as 3.73398 s, considering v = 27.7778 m/s and u = 19.444 m/s. This time is considered for the braking force achieved using turbotor system entirely as shown previously i.e. 2477.27025 N. The heat energy absorbed by the turbo amounts to 128.47335 kJ. And each turbotor at front absorbs 38.3493 kJ.Therefore the temperature change in turbotor is obtained as 20.18384 K with m = 4 kg, = 0.475 KJ/kg-K.
VII.
CONCLUSION
From the mathematical results obtained, when a nozzle diameter of 0.04 m (which is a practical figure) is considered the DOT 5.1 brake fluid is able to impart necessary force i.e. 2.35128 kN, which is close to the desired performance value of 2.47727025 kN (35 per cent of existing braking force from conventional braking system), on the turbotor blades there by causing deceleration of the vehicle as a whole and bringing about effective braking assistance to the existing braking system. This reduces the work done by the conventional braking system at high speeds. Hence Turbotor can be practically employed with the existing braking system as effective braking assistance. Another important inference from the mathematical study is that the velocity of brake fluid remains constant at any ratio of braking power and braking force. By varying the nozzle diameter and angle of deflection, the braking force of the brake fluid can be varied accordingly.
REFERENCES
[1]. Richard Stone and Jeffrey K. Ball (2004), Automotive Engineering Fundamentals, SAE International, Warrendale, Pa. [2]. Dr.R.K.Bansal (2005), A Textbook of Fluid Mechanics and Hydraulic Machines S.I.Units, Laxmi Publications (P) Ltd., New Delhi. [3]. Heinz Heisler (2002), Advanced Vehicle Technology, Butterworth-Heinemann, Woburn, MA. [4]. Yuvan.S.W, Foundation of Fluid Mechanics, Prentice Hall, 1998. [5]. Ellis.J.R, Vehicle Dynamics, Business Books Ltd., London, 1991. [6] .Rogers, G.F.C. and Mayhew, Y.R. (1967), Engineering Thermodynamics Work and Heat Transfer, Longman. [7]. Limpert, R. (1999), Brake Design and Safety, Society of Automotive Engineers, Warrendale, Pa. [8]. SAE Handbook (1989), Volume 2, Society of Automotive Engineers, Warrendale, Pa. [9]. Gillespie, T.D. (1994), Fundamentals of Vehicle Dynamics, Society ofAutomotive Engineers,Warrendale, Pa. Authors G. Ram Kumar was born in Chennai, Tamil Nadu, India, in 1991. He is currently pursuing Bachelor of Engineering degree in sixth semester with the Department of Mechanical Engineering, in Thiagarajar College Of Engineering, Madurai, Tamil Nadu, which is affiliated to Anna University, Tamil Nadu. His research interest includes Automotive Technology and Engineering, Thermal Engineering, blended wing design in aircrafts, turbo machinery and theoretical physics.
Amrit Om Nayak was born in Bhubaneswar, Odisha, India, in 1991. He is currently pursuing Bachelor of Engineering degree in eighth semester with the Department of Mechanical Engineering, in Thiagarajar College of Engineering, Madurai, Tamil Nadu, which is affiliated to Anna University, Tamil Nadu. His research interests includes Automotive Technology and Engineering, Thermal Engineering, blended wing design in aircrafts, turbo machinery and theoretical physics.
353
Vol. 3, Issue 1, pp. 345-353
STRUCTURAL REFINEMENT BY RIETVELD METHOD AND MAGNETIC STUDY OF NANO-CRYSTALLINE CU-ZN

FERRITES
K.S. Lohar1, S.M. Patange1, Sagar E. Shirsath2, V.S.Surywanshi3, S. S.Gaikwad1, Santosh S. Jadhav4 and Nilesh Kulkarni5
Materials Research Laboratory, Shrikrishna Mahavidyalaya, Gunjoti, MS, India 2 Department of Physics, Vivekanand College, Aurangabad, MS, India 3 Department of Chemistry, S.C.S. College, Omerga, MS, India 4 Department of Physics, D.S.M. College, Jintur, MS, India 5 Department of Condensed Matter Physics and Materials Science, Tata Institute of Fundamental Research. Mumbai, MS, India
1
ABSTRACT
A series of nanocrystalline spinel ferrites with chemical compositions Cu1-xZnxFe2O4 (x = 0.0, 0.2, 0.4, 0.6, 0.8 and 1.0) synthesized by wet-chemical co-precipitation method, to investigate their structural and magnetic properties. The structural parameters were measured by applying the full pattern fitting of Rietveld method using Full Prof Program. The crystallite size was calculated by using Scherer formula and Williamson Hall analysis method, and also confirm from by transmission electron microscopy (TEM). The lattice parameter is found to increases with increasing zinc ion concentration. X-ray intensity ratios were calculated in order to determine the possible cation distributions among the tetrahedral (A) and octahedral [B] site. The saturation magnetization (Ms) and Bohr magneton per formula unit (nB) was measured at room temperature using vibrating sample magnetometer. The Y-K angle increases gradually with increase in Zn2+ concentration 'x'. Magnetic AC susceptibility measurements were carried out as a function of temperature in order to measure the Curie temperature (TC).
PCAS: 75.50.Gg, 75.75.+a, 61.46.Df, 75.40.Cx KEYWORDS: Ferrites, Nano-particle, X-ray diffraction, AC susceptibility.
I.
INTRODUCTION
Nano-crystalline ferrites are widely used in many electronic devices such as antenna cores, high frequency inductors, transformer cores, read write heads for high-speed digital taps or disc recordings etc. These are preferred because of their high electrical resistivity, low eddy current losses, high Curie temperature, mechanical hardness, chemical stability and reasonable cost [1-3]. The structural and magnetic properties of ferrites mainly depend upon chemical composition, methods of preparation, sintering temperature and sintering time [4]. A small amount of dopant ion can change structural and magnetic properties of ferrites [5-8]. Among the ferrites, copper ferrites (CuFe2O4) are extensively investigated because of their interesting cryptographic and magnetic properties [9]. Copper ferrite crystallizes either in a tetragonal (T) or cubic (C) symmetry depending on the cation distribution among sites in its spinel structure [10-11]. The Fe3+ ions occupy both tetrahedral (A) and octahedral [B] site. The copper ions occupy only [B] site and causes the tetragonal structure distortions as per Jahn-Teller effect [10-11]. The Cu-Zn ferrites have been studied in the literature due its technological importance [12-15]. The present work deals to investigates the structural and magnetic properties of Cu1-xZnxFe2O4 (x = 0.0 to x =1.0) system via wet-chemical method.
354
Vol. 3, Issue 1, pp. 354-361
II.
EXPERIMENTAL
Nano-ferrites Cu1-xZnxFe2O4 (x = 0.0 to 1.0 in steps of x = 0.2) were synthesized by wet-chemical coprecipitation method. In order to obtain desired composition, stoichiometric amount of corresponding AR grade metal sulphates, were dissolved in distilled water and 2M sodium hydroxide solution was added till pH become 11.5 with constant stirring at a temperature of 60 C in oxygen atmosphere. The precipitate was digested for 3 h. at same temperature then filtered and thoroughly washed with distilled water until it becomes free from sodium ions and dried in oven under vacuum. The dried powder mixed homogeneously and annealed at 700 C for 6 h. The annealed samples were X-ray examined on Jeol- JDX-8030 at room temperature by using Cu- K radiation and TEM analysis was carried out on Philips CM-12 transmission electron microscopy. The refinement of the structure was confirmed by using software WinPLOTR (version LIB-LCSIM). The magnetization and Curie temperature were measured by using vibrating sample magnetometer and AC susceptibility instrument respectively.
III.
RESULT AND DISCUSSION
The X-ray diffraction (XRD) patterns refinement was continuously done until to get goodness factor very close to one. The values of discrepancy factor (Rwp) and expected values (Rexp) with goodness index are listed in Table 1.
Table 1. The value of discrepancy factor (Rwp) goodness factor (2), expected values (Rexp), lattice constant (a), phase, particle size (t) of Cu1-x Zn xFe2O4 system a ( ) t (nm) Comp. Rwp Rexp 2 Treor Phase Scherers Williamson x Braggs a c formula Hall Analysis 0.0 5.15 2.80 1.84 8.400 8.404 9.504 T 25.6 31.4 0.2 6.87 3.42 2.01 8.412 8.411 8.411 C 27.0 24.6 0.4 6.20 3.52 1.76 8.417 8.417 8.416 C 28.0 28.3 0.6 7.90 3.82 1.64 8.424 8.426 8.403 C 25.2 24.0 0.8 5.94 3.86 1.54 8.431 8.433 8.406 C 27.7 28.1 1.0 3.76 2.54 1.48 8.445 8.445 8.445 C 28.2 28.5
The Rietveld refined XRD pattern for the typical samples (x = 0.0 and 0.2) of Cu1-xZnxFe2O4 systems are shown in Fig. 1. Rietveld refined patterns illustrates the observed and calculated X-ray pattern as well as for their difference for sample x = 0.0 and x = 0.2.
355
Vol. 3, Issue 1, pp. 354-361
Fig. 1.Observed ( ) and calculated ( - ) Xray diffraction pattern for sample (x = 0.0 and 0.2)
Analysis of XRD diffraction pattern of all samples confirm the formation of tetragonal structure for x = 0.0, for samples x = 0.2-1.0 it revels the single-phase cubic structure. The tetragonality decreases slowly with increase in Zn content x [16]. The values of lattice constant (a) calculated by using Braggs law and by Powder X program for Zn substituted copper ferrite are listed in Table 1. Both the values of lattice parameter are good in agreement to each other. Table 1 illustrates that there is an increase in lattice constant with Zn2+ ion concentration x, the addition of Zn2+ ion in copper ferrite, causes the Fe3+ ions migrate from (A) site to [B] site, the Zn2+ ion being larger ionic radius (0.81 ) than that of Fe3+ ion (0.67 ). Zn2+ ions replaces Fe3+ from (A) site and its results there is expansion in lattice constant [12]. The tetragonal distortion of copper ferrite from cubic spinal structure, due to absence of Fe3+ ions at tetrahedral sites, this distortion explain by Jahn- Teller theory according to it, a non liner molecule in ground state is unstable in the symmetric configuration, it will distort it self to lower its energy. The ions with d4 or d9 electronic configuration, when situated at octahedral site causes a strong tetragonal distortion (c/a.>1). Similarly the ions with d3 and d8 configuration and located at tetrahedral site produce a tetragonal distraction with (c/a.>1), at tetrahedral site the ions with d3 and d8 configuration also produce tetragonal distortion with c/a>1 [13].
Fig. 2. Scanning electron micrograph for sample (x = 0.6)
356
Vol. 3, Issue 1, pp. 354-361
Figure 2 shows scanning electron micrographs (SEM) of the typical sample (x = 0.6). The SEM image indicating the fracture surface, further the texture was homogenous structure constiting polyhedral grain with well defined grain boundaries. The average crystallite size for each sample was determined using Scherrer formula [17] and also calculated by Williamsons Hall analysis [18]. Crystallite size obtained by both the method shows good agreement with each other (Table 1). The morphology of synthesized sample is studied by TEM as shown in Fig. 3 (x = 0.4). It observed that the particle obtained in the range of 25 to 32 nm.
Fig. 3.Transmission electron micrograph for sample (x = 0.4)
The cation distributions for the system Cu1-xZnxFe2O4 were determined by X-ray diffraction method. In order to determine the cation distribution. The X-ray intensity was calculated using the formula suggested by M.G. Burger [19]. The X-ray intensity ratios I(220)/I(400) and I(422)/I(440) were calculated for various possible compositions of cations and compare with observed intensity ratios. The combinations of cation for which observed and calculated intensity ratio agree close to each other are taken to be corrected cation distributions which are listed in Table 2.
Table 2. Cation distribution and X- ray intensity ratio for oCu1-x Zn xFe2O4 system. Intensity ratio Cation distribution I220/I400 I422/I440 A- site B- site Obs. Cal. Obs. Cal. (Cu0.0Fe1.0) (Cu0.0Zn0.2Fe0.8) (Cu0.0Zn0.4Fe0.6) (Cu0.0Zn0.6Fe0.4) (Cu0.0Zn0.8Fe0.2) (Cu0.0Zn0.5Fe0.5) [Cu1.0Fe1.0] [Cu0.8Zn0.0Fe1.2] [Cu0.6Zn0.0Fe1.4] [Cu0.4Zn0.0Fe1.6] [Cu0.2Zn0.0Fe1.8] [Cu0.0Zn0.5Fe1.5] 1.1428 1.1785 1.3461 1.6500 1.3636 1.5714 1.1646 1.4700 1.4981 1.7143 1.9485 1.5854 0.3809 0.3260 0.3061 0.3809 0.3500 0.2916 0.1960 0.2099 0.2128 0.2395 0.2560 0.2356
`x` 0.0 0.2 0.4 0.6 0.8 1.0
Table 2 indicates that the Zn ions occupy (A) site for the samples x = 0.2-0.8. The Zn cation prefers to situate at (A) site as the Cu concentration decreases from [B] site, at the same time Fe cation shifts from tetrahedral (A) site to octahedral [B] site in order to balance the relative occupancy by the space group [20]. For x = 0.2 to x = 0.8, zinc shows a preference for tetrahedral (A) site where their 4S, 4P or 5S, 5P electron respectively can perform covalent bond with the 2 P electron of oxygen ions. The magnetic properties of Cu1-xZnxFe2O4 system were studied by hysteresis loop techniques and AC susceptibility measurements. The saturation magnetization (Ms) and magneton number (nB) were
357
Vol. 3, Issue 1, pp. 354-361
measured at room temperature with the help of hysteresis loops. The observed magneton number (nB Obs.) were calculated using the relation,
nB =
where Mw is molecular weight of the composition and Ms is saturation magnetization. According to the Neels two sublattices model of ferrimagnetisms, the calculated magnetic moment N (nB Cal.) per unit formula in B, n B is expressed as,
N n B = M B ( x) M A ( x)
Mw Ms 5585
(1)
(2)
where, MA and MB are the A and B sublattice magnetic moment in B.

Table 3. Compositional variation of Saturation magnetization (s), Magneton number (nB) and Y-K angle (YK) of Cu1-x Zn xFe2O4 system
Comp. x 0.0 0.2 0.4 0.6 0.8 1.0
s (emu/gm) 38.85 62.16 108.78 62.16 38.85 23.31
Obs. 1.66 2.66 4.67 2.67 1.67 1.00
nB (B)
cal 1.0 2.8 4.6 6.4 8.2 5.0
Y-K (degree) -----56.22 73.13 62.18
Fig.4. Variation of magneton number (nB), observed and calculated with Zn content (x).
N The values of Neels magnetic moment n B were calculated by taking ionic magnetic moment of Fe, Cu and Zn. The values of saturation magnetization Ms and magneton number nB are represented in Table 3. It is observed from Table 3 and Fig. 4 that both the values Ms and nB increases up to x = 0.4 then decrease as x increases. Since the diamagnetic Zn2+ ions replaces Fe3+ ions at tetrahedral (A) site due to the prominent inter-sub lattice A-B interaction up to x = 0.4 the net magnetic moment is increased. The behavior of magnetic moment nB is also explained on the basis of Neels theory [21] and the calculated nB value of spin moment and also by Neels two sub-lattice model. The observed and calculated magnetic moment values nB are similar to each other for x = 0.4, estimating that the structure is collinear, then x = 0.4 the observed nB values are lower then calculated values. The reason for this decrease in magnetic moment after x = 0.4 can explain on basis of Yafet-Kittle angle [22]. These low magnetic moments can be explained in term of non-collinear spin arrangement that is
358
Vol. 3, Issue 1, pp. 354-361
the presence of a small canting of [B] site moment with respect to distortion of (A) site. [23]. The values of Yafet-Kittle angle were calculated using relation discussed elsewhere [24], and their values are presented in Table 3. The temperature dependence AC susceptibility measurements for all the samples are shown in Fig. 5. Theoretical calculation of Curie temperature for Zn substituted copper ferrites was obtained by Upadhyas model [25]. The Curie temperature is depends upon the number of active magnetic linkages per magnetic ion per formula unit. Curie temperature calculated by using relation [26].
TC =
M ( x = 0) Tc ( X = 0) n x n( x = 0 ) M x
(3)
where M(x) is the relative weighted total magnetic ions per unit formula, calculated by considering the weighing of magnetic ion M to that of ions as M/Fe where M and Fe are magnetic moments of M and iron ions respectively.
Fig. 5. Variation of T/RT versus temperature. The inset is the variation of Curie content (x).
temperature versus Zn
The observed Curie temperature and Curie temperature calculated theoretically both are given in the inset of Fig. 5. It is observed that T / RT fall sharply, when magnetic state of sample changes from ferri-magnetic to para-magnetic. The magnetic properties of Cu1-xZnxFe2O4 system strongly depends upon Zn2+ ion concentration. The change in Curie temperature is due to influences of cation stoichiometry and their specific sites. The decrease in Curie temperature with the increase in Zn2+ may be explained by A-B exchange interactions. The decrease in Curie temperature due to weak A-B interactions, low magnetic linkages, and can be attributed to increase in distance of magnetic moment of (A) and [B] sites. The observed and calculated Curie temperature values agree with each other.
IV.
CONCLUSIONS
Homogeneous nano-ferrite Cu1-xZnxFe2O4 system was synthesized by simple and economical wet chemical method. In the present study Zn2+ ions causes change in structural and magnetic properties of copper ferrite, initially CuFe2O4 is tetragonal then its phase changes to cubic as Zn content x increases. The unit cell parameter also increases with Zn content x. The particle sizes calculated by different methods are fairly agrees with each other. The Y-K angle increases with Zn content x. The calculated and observed values of magneton number are not agreed close to each other for x > 0.4, suggesting that the structure is non-collinear. The decrease in Curie temperature with Zn content x is due to the decreases in magnetic linkages.
REFERENCE
[1] P. Ravindranathan and K. C. Patil Novel Solid Solution Precursor Method for the Preparation of Ultrafine Ni-Zn Ferrites J. Mater. Sci. 22, 3261-3264. 1987.
359
Vol. 3, Issue 1, pp. 354-361
[2] [3] [4] Ramesh Peelamedu,Craig Grimes, Dinesh Agrawal, and Rustum Roy, Ultralow Dielectric Constant NickelZinc Ferrites Using Microwave Sintering J. Mater. Res.,18, 2292- 2295,2003 A. K. M. Akther Hossain, M. Seki, T. Kawai, and H. Tabata Colossal Magnetoresistance In Spinal Type Zn 1-xNix Fe2O4, J. of Appl.Phy.96,1273-1275,2004. Sagar E. Shirsath, R. H. Kadam, Anil S. Gaikwad, Ali Ghasemi, and Akimitsu Morisako, Effect of Sintering Temperature and The Particle Size On The Structural And Magnetic Properties of Nanocrystalline Li0.5Fe2.5O4 J. Magn. Magn. Mater. 323, 3104-3108,2011 H.M. Zaki and S.F. Mansour, X-Ray and Ir Analysis Of Cu-Si Ferrite J. Phys. Chem. Solids, 67,16431648,2006. R.G. Kulkarni, S. Bimal, H.H. Joshi, G.J. Baldha, Magnetic Properties Of Copper Ferrite Aluminates J. Magn. Magn. Mater. 159,375-380,1996. A. Mazln, H.M. Zake, Ti4+ And Ge4+ Ionic Substitution In Cu-Ferrite, Electrical Conductivity and Thermoelectric Power J. Magn. Magn. Mater. 248,200-215,2002. S.S. Shinde, K.M. Jadhave, G.K. Bichile, B.S. Trivedi, R.G. Kulkarni, Magnetic Properties Of The Mixed Spinel Co1+XSixFe22xO4 Bull. Mater. Sci. 21,409.-413,1998. V.V. Parfenov R.A. Nazipov, Effect Of Synthesis Temperature On The Transport Properties Of Copper Ferrites Inorg. Mater. 39,1315-1319,2003. B.J. Evans,and S.S. Hanfier, Mssbauer Resonance Of Fe57In Oxidic Spinels Containing Cu And Fe.J. Phys. Chem. Solids 29,1573 -1588,1968. R.A. Mecurrie, Ferromagnetic Material Structure And Properties, (Academic London 1994). R. G. Kulkarni and V. U. Patil, Magnetic Ordering In Cu-Zn Ferrite. Mater. Sci. 17, 843-848,1952 V.U. Patil,and R.G. Kulkarni, Magnetic Properties Of Cu-Zn Ferrite Investigated By Mssbauer Spectroscopy sol. State Comm. 31, 551-555,1979. Muhammad Ajmal ,and Asghari Maqsood, Structural, Electrical And Magnetic Properties Of Cu1XZnxFe2O4 Ferrites (0 X 1)J. Alloy. Compd. 460,54-59,2008. Yen-Pei Fu, Ko-Yin Pan, Cheng-Hsiung Lin, Microwave-Induced Combustion Synthesis of Ni0.25Cu0.25Zn0.5 Ferrite Powders And Their Characterizations Mater. Lett. 57,291-296, 2002 D. Markovec, M. Drofenik, A. Znidarsic, Hydrothermal Synthesis Of Manganese Zinc Ferrite Powders From Oxides J. Amm. Ceram. Soc. 82 , 1113 -1120.,1999. S.A. Patil, B.L. Patil, S.D. Lotke, R.N. Patil, Stability Of Phases In Copper Ferrite Phase Trans. 56,2127,1996 I. H. Gul, A.Z. Abbasi, F Amin, M. Anis-Ur Rehman, A. Maqsood, Physical, Electrical And Magnetic Properties Of Nanocrystalline Zr-Mn-Co Prepared By Co-Precipitation Route Scr. Mater 56,497 500.,2007 P.U. Sharma, K.B. Zankat, V.V. Sivakumar, Ravi Kumar and K.B. Modi ,Dielectric Performance of Nanocrystalline Yttriun Iron Garnet, Proceeding Of 52 Dae India 293-294,2007 M.G. Buerger, Crystal Structure Analysis, (Wiley, New York 1960). A.K. Singh, T.C. Goel, R.G. Mendiratta Effect of Cation Distribution on the Properties of Mn0.2ZnxNi0.8x2Fe2O4 Solid State Comm. 125, 121-129.2003 Neel L. Ann. Ferromagnetism And Anti-Ferromagnetism .Physique 3,137. 1948 Y. Yafet and Kittle, C X-Ray Diffraction and Magnetic Measurements of the Fe-Cr Spinels Phys. Rev. 87, 290.,1952 J.B. Goodenoug. Metalic Oxides Prog. Solid State. Chem. 5, 146 -399,1971. R.V. Upadhya, G.J. Baldha A Simplified Model to Calculate the Curie Temperature Of Ferrimagnetic Spinels Ind. J. Phys. A 63(8) 835.1989. S.M. Patange, Sagar E. Shirsath, B.G. Toksha, S.S. Jadhav, S.J. Shukla, K.M. Jadhav, Cation Distribution By Rietveld, Spectral And Magnetic Studies Of Chromium-Substituted Nickel Ferrites J. Appl. Phys. A 95),429-434,2009
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
[19] [20] [21] [22] [23] [24] [25] [26]
Authors
Kishan Lohar is working at Shri Krishna Mahavidyalaya, Gunjoti as Associate Professor and Head Department of Chemistry He completed M.Sc. (1990) and Ph.D. (2005) from Dr. Babasaheb Ambedkar Marathwada University, Aurangabad. His research interests are Organic synthesis, Material science, Nanomagentic material, Green Synthesis. He has published Twenty two research papers in reputed international, national journals and in national and international conferences.
360
Vol. 3, Issue 1, pp. 354-361
Sunil Patange is currently working at Shri Krishna Mahavidyalaya, Gunjoti as Associate Professor and Head Department of Physics. He completed M.Sc. (1992) and Ph.D. (2007) from of S.R.T.M. University, Nanded. His research interest Material science, Nanomagentic material, composite material. He has published Twenty five research papers in reputed international, national journals and in international and national journals conferences. Sagar E. Shirsath is currently working at Vivekanand College Aurangabad, Department of Physics. He completed M.Sc. (2001) and Ph.D. (2010) from Dr. Babasaheb Ambedkar Marathwada University, Aurangabad He working as fourteen Journal referee and published fifty-two papers in reputed international and national journals His research interest is in Material science, Nanomagentic material, composite material.
361
Vol. 3, Issue 1, pp. 354-361
IMPLEMENTATION AND CONTROL DIFFERENT MULTILEVEL INVERTER TOPOLOGIES FOR CURRENT WAVEFORM IMPROVEMENT
T. Shanthi and K. Karthikadevi
Department of Electrical and Electronics Engineering, Kumaraguru College of Technology, Coimbatore, Tamil Nadu, India
ABSTRACT
Multilevel converters offer advantages in terms of the output waveform quality due to the increased number of levels used in the output voltage modulation. This advantage is particularly true for cascade H-bridge (CHB) converters that can be built to produce a large number of levels owing to their modular structure. In this paper, the NPC connected with the H-bridges (HBs) is used for improving the voltage waveform quality. Further more, NPC converter can be replaced with flying capacitors to get an improved output current waveform. This topology requires only a main dc-link voltage and presents a easy way to increase the output voltage levels by increasing the number of cells. The experimental results show the good performance of the proposed modulation technique and the resulting improvement in the output current waveform are provided.
KEYWORDS: Current control, harmonic distortion, power electronics
I.
INTRODUCTION
The multilevel topologies can be classified into three main categories: the neutral point clamped (NPC)[1], the flying capacitors[2], and the cascade H-bridge (CHB) converters. The three-level NPC bridge is probably the most widely used topology for medium-voltage ac motor drives and pulsewidth modulation (PWM) active rectifiers[3],[4].NPC converters with more levels are also possible, although there are significant problems in the balancing of their dc link capacitor voltages[5],unless modified modulation strategies or an additionally circuitry is used. In applications with no active power transfer, such as in reactive power compensation, where the converter can operate without the rectifier front end, the CHB is a highly attractive Solution. In recent years, an increased interest has been given to hybrid topologies integrating more than one topology in a single converter. In this topology, the NPC is used to supply the active power while the H-bridges (HBs) operate as series active filters, improving the voltage waveform quality by only handling the reactive power. In this way, this topology reduces the need for bulky and expensive LCL passive filters, making it an attractive alternative for large power applications [6],[7]. In this paper, the control strategy for the FLYING CAPACITOR- HB s hybrid converter, previously introduced is experimentally verified which improves the quality of the output current waveform.
II.
HYBRID POWER CIRCUIT
2.1 Power circuit

The 3-phase, 3-level NPC multilevel inverter consists of twelve IGBTs, six clamping diodes and two capacitors. The clamping diodes are connected in such a way that it blocks the reverse voltage of the capacitor. Two capacitors have been used to divide the DC link voltage into three voltage level i.e. +Vdc, 0V and Vdc, thus the name of 3-level.In this work, twelve triggering signals are needed for
362
Vol. 3, Issue 1, pp. 362-370
the 3-level NPC inverter. These signals should be synchronized with the AC supply voltage. With this NPC circuit the H-BRIDGE was connected. The considered hybrid topology is composed by a traditional three-phase three-level NPC inverter, connected with a single phase HB inverter in series with each output phase[6]-[8] .The power circuit is illustrated in Fig. 1, with only the HB of phase a shown in detail.
Fig. 1. Hybrid topology power circuit.
In the hybrid topology considered, the NPC inverter provides the total active power flow. For a highpower medium-voltage NPC, there are advantages to using latching devices, such as integrated gatecommutated thyristors (IGCTs), rather than insulated-gate bipolar transistors (IGBTs) due to their lower losses and higher voltage blocking capability[6],[8],[9], imposing a restriction on the switching frequency. In this paper, an NPC operating at a low switching frequency (of 250 Hz) is considered. In contrast, the HBs are rated at a lower voltage and need to be commutated at a higher frequency for an effective active filtering effect. This calls for the use of the IGBT. The proposed converter, shown in Fig. 1, can be analysed from two different points of view. The first interpretation is as a single hybrid multilevel inverter with a nine-level phase voltage, achieved by the cascade connection of a three-level NPC leg and an HB per phase. The second interpretation is as an NPC converter with a series active filter that compensates for the harmonic content introduced by the low switching NPC stage. For the modulation of the NPC inverter, the selective harmonic elimination (SHE) method has been selected. This method has the advantage of very low switching frequency and, hence, low switching losses while eliminating the low order harmonics.
Fig.2.Three-level NPC-SHE phase voltage waveform.
2.2 NPC SHE

A qualitative phase output voltage waveform is presented in Fig. 2 considering a five-angle realization, so five degrees of freedom are available. This enables the amplitude of the fundamental component to be controlled and four harmonics to be eliminated. Since a three-phase system is
363
Vol. 3, Issue 1, pp. 362-370
considered, the triple harmonics are eliminated at the load by connection, and hence, they do not require elimination by the modulation pulse pattern. Thus, the 5th, 7th, 11th, and 13th harmonics are chosen for the elimination. For variable frequency drive applications, the number of angles must be varied in order to maintain a near constant switching frequency at any operation point [10].
III.
CONTROL STRATEGY
3.1 HB CONTROLLER
Each series HB converter is independently controlled by two complementary references, as shown in Fig. 3 .The first reference corresponds to the inverse of the harmonics remaining from the SHE pulse pattern, calculated as described in the previous section from the difference between the NPC pulsed voltage pattern and its sinusoidal voltage reference. This calculation provides a fast and straightforward distortion estimation allowing for simple feed-forward compensation. Moreover, this voltage does not have a fundamental voltage component, and hence, it does not affect the floating average dc-link capacitor voltage. Nevertheless, to achieve a start-up capacitor charge and to compensate the voltage drift due to transient operation, an additional reference component for dclink voltage control is included. This second component of the voltage reference vaa (f1) corresponds to a signal in phase with the load current. During operation, the fundamental load current is generated by the NPC converter. In order to synchronize the voltage reference vaa (f1) with this current, a phase lock loop (PLL) algorithm is used, which guarantees a zero phase shift between both signals and therefore maximizes the active power transfer to the capacitors for any power factor. The magnitude of this voltage reference is obtained from the dc-link voltage controller shown in Fig. 3.
Fig. 3 . HB control diagram for phase a.
For the design of this voltage controller, the dynamic model(1) of the dc-link voltage vHa as a function of vaa_ is used.
(1)
An undesirable characteristic of (1) is its nonlinearity with respect to vHa. This can be dealt with by linearization or by simply introducing the auxiliary variable x = v^2Ha and controlling x directly.
IV.
SIMULATION DIAGRAM OF NPC WITH H-BRIDGE INVERTER
364
Vol. 3, Issue 1, pp. 362-370
Fig 4.Simulation diagram of NPC with H-BRIDGE inverter.
V.
SIMULATION RESULTS OF NPC WITH H-BRIDGE INVERTER
Fig 4.a Observed output waveform-NPC
Fig 4.b. load voltage waveform
365
Vol. 3, Issue 1, pp. 362-370
Fig 4.c.load current waveform
VI.
THD ANALYSIS OF NPC AND H-BRIDGE
Fig.5. Harmonic analysis of NPC with H-BRIDGE
The operation of the hybrid converter shows almost a complete elimination of these characteristic harmonics, resulting in a current THD of 2.27%.
VII.
IMPLEMENTATION OF ANOTHER TOPOLOGY WITH FLYING CAPACITORS
Fig .6. Power circuit with flying capacitors and H-bridge inverter
366
Vol. 3, Issue 1, pp. 362-370
The operation of the above circuit is almost similar to the first topology. The capacitor connected across the switches charges when Sa1 is ON and Sa2 remains OFF. At this condition the output voltage is Vdc/2.Then the capacitor starts discharging when Sa1 turned OFF with the output voltage of Vdc/2.
A. External current control loop
Fig. 7. Simplified current control loop for the proposed topology, including SHE for the flying capacitor (the control loops for the HBs are not shown).
It is important to note that, in applications with low frequency switching patterns, such as the SHE modulation, the use of direct synchronous sampling of the currents is not adequate to obtain the fundamental current because the switching harmonics do not cross zero at regular intervals. Instead, the observers are needed to extract the fundamental current values; otherwise, complex nonlinear control schemes are required. In this paper, this problem is overcome by the compensating effect of the series-connected HBs. This effectively simplifies the outer load current control loop design, resulting in a standard dq frame linear current regulator a. In a regenerative operation, such as active front end applications for regenerative drives, the power flow needs to be controlled bidirectionally.
VIII.
SIMULATION DIAGRAM INVERTER
OF FLYING
CAPACITOR
WITH H-BRIDGE
Fig 8.Simulation diagram of FLYING CAPACITOR with H-BRIDGE inverter.
367
Vol. 3, Issue 1, pp. 362-370
Fig.8.a Observed output waveform - flying capacitor
Fig.8.b. load voltage waveform
Fig 8.c.load current waveform
IX.
THD ANALYSIS OF FLYING CAPACITOR AND H-BRIDGE
The operation of the Flying capacitor with H-Bridge converter shows almost a complete elimination of these characteristic harmonics, resulting in a current THD of 1.36%.
368
Vol. 3, Issue 1, pp. 362-370
Fig 9.Harmonic analysis of FLYING CAPACITOR with H-BRIDGE
X.
CONCLUSION
This paper has presented the series connection of a SHE modulated NPC and an HB multilevel inverter with a control scheme to control the floating voltage source of the HB stage. The addition of the HB series active filter or additional converter stage is not intended to increase the power rating of the overall converter. Rather, the main goal is to improve, in a controllable or active way, the power quality of the NPC Bridge which may have a relatively low switching frequency. And in the proposed topology the flying capacitor with H-Bridge converter was connected and the current waveform was improved and the THD was reduced to 1.36%.
REFERENCES
[1] J. S. Lai and F. Z. Peng, Multilevel convertersA new breed ofpower converters, IEEE Trans. Ind. Appl., vol. 32, no. 2, pp. 509517,May/Jun. 1996. [2] T. Meynard and H. Foch, Multi-level choppers for high voltage applications,Eur. Power Electron. J., vol. 2, no. 1, pp. 4550, Mar. 1992. [3] J. Rodrguez, J. Pontt, G. Alzamora, N. Becker, O. Einenkel, and A.Weinstein, Novel 20 MWdownhill conveyor system using three-levelconverters, IEEE Trans. Ind. Electron., vol. 49, no. 5, pp. 10931100,Oct. 2002. [4] A. Yazdani and R. Iravani, A neutral point clamped converter system for direct drive in variable speed wind power unit, IEEE Trans. EnergyConvers., vol. 21, no. 2, pp. 596607, Jun. 2006. [5] J. Pou, R. Pindado, and D. Boroyevich, Voltage-balance limits in fourlevel diode-clamped converters with passive front ends, IEEE Trans. Ind.Electron., vol. 52, no. 1, pp. 190196, Feb. 2005. [6] F. Z. Peng, J.-S. Lai, J. W. McKeever, and J. Van Coevering, A multilevel voltage-source inverter with separate dc sources for static VAr generation, IEEE Trans. Ind. Appl., vol. 32, no. 5, pp. 11301138,Sep./Oct. 1996. [7] M. Veenstra and A. Rufer, Control of a hybrid asymmetric multilevel inverter for competitive mediumvoltage industrial drives, IEEE Trans. Ind. Appl., vol. 41, no. 2, pp. 655664, Mar./Apr. 2005. [8] P. Steimer andM.Manjrekar, Practical medium voltage converter topologies for high power applications, in Conf. Rec. IEEE IAS Annu. Meeting,2001, pp. 17231730. [9] T. Gopalarathnam, M. Manjrekar, and P. Steimer, Investigations on a unified controller for a practical hybrid multilevel power converter, in Proc. IEEE APEC, 2002, pp. 10241030 [10] S. Bernet, R. Teichmann, A. Zuckerberger, and P. Steimer, Comparison of high-power IGBTs and hard-driven GTOs for high-power inverters, IEEE Trans. Ind. Appl., vol. 35, no. 2, pp. 487495, Mar./Apr. 1999. [11] C. Silva, P. Kouro, J. Soto, and P. Lezana, Control of an hybrid multilevel inverter for current waveform improvement, in Proc. IEEE ISIE, 2008,pp. 23292335.
369
Vol. 3, Issue 1, pp. 362-370
Authors Biography
T. Shanthi received her bachelors degree from Institute of Road and Transport Technology, Erode, Tamilnadu, India in Electrical and Electronics Engineering during 1999. She received her Masters degree from National Institue of Technology, Tiruchirappalli, India during 2007. She is currently working as Assistant Professor at the department of Electrical and Electronics Engineering in Kumaraguru College of Technology, Coimbatore, India.
K. Karthikadevi was born in Erode India, in 1989. She received the B.E (Electronics and Instrumentation) degree in Vivekanandha College of Engineering for women, Thiruchengode, India in 2010. Now she is doing M.E (Power Electronics and Drives) in Kumaraguru College of Technology, Coimbatore, India. Her areas of interest are Power Electronics and Drives, Power Quality, and Renewable Energy.
370
Vol. 3, Issue 1, pp. 362-370
PERFORMANCE EVOLUTION OF HIGH SPEED NETWORK CARRYING MULTIMEDIA

Ehab Aziz Khalil
Department of Computer Science & Engineering, Faculty of Electronics Engineering, Menoufiya University, Menouf-32952, EGYPT
ABSTRACT
The performance evolution of a high-speed network, called Fiber Distributed Data Interface (FDDI), carrying video and data traffics is tackled. Also, a proposal aspect to improve the FDDI performance for supporting the asynchronous traffic that suffers low throughput and excessive delay is presented. It is well known that the FDDI network is based on Timed Token Rotation (TTR) protocol for controlling the token rotation time. With TTR protocol, packets are grouped into two classes: synchronous class, which represents video, and asynchronous class, which represents the ordinary data. The study of the performance evolution and the proposal aspect is done by means of computer simulation. The simulation results confirm that the proposal aspect provides FDDI network more powerful for carrying video and data traffics.
KEYWORDS: FDDI, FDDI`s MAC protocol, video and data traffics.
I.
INTRODUCTION
Many multimedia applications are distributed in nature and involve networking and communications. Examples of such applications are video on demands (VoD) for training and education, conferencing, and computer supported collaborative work. There is a great interest in carrying these distributed multimedia applications over existing network (which have been designed originally to carry conventional data applications, such as database access, file transfer, etc.) [1]. It is a known fact that the real-time (video) communication requires high-speed transmission and performance guarantees, and end-to-end delay to be less than 100-200ms [2,3]. The FDDI high speed network is the suitable for real-time communication because of its high speed transmission and performance guarantees it provides [4-8]. The FDDI is also overcome the bandwidth saturation problem of the 10Mbps Ethernet and the 4 16 Mbps Token Rings due to its high speed transmission [5]. FDDI is one of the HighSpeed Local Area Networks (HS-LANs) standard [9], and provides a data rate of 100Mbps[8]. The basic topology is a dual (counter-rotating) ring consisting of a primary ring for transfer and a secondary ring for redundancy. Access to the ring is by a timed-token mechanism similar to that used for token bus [11]. The Media Access Control (MAC) layer is one of the FDDI`s four layers, defines addressing, scheduling, and routing of data. It is also communicate with higher layer protocols, such as TCP/IP, SNA, IPX, and DECnet. It operates according to the so-called TTR Protocol. During ring initialization process, a protocol parameter called Target Token Rotation Time (TTRT) is determined, and that indicates the expected target token rotation time. Each station is assigned a fraction of the TTRT known as Synchronous Allocation (SAi), and that represents the maximum permitted time for a station number i to transmit its real-time messages (i.e., digital video) whenever it receives the token. Meanwhile, the Token Rotation Time (TRT) controls the transmission of non-real time messages (i.e., data packets) by measuring the time between successive token arrivals at a station. When a station receives the token, the value of TRT is copied into Token Holding Time (THT), that is only if the token arrived early, then the station can transmit its non-real time messages until the THT expires or all its data packets are transmitted, whichever occurs first [12,13]. As mentioned above, the FDDI network provides both bounded transmission delays and guaranteed bandwidth for real-time traffic. This is because the transmission delay controlled by the rings TTRT, which guarantees each station to have a chance to transmit its real-time messages at least once every (2*TTRT) units of time (in the
371
Vol. 3, Issue 1, pp. 371-380
worst case). It also assures the average time between two consecutive token`s visits to a node not to exceed the TTRT. The bandwidth of real-time traffic for a station i is guaranteed by SAi.
II.
A RELATED WORK
An extensive research has been conducted on FDDI networks, where the objectives were mainly focused on maximizing the throughput and minimizing the delay to guarantee meeting the individual message deadline. E. A. Khalil in [14] proposed a new protocol of FDDI's MAC to control the maximum TTRT in the worst case, the new protocol of FDDIs MAC increasing both the throughput of asynchronous traffic and the capacity of synchronous sources. Wang, and etc. [15,16] proposed a new approach for allocating synchronous bandwidth, which can be easily implemented in practice. A. P. Jayasumana, and etc. [17] presented a model can be used to evaluate the throughput of FDDI networks for different priority threshold values, the model can also be used to determine the control parameters of the priority scheme to meet the throughput requirements of different classes of traffic. Sijing Zhang and etc in [18-27] have done extensive research in which conducted on using the timed token protocol to guarantee timely transmission of messages in a communication environment. In [28] Ching-Chih Han and etc., were formally proven that there did not exist any optimal local Synchronous Bandwidth Allocation (SBA) scheme, they also proposed an optimal global SBA scheme. Daoxu Chen and etc. in [29] first they proposed a new taxonomy of SBA schemes based on the underlying strategy used to partition the synchronous bandwidth, second they presented an analysis of the timing properties of the FDDI-M protocol, using the Worst Case Achievable Utilization (WCAU) as the performance metric. Y Ofek and etc. in [30] described and evaluated a protocol for integrating two types of traffic on the MetaRing architecture, the integration mechanism is functionally equivalent to the timed-token function in FDDI, which is a shared media ring protocol. There are many related works and researches were studied different applications of allocating synchronous and asynchronous bandwidths in FDDI networks, and related researches were confirmed that the TTRT is the key network parameter that network manager can use to optimize the performance of the FDDI ring network [31-62]. In this paper the performance characteristics of the FDDI network carrying video and data traffics is studied. The effect of the system parameters such as video packet length, data traffic intensity on the network performance is discussed. The maximum number of real-time traffic the can be carried by the network is also discussed. The proposal aspect for improving service to data traffic is included. The next section describes network architecture. The traffic models and assumptions are discussed in section 3. Section 4, explains the proposal aspects, followed by the simulation results in section 5, finally section 6, and presents the conclusion.
III.
NETWORK ARCHITECTURE
3. A. Assumptions
The network is simulated for FDDI ring network with N stations and 92Km length. All stations are uniformly distributed on the ring, i.e., the station to station propagation delay is the same for all station. Each station has a latency of 0.6s and the latency of the link is the same as the propagation delay of light through the fiber which is 5.085s/Km [63]. According to the FDDI standard, the channel speed (C) is 100Mbps and the maximum packet size is 4500 bytes including header (28 bytes), and thus, the maximum packet transmission time is (36Kb/100Mbps) 0.36ms [64].
3. B. Network Configuration
The network consists of N nodes connected as point-to-point links forming a ring as shown in Figure 1. A special bit pattern called token circulates around the ring providing access control among the active nodes. We consider that the ring latency is Trig , which represents the movement time of the token around the ring without any disturbance by any nodes on the network.
372
Vol. 3, Issue 1, pp. 371-380
Figure 1 FDDI Network
3.C. FDDI`s Medium Access Control Operation

The FDDI`s MAC operates for N stations numbered 1 to N using the TRT mechanism. The stations access to the ring is controlled by the following protocol: 1-With the network`s initialization process, each station is announced a Target Token Rotation Time (TTRT), which equals one half of the request transmission delay (Treq) of its synchronous traffics. The lowest value of (Treq) amog al lstatons is assigned to ring`s TTRT. Then, each station have synchronous traffic should assign a portion of the TTRT to transmit its synchronous packets. The portion of TTRT, denotes bySAi, is greater than or equal to zero. 2- There are two timers at each station namely, Token Rotation Time (TRT), and Token Holding Time (THT). With the first rotation of the token, there is no chance for transmitting any packet. All stations are initialized the following parameters: TRTi = TTRT, THTi = 0, and LCi = 0, (Where i indicaes the station`s number, and LC denotes late counter). 3- Both TRTi and THTi counters count down. However, when TRTi reaches zero before the token arrives at station i, the TRT is rest toTTRT and the token is marked as late (i.e., TRTi = TTRT, and LCi = LCi+1). Then the TRTi starts again to count down and LCi begin incremented by one whenever TRTi expires. Noted that, if LCi > 1, the ring should reinitiate its process [12]. 4- The station which captures the token is the only sation has the right to transmit its packets. The transmission time of the packet is controlled by the timers, meanwhile the packet in transmission progress will not be interrupted until the transmission of the packet is completed. The response of the station that seize the token depends on the arrival time of the token which may be early or late. It is considered early if LCi = 0, and late if LCi > 0 at the time of its arrival. In case of early arrival, the following actions take place: i) THTi = TRTi; ii) TRTi= TTRTi; iii) For a duration equals to SAi the station i tansmits its synchronous traffics, or upto the transmission of the hole message, whichever occurs first.; and iv) The station i enables its THTi to count down, and begins transmitting asynchronous traffics (if any) until THTi reaches zero or the transmission of the whole packets, whichever occurs first. It is important to mention here that if, there is a bandwidth (SAi) belongs to synchronous traffic not used by the station, that bandwidth can be used to transmit asynchronous traffics, only if the token arrives early. In case of late arrival, the following action take place: i) LCi = 0; ii) TRTi counts down upto expire (it is to be noted here that TRTi not reset as in the case of token arrives early); iii) For a duration equals to SAi the station i trasmts its synchronous traffics (if any) or upto the transmission of the whole packets, whichever occurs first; and iv) The asynchronous traffic will not transmitted. 5- Station i passes the token to the next station (i + 1), and so on.
IV.
TRAFFIC MODELS AND ASSUMPTIONS
4.A. Stream Traffic

Figure 2 illustrates the block diagram of a station, which encodes and sends a video stream over a communications network [1,2,65]. A frame is taken in the video camera, and sent as an analog signal into the frame grabber, where it is digitized. Then it is compressed by the encoder. The data produced by the encoder is passed to the station which packetizes the encoded video stream, and sends it over
373
Vol. 3, Issue 1, pp. 371-380
the network. We consider that video streams are encoded according to a coding standard such as H.261 [66], or MPEG [67]. The encoding rate (V) is considered to be V= 1.5Mbps for high quality video conferencing. The sending station uses a fixed video packet length denotes as Pv in bits. Consequently the packets are Generated regularly every Gp second (where Gp = Pv/V sec.).
Figure 2 Block Diagram of a Station Which Encodes and Sends a Video Stream
The maximum queuing delay until a video packet reaches the top of its queue is Qv-max. The transmission time of a video packet is Tv second (where Tv=(Pv + H)/C sec., H is the packet header which equals 28bytes, and C is the data rate which equals 100Mbps). The maximum transmission delay of a video packet (Dv-max) can be given by the following: Dv-max = Gp + Qv-max + Tv + Ds,d sec, where Ds,d represents the propagation delay between source (s) and destination (d) stations. Here, it is to be mention that the video packets are required delivery within a bounded delay (Dbound). This means that, the video packets that exceeding the bounded delay considered lost. The probability of loss (Ploss) of a video packet is given by: Ploss =prob (Dv-max > Dbound).
4.B. Data Traffic

As described in [2,65] a data source is characterized by its message size (Md) and its average message interval time (d). We assume that, a data source generates fixed messages with uniformly distributed interval times. If a message is longer than a maximum data packet size (Pd), it is divided into as many packets as needed. In that case, the packets belonging to the message are placed into the transmit queue as a bulk arrival. The total load of the network can be adjusted by varying the interarrival time (d). That is, if the total number of data sources is Nd, and the arrival rate of data message is in message per second (message/sec.), then the total data load (Td) over the network can be given by: Td =* Nd* Md b/s [68,69]. The maximum data packet transmission delay (Dd-max) can be given by: Dd-max = Qd-max + Td + Ds,d sec. Where Qd-max is the maximum queuing delay faced the target data Packet reaches the top of its queue. Td is the transmission time of a data packet (where Td =(Pd+H)/C sec.). The normalized data throughput (d) (data traffic intensity) of the network can given by: d = Td /C = ( * Nd * Md)/C .
V.
THE PROPOSAL ASPECT
In the proposal aspect, we assume that there are N stations on the ring numbered from 0 to N-1. The token circulates among all the N stations of the ring in a round robin fashion. The stations` access to the ring is then controlled by the following protocol [70]: 1Each station declares its TTRT which equals to Treq, instead of one half of it. The smallest value among them is selected as the ring`s TTRT. Each station has synchronous traffic is then assigned a portion of TTRT to transmit its synchronous messages. Let SAi 0 denotes the portion of TTRT that assigned to station i to transmit its synchronous messages. 2Each station has two timers: Token Rotation Timers (TRT) and Token Holding Ring (THT). During the first token rotation, no packets are allowed to be transmitted and the following parameters are initialized at each stations: TRTi = TTRT, THTi = 0, and LCi = 1. Note that LCi is set to 1 in order to ensure that there is no transmission of asynchronous packets during the second token rotation. 3Set a new parameter called total Asynchronous Time (Ta), represents the total time used to service asynchronous packets through a complete cycle of the token to be Ta TTRT -Tring. This
374
Vol. 3, Issue 1, pp. 371-380
can be realized by setting Ta =TTRT-Ts -Tring, (where Ts is the bounded time used for transmitting asynchronous packet Ts= SAi). If Ts(1-) TTRT, otherwise Ta=TTRT-Tring. Where is a reserved portion of the TTRT allowed for transmitting asynchronous packets, in the worst case, plus the time requested to circulate the token around the ring once without transmitting any packet, such that (Tring/TTRT) 1/2. 4- The TRTi counter always counts down, meanwhile, the THTi counter counts down only when the station i is transmitting or forwarding asynchronous messages. Whenever the THT counter of any station reaches zero before the token arrives it, the token is marked as late by incrementing the station`s late counter (LCi) by one. Also, whenever the TRT counter of any station reaches zero before the token arrives it, the ring recovery process is initialized. 5- Any station which has the token is eligible to transmit its packets. The packet transmission time is controlled by the timer. The station is permitted to transmit a packet only if the transmission can be completed before the timer (THTi or SAi) expires. Whenever station i receives the token, its response depends on whether the token is early late. The token is considered to arrive early to station i if its LCi = 0, and late if its LCi > 0 at the time of its arrival. AIf the token arrives early, the following status takes place: i) TRTi = TTRT. ii) If the station has synchronous packets, it transmits them either within a span of time equals to SAi , or until all the synchronous packets are transmitted, whichever occurs first. iii) The station i enables its THTi counter (i.e., it starts counting down) and begins transmitting asynchronous packets until THTi reaches zero or all its asynchronous packets are transmitted, whichever occurs first. Note that, if a station doesn`t use the guaranteed synchronous bandwidth (SAi), the remaining bandwidth can be used to transmit asynchronous packets. iv) THTi is reset to Ta in order to compute the total time used to service asynchronous packets during the next rotation up to receive the token. B- If the token arrives late, the following status takes place: i) TRTi = TTRT. ii) LCi = 0, iii) The station i can transmit synchronous packets (if any) for either a maximum time SAi, or until all the synchronous packets are transmitted, whichever occurs first. iv) The asynchronous packets will not be transmitted. v) THTi is set to Ta (i.e., THTi = Ta). 6- The station i passes the token to the next station (i + 1).
VI.
SIMULATION RESULTS
An extensive simulation study has been carried out. However, in this paper the focus only on the effects of packet length and number of video sources as parameters on the various network performances. All the measures are for the given values of MPEG (Moving Picture Expert Group) coding rate V=1.5Mbps, maximum bounded delay Dv-max = 30ms, and audio packet loss probability Ploss 0.1%. Figure 3 shows the delay versus Pv for various d. Initially with short Pv, the delay is very high this is because the huge number of packets are generated and waiting for transmission with the short packet length. However, as the Pv increases the delay decreases upto the minimum delay (which corresponding the optimal packet length) reaches the delay increases as Pv increases due to the long delay for the generation of video packet. Figure 4 indicates Nv-max versus Pv, for various bounded delays Dv-max, and packet loss probability Ploss 0.01%. It is to be mentioning that as the Dv-max increases the Nv-max is also increase. This is, because high delay resulting from heavy load, so as Dv-max increases Nv-max is also increase.
375
Vol. 3, Issue 1, pp. 371-380
Figure 3 Average Transmission Delays (Nd=20, Nv=40, Md=1KB, Pd=1024B)
Figure 4 Maximum Number of Video Sources versus Pv.
Figure 5 shows the Nv-max versus Pv for various d (data traffic intensity) and fixed bound delay Dvmax = 30ms. Nv-max increases as the Pv increases upto the saturation limit of the network reaches (depends upon the value of d), Nv-max decreases with the increase of Pv. This is because the optimal packet length has sensitive effect on the maximum network capacity, however, initially with the short Pv, Nv-max increases upto the optimal packet length reaches (correspond to the saturation limit), after that, Nv-max decreases as Pv increases. The effect of d is very clear as d increases Nv-max decreases, this is due to the contribute of data traffic with the video traffic. It is to be mention here that, the bounded delay Dv-max = 30ms the maximum number of the video sources can be carried by the network reaches to 63-64 audio sources and this compared well with the standard FDDI, which can carried up to 66 video sources (100Mbps/1.5Mbps). Figure 6 shows the data throughput versus Nv for various values of d . It is to be noted that as d increases the data throughput is also increase. Meanwhile, as Nv increases the data throughput decreases this is, because the video traffic has highest priority for transmission than data traffic; and at the ring initialization process, the FDDI guarantees a pre-specified value of SAi at each synchronous, however, the total bandwidth for video traffic increases as Nv increases, resulting in the bandwidth for data traffic decreases, that problem has been solved by our proposal aspect.
Figure 5 Maximum Number of Video Sources versus for Various d
Figure 6 Data Throughput versus Nv for Various Values of d
Pv
Figure 7 illustrates the throughput characteristics of asynchronous traffic for standard FDDI and our proposal aspect. The Figure shows that the achievable throughput of our proposal aspect is higher
376
Vol. 3, Issue 1, pp. 371-380
than that of the standard protocol. However, with our proposal aspect as increases, the achievable throughput is also increase. This is because the reserved portion of the TTRT for transmitting asynchronous packets increases as increases. This performance can be seen more clearly in Figure 8, which shows the average token rotation time for the standard FDDI and our proposal aspect under the same conditions of Figure 7. Obviously, the shorter average token rotation time, the longer transmission bandwidth that used for the token passing, resulting in, short bandwidth for asynchronous traffic. The difference between asynchronous throughput is more pronounced with the increasing of the average token rotation time.
Figure 7 Throughput Characteristics of Asynchronous Traffic for Standard FDDI and Modified FDDI
Figure 8 Average Token Rotation Time of Standard FDDI and Modified FDDI Protocol
VII.
CONCLUSION
This paper presents the performance evolution of FDDI network carrying video and data traffics. The results compared well with the standard FDDI network, and show that the network parameters such as packet length, tolerable bound delay, data traffic intensity, and loss constraints can be adjusted to increase the FDDI`s network capability for supporting multimedia applications. The simulation results are also confirmed that the FDDI is suitable for synchronous traffic because of high speed and performance guarantees it provides. However these guarantees are achieved at the expense of asynchronous traffic, because synchronous traffics are given higher priority over asynchronous traffics, resulting in excessive delays and low throughput for asynchronous traffics. Our proposal aspect has tackled these problems, and confirms that the FDDI is suitable not only for the synchronous traffics but also for asynchronous traffics. Just looking at the high throughput of asynchronous traffic for confirmation.
REFERENCES
[1] F.A.Tobagi, and I.Dalgic, ``Performance Evaluation of 10Base-T and 100Base-T Ethernets Carrying Multimedia Traffic,`` IEEE J. of SACs, Vol.14, No.7, Sept. 1996. [2] I.Dalgic, W.Chien, and F.A.Tobagi, `Èvaluation of 10Base and 100Base-T Ethernets Carrying Video, Audio, and Data Traffic,`` IEEE INFOCOM`94 Procs., Toronto, Canada, June 1994. [3] F.Fluckiger, `Ùnderstanding Networked Multimedia Applications and Technology,`` Prentice Hall, 1995. [4] M.Hamdaoui, and P.Ramanathn, ``Deferring Real-Time Traffic for Improved Non-Real-Time Communication in FDDI Networks,`` IEEE Trans. on Computers, Vol.44, No.12, Dec. 1995. [5] K.G.Shin, and Q.Zheng, ``FDDI-M: A Scheme to Double FDDI`s Ability of Supporting Synchronous Traffic,`` IEEE Tran. On Parallel Distributed Systems, Vol.16, No.11, pp.1125-1131, Nov.1995. [6] A.Shah, and G.Ramakrishnan, ``FDDI: A High Speed Network,`` PTR Prentice Hall, Inc. 1994 [7] M.Yabe, T.Ideguchi, T.Fuji, Y.Koui, Y.Baba, and N.Okazaki, ``Proposal of a Data Transfer Protocol Suitable for Multimedia Communications on Shared Media LAN,`` Procs. APCC/OECC99, Beijing, Chaina, Oct. 1999. [8] M.Yabe. T.Fuji, Y.Koui, T. Ideguchi, Y.Baba, and N.Okazaki, ``Proposal of Synchronous Data Transfer Protocol Suitable for FDDI With RSVP Mechanism,`` Procs. IEEE APCC/ICCC`98, Singapore, Nov.1998. [9] FDDI Token Ring Media Access Control (MAC), ANSI X3T9.5, Doc.X3.139/1987 (ISO 9312-2/ 1989).
377
Vol. 3, Issue 1, pp. 371-380
[10] H.P.Davids, T.Meuser, and O.Spaiol, ``FDDI: Status and Perspectives,`` Computer Networks and ISDN Systems, Vol.26, No.6-8, March 1994. [11] H.R.van As, ``Media Access Techniques: The Evolution Towards Terabits/s LANs and MANs,`` Computer Networks and ISDN Systems, Vol.26, No6-8, March 1994. [12] C.M.Bakes, and F.N.Goldberg, `Àpplications of Fibre Optic Networks in High Technology Research,`` Intel. J. of Computer Applications in Technology, Vol.8, No.3-4, 1995. [13] G.Agrawal, B.Chen, W.Zhao, and S.Davari, ``Guaranteeing Synchronous Messages Deadlines With Timed Token Medium Access Control Protocol,`` IEEE Trans. On Communications, Vol.43, No.3, March 1994. [14] E. A. Khalil, "High Throughput of Asynchronous Traffic and Maximum Capacity of Synchronous Sources With the New FDDI's MAC Protocol," IJESET, Vol. 1, Issue, 2, pp. 27-36, February, 2012. [15] Wang, J. Zhang, S. Maple, C. " Guaranteeing the Timely Transmission of Periodic Messages with Arbitrary Deadline Constraints Using the Timed Token Media Access Control Protocol ," In the Communications, IET, Volume: 5 Issue: 4, pp. 519 533, March 2011. [16] Jun Wang, Sijing Zhang, Carsten Maple, Zhengxu Zhao, "Guaranteeing Hard Real-Time Traffic with Legitimately Short Deadlines with the Timed Token Protocol," In the Journal of Computer Standards & Interfaces , Volume 31 Issue 3, March 2009. [17] A. P. Jayasumana and P. N. Werahera, "Performance of Fibre Distributed Data Interface Network for Multiple Classes of Traffic," IEEE Proc., pt. E, vol. 137, no. 5, pp. 401-408, Sept. 1990. [18] Sijing Zhang, Alan Burns, Jing Chen, E. Stewart Lee, "Hard Real-Time Communication with the Timed Token Protocol: Current State and Challenging Problems," Published in: Journal Real-Time Systems, Volume 27 Issue 3, September 2004. [19] Sijing Zhang, Alan Burns, Tee-Hiang Cheng, "Cycle-Time Properties of the Timed Token Medium Access Control Protocol," Journal IEEE Transactions on Computers, Volume 51 Issue 11, November 2002. [20] Sijing Zhang, Alan Burns, Ahmed Mehaoua, E. Stewart Lee, Hongji Yang, "Testing the Schedulability of Synchronous Traffic for the Timed Token Medium Access Control Protocol," Journal Real-Time Systems, Volume 22 Issue 3, May 2002. [21] Sijing Zhang, E. Stewart Lee, " Efficient Global Allocation of Synchronous Bandwidths for Hard Real-Time Communication with the Timed Token MAC Protocol," In the Proceeding of RTCSA '99 Proceedings of the Sixth International Conference on Real-Time Computing Systems and Applications, December 1999. [22] Sijing Zhang, E. Stewart Lee, " The Worst-Case Scenario for Transmission of Synchronous Traffic in an FDDI Network," In the Proceeding LCN '99 Proceedings of the 24th Annual IEEE Conference on Local Computer Networks, October 1999. [23] Sijing Zhang, Alan Burns, "Timing Properties of the Timed Token MAC Protocol," In the Proceeding of IC3N '97 Proceedings of the 6th International Conference on Computer Communications and Networks, September 1997, IC3N '97. [24] Sijing Zhang, Tee-Hiang Cheng, K. R. Subramanian, Liren Zhang, Chee Heng Tan, "Testing the Feasibility of Synchronous Bandwidth Allocation for Time-Critical Communication in FDDI Networks," Proceeding of the IC3N '97 Proceedings of the 6th International Conference on Computer Communications and Networks, September 1997, IC3N '97. [25] Sijing Zhang, Tee-Hiang Cheng, Hongji Yangt, K.R. Subramanian, Chee Heng Tan and Liren Zhang, "Improving the Normalized Proportional Allocation Scheme for Time-Critical Communication in FDDI Networks," International Conference on Information, Communications and Signal Processing ICICS '97 Singapore, pp. 1436-1440, 9-12 September 1997. [26] Sijing Zhang, Alan Burns, Andy Wellings, "An Efficient and Practical Local Synchronous Bandwidth Allocation Scheme for the Timed-Token MAC Protocol," In the Proceeding of the INFOCOM'96 Proc. of the 15th annual joint conference of the IEEE computer and communications societies conference on The conference on computer commun. Vol. 2, March 1996 INFOCOM'96. [27] Sijing Zhang and Alan Burns, "An Optimal Synchronous Bandwidth Allocation Scheme for Guaranteeing Synchronous Message Deadlines with the Timed-Token MAC Protocol ," IEEWACM TRANSA~IONS ON NETWORKING, Vol. 3. No. 6 , pp. 729-741, December 1995. [28] Ching-Chih Han; Shin, K.G.; Chao-Ju Hou, " Synchronous Bandwidth Allocation for Real-Time Communications with the Timed-Token MAC Protocol," IEEE Transactions on Computers, Volume: 50 Issue:5, pp. 414 431, May 2001. [29] D. Chen, V.C.S. Lee, E. Chan, "On the Ability of the FDDI-M Protocol to Support Real-Time Traffic," In the Fifth International Conference on Real-Time Computing Systems and Applications (RTCSA'98), Hiroshima, Japan, Page 51, October 27-29, 1998. [30] Y Ofek, K Sohraby, Ho-Ting Wu, "Integration of Synchronous and Asynchronous Traffic on the MetaRing and its Performance Study," In IEEEACM Transactions on Networking'97, Vol.5, Issue 1, pp. 111-121, 1997. [31] Nicholas Malcolm and Wei Zhao, "The Timed-Token Protocol for Real-Time Communications," In the IEEE Computer, pp.35-41, January 1994.
378
Vol. 3, Issue 1, pp. 371-380
[32] G. Agrawal, B. Chen, B. Chen, S. Davari, "Guaranteeing Synchronous Message Deadlines with the Timed Token Medium Access Control Protocol ," In the IEEE TRANSACTIONS ON COMPUTERS, Vol. 43 No. 3, pp. 327-339, March 1994. [33] Chug-Ju Chang, AND Hsin-Huei Wu, "Analysis of Synchronous and Asynchronous Traffic in an FDDI Network," TENCON '93. Proceedings. Computer, Communication, Control and Power Engineering.1993 IEEE Region 10 Conference on 19-21 Oct 1993, IEEE TENCON '93 / Beijing, Vol.3, pp. 187-189, China, 1993. [34] Qin Zheng, Kang G. Shin, "Synchronous Bandwidth Allocation in FDDI Networks," TR93-06 December 1993, MITSUBISHI ELECTRIC RESEARCH LABORATORIES, Cambridge Research Center, http://www.merl.com. [35] B. Chen, G. Agrawal, and W. Zhao, "Optimal Synchronous Capacity Allocation for Hard Real Time Communications With the Timed Token Protocol," in Proc. Real-Time Systems Symposium. IEEE, Dec. 1992. [36] G. Agrawal, B. Chen, W. Zhao and S. Davari, "Guaranteeing Synchronous Message Deadlines with Timed Token Medium Access Control Protocol," Tech. Rep. 92-006, Comput. Sci. Dept., Texas A&M Univ., Apr. 1992. [37] Willebeek-LeMair, M.; Schaffa, F.; Patel, B., " Isochronous versus Synchronous Traffic in FDDI," In the Proc.., of 17th Conference of Local Computer Networks, pp. 100 - 109, 13-16 Sep. 1992. [38] Raj Jain, "Performance Analysis of FDDI Token Ring Networks: Effects of Parameters and Guidelines for Setting TTRT," In the IEEE, Letter, pp.16-22, May 1991. [39] M. Tan emann and K. Sauer, "Performance Analysis of the Timed Token Protocol of FDDI and FDDI-II," IEEE Globelcom, pp. 271-278, 1991. [40] Y. Y. Yang and R. Sankar, "Maximizing FDDI Network Performance by Parameter Tuning," IlEE Globecom, pp. 1431-1439, 1991. [41] H. Takagi, "Effects of the Target Token Rotation Time on the Performance of a Timed-Token Protocol," in Proc. Performance '90, Edinburgh, Scotland, pp.363-370, Sept. 1990. [42] Amarvir Singh,"Throughput Analysis of Ethernet and Fiber Distributed Data Interface Using OPNET IT Guru Academic Edition 9.1," Research Cell: An International Journal of Engineering Sciences ISSN: 2229-6913 Issue,Vol. 4. pp.455-461, Sept 2011. [43] M A Al-Shargabi, A S Ismail, "Utilizing Optical Burst Switching Networks for the Internet Real-Time traffic," 2008 International Symposium on High Capacity Optical Networks and Enabling Technologies, , pp.63-68, HONET 2008 (2008). [44] Ping Wang, Weihua Zhuang, "A Token-Based Scheduling Scheme for WLANs Supporting Voice/Data Traffic and its Performance Analysis," IEEE Transactions on Wireless Communications (2008) Vol.7, Issue 5, pp.1708-1718, Publisher: IEEE 2008. [45] J Wyatt, D Habibi, I Ahmad, H Zen, "Providing QoS for Symmetrical Voice/Video Traffic in Wireless Networks ," ICON 2007 Proceedings of the 2007 15th IEEE International Conference on Networks, pp.312317, (2007). [46] Ping Du, Shunji Abe, "Burst Assembly Method with Traffic Shaping for the Optical Burst Switching Network ," IEEE Globecom (2006). [47] Ruyan Wang, Dapeng Wu, Keping Long, Weile Zhu, "A Low-Latency Burst Assembly Technique in Optical Burst Switching Networks," 2006 International Conference on Communications Circuits and Systems (2006) Volume: 00, pp.1865-1868, Publisher: IEEE 2006. [48] L.Pushparani Devi, S. Birendra Singh and Sanasam Bimol, A new Analysis of System Utilization Performance for Ethernet Networking System, IEEE 802.11e EDCA, IEEE Transactions on Comm., 2005. [49] George N Rouskas Teng Jing, "Traffic Engineering Approach to Path Selection in Optical Burst Switching Networks ," Journal of Optical Networking (2005), Vol.4, Issue 11, pp.759-777, 2005. [50] Y. Y. Yang and R. Sankar, Performance Analysis of FDDI, IEEE Network, September 2004. [51] William J. Cronin, Jerry D. Hutchison, K.K. Ramakrishnan and Henry Yang, A comparison of High Speed LANs, Proc. of IEEE Inte'l Symposium on Network Computing and Applications, 2003. [52] Jiangtao Luo, Qingji Zeng, Hao Chi, Zhizhong Zhang, Huandong Zhao, "The Impacts of Burst Assembly on the Traffic Properties in Optical Burst Switching Networks," Inte'l Conference on Comm. Technology Proceedings 2003 ICCT'03, Issue 69990540, pp. 521-524, 2003. [53] S Dharmaraja, A S Alfa,"Performance Analysis of 3G Wireless Networks Supporting Voice, Data and Video Traffic," Proc. of The Inte'l Conference On Wireless Networks 2003, pp.495-498, 2003. [54] Shriram Patwardhan, High Speed Optical Networking with Ethernet, proceedings of 24th conference on Local Computer Networks (LCN02), 2002. [55] Timothy J King, The fiber distributed data interface (FDDI) A New Generation Standard for Local Area Networks, http://www.citeseerx.edu/
379
Vol. 3, Issue 1, pp. 371-380
[56] Joseph Kee-Yin Ng, Victor Chung-sing Lee, "Performance Evaluation of Transmission Schemes for RealTime Traffic in a High-Speed Timed-Token MAC Network ," Journal of Systems and Software (2000), Vol. 54, Issue 1, pp.41-60, 2000. [57] Nazy Alborz, Maryam Keyvani, Milan Nikolic, and Ljiljana Trajkovic, Simulation of Packet Data Networks Using OPNET, http:// www.citeseerx.edu. [58] G. Bucci, A. Del Bimbo and S. Santini, Design and Performance Evaluation of High Throughput Ethernet/FDDI Interconnect, Proceedings of Seventh International Conference on Computer Communications and Networks (ICCCN 98), 1998. [59] S Kamat, Wei Zhao, "An Efficient Optimal Reconfiguration Algorithm for FDDI-Based Networks, " IEEE Transactions on Parallel and Distributed Systems'96, Vol.7, Issue 4, pp.411-424, Apr 1996. [60] G Bucci, A Del Bimbo, S Santini, "Performance Analysis of two Different Algorithms for Ethernet-FDDI Interconnection ," IEEE Transactions on Parallel and Distributed Systems (1994), Vol.5, Issue 6, pp.614-629, 1994. [61] Sanjay Kamat, Nicholas Malcolm, Wei Zhao, "Performance Evaluation of a Bandwidth Allocation Scheme for Guaranteeing Synchronous Messages with Arbitrary Deadlines in an FDDI Network ," IEEE Real-Time Systems Symposium, 1993, pp.34-43, 1993. [62] Th Welzel, Peter Martini, S Rudloff, "Performance Analysis of FDDI and Multiple Token Ring Backbones in a Mixed Traffic Environment," High Speed Local Area Networks II 1990, pp.53-68, 1990. [63] B.Albert, and A.P.Jayasumana, ``Performance Analysis of FDDI LANs Using Numerical Methods,`` Procs. Of the 20th Conf. On Local Computer Networks, Minneapolis, MN, USA, 1995. [64] W.Stallings, ``Networking Standards A Guide to ISDN, LAN, and MAN Standards,`` Addison Wesely Publishing Co. 1993. [65] I.Dalgic and F.A. Tobagi, ``Constant Quality Video Encoding,`` Procs. Of IEEE ICC`95, Seattle, Washington, June 1995, [66] ITU/CCITT Rec. H.261, ``Video CODEC For Audiovisual Services ar px64 Kbbits/s,`` Geneva, 1990. [67] International Organization for Standardization (ISO), ``MPEG: Coding of Moving Pictures and Associated Audio for Digital Storage Media at up to about 1.5Mbps,`` Nov. 1991. [68] P.Wong, and T.Peter, `Àn Integrated Services Token Controlled Ring Network,`` IEEE J. SACs, Vol.7, No.5, June 1989. [69] T.Suda, and T.T.Bradley, ``Packetized Voice/Data Integrated Transmission On a Token Passing Local Area Network,`` IEEE Trans. On Communications, Vol.37, No.3, March 1989. [70] E.A.Khalil, and G.M.Ali, `Àn Efficient FDDI`s MAC For Increasing The Throughput of Asynchronous Traffic and The Capacity of Synchronous Traffic,`` Accepted For Publication in the Eighth IEEE IC3N`99,Boston, MA, USA, Oct. 1999.
Author
Ehab Aziz Khalil, (B.Sc78 M.Sc.83 Ph.D.94), Ph.D. in Computer Network and Multimedia in the Dept. of Computer Science & Engineering, Indian Institute of Technology (IIT) Bombay-400076, India in July 1994, Research Scholar from 1988-1994 with the Dept. of Computer Science & Engineering, Indian Institute of Technology (IIT) Bombay-400076, India, M.Sc in the Systems and Automatic Control, Faculty of Electronic Engineering, Minufiya University, Menouf 32952, EGYPT, Oct. 1983, B.Sc. in the Dept. of Industrial Electronics, Faculty of Electronic Engineering, Minufiya University, Menouf 32952, EGYPT, May 1978. Since July 1994 up to now, working as a Lecturer, with the Dept. of Computer Science & Engineering, Faculty of Electronic Engineering, Minufiya University, Menouf 32952, EGYPT.. Participated with the TPC of the IASTED Conference, Jordan in March 1998, and With the TPC of IEEE IC3N, USA, from 2000-2002. Consulting Editor with the Whos Who? in 2003-2004. Member with the IEC since 1999. Member with the Internet2 group. Manager of the Information and Link Network of Minufiya University, Manager of the Information and Communication Technology Project (ICTP) which is currently implementing in Arab Republic of EGYPT, Ministry of Higher Education and the World Bank. Published more than 85 research papers and articles review in the international conferences, Journals and local newsletter.
380
Vol. 3, Issue 1, pp. 371-380
APPLICATION OF NEWTON-BASED OPF IN DEREGULATED POWER SYSTEMS

2
Department of EE Engineering, VNR VJIET, Hyderabad, India. Department of EE Engg., Vasireddy Venkatadri Institute of Tech., Nambur, A.P., India. 3 Department of Electrical Engineering, J.N.T.U College of Engg., Kakinada, India.
T. Nireekshana1, G. Kesava Rao2 and S. Siva Naga Raju3
ABSTRACT
In this work, a Newton-based optimal power flow (OPF) is developed for implementation into a power system simulation environment. The OPF performs all system control while maintaining system security. System controls include generator megawatt outputs, transformer taps, and transformer phase shifts, while maintenance of system security ensures that no power system components limits are violated. Special attention is paid to the heuristics important to creating an OPF which achieves solution in a rapid manner. Finally, sample applications of the OPF are discussed. These include transmission line overload removal, transmission system control, available transfer capability calculation (ATC), real and reactive power pricing, transmission component valuation, and transmission system marginal pricing.
K EYW ORDS: OPF, Available Transfer Capability (ATC), Short Run Marginal Cost (SRMC), Marginal Pricing, Power World Simulator.
I.
INTRODUCTION
Throughout the entire world, the electric power industry has undergone a considerable change in the past decade and will continue to do so for the next several decades. In the past the electric power industry has been either a government-controlled or a government-regulated industry which existed as a monopoly in its service region. However, countries have begun to split up these monopolies in favor of the free market which can be referred as Restructuring. One of the cornerstones of any restructuring plan is the ability to operate the transmission system in a manner which is fair to all participants in the industry. As stated in [1], participants in wholesale power markets will have non-discriminatory open access to the transmission systems of public utilities. In order to achieve the ideal of open access, many outstanding engineering problems will need to be investigated and tools created for their solution. It is very important that these problems be addressed early in the restructuring process. If these engineering problems become overshadowed by short term economic concerns, then the result could be decreased electricity reliability. The primary goal of a generic OPF is to minimize the costs of meeting the load demand for a power system while maintaining the security of the system. The costs associated with the power system may depend on the situation, but in general they can be attributed to the cost of generating power (megawatts) at each generator. It should be noted that the OPF only addresses steady-state operation of the power system. A secondary goal of an OPF is the determination of system marginal cost data. This marginal cost data can aid in the pricing of MW transactions as well as the pricing ancillary services such as voltage support through MVAR support. In solving the OPF using Newtons method, the marginal cost data are determined as a by-product of the solution technique. A first comprehensive survey regarding optimal power dispatch was given by H.H.Happ [41] and subsequently an IEEE working group [42] presented bibliography survey of major economic-security functions in 1981. Thereafter in 1985, Carpentair [43] presented a survey and classified the OPF
381
Vol. 3, Issue 1, pp. 381-394
algorithms based on their solution methodology. In 1990, Chowdhury [44] did a survey on economic dispatch methods. In 1999, J.A.Momoh et al. [45] presented a review of some selected OPF techniques. S.Chen et al. [49] proposed a new algorithm based on Newton-Raphson (NR) method with sensitivity factors incorporated to solve emission dispatch in real-time. Execution time is lesser than that of the conventional one. K.L.Lo et al. [50] proposed two Newton-like load flow methods, the Fixed Newton method and the modification of the right-hand-side vector method for line outage simulation that is a part of contingency analysis. These two methods are compared with the Newtonbased full AC load flow method and the fast decoupled load flow to show their better convergence characteristics. X. Tong et al. [51] presented the semi smooth Newton-type algorithms for solving OPF problems. It treated general inequality constraints and bounded constraints separately. The method saves computing cost. In this paper after giving the introduction to the title, in section II, the development of Newton based OPF is explained with Problem formulation considering equality and Inequality constraints. The applications of Newtons method of OPF is mentioned and given the flow chart. In section III, heuristics of the OPF solution is explained with implementations of sparse matrix technique and in section IV, comparisons are made with different considerations of the power system conditions to explain the uses of an OPF in a Power system simulation environment.
II.
DEVELOPMENT OF NEWTON BASED OPF
2.1 Problem Statement Newtons method is a very powerful solution algorithm because of its rapid convergence near the solution. This property is especially useful for power system applications because an initial guess near the solution is easily attained. System voltages will be near rated system values, generator outputs can be estimated from historical data, and transformer tap ratios will be near 0.1 p.u. A general minimization problem can be written in the following form. The objective function Minimize f(x) ----------------------------- (1) Subject to: hi(x)=0, i=1,2,.. m Equality constraints -------------------- (2) gj(x) 0, j=1,2,.. n Inequality constraints -------------------- (3) There are m equality constraints and n inequality constraints and the number of variables is equal to the dimension of the vector x. 2.2 Application of Newtons Method to OPF The goal of the OPF is to minimize the costs of meeting the load demand for a power system while maintaining the security of the system. This section of the thesis will discuss the application of Newtons method in a manner that will achieve this desired goal. First, the objective function, f(x), will be introduced. It will reflect the desire to minimize the costs of the system. Then the equality and inequality constraints will be discussed. These constraints model the physical laws of the power system as well as the need to maintain system security. The objective function The objective function for the OPF reflects the costs associated with generating power in the system. The quadratic cost model for generation of power will be utilized: CPgi = ai + biPGi + ciPGi 2 ---------------------- (4) Where PGi is the amount of generation in megawatts at generator i. The objective function for the entire power system can then be written as the sum of the quadratic cost model at each generator. f ( x) = (ai + biPGi + ciPGi 2 ) ---------------------- (5)
This objective function will minimize the total system costs, and does not necessarily minimize the costs for a particular area within the power system. Equality constraints The equality constraints of the OPF reflect the physics of the power system as well as the desired voltage set points throughout the system. The physics of the power system are enforced through the
382
Vol. 3, Issue 1, pp. 381-394
power flow equations which require that the net injection of real and reactive power at each bus sum to zero.
Pk = 0 = Vk [Vm[ gkm cos( k m ) + bkm sin( k m)]] PGk + PIk ------------- (6)
m =1
Qk = 0 = Vk [Vm[ gkm sin( k m) bkm cos( k m)]] QGk + QIk ------------ (7)
m =1
VGi VGisetpo int = 0 ---------------------- (8) For multi area power systems, a contractual constraint requires that the net power interchange be equal to the scheduled power interchange. This adds an equality constraint for all but one area. P int erchange Pscheduled int erchange = [ Plow] Pscheduled int erchange = 0 ------------- (9)
tielines
This last area must not have the equality constraint and essentially becomes a slack area.
Figure 1: Multi area System with Scheduled Interchanges
Inequality constraints Generators have maximum and minimum output powers and reactive powers which add inequality constraints. PGi min PGi PGi max --------------------- (10) QGi min QGi QGi max --------------------- (11) Load tap changing transformers have a maximum and a minimum tap ratio which can be achieved and phase shifting transformers have a maximum and a minimum phase shift, which can be achieved. Both of these create inequality constraints. tkm min tkm tkm max ---------------------- (12) km min km km max ---------------------- (13) 2 2 | Skm | | Skm max | 0 ---------------------- (14) To maintain the quality of electrical service and system security, bus voltages usually have maximum and minimum magnitudes. These limits again require the addition of inequality constraints. Vi min Vi Vi max ----------------------- (15) One issue sometimes encountered when trying to solve a minimization problem is the nonexistence of a feasible solution. Essentially this means that too many constraints have been added to the problem and no solution exists which obeys all of the constraints. A well-suited penalty function for use in Newtons method is the quadratic penalty function [3] which meets the requirements of a penalty function and is also easily differentiated for use by Newtons method. In the OPF presented in this thesis, soft inequality constraints are available for transmission line MVA limits as well as bus voltage limits. Penalty Functions Wkm = k (| Skm |2 | Skm max |2 ) 2 ------------------- (16)
Wi = k (Vi min Vi )2 ;Vi < Vi min Wi = k (Vi Vi max) ;Vi > Vi max Wi = 0;Vi min Vi Vi max
2
------------------- (17) ------------------- (18) ------------------- (19)
383
Vol. 3, Issue 1, pp. 381-394
Figure 2: Penalty Function for Bus Voltage
Figure 3: Penalty Function for Line MVA Flow Limit
After doing trial and error experimentation, values for k were chosen for use in the OPF of this work. For soft bus voltage constraints, k was chosen to be $200/V2. For soft transmission line constraints, k was chosen to be $100/MVA. 2.3 Solution of the optimal power flow The Newtons method algorithm is summarized in the flowchart.
Figure. 4: Flow chart of Newtons method
III.
HEURISTICS OF THE OPF SOLUTION
3.1 Classification of OPF Variables While writing software to perform an OPF solution, a primary concern is identification of variables during the process. Because of this, in order to handle the variables in the OPF problem efficiently, it is convenient to separate them into three categories: controls, states, and constraints. The control
384
Vol. 3, Issue 1, pp. 381-394
variables correspond to quantities that can be arbitrarily manipulated, within their limits, in order to minimize the costs.
3.2 Implementation of Sparse Matrix Techniques At the heart of the solution of the OPF using Newtons method is the solution of the linear system of equations, [H]Dz = L(z). During the Newtons method solution process, this system of equations is solved repeatedly. Because of this, and because one of the primary objectives of an OPF is to find its solution in a short amount of time, the speed of the solution of this linear system of equations is essential to a successful OPF solution. Fortunately, as is the case with many power system matrices, the Hessian matrix is extremely sparse. By implementing sparse matrix routines, the equations can be quickly solved. Much of the structure of the Hessian matrix for the OPF problem is known prior to any calculation. In general, for any Newtons method routine, the Hessian takes the following form
W H = J
JT 0
---------------------- (20)
Already, it is seen that the lower right quadrant of the matrix has entries of zero. By a closer examination of the Hessian structure for the OPF problem, advantage can be taken of this sparsity. The first step was to reorder the variables to include the generator megawatt controls first, Followed by the remaining controls, states, and then constraints. This resulted in the following structure.
W PG H = 0 a 31 0 W rem aining J rem aining n PG T J rem aining 0
T
--------------------- (21)
where WPG is a purely diagonal matrix with entries of 2ci, nPG is a matrix with a (-1) entry in each column corresponding to the net power injection constraint, and possibly an entry of (1) in a column corresponding to a generator with a maximum or a minimum MW constraint being enforced, and Wremaining and Jremaining are the remaining parts of the Hessian matrix. Due to this structure, the sparse LU factorization up to the rows of Wremaining, using Gausss method [24], will result in the following matrix.
H
m o d ified
W PG = 0 n PG
0 W rem a in in g J rem a in in g
W PG 1n PG T J rem a in in g T n PGW 1n PG T
---------------------- (22)
Wremaining and Jremaining are unaffected by these steps and fills are only created by the term nPG W-1PG nPG T . In order to determine the fills created by this matrix, consider only its zero/nonzero structure. One will find that most fills created by this term are actually desired.
W PG 0 H = 0 0 n PG 0 I W rvset W rvset J rvset 0 0 I 0 0 0 0 W rvset T Wr J rr n PG T 0 J rvset T J rr T 0
--------------------- (23)
Given this ordering, the sparse LU factorization up to the rows of Wr, using Gausss method [4], will result in the following matrix.
W PG H
m o d ified
0 I W rv set W rv set J r vse t
0 0 I 0 0
0 0 W r vse t Wr J rr
T
1 W PG n P G T
0 0 0 n PG
0 J rv set T J rr T 1 n P G W PG n P G T
--------------- (24)
The fills created by this method would be identical to those created by the previous ordering. However, in addition, the sparse matrix routines could also take advantage of simply skipping over the processing of the second- and third-row partitions because the only operations required by Gausss method would be divisions by 1.
385
Vol. 3, Issue 1, pp. 381-394
3.3 Solution of an OPF Repeatedly Over Time By keeping the solved OPF solution, the speed of solution in the simulation environment will be greatly increased. These modifications will greatly enhance the power system simulation over time. In calculating the Hessian a time step, further advantage could be gained by implementing partial refactorization schemes for the Hessian as seen in [5]. These schemes only update those elements of the L and U factors of the Hessian that are affected by changes to the Hessian. While this holds great promise for future work, these schemes have not yet been implemented into the OPF written for this work.
IV.
USES OF AN OPF IN A POWER SYSTEM SIMULATION ENVIRONMENT
In this environment, simulation of a system over time can be done while maintaining it at its optimal condition. In this way, a vast amount of economic data can be gleaned from the simulation. The OPF code as implemented into the power system simulator, POWERWORLD [25].
4.1 Line Overload Removal A simple power system not operating under the OPF control is shown in Figure 5.
Figure 5: Six-bus, Single-Area System Not on OPF Control
In order to remove the line constraint, the OPF control is turned on, and the line overload is removed as shown in Figure 6.
Figure 6: Six-bus, Single-Area System on OPF Control
As can be seen, the generators have redispatched themselves in order to remove the line overload. 4.2 Use for Bus Real and Reactive Power Pricing To illustrate the real and reactive power pricing potentials of the OPF solution, the same system shown in Figure 6 is placed on OPF control with its line limit raised. See Figure 7. Note that the OPF results in the same dispatch seen in Figure 6 now that the line limit has been removed.
386
Vol. 3, Issue 1, pp. 381-394
Figure 7: Line Limit from Bus 5 to 4 raised to 100 MVA
4.3 Use for Area Real Power Pricing The OPF solution method may also be used with multi area power systems. The OPF will enforce the scheduled area interchange in these systems. In Figure 5, the simple six-bus system from before is split into two areas as shown.
Figure 8: Six-bus, Two-Area System on OPF Control
As can be seen, for this case the generation in Area Two is less expensive than in Area One; therefore, it would be advantageous for both areas if Area One were to purchase some power from Area Two. Using the capabilities of the POWERWORLD Area Transactions/Information display [2], scheduled transactions can be set up between the two areas to optimize their costs.
Table 1: summarizes several possibilities.
From table 1, the least expensive scenario for the sum of the areas is when an interchange of 65.5 MW is undertaken. This transaction scenario is shown in Figure 9.
387
Vol. 3, Issue 1, pp. 381-394
Figure 9 : Transaction of 65.5 MW Undertaken
4.4 Transmission Line Valuation by Time-Domain Simulation One of the most intriguing potential uses of an OPF is its use as a pricing tool by doing time domain simulations of a power system. As a simple example, consider again the six-bus, one area power system. Assume that the loads of the system vary as shown in Figure 10. The load at a given bus is then equal to its base value multiplied by the load factor.
Figure 10: Load Factor Curve for six-bus System
Entering this load factor curve into the POWERWORLD simulation software and simulating the power system for a full 24 hours yield a total cost in dollars for operating the power system over that period of time. By re-simulating the system repeatedly with variations in system structure, one can gain useful economic insights from the comparison of total system costs. Note that the case was simulated at 600 times real time for the first 23 hours and at 60 times real time for the last hour. The POWERWORLD simulation software utilizes trapezoidal rule integration to calculate the total system cost from the incremental costs at each time point [18]-[20]. Using this technique, the six-bus system was simulated with various transmission lines removed. The results are summarized in Table 2.
Table 2: Simulation data for six-bus system with transmission lines removed
By comparing the total system cost with all lines connected to the cost with a line removed, one can find the approximate cost a company would incur per day while taking a line out for maintenance. It is very important to note, however, that this does in no way take into account the possibility that another line may unexpectedly go out, in which case all lines may be needed to take up the slack. In other words, the pricing of redundancies in the transmission system is not considered in this methodology.
4.5 Limit on Available Transfer Capability (ATC) Due to a Voltage Constraint [6]
388
Vol. 3, Issue 1, pp. 381-394
Another use for the OPF algorithm is for the calculation of the available transfer capability for a system. The OPF is able to take into account system security concerns including voltage limits and transmission line limits while calculating the ATC [31]-[35]. The three-bus system of Figure 11 can be used to illustrate and examine the value of reactive power support at a system load bus. Area one consists only of the generator and load at bus 1 plus two tie lines. Area two consists of the generator and load at bus 3 plus the load at bus 2, and the line from bus 2 to bus 3. Area one has expensive generation, while area two has cheaper generation. For the indicated loads and no area transfers the cost of operation for the total system is $25,799/hr.
Figure.11: Three-Bus Base Case with No Area Power Transfer
Note that the incremental cost for reactive power at a generator bus is $0.00. This is true as long as no MVAR limits are reached because no cost is associated with generation of MVARs by a generator in the OPF formulation of this work. Consider the situation where area one wants to purchase power at considerable savings from area two. Similarly, area two wants to sell power to area one. We first examine the available transfer capability of this system for transfer from area two to area one. An OPF solution with bus voltage constraints is used to find the maximum power which can be transferred from area two to area one. The voltage constraint for power quality used in this example is 0.96 Vi 1.04. Both generators have their excitations set to give 1.04 p.u. voltages for serving their local area load within the voltage constraints. A global OPF solution which minimizes the total cost of providing the total load will attempt to increase area-two generation and decrease area-one generation. This power transfer will change the voltage at bus 2. When the transfer is such that the voltage reaches its lower limit (0.96), the OPF solution will stop at that constraint.
4.6 Transmission System Pricing Through Short-Run Marginal Costing (SRMC) In the restructured environment of the future, it may be necessary to determine the cost incurred to the transmission system due to a power transaction between two companies. This cost would be charged to the two companies undergoing the transaction and would be paid to the operator of the transmission system. Two methods proposed for the determination of this cost is through short-run marginal costing (SRMC) and long-run marginal costing (LRMC). These two topics are extensively discussed in both [7] and [8]. A brief, albeit simple, explanation of these follows. The SRMC takes into account only the operating costs of the power system. The operating costs include fuel costs, losses, as well as system constraints such as transmission line limits. The LRMC, however, includes both the operating costs and the capital costs of the power system. The capital costs include the costs of future expansion to the transmission system as well as the generation capacity. Through the use of the OPF, the SRMC technique can easily be implemented as discussed in [8]. The SRMC of a power transaction can be estimated by solving the OPF for both the system with the transaction and the system without the transaction in place. After these solutions are obtained, the SRMC can be found as follows. Define BMCi = (marginal cost of power at bus i after the Transaction) P i, transaction = the net power injection at bus i due to the Transaction = (- power generation at bus I Before Transaction + power generation at bus i
389
Vol. 3, Issue 1, pp. 381-394
after transaction) From these definitions a definition for SRMC can then be made [8].
SRMCtransaction =
allbuse sin transaction
BMCi *Pi , transcation --------------- (26)
As a simple example consider the six-bus, two-area system with no transactions Figure 8, and with a transaction of 65.5 MW undertaken between the two areas Figure 9. Taking the data from these two figures, a calculation of the SRMC for the transaction can be made. This calculation is summarized in Table 4.
Table 4: SRMC calculation for six-bus, two-area system
Thus, the SRMC for this transaction is (- $132.13)/hour. As mentioned in [7], SRMC can be a negative value. For comparison, consider the same system undergoing the same 65.5 MW transaction, but with the line between buses 4 and 5 doubled to 100 MVA from 50 MVA. The simulation of this system is shown in Figure 12.
Figure.12: Six-bus, Two-Area System Undergoing Transaction with Line Limit Doubled
Again, Figure 12 shows this system with no transactions, and Figure 9 shows the system with a transaction of 65.5 MW undertaken between the two areas. Taking the data from these two figures, a calculation of the SRMC for the transaction can be made for the system where no line limit will be encountered. This calculation is summarized in Table 5.
Table 5: SRMC calculation for six-bus, two-area system where no limit is encountered
Thus, the SRMC for this transaction is (- $146.81)/hour. This is not unexpected. The previous system encountered a transmission line constraint, while this one has not; therefore, the incentive to do the
390
Vol. 3, Issue 1, pp. 381-394
transaction should be large, i.e., the SRMC should be less. While this discussion is not entirely complete, it does display the basic use of the OPF to calculate the SRMC. More study will be done on this topic in the future.
V.
A six- bus single area with not an OPF control and on OPF control are simulated and compared. Comparison of the OPF solutions in Figure 6 and Figure 7 yield valuable insight. The total system cost without the line limited as in Figure 6 is $7824/hr. With the line limited as in Figure 7 this costs increases, as would be expected, to $7895/hr. Also note the differences between the bus MW marginal costs in Figure 7 and Figure 6. Because the generators at buses 2 and 4 were forced to decrease their output in order to remove the overload, their bus MW marginal costs also decreased. Conversely, the bus MW marginal costs at buses 1, 3, and 5 increased. As might be expected, the largest changes occurred at the ends of the limited line, buses 4 and 5. Comparing the OPF solution in Figure 6 and Figure 9 shows the same solution. This is of course not unexpected, because the two-area economic negotiations should yield the same solution that an OPF solution disregarding areas would. This must happen so that all areas are at their optimal solution. SRMC calculation for 6-bus 2-area with line limit double and where no limit is encountered are compared and concluded as When the capacitor bank at bus 2 is increased from a nominal 15 MVAR rating to a nominal 30 MVAR rating, the OPF solution is given. The available transfer capability for these constraints is increased to 325 MW. The cost of operation for the total system is $20,890/hr. Thus the increase in available power transfer capability has resulted in an additional reduction of total cost by $451/hr. Again, additional transfers are not possible without violation of the voltage constraint at bus 2. Note that the incremental cost for MVARs has reduced as the OPF approaches a more economical dispatch. When the capacitor bank at bus 2 is increased from a nominal 30 MVAR rating to a nominal 45 MVAR rating, the OPF solution is given. The work presented in this paper utilizes an optimal power flow program, OPF, as the tool for solving these problems. The OPF is a natural choice for addressing these concerns because it is basically an optimal control problem. The OPF utilizes all control variables to help minimize the costs of the power system operation. It also yields valuable economic information and insight into the power system. In these ways, the OPF very adeptly addresses both the control and economic problems. After creating the OPF program, the user-interface and simulation problems may also be addressed by implementing the OPF into a power system simulator. In this way, the results of the economic and control operations of the OPF can easily be utilized by the user of the program. This paper will explore the application of Newtons method to the OPF problem. Specifically, it will explore the implementation of a Newtons method based OPF in the power system simulator POWERWORLD [25].
VI.
CONCLUSION
Minimization of system costs, while maintaining system security, was accomplished through the implementation of Newtons method to the OPF problem. Newtons method has proven to be very adept at solving the OPF problem. The OPF performs generator control and transmission system control while taking into account system limits. The marginal cost data from the OPF were shown to aid in the available transfer capability (ATC) calculation, real and reactive power pricing, transmission system pricing, and transmission system component valuation. The author would like to reiterate that while the applications of the OPF as discussed in this thesis are extremely valuable, they do not take into account the possibility of random outages of transmission line components or generators. Also, the pricing schemes discussed do not take into account the need for future expansion of the power system.
REFERENCES
[1]. Federal Energy Regulatory Commission of the United States of America, Promoting wholesale competition through open access non-discriminatory transmission services by public utilities, Docket No. RM95-8-00, March 1995.
391
Vol. 3, Issue 1, pp. 381-394
[2]. T. J. Overbye, P. W. Sauer, C. M. Marzinzik and G. Gross, A User-Friendly Simulation Program for Teaching Power System Operations, IEEE Transactions on Power Systems, Vol. 10, No. 4, November 1995, pp. 1725-1733. [3]. D. G. Luenberger, Linear and Non-Linear Programming, Reading, MA: Addison-Wesley Publishing Company, 1984, pp. 295-392,423-450. [4]. G. E. Forsythe and C. B. Moler, Computer Solution of Linear Algebraic Systems, Englewood Cliffs, N.J.: Prentice-Hall, 1967, Chapters 9 and 16. [5]. S. M. Chan and V. Brandwajn, Partial Matrix Refactorization, IEEE Transactions on Power Systems, Vol. PWRS-1, No. 1, February 1986, pp. 193-200. [6]. J. D. Weber, M. J. Laufenberg, T. J. Overbye and P. W. Sauer, Assessing the Value of Reactive Power Services in Electric Power Systems, Conference on Unbundled Power Quality Services, Key West, Florida, November 17-19, 1996. [7]. R. D. Tabors, Transmission System Management and Pricing: New Paradigms and International Comparisons, IEEE Transactions on Power Systems, Vol. 9, No. 1, February 1994, pp. 206-215. [8]. D. Shirmohammadi, X.V. Filho, B. Gorenstin and M. Pereira, Some fundamental technical concepts about cost based transmission pricing, IEEE Transactions on Power Systems, Vol. 11, No. 2, May 1996, pp. 1002-1008. [9]. A.F. Zooba, W.J. Lee, The Globalization of Energy Markets in Asia, IEEE PES General Meeting 2006, pp. 1-6, Oct. 2006. [10]. S.N. Singh, S.C. Srivastava, Electric Power Industry restructuring in India: Present Scenario and Future Prospect, IEEE International Conference on Electric Utility Deregulation, Restructuring and Power Technologies (DRPT2004), vol.1, pp. 20-23, April 2004. [11]. P. Bajpai, S.N. Singh, An Electric Power Trading Model for Indian Electricity Market, IEEE PES General Meeting 2006, Oct. 2006. [12]. J. Wernerus, A Model for studies related to Deregulation of an Electricity Market, KTHStockholm, Master Thesis TRITA-EES-9503, November 1995. [13]. S. Mohammad & A. Muwaffaq, Restructured Electric Power systems- Operation, Trading & Volatility, Markcel Dekker, Inc., Basel, Switzerland, 2001. [14]. S.N. Singh, S.C. Srivastava, Electric Power Industry restructuring in India: Present Scenario and Future Prospect, IEEE International Conference on Electric Utility Deregulation, Restructuring and Power Technologies (DRPT2004), vol.1, pp. 20-23, April 2004. [15]. P. Wolfgang, Preface: Competition and Long-Term Dimensions of Electricity Supply, Jacobs University Bremen, Page (29- 36). [16]. http://en.wikipedia.org/wiki/Electricity_tariff [17]. E. Hirst & B. Kirby, The Function, Metrics, and Prices for Three Ancillary Services, Edison Electric Institute, Washington, DC, October 1998. [18]. S. G. Kim, LMP as Market Signal for Reserve Supply in Energy and Reserve Integrated Market, IEEE PES PSCE 2006, pp.896-900, Nov. 2006. [19]. B. Kankar, B.H.J. Math & E.D. Jaap, Operation of Restructured Power systems, Kluweir Power Systems series Series Editor: M.A. Pai, 2001. [20]. B. Joseph, The PJM Market, Chapter 13, pp. 451-477, PJM Interconnection LLC, Pennsylvania, USA, 2006. [21]. G.Terge, G. jan Ove, F. Roger, Ancillary Services in Deregulated Power Systems; What are they; Who need them and who pays, IEEE POWERCON98, vol.1, pp. 704-709, August 1998. [22]. Kwon W. Cheung, Ancillary Service Market Design and Implementation in North America: From Theory to Practice, IEEE Third International Conference on Electric Utility Deregulation, Restructuring and Power Technologies (DRPT2004), pp. 66-67, April, 2008. [23]. S.Mohammad & S. Muwaffaq, Restructured Electrical Power Systems- Operation, Trading & Volatility, Marcel Dekker, Inc., NY, 2001. [24]. S.N. Singh, S.C. Srivastava, Electric Power Industry restructuring in India: Present Scenario and Future Prospect, IEEE International Conference on Electric Utility Deregulation, Restructuring and Power Technologies (DRPT2004), vol.1, pp. 20-23, April 2004. [25]. http://www.powerworld.com/products.asp [26]. IEEE Reliability Test System, IEEE transactions on power apparatus and systems, Vol.PAS-98, No.6, Nov/Dec. 1979. [27]. Power Electronics in Electric Utilities: Static Var Compensators Proceedings of the IEEE, VOL. 76, NO. 4, APRIL 1988. [28]. V. Ajarapu, and C. Christie, The continuation power flow: a practical tool for tracing power system steady state stationary behavior due to the load and generation variations, IEEE Trans. Power Systems, Vol. 7, No. 1, Feb. 1992,416-423.
392
Vol. 3, Issue 1, pp. 381-394
[29]. M.Noroozian, G.Anderson, Power Flow Control by Use of Controllable Series Components, IEEE Transactions on Power Delivery, VO1.8, No.3, July 1993, 1420-1428. [30]. Vladimiro Miranda, J.V. Ranito, L. M. Proenca, Genetic Algorithms in Optimal Multistage Distribution Network Planning, IEEE Trans. Power Systems, Vol. 9, o. 4, Nov. 1994, 1927-1933. [31]. Transmission Transfer Capability Task Force, Available transfer capability Definitions and determination, North American Electric Reliability Council, NJ, June 1996. [32]. G. C. Ejebe, J. Tong, J. G. Waight, J. G. Frame, X. Wang, and W. F. Tinney, Available transfer capability calculations, IEEE Trans. Power Systems, Vol. 13, No.4, Nov.1998, 1521-1527. [33]. Y. Dai, J. D. McCalley, V. Vittal, Simplification, expansion and enhancement of direct interior point algorithm for power system maximum loadability, Proc. 21st Int. Conf. Power Ind. Comput. Applicat, pp. 170-179, 1999. [34]. Richard D. Christie, Bruce F. Wollenberg and Ivar Wangensteen, Transmission Management in the Deregulated Environment, Proceedings of the IEEE, Vol.88, No.2, February 2000. [35]. Y. Xiao, Y. H. Song, Y. Z. Sun, Available Transfer Capability Enhancement Using FACTS Devices, Proceedings of the 2000 IEEE/PWS Summer Meeting,, Seattle, vol.1, pp. 508-515, July 2000. [36]. Wang Feng, G.B.Shrestha Allocation of TCSC Devices to Optimize Total Transmission Capacity in a Competitive Power Market IEEE Trans. Power Systems, Feb 2001, 587-592. [37]. Stephane Gerbex, Rachid Cherkaoui, and Alain J. Germond, "Optimal Location of Multi-Type FACTS Devices in a Power System by Means of Genetic Algorithms," IEEE Trans. on Power Systems, vol. 16, no. 3, August 2001. [38]. Dr. Tapan Kumar Saha Required Shunt Compensation for an EHV Transmission Line The University of Queensland, Australia, May 2003. [39]. R.C. Bansal, A.F.Zobaa, K.A.Nigim Technology of VAr Compensators for Induction Generator Applications in Wind Energy Conversion Systems Journal of Electrical System, 2006, 172-184. [40]. MATLAB Genetic Algorithm Tutorials [41]. H. H. Happ, Optimal power dispatch-A comprehensive survey, IEEE Trans. Power Apparat. Syst.,vol. PAS-90, pp. 841-854, 1977. [42]. IEEE working group, Description and bibliography of major economic-security functions part-II and III, IEEE Trans. Power Apparat. Syst., vol. PAS-100, pp. 215-235, 1981. [43]. J. Carpentier, Optimal power flow, uses, methods and development, Planning and operation of electrical energy system Proc. Of IFAC symposium, Brazil, 1985, pp. 11-21. [44]. B. H. Chowdhury and Rahman, Recent advances in economic dispatch, IEEE Trans. Power Syst., no. 5, pp. 1248-1259, 1990. [45]. J. A. Momoh, M. E. El-Harwary and Ramababu Adapa, A review of selected optimal power flow literature to 1993, part- I and II, IEEE Trans. Power Syst., vol. 14, no. 1, pp. 96-111, Feb. 1999. [46]. T. S. Chung and Ge Shaoyun, A recursive LP-based approach for optimal capacitor allocation with cost-benefit consideration, Electric Power Syst. Research, vol. 39, pp. 129-136, 1997. [47]. E. Lobato, L. Rouco, M. I. Navarrete, R. Casanova and G. Lopez, An LP-based optimal power flow for transmission losses and generator reactive margins minimization, in Proc. of IEEE porto power tech conference, Portugal, Sept. 2001. [48]. F. G. M. Lima, F. D. Galiana, I. Kockar and J. Munoz, Phase shifter placement in largescale systems via mixed integer linear programming, IEEE Trans. Power Syst, vol. 18, no. 3, pp. 10291034, Aug. 2003. [49]. S. D. Chen and J. F. Chen, A new algorithm based on the Newton-Raphson approach for real-time emission dispatch, Electric Power Syst. Research, vol. 40, pp. 137-141, 1997. [50]. K. L. Lo and Z. J. Meng, Newton-like method for line outage simulation, IEE proc.- Gener. Transm. Distrib., vol. 151, no. 2, pp. 225-231, March 2004. [51]. X. Tong and M. Lin, Semismooth Newtontype algorithms for solving optimal power flow problems, in Proc. of IEEE/PES Transmission and Distribution Conference, Dalian, China, 2005, pp.1-7.
Biography of Authors T. Nireekshana obtained her B.Tech and M.Tech from JNT University in 2002 and 2005 respectively. She is Assistant professor in EEE, VNR VJIET Engineering College, Hyderabad. She has been pursuing Ph.D work, part-time with JNTU.
393
Vol. 3, Issue 1, pp. 381-394
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 22311963
G. Kesava Rao obtained his PhD from Moscow Power Engineering Inst. Moscow, U.S.S.R. He worked in Institute of Technology at Banaras Hindu University, Varanasi, India in various administrative and academic positions. Currently, he is working at Vasireddy Venkatadri Institute of Technology, Nambur, A.P., India. His fields of interest are power system deregulation and renewable energy sources. S. Sivanagaraju, is graduated in 1998, Masters in 2000 from IIT, Kharagpur and did his Ph.D from J.N.T.University in 2004. Working as Associate Professor in the Department of Electrical Engineering, J.N.T.U College of Engg. (Autonomous), Kakinada, Andhra Pradesh since November 2006. He has received two National awards (Pandit Madan Mohan Malaviya memorial prize award and Best Paper Prize award) from the Institute of Engineers (India) for the year 2003-2004. He is Referee for IEE Proceedings- Generation Transmission and Distribution and International Journal of Emerging Electric Power System. About 40 publications in National and International Journals and Conferences to his credit. His areas of interest are in Power Systems, Distribution Automation, Genetic Algorithm applications to distribution systems and power system.
394
Vol. 3, Issue 1, pp. 381-394
SECURED ENERGY EFFICIENT HYBRID ROUTING IN WIRELESS SENSOR NETWORKS

Deepika Srikumar and Seethalakshmi Vijaykumar
Department of Electronics and Communication Engineering, Sri Shakthi institute of Engineering and Technology, Coimbatore, India
ABSTRACT
Energy is the scarcest resource of sensor nodes and it determines the lifetime of wireless sensor networks. The main challenge is to minimize the energy required for wireless communication by developing application independent routing algorithm for reliable and efficient data delivery. In addition to the limited resources, security is a main concern of wireless sensor networks. In this paper the proposed scheme is to revolutionize the contemporary technical arena based on the hybrid routing. The hybrid routing is derived from fundamental radio energy model considering various parameters such as number of hops, threshold energy level, distance between number of hops and Packet Reception Rate (PRR) and also it protects the network against worm hole attacks. This paper evaluates the performance of existing energy efficient based routing algorithms in wireless sensor networks such as Greedy, ODGR, and PEGASIS and it is compared with the proposed EEHR algorithm in terms of energy efficiency. Simulation is done using NS-2(version NS-2.33). The proposed scheme is more efficient than existing methods since it reduces average energy consumption for multihop communication even in the presence of high node density, high throughput, improves system life time and reduces the average latency leads to higher Packet Deliver Ratio(PDR) and provides the defense technique against worm whole attacks.
K EY W ORDS:
Average Latency, Hybrid Routing, PRR, Reliable Link, Threshold Energy level
I. INTRODUCTION
Wireless sensor networks are core technologies in various applications such as Greenhouse monitoring, structural monitoring in industry, area monitoring in military etc. In wireless sensor network spatially distributed sensors are deployed either inside the phenomenon or very close to it. Nodes in the sensor environment have restricted storage, computational and energy resources. The conventional wireless sensor networks support IP style of addressing of sources and destination. They also use intermediate nodes to support end-to-end between attribute nodes in the network. Each node in the wireless sensor networks maintains the neighbor information. The reusability of maintained information widely increases the performance of the wireless sensor network [1] [14]. The many routing algorithms have been proposed among them adaptive algorithms are very popular as it adapts itself to the shortest hop count but it non reliable link quality. The Communication may occur less reliably with nodes faraway, but there are many distant nodes and a few of them are likely to have strong connectivity. Generally, many of the links are lossy and the loss rate may change dynamically with environmental factors or due to contention arising from the highly correlated behaviour of the application [16]. The energy efficient route allocation is found in few works and in majority, it does not evaluate in wireless sensor networks context, but analysis the parameters like shortest path routing, alternate path routing during link breakage etc. Beyond the above mentioned parameters the most important criteria in wireless sensor networks is threshold energy level and residual energy level. If the energy in the network falls below the threshold value during transmission or reception, it may leads to the loss of packets and error in receiving information. For long distance communication battery power needed for sensor node is
395
Vol. 3, Issue 1, pp. 395-404
more but for the short distance it is less. Example the power required to transmit 10m is 1mw but the power required to transmit 100m is 100m [8]. This paper also addresses the security concerns in wireless sensor networks. More specifically, we address the wormhole attack, which is a severe attack in wireless sensor networks whereby an attacker stores transmitted packets and then replays them into the network. Defending against such an attack is challenging because it can be launched even if all network communication is authentic and confidential. Existing algorithms either lead to long distance communication or wastage of energy resources. To overcome the various drawbacks in the existing schemes, the proposed Energy Efficient Hybrid Routing (EEHR) scheme evaluates following parameters in priority manner to forward a packet as follows: 1. Choosing type of transmission either direct transmission or multihop from the radio energy model. 2. If direct transmission follow step5 procedure else following the next step. 3. Optimal number of hops between source and destination. 4. Optimal distance between numerous motes during intermediate transmission. 5. Calculating threshold energy level from initial energy of each node and minimum energy required for efficient operation. In addition to the above process the hybrid routing also provides the defense mechanism against the worm hole attack in network. The efficient network working is analyzed from above parameters and blacklisting the inefficient route. The rest of the paper is organized as follows. In Section 2, the related work is explained. The new geographic routing algorithm in WSNs is presented in section 3. After that, the results of simulation study that evaluates the performance of proposed method and others are described in Section 4. Finally, conclusion of the proposed work and future extensions of this study are described in Section 5.
II. RELATED WORK

2.1 Greedy Forwarding
Greedy Forwarding is the key mechanism in geographic routing protocols. In Greedy Forwarding, each node knows the location of neighbor nodes for data transmission. Neighbor nodes are located in 1-hop distance of source node. In this way, source node uses the location of neighbor nodes for data transmission. Additionally, it knows the location of destination node. Source node selects a candidate-node in neighbor nodes. A candidate-node is the closest to destination node in neighbor nodes. For forwarding data packets to destination node, Greedy Forwarding repeats this pattern until data packets reach destination node [9]. However, in Greedy Forwarding, source node forwards data packets to a candidate-node without considering wire-less link-quality [6]. Therefore, this method has a problem that delivery rate is dropped without retransmission. If the retransmission is possible, the main problem is that energywastage is increased.
2.2. PEGASIS
The main idea in Power-Efficient Gathering in Sensor Information Systems (PEGASIS) is for each node to receive from and transmit to close neighbors and take turns being the leader for transmission to the BS.
2
1
B S
5 4 6
Figure.1. Routing in PEGASIS
396
Vol. 3, Issue 1, pp. 395-404
This approach will distribute the energy load evenly among the sensor nodes in the network [15]. The nodes are randomly located in network. The nodes will be organized to form a chain, which can either be accomplished by the sensor nodes themselves using a greedy algorithm starting from some node. Alternatively, the BS can compute this chain and broadcast it to all the sensor nodes [5]. PEGASIS improves on LEACH by saving energy in several stages. First, in the local gathering, the distances that most of the nodes transmit are much less compared to transmitting to a cluster-head in LEACH. Second, the amount of data for the leader to receive is at most two messages instead of 20 (20 nodes per cluster in LEACH for a 100-node network) [21]. Finally, only one node transmits to the BS in each round of communication. The main drawback of this method is, it lead to large delay.
2.3. ODGR
Optimal distance geographic routing (ODGR) is an algorithm that uses geographic information and power control in the transmission scheme to dynamically explore the optimal routing path. There are three main key mechanisms in routing techniques. They are (i) Direct transmission between source and destination. (ii) Mutihop transmission through neighbor nodes. (iii) Only few numbers of motes selected between source and destination.
Multiple numbers of hops Direct transmission between source & destination - - - - - - -> Selected hops
Figure.2. An example of wireless network scenario.
The fundamental radio energy model is given by the following equation. E(d)=2lelec +lfs-amp d2 if d < d0 (1) =2lelec +lmp-amp d4 if d > d0 (2) where l is the number of bits in a packet, h is the distance between the transmitter and the receiver nodes and elec is the electronics energy for the transceiver circuitry. fsamp and mpamp denote power amplifier energy dissipations at free-space and multipath modes, respectively, and d0 is the threshold distance [2]. ODGR considers two important physical properties of wireless sensor networks, geographic location and power control. The ODGR algorithm is developed based on the optimal distance analysis and optimal number of hops. ODGR requires each node to know the locations of the neighbor nodes within a specific distance. This distance is set as 1.5 hopt, because for a given distance d and number of hops n, the transmit distance per hop h can be calculated as n = d/h [16].
397
Vol. 3, Issue 1, pp. 395-404
It is applicable for any node-to-node communication in wireless sensor networks. It can be used, but not limited, in a form of data collection or dissemination among sensor nodes and base station. In an intelligent environment with both sensors and actuators, node-to-node unicast communication may be necessary for task coordination.
Figure.3. Simulation results of ODGR
The above routing shows the good results by implementing ODGR algorithm. The main constraint in this algorithm is, this algorithm does not consider the threshold energy level in the intermediate nodes. This leads to wastage of energy resources.
III. PROPOSED SCHEME

The proposed EEHR is suitable for geographic routing. It is a new multicast routing algorithm. The minimum numbers of sensor nodes are used to route message to destination. If source need to forward packets to destination which is in the transitional region. It forwards through the neighbor nodes. If residual energy in the neighbor node is not enough to forward the data packets, only few packets are forwarded till its total energy drops in that sensor node, remaining data packets are dropped. The threshold energy level of neighbor node should be calculated before forwards to that particular node. To achieve above goals the EEHR was proposed [19].
3.1 Network Model

EEHR assumes a general wireless sensor network environment. Our method initially decides whether source to destination is direct communication or multihop communication. For example, energy required for direct transmission is 7J and energy required for multihop communication is 8J, i.e. 4J for each hop. From above example we conclude that direct transmission should be done for that case. This follows to optimal number of hops and optimal distance between source to destination. Finally the PRR and threshold energy level are analyzed. In this paper grid network is considered, i.e. distance between neighbor nodes are equal. Nodes are placed 5m distances each [13].
3.2 Link Loss Model

The link loss model of real environment is required to simulate reliable data transmission. PRR is used in the research. PRR is the link quality between two nodes. It varies from 0 to 1. In the
398
Vol. 3, Issue 1, pp. 395-404
connected region packet reception rate is 1(from 0 to 8m) because in this region nodes can transmit data packets perfectly. The link qualities between two nodes for the transitional region (8m to 35m) are observed by [4] and [8].
3.3 Security Concerns

The hello packets are sends to neighbor nodes from the all nodes in symmetric environment to discover the neighbor nodes and protect against the wormhole attack. A hello packet consists of ID of the particular node that originate hello packet and hop count to destination. The sensor node which receives the hello packet inserts the new entry in its hello packet list that contains all the ID of nodes which sends hello packet to that particular node. The hello packet list sorted the various ID and made the ID which has lowest number of hops as higher priority. As soon as the node receives the hello packet, acknowledgment timer is set to expire over a particular period. When the acknowledgement timer got expired the node send acknowledgement packet (ACK_packet) to source node. The ACK_pack consists of source node ID, destination ID and hop count. The updates are also made in ACK_table (Acknowledgement table). The ACK_table consists of destination ID, hop count, No_ACK_pack (total number of acknowledgement packets) and Rec_Accept (total number of received accepted packets). The destination ID and hop count are initialized in ACK_table but No_ACK_pack and Rec_Accept are made zero. Check timer is set; it will expire after a period of transmission of ACK_packet. As soon as the check timer got expired increment is made in the No_ACK_pack field of ACK_table. Source node which receives the ACK_packet inserts that particular destination ID, hop count and number of identical ACK_packets (ID_ACK_packets) is incremented or it is set to one for newly received packet in acknowledgement list (ACK_list) [21].
Tab 1.Acknowledgement Table
NODE ID HOP COUNT No_ACK_pack Rec_Accept Accept timer is set upon the reception of the first ACK_packet, when accept timer got expired the source node sends the Acc_packet (accept packet) to the destination for the equivalent entries in ACK_list. Nodes upon receiving the Acc_packet checks whether the source ID is same as the source ID in the ACK_table. If source ID got match update made in ACK_table by incrementing Rec_Accept, if not attacker stores the ACK_packet and reply the Acc_packet to destination. The number of reply in the Acc_packet should be one greater than the No_ACK_pack in ACK_table. If the above condition is not satisfied neglect the Acc_packet and include the ID in the accept packet neglected accepted packet table, set the ACK_table values to zero. Send another ACK_packet that corresponds to next priority in the ACK_list or wait for another ACK_packet if not available [11]. 3.4. Design Concerns If source A is located at (xA, yA) and destination B is located at (xB, yB), the EEHR routing algorithm works as follows. (i) Set the current node location (xi, yi) as the source node location: xi = xA; yi = yA. (ii) Computes the Euclidean distance from the current node to destination. If the distance d is less than hopt, jumps to Step (ix), where
d=
(xi x B ) 2 + ( yi yB ) 2
(3)
(iii) Computes nopt = d/hopt. The distance per hop h is determined as d/n.
399
Vol. 3, Issue 1, pp. 395-404
(iv) If n = 1, goes to Step (ix). Otherwise, computes the estimated location of the next hop node j as, xi =xi + x =xi + (( xB - xi )/n) yi = yi + y = yi + (( yB - yi )/n) (5) (4)
(v) For the estimated location, calculate the optimal number of hops and optimal distance between source and destination in priority order for four indirect transmission paths. Choose the one route in priority order. (vi) Blacklisting the neighbor nodes based on PRR and threshold energy value. A node has initial energy is 12J. The node requires minimum 1.75608J (Emin) energy to operate properly. The threshold energy is mainly based on minimum energy required for the proper operation of the node which depends on the parameters like transceiver energy, bandwidth, bit rate, signal to noise ratio, antenna gain, transmitter efficiency as per mica 2 model. The threshold energy is calculated as follows, Ethreshold = Emin / Einitial =0.14634J Therefore current forwarding node omits neighbor nodes which have lower Energy-level than threshold energy level and go to previous step to choose the next optimal path. In (7) EX (residual) is the residual energy of node X and EX (initial) is the initial energy of node X. The Energy_level of neighbor nodes should be taken into account to analyze the energy of the node. The Energy_level is normalized value between 0 and 1. Energy_level is given by: Energy_level=EX(residual) / EX(initial) (7) (6)
(vii) After blacklisting has done consider the node which has the largest priority value PV. Priority Value=W1 *CSL+ W2 *Energy_level (8)
Priority value is analyzed from the Change in Status of Link quality (CSL), Energy_level of neighbor node and weighting factors (W1 & W2 ) CSL is the value that explains link quality between source and neighbor node. Modified Status of Link quality of the data transmission based on size of data packets and ACK message packet. The two factors are applied here which decides the size of the packet (Sdata & SACK) for various sizes of data packets. CSL=(Sdata/(Sdata+SACK )*PRRS-N+(SACK/(Sdata + SACK )* PRRN-S (9)
W1 and W2 are weighting factors. Weighting factors are defined by Energy_level of the neighbor nodes. If neighbor nodes have sufficient energy to transmit data or receive data, there is no need to consider the relative energy. Concentrating on link quality and optimal number of hops provides good result.
W1 =n i=n Energy_levelnbr(i) n (10)
400
Vol. 3, Issue 1, pp. 395-404
(viii) Sends the packet to the next hop node if it is not in the exclusion list of the current node, and waits for an acknowledgement (ACK). If the ACK is successful, sets the next hop node as the current node and goes to Step (ii) to continue. If the next hop node is in the exclusion list, no matter whether it is discovered from the most recent ACK or from information obtained during previous search, it goes back to Step 5 to choose a different node. If all available neighbor nodes are in the exclusion list, the algorithm terminates with an announcement that no neighbor nodes have energy higher than threshold value and PRR is less than PRR threshold. (ix) Transmits packets directly (direct transmission) from the current node to the destination. The EEHR algorithm is completed.
3.4. Blacklisting Nodes

The symmetric environment is assumed, nodes are placed equidistance from each other. The value of PRR threshold compares with actual PRR in each data transmission. Nodes should have appropriate Energy_level for data transmission. If the node has lower Energy_level than the Energythreshold in the optimal path transmission that particular optimal path is discarded and transmission is done using next optimal path [7].
3.5 Data Transmission

The current node omits some neighbor nodes using the blacklisting method in data transmission. Then, the current node sends data packets to a candidate node that has highest priority value of neighbor nodes. If the current node does not receive ACK packet from candidate node perfectly, the current node should retransmit. Data delivery will fail, if the number of retransmission time exceeds ARQ. Each node updates the Energy_level of neighbor nodes at the end of data transmission. Each node repeats these steps until the destination node receives data packets.
IV. ANALYSIS OF EXPERIMENTAL RESULTS

4.1 Simulation Environment
The simulations for existing and proposed scheme are performed using the NS2 simulator. The symmetric environment is created in the proposed scheme consists of 7 rows and 7 columns of nodes totally 49 nodes, which is placed 5m equidistance from neighbor nodes. The maximum initial energy for each node is set as 12J. Energy required for transmission and reception of 100 data bytes is 1762.5J and ACK packet size is 193.875J. The various parameters in proposed scheme and existing schemes are analyzed, and finally comparing each other performance.
4.2 Parameter Analysis

Among the various parameters the most important parameters are analyzed below. Node density is number of nodes within the radio range. During routing process the proposed routing mechanism checks the whether any wormhole attack occurs. If wormhole attack occurs during hybrid routing process the defense mechanism against that attack and recover the network from that attack. Hence routing through the false path is aborted and protects the wireless sensor network from the wormhole attack. Throughput is defined as the ratio of number of packets received to the time seconds. The energy consumed by the sensor node varies depends upon the transmission distance source and destination. The transmitter amplifier energy required for transmission is 100pJ/bit/m. The energy required for idle operation is 40nJ/bit.
401
Vol. 3, Issue 1, pp. 395-404
Throughput= Total number of packets received Time(sec)
Figure.4. The above simulation result shows direct communication
Figure.5. The above simulation result shows multihop communication
In this simulation, all the nodes used to transmit data packets repeatedly. Therefore, we cannot find any advantage at low node densities. As the node density gets higher, the network lifetime gets longer. The node16 is within the connected region so direct transmission is possible with lower energy consumption. The Fig(8) shows that node16 directly transmits the data packets to base station. In Fig(9) the source node 6 sends the data packets to base station through node12 and node18. This shows optimal number of hops and optimal distance between source and destination.
Figure.6. The above simulation results shows the proposed scheme consumes less energy than existing scheme
Figure.7. The above simulation results shows the proposed scheme consumes less energy than existing scheme
Comparing with other existing schemes our proposed scheme has high throughput for the same environment of sensor nodes. The Fig (10) shows that EEHR consumes lower energy than the existing algorithms. Energy is measured in joules.
402
Vol. 3, Issue 1, pp. 395-404
V. CONCLUSION
The simulation results show that this type of routing enhances network lifetime considering limited sensor nodes energy. It will be efficiently used routing protocol in future, since it considers both energy level in each node and optimal no of hops. Thus hybrid routing reduces the average latency and energy needed for multihop communication. A great advantage of detection and defense mechanism against wormhole attack is, it doesnt require any geographical information about the sensor nodes, and doesnt take the time stamp of the packet as an approach for detecting a wormhole attack, which is very important for the resource constrained nature of the sensor nodes. In this paper we have evaluated the performance of different routing algorithms for energy efficient in sensor networks. Graph results shows that EEHR consumes minimum energy compared to PEGASIS, Greedy and ODGR. PEGASIS seems to have slightly more balanced energy consumption between nodes. The proposed scheme EEHR is analyzed as the best routing algorithm compared to PEGASIS, Greedy and ODGR, when energy efficiency is taken into consideration.
REFERENCES
[1] Gaafar A Elrahim, Hussein A Elsayed, Salwa E Ramly, Magdy M. Ibrahim, (NOV 2010), An Energy Aware WSN Geographic Routing Protocol, Universal Journal of Computer Science and Engineering Technology, pp.105-111. [2] Jabbar S, Butt A.E, Sahar N, Minhas A.A, (2011), Threshold based load balancing protocol for energy efficient routing in WSN, 13th International Conference on Advanced Communication Technology (ICACT),pp.196 201. [3] Jaewan Seo, Moonseong Kim , In Hur ,Wook Choi and Hyunseung Choo, (2010), DRDT: Distributed and Reliable Data Transmission with Cooperative Nodes for Lossy Wireless Sensor Networks, ISSN,pp.2793-2811. [4] Karim Seada, Marco Zuniga, Ahmed Helmy, Bhaskar Krishnamachari, (NOV 2004), Energy Efficient Forwarding Strategies for Geographic Routing in Lossy Wireless Sensor Networks, Sensys 04, Baltimore, Maryland, USA. [5] Lindsey S and Raghavendra C S, (2002), PEGASIS: power-efficient gathering in sensor information systems, Proceedings of the IEEE Aerospace Conference, Big Sky, MT, March, pp.11251130. [6] Maskooki A, Cheong Boon Soh, Gunawan E.; Kay Soon Low, (2011), Opportunistic routing for body area network, Consumer Communications and Networking Conference (CCNC), IEEE ,pp. 237 241. [7] Myung Kyun Kim, Ngo, Hoai Phong,(2011), A reliable and energy efficient routing protocol in industrial wireless sensor networks, International Conference on Advanced Technologies for Communications (ATC),pp.32 -35. [8] Organisation For Economic Co-Operation And Development, OCED, (DEC2009), Smart Sensor Networks: Technologies and Applications for Green Growth. [9] Sajjad Ahmad Madani, Daniel Weber, Stefan Mahlknecht, (2010), Position based Routing Protocol for Low Power Wireless Sensor Networks, Journal of Universal Computer Science, vol. 16, pp.1215-1233 [10] Idris M. Atakli, Hongbing Hu, Yu Chen, Wei-Shinn Ku, Zhou Su,(2008), Malicious node detection in wireless sensor networks using weighted trusted evaluation , The symposium on simulation of system security,(SSSS 08). [11] Khin sandar win, Pathin Gyi,(2008), Analysis of detecting wormhole attack in wireless sensor networks, World academy of science Engineering and Technology, pp.422-428. [12] Shahzad Ali and Sajjad Madani, (July 2011), Distributed Efficient Multi Hop Clustering Protocol for Mobile Sensor Networks, The International Arab Journal of Information Technology, Vol. 8, No. 3, Page No: 320 309. [13] Nauman Aslam, William Phillips, William Robertson, Shyamala Sivakumar, (2010), A multi-criterion optimization technique for energy efficient cluster formation in wireless sensor networks, Information Fusion - Published by Elsevier. [14] Banner Engineering (March 2009), Application Notes. [15] M. Botts, G. Pecivall, C. Reed, and J. Davidson, (2008) OGC sensor web enablement: Overview and high level architecture, in Lecture Notes in Computer Science, Geo Sensor Networks. New York: Springer, pp. 175190. [16] Mustard, S. (2007), Unraveling today's networks tangle wireless process control network, Process Engineering, vol.88, No.11, pp.154-163. [17] Y. Zhang, R. Yu, S. Xie, W. Yao, Y. Xiao, and M. Guizani, (April 2011), Home M2M networks: Architectures, Standards, and QoS improvement, Vol.49, pp 43-52.
403
Vol. 3, Issue 1, pp. 395-404
[18] Zhou Yan-li, Fan Xiao-ping, Liu Shao-qiang, Xiong Zhe-yuan (2010), Improved LZW Algorithm of lossless Data Compression for WSN, IEEE ICCSIT. [19] Junguo Zhang; Wenbin Li; Xueliang Zhao; Xiaodong Bai; Chen Chen; , "Simulation and Research on Data Fusion Algorithm of the Wireless Sensor Network Based on NS2," Computer Science and Information Engineering, 2009 WRI World Congress on , vol.7, no., pp.66-70, March 31 2009-April 2 2009 [20] Gutierrez, J.A., Naeve,M., Callaway, E., Bourgeois, M., Mitter, V., and Heile, B, IEEE 802.15.4: A developing standard for low-power low-cost wireless personal area networks, in network,IEEE,2001,p.12. [21] C. Karlof and D. Wagner, Secure Routing in Wireless Sensor Networks: Attacks and Countermeasures, Proc. 1st IEEE Intl. Wksp. Sensor Network Protocols and Applications (SNPA03), May 2003.
Authors Biographies
Deepika Srikumar is currently doing M.E in Communication Systems at Sri Shakthi Institute of Engineering & Technology, Coimbatore. She has done her B.E in Electronics and Communication Engineering from Anna University of Technology, Coimbatore, Diploma in Electrical and electronics Engineering from PSG Polytechnic college, Coimbatore. Her research interest is Senor Networks. She has presented various papers in symposium, national conference, international conferences and International journal. Seethalakshmi Vijaykumar is with the ECE department in Sri Shakthi Institute of Engineering & Technology, Coimbatore as Associate professor. She has done her B.E in Electrical and electronics Engineering from PSG college of Technology, Coimbatore, Tamilnadu, M.Tech in Electronics and Communication Engineering from PTU university, Punjab and pursuing PhD under Anna University of Technology, Coimbatore. Her research interest is Network Routing. She has rich 16 years of experience in industry as well as teaching. She has presented 16 papers in national conference and 11 papers in international conferences. She has published 4 papers in international journals.
404
Vol. 3, Issue 1, pp. 395-404
CO-OPERATIVE P2P INFORMATION EXCHANGE (CPIE) USING CLUSTERING APPROACH IN WIRELESS NETWORK
Department of Computer Science, School of Engineering, Pondicherry University, Puducherry, India 2 Department of Computer Center, Pondicherry University, Puducherry, India.
1
S. Nithya1 & K. Palanivel2
A BSTRACT
Wireless P2P networks such as ad hoc network, mesh networks, and sensor networks, have become a popular technology due to their potential to support huge volume of data with reliability and fault tolerance. Cooperative Peer to Peer Information Exchange (cPIE) is an innovative technique to improve the data availability using data present within the peers in the network. But the performance degrades gradually with increase in the number of peers in the network. Clustering approach is proved to be more efficient for solving the scalability problem in wireless networks. In this paper, we propose a cooperative PIE technique for large wireless P2P networks using clustering approach. Simulation result shows that the proposed technique improves the performance than the existing PIE techniques in network with large number of peers.
K EYW ORDS:
P2P; cooperative; information exchange; wireless networks; cluster.
I.
INTRODUCTION
Regarding the perception of communication, there subsist a huge number of networks which we rely on such as internet, mobile networks, peer-to-peer (P2P) networks, wireless networks, etc. Efficient research has been concentrated on the operation and management of communication networks with the essential intention of information transfer in terms of high reliability, security, robustness, and throughput. Wireless P2P networks are more comfortable for many applications since it offers a higher level of self independence and also managing the services what they exploit. Normally in P2P networks the clients wont be cooperative. It highly depends on the way system is built, if it is not proper it leads to dreadful conditions of the service and also the entire system will be collapsed. Thus, making the wireless P2P network nodes work cooperatively improves the overall efficiency of the application and the network. On the other hand, distinct sending sequences or scheduling strategies will have a straight impact on the overall network due to the shared wireless channel and the de facto half-duplex transmission quality. In some cases, the gap between the optimal and the worst is vast. Therefore, the information exchange among a group of peers still faces one more scheduling problem, which is unlike from the block scheduling problem. In this paper, we recognize such a problem of determining peer sending sequences as the peer scheduling problem. In other words, the peer scheduling problem is how to determine the scheduling policy among a group of potential senders to achieve the maximal utilization of limited wireless resources. A peer scheduling problem means between the groups of capability senders how to resolve the scheduling procedure to accomplish the maximal deployment of restricted wireless resources. We illustrate the peer scheduling problem using the following simple situation, consider in a wireless network which holds seven peers, nodes 1, 2, 3,4,5,6 and 7. And these peers communicate each other through a shared wireless channel. The main aim of this exchange is that each peer should have all the seven blocks and we are guessing to achieve this aim with lowest number of transmissions through
405
Vol. 3, Issue 1, pp. 405-417
arranging the sending sequence. These peers plan to exchange with each other a part of information which is consisting of seven blocks.
Fig 1.Cooperative Peer-to-Peer Information Exchange in Wireless P2P networks Table 1 Peer Sending Sequence IN-EFFECIENT EFFICIENT I) 1 < A, F, G > I) 1 <A, F,G> II) 2 < B, G > II) 6 <B ,D,E> III) 3 < D, F,G> III) 5 <C> IV) 4 <A, B, E> V) 5 < C, G>
Presently, peer 1 owns block A, F and G; peer 2 owns block B and G; peer 3 owns block D, F and G; peer 4 owns block A, B and E; peer 5 owns block C and G; peer 6 owns block B, D and E; peer 7 owns block C and F. In this example we can confirm through our reconsidered principles that peer sending sequence of <1, 6, 5> is a finest solution but <1, 2, 3, 4, 7 > is not finest solution. The solution is given in terms of transmission efficiency under the account of no packet loss and the peer should always send out packet with all the blocks it holds. Since each wireless nodes help mutually, we can consider this problem as Co-operative Peer-to-Peer Information Exchange (cPIE) approach. When the number of nodes in the network increases, the network performance decreases. It is evident from the theoretical analysis that even under the optimal circumstances, the throughput of each host decreases towards zero rapidly. Despite of the various solutions available, cluster formation seems to be more efficient for solving the scalability problem in adhoc networks. Clustering is the concept of grouping the network hosts which are in physical proximity. Thus clustering will help us to get scalability and it increases the network performance. Clustering algorithms designed at producing the minimum number of clusters that maximize the network lifespan and fault tolerance and provide load balancing and data throughput. From the above trace, we can justify that the cooperative approach among the peers improves the performance of cPIE application in wireless environment. But the throughput of the performance of the network degrades with increase in the number of nodes in the network. Thus in this paper, we propose an efficient Co-operative Peer-to-peer Information Exchange (cPIE) technique with an effective algorithm using clustering approach. The rest of the paper is organized as follows section 2 discusses about the existing work related to the proposed system. Section 3 presents the proposed methodology for cluster formation in wireless environment with critical concerns. In section 4, Cooperative PIE Mechanism has been proposed and in section 5, the performance of the proposed cPIE scheme has been evaluated in terms of time taken
406
Vol. 3, Issue 1, pp. 405-417
and message exchange through extensive simulation. Finally, section 6 concludes the paper and presents direction for future work.
II.
RELATED WORK
Wireless P2P environment becomes very significant research area with vast approaches to tune its critical performance criteria to make the system to get intended throughput. Regarding these P2P approaches and algorithms, there is a huge improvement in enlarging theoretical understanding. For example [1-4] [6] developed varieties of model to understand about throughput capacity of P2P algorithms beneath different models hypothesis [5,7] and others build up stochastic system models to estimate the managing rules with dynamic peer populations. These studies afford an excellent theoretical basis and several insights to why the P2P systems range so well. In P2P overlay building and P2P scheduling; a review of the best ways has been made which was given in [8]. A detailed study of structured and unstructured P2P networks and protocols is given in [9]. Popularities specified in [10-12] [25], certain file shows clustering in mainly real-world circumstances, system performance can be developed by regulating the fundamental overlay topology to imitate the clustering (e.g. . having a search overlay topology corresponding the data-sharing graphs of [10]). Yanfei Fan et al proposed a [13] cooperative Peer-to-peer Information Exchange (PIE) scheme with an efficient and light-weight scheduling algorithm which is suitable for a wireless network with very less number of nodes and notable drawback of the work is that the complete broadcast nature of wireless channels has not been fully developed. Clustering has ample benefits. It can simplify the range of the routing table saved at the unique node by restricting the route set up within the cluster [14] [21-23]. Network clustering groups collectively the hosts that are in physical propinquity. Each cluster is methodized of a Cluster-Leader (CL) and a number of cluster members. Cluster head is chargeable for organizing the fundamental operations of the cluster members such as channel access scheduling, power measurements, and coordination of intra and inter-cluster communications [15]. Clustering can also preserve the communication data transmission rate because it margin the scale of inter-cluster interactions to CLs and evade unnecessary exchanges of datas among the nodes [16]. By using clustering technique [17] [24] the energy efficiency will be significantly developed and the all over network traffic can be decreased. Its not efficient to recluster the entire network periodically as it is a heavy burden on use of network [1820, 15]. Hence the proposed cluster formation algorithms can increase the network efficiency. Because of this algorithm can reorganize the cluster whenever and wherever it is required, rather than doing it periodically. Observation from the Studied Related Works: To summarize, Clustering is the best approach to improve the performance criteria such as throughput, fault tolerance and load balancing of large P2P networks. Thus in this paper we are going to utilize the various advantageous factors of clustering approach in large wireless P2P environment. The proposed work consists of two folds, first cPIE which provide efficient outcome for the problem explained in the previous section. Second layer is clustering of large network which segregate the overall computation burden of the technique among different clusters which maximize the effectiveness of the cPIE by applying excellent way to schedule the peers to use the wireless medium to completely utilize broadcast nature of wireless channels.
III.
PROPOSED SYSTEM
In the proposed system section, we present our analytic framework with two levels of techniques, which are modeling the wireless peer-to-peer networks into clusters and to effectively achieve cPIE in the corresponding wireless environment.
3.1. Network infrastructure

In this work, we consider a peer-to-peer network model comprised of several wireless nodes which are also called peers. These peers share a common wireless channel and can communicate with each other directly. Thus, all peers can only communicate in actually a half-duplex mode. In other words, if two peers are transmitting at the same time, their signals will interfere with each other and no peer can correctly receive the transmitted signal.
407
Vol. 3, Issue 1, pp. 405-417
Consider a wireless peer-to-peer network model which consists of numerous wireless nodes and this can also be called as peers. These peers communicate explicitly with each other by sharing the common wireless channel. In fact all peers in this network model communicate in a half-duplex mode. Otherwise stated as, no peer can exactly obtain the broadcast signal if two peers are broadcasting at the equal time since their signals will impede with each other. However, alternately each peer can fetch the wireless channels due to the transmit character of their wireless channels and regain the frames perfectly when one and only one is broadcasting. With reference to a realistic application, we highlight the subsequent framework as shown in Fig. 2. A remote Base Station (BS) transmits a pack of packets (blocks) to the nodes in the network. Owing to weakening and mobilization of wireless channels, each and every peer gets some (maybe all or none) of these blocks. In order to modulate the overflow of downlink channels from the base station to those nodes and liberate the chokepoint of the base station as a network gateway, the nodes can share their received blocks with each other through local wireless networks.
Fig.2 Wireless P2P Networks and Cluster formed Wireless P2P network
3.2. Cluster construction

It is presumed that nodes in the wireless networks are not uniformly distributed, and every node in cluster has a unique identity number which is called as node id. These nodes are limited by electric power, memory space, computation capacity, and communication range. A node can communicate directly with other nodes in the same cluster.
Table 2 Notations Used NOTATION DID SID Enode Data MyID MyCL ClusterMemberArray ***_msg $$$(***_msg) DESCRIPTION Destination address field of message Source address field of the message Energy Level of the current node Information field of the message Address of current node Address of nodes CL (initially null) Array maintained by CL to hold its member addresses message to perform some operation $$$ field value of ***_msg
The base station can be positioned at any place of the network, and it is not limited by electric power, memory space, or data-processing capacity. It has virtually unlimited computational and
408
Vol. 3, Issue 1, pp. 405-417
communication power, unlimited memory storage. The base station serves as the gateway for external communication. The following description elaborate about the various operations performed in the wireless network such as Cluster Leader (CL) selection, Cluster formation for every CL and real-time node join/leave in the cluster. 3.2.1. Cluster leader selection Stage 1: Choosing cluster leader Step 1: Action performed by participating nodes SID = Current Node ID Energy= Enode DID = Address of Base Station Create data_msg = (Energy + Source ID) Send data_msg to DID Step 2: Action by BS to select CL on receiving cluster data message (data_msg) from group of nodes IF (Energy (data_msg) = High Level AND SID (data_msg) = Least ID) Create ack_msg=you are selected as CL DID = SID (data_msg) Send ack_msg to SID (data_msg) ELSE Discard data_msg ENDIF Pseudo Code for Cluster Leader Selection Step 1: Action performed by participating nodes For the effective transmission of data from the base station, a cluster leader should be selected. So this phase is for updating information of the entire nodes in the cluster. During this process of selecting cluster head, all the nodes of the cluster transmit a data message packet with message of their energy level which is denoted as Enode along with their source id (Energy + SID) to the destination id which is the base station. In order to reduce packet collisions, the nodes use random back-offs before sending the update packets. These nodes must exchange their data message only once, at the beginning of the network lifetime. Step 2: Action by BS to select CL During the selection phase of cluster head the nodes in the cluster send their energy data message to the base station. The base station can receive the energy data msg only for a finite duration of time which is called activation time. Beyond the activation time the base station cannot accept the energy data msg. Now the base station processes the data message and selects the cluster head that is having highest energy level. If a condition occurs where two nodes have same energy level, then the cluster head is selected on the basis of least ID. Least ID is the ID given to the node to provide user oriented priority such as node location, resource availability etc. After selecting the cluster head, an ack msg stating You are selected as CL is created in the base station. This ack msg is sent to the corresponding cluster head. If this process is not successful then the message is discarded. 3.2.2. Cluster formation Stage 1: Cluster Leader operation Step 1: At the end of the cluster leader selection process, base station disseminates the ack_msg to the corresponding selected CL node. On the reception of ack msg the node checks if the field value DID in the message just received for match with its own id. If the message is intended to itself then, the node declares itself as the cluster head otherwise just discards the message. On success, CL creates an invitation message stating I am CL, Join me and this invite message is broadcasted by the cluster leader to all adjacent nodes.
409
Vol. 3, Issue 1, pp. 405-417
Stage 1: Cluster Leader operation Step 1: On receiving acknowledge message (ack_msg) from BS Check IF (DID (ack_msg) = MyID) THEN Create invite_msg = I am CL join me SID=DID (ack_msg) Broadcast invite_msg ENDIF Stage 2: Joining member nodes Step 1: On receiving invite message (invite_msg) from any CL then Check IF (MyCL = Null) THEN Set MyCL= SID(invite_msg) Create join_msg = To join as Cluster Member DID =MyCL Send join_msg to DID ELSE Discard invite_msg ENDIF Step 2: On receiving join message (join_msg) from any node Add SID(join_msg) into ClusterMemberArray Ack_msg = Successfully added into Cluster DID =SID(join_msg) Send Ack_msg to DID ENDIF Step 3: On receiving join acknowledge message (ack_msg) from its CL Confirm MyCL and Discard ack_msg ENDIF Pseudo Code for cluster formation Stage 2: Joining member nodes operations Step 1: The nodes receiving the invite message check whether its MyCL field is null which means that it is not under any CL. If yes, it set its MyCL field as SID field of received invite message and send join message to its CL otherwise discard the received invite message. Step 2: On receiving the join message, the cluster leader processes the received information and adds the field SID of the join message to its cluster member array which holds all the member nodes of corresponding CL. To acknowledge CL creates an ack message mentioning Successfully added into Cluster to the newly added member node. Step 3: After receiving the join acknowledge message from the CL, new member nodes confirms its MyCL and discards the ack message. At the end of this step, clusters are formed completely and the cluster formed Wireless P2P network has been shown in fig 2. 3.2.3. Intra Cluster Routing Step 1: To send a data message to any node (say x) within its Cluster Create data_msg=Some data SID= MyID DID = Destination node address (of node x) Send Data_msg to Destination ID Step 2: On receiving data message (data_msg) from any node If Receive Data message from Node (within its cluster) then Check IF (SID(data_msg) = MyID) THEN
410
Vol. 3, Issue 1, pp. 405-417
// Some data received ELSE WHILE (Enode 0) Send data_msg to DID(data_message) END WHILE ENDIF Pseudo Code for Intra Cluster Routing Step 1: Intra cluster routing is a form of communication by which exchange of information happens within the cluster. If a data message is to be sent to any node (say x) within the cluster, the source node (say node A) creates the data message. Since node A is the sender of the data, the source id SID is assigned as MyID which is the ID of node A. The destination id field of the message is assigned as the address of node x and forwarded. Step 2: When a particular node in the cluster receives a data message from any other node, it checks if the destination id field of message, DID matches with its own id. If so, the message is intended to the current node otherwise, it forward the data message to the destination id given in the data message provided energy of the node is not zero. 3.2.4. Inter-Cluster Routing Stage 1: Operations on Home Cluster Step 1: The inter cluster routing is a type of communication through which information is exchanged from one cluster to other cluster. A node (say y) in one cluster can communicate with another node (say x) in any other cluster using the following steps. The node y creates the data message with the SID as the address of the node y and the DID as the address of node x. The address of cluster leader of node y which is kept as MyCL is assigned to CLID field of the message and forwarded to MyCL. Step 2: After receiving data message from any node of its own cluster the cluster leader checks if the CLID in the data message matches with its own id. If so, CL understood that the message is for a node in some other clusters thus it multicasts the data message to all other cluster in the network. Stage 1: Operations on Home Cluster Step 1: To send a data message to any node (say x) outside its Cluster Create data_msg=Some data Source ID= SID DID = Destination node address (of node x) CLID=MyCL Send data_msg to MyCL Step 2: Home CL operation on receiving data message from any node If Receive data message (data_msg) from node (from its cluster) then Check IF (CLID = MyID) THEN Multicast data_msg to all other CLs in the network ENDIF Stage 2: Operations on Destination Cluster Step 1: Destination CL operation on receiving data message (data_msg) from any CL Check IF (ClusterMemberArray contains DID(data_msg)) THEN Send data_msg to DID(data_msg) ELSE Discard data_msg ENDIF Step 2: On receiving data message (data_msg) from its CL Check IF (DID(data_msg) = MyID) THEN // Some data received
411
Vol. 3, Issue 1, pp. 405-417
ELSE WHILE (Enode 0) Send data_msg to DID(data_msg) END WHILE ENDIF Pseudo Code for Inter-Cluster Routing Stage 2: Operations on Destination Cluster Step 1: All the CLs in a network receive the data message which is multicasted by the source CL. Each cluster leader checks if the DID field of the message points to any node that comes under its cluster member. If the cluster member array contains the DID of the data message then its sends the data message to the destination id otherwise the message is discarded. Step 2: When a particular node in the cluster receives a data message from any other node through its CL, it checks if the DID of the data message matches with its own id. If it matches, then the node gets to know that it has received some data. If the destination id of the data message does not match with its own id, it forwards the data message to the destination id given in the data message provided energy of the node is not zero. 3.2.4. New Node to Join: When any node wants to leave from its existing cluster then it follow the following operations. Step 1: When a new node wants to join a cluster, then it checks the cluster leader of its nearest node by creating a check message stating inform me your CL. The SID field of the check message is set with the ID of new node and is broadcasted to all its adjacent nodes. Step 2: In response to the check message, the nearest node creates an acknowledge message saying Im under given CL. The SID, DID and CLID fields of the acknowledge message are assigned with IDs of the node given response, the new node and the MyCL field of the response node respectively and forward to DID. //When a new node wants to join into the cluster Step 1: To check the CL of its nearest node Create check_msg=Inform me your CL SID= MyID Broadcast check_msg Step 2: Any node receives Check message (check_msg) Create ack_msg=I am under given CL SID= MyID DID = SID (check_msg) CLID=MyCL Send ack_msg to DID Step 3: when a node receive acknowledge message (ack_msg) for its check message Check IF (DID (ack_msg) = MyID) THEN Set MyCH= MyCLID (ack_msg) Create join_msg = To join as Cluster Member DID =MyCH Send join_msg to DID ENDIF Pseudo Code for New Node Join Step 3: Now the new node receives the acknowledge message for its check message from one of its nearest node. It checks if the DID of the ack message matches with its own id. If it matches, then it sets
412
Vol. 3, Issue 1, pp. 405-417
its MyCL field as the CLID field in the ack message and creates a join message saying to join as cluster member and forwarded to its new CL. 3.2.5. Any Node to Leave: When any node wants to leave from its existing cluster then it follow the following operations. //When any node wants to leave from the cluster Step 1: The node want to separate from cluster send the leave message to its CL Create leave_msg=To leave from your cluster SID= MyID DID = MyCL Send leave_msg to DID Step 2: The CL receives Leave message (leave_msg) perform following operation Check IF (SID (leave_msg) = MyID) THEN Delete SID from ClusterMemberArray ENDIF Pseudo Code for Leave in lively manner Step 1: The node which wants to get separate from its current cluster creates the leave message and forwards it to its CL. Step 2: When the CL receives leave message from any node of its cluster. It just removes the ID of the requester node from its Cluster Member array.
IV.
COOPERATIVE PIE MECHANISM
In the previous chapter, it is elaborated how the clusters are being formed and each cluster will be having exactly one Cluster Leader (CL) and more Cluster Members (CM). The Total number of peers in the network can be given as, Where, n is the number of clusters in the network CMi is the number of Cluster Members in the ith cluster otal number of peers= +
We have introduced a peer scheduling problem in the introduction part of the paper which deals about the concept of loss of packets which is broadcasted by the tower, at one time, to all the peers in the network. In this phenomenon, we can assume two facts, Fact 1: Each packet sent by the tower should be received by at least one peer. Fact 2: No peer shall receive all the packets that are sent by the tower. These facts imply that it is possible to find each packet sent by the tower within the network instead of requesting to broadcast all the packets again which will be tedious. To handle the problem effectively a technique introduced called Co-operative Peer to Peer technique (cPIE). By this technique, the peer can receive the packets directly from other peers which received it errorless rather than getting the missing packets from the tower. Here we elucidate the process to effectively achieve cPIE technique using the clusters which are already formed in the network in three stages. Stage 1 Sharing packets within single cluster Stage 2 Sharing packets among the CLs Stage 3 Sharing new packets within the cluster which are received from other CLs Stage 1:
413
Vol. 3, Issue 1, pp. 405-417
Each node maintains a vector which is known as packet vector used to store the packets it received from the tower without any errors. After receiving each errorless packet from the tower, peers save it in the vector and till the completion of the broadcast process of the tower. Once the process gets finish, each CM sends its packet vector to its corresponding CL. Each CL waits until it receives packet vector from all of its k CMs, where k is the size of ClusterMemberArray of that CL. When a CL receives the packet vectors from all of its CMs, it starts to form a packet matrix called block matrix which is comprises of the number of CMs and packets received errorless. The block matrix of CL of i'th cluster can be represented as BMi [0.x, 0.y] where x is the number CMs in the ith cluster CL and y is the total number of different errorless packets received by CMs in the network. It can be explained as, If BM [a,b] = content of the packet, then the peer a received the packet b without any error and if BM [a,b] = 0, then the peer a received the packet b with error. Based on the BM matrix values, each CL identifies which are the packets are not received correctly by each CM (i.e.) if BM [a,b] = 0. As a result, CL retrieves the packet from any peer which received it errorless and forwards it to all the peers which didnt receive or received with error. Stage 2: After the completion of Stage 1, each CLs share the packets they have with each other. This is because some packets are received by CMs of any one cluster, to make it available to all other clusters stage 2 is performed. At the success of stage 2, each CL has all the packets, without any error, broadcasted by the tower. Stage 3: At this stage each CLs share the new packets, which they received during stage 2, with their CMs. This makes sure that each CMs of each cluster has all the packets that are broadcasted by the tower without any error. At the successful completion of all these three stages, each peer in the network holds equal and all the packets broadcasted by the tower without asking for re-broadcasting.
V.
SIMULATION RESULTS AND ANALYSIS
In this section, we evaluate the performance of the proposed cPIE technique using clustering approach by conducting extensive simulation using OMNeT++ simulation tool. To justify the effectiveness and efficiency of the proposed technique, we assessed performance parameters such as size of the network, time taken to perform cPIE and the total number of message transfer between peers are compared against the PIE technique proposed in [13] (hereafter it will be mentioned as base PIE technique). We have conducted evaluation in two stages, the change in the size of the network vs. time taken for completion of cPIE technique and change in the size of the network vs. the total number of message transfer between peers in the network. Stage I: Evaluation in terms of Change in the Size of the Network vs. Time Taken for Completion of cPIE Technique In this phase, assessment has been carried out to justify that the proposed system is highly scalable. The time taken for performing cPIE operation increases optimally, whereas the same will increases tremendously with the increase in the size of the network in the base technique. From the simulation, the performance of the system with change in the size of the network vs. time taken for completion of cPIE technique has been assessed and tabulated in Table 3.
414
Vol. 3, Issue 1, pp. 405-417
Table 3. Performance measures assessed during stage - 1 part of analysis
Base PIE technique Sl. N o Size of the networ k (peers) 10 100 500 1000 5000 Total packets broadcaste d 100 500 1000 5000 5000 Numbe r of peers involve d 4 39 68 121 129 Time taken for complete in seconds(a ) 0.3 1.52 2.89 12 32 Total packets broadcaste d 100 500 1000 5000 5000 cPIE technique Numbe r of peers involve d 6 59 86 137 143 Time taken for complet e in ms (b) 0.23 0.63 1.87 6.25 21.2 Efficiency improvemen t in terms of time taken (a/bx100) 30.43% 41.26% 54.54% 92% 50.94%
Packets exchange d 30 152 289 1201 3204
Packets exchange d 23 63 187 625 2120
1 2 3 4 5
The tabulation is prepared as comparison between the base and the proposed cPIE techniques in terms of the total packets broadcasted from the source tower, the total number of peers involved in the information exchange, total number of packets exchanged and the time taken for completion of the technique in seconds. Fig.3 shows the comparison graph for which elucidates that the proposed technique out-performs the base technique at large number of peers in the network. Stage II Evaluation in terms of Change in the Size of the Network vs. The Total Number of Message Hops between Peers In this phase, assessment has been performed to validate that the total number of message hops required for each message exchange required to cPIE is highly minimized using proposed system in network with large number of peers. Number of hops required for message exchange will be higher for network with sparsely distributed nodes. From the simulation results, the performance of the system with change in the size of the network vs. average message hops per message exchange has been assessed and tabulated in Table 4. The tabulation is prepared as comparison between the base and the proposed cPIE techniques in terms of the total packets broadcasted from the source tower, the total number of peers involved in the information exchange, total number of packets exchanged and the total number of message hops performed and the average message hops taken per message exchange. Fig.4 shows the comparison graph for which make clear that the proposed technique with clustering out-performs the base technique at large number of peers in the network.
Table 4. Performance measures assessed during stage - 2 part of analysis
Size of the netw ork (peer s) 10 100 500 1000 5000 Base PIE technique Total packets broadca sted 100 500 1000 5000 5000 Number of peers involve d 4 39 68 121 129 Total message hop performe d 62 762 8247 34536 134566 Average hops per message exchange (a) 2.07 5.01 28.54 28.76 42.00 Total packets broadca sted 100 500 1000 5000 5000 cPIE technique Total message hop perform ed 43 178 643 4563 13255 Average hops per message exchange (b) 1.87 2.83 7.39 7.30 6.25 Efficiency improveme nt in terms of hops (a/bx100) 10% 77.03% 286.19% 293.97% 572%
Sl. No
Packets exchange d
Number of peers involved
Packets exchange d
1 2 3 4 5
30 152 289 1201 3204
6 59 86 137 143
23 63 87 625 2120
415
Vol. 3, Issue 1, pp. 405-417
50000 100000 150000
35 Time taken to accomplish in Sec Total message hops performed 30 25 20 15 10 5 0 10 100 500 1000 5000 Base PIE technique cPIE technique
Base PIE technique cPIE technique
0 10
Size of the network
100 500 1000 Size of the network
5000
Fig 3. Comparison of parameters assessed at stage - 1 of analysis analysis
Fig 4. Comparison of parameters assessed at stage - 2 of
VI.
CONCLUSION
Cooperative Peer to Peer Information Exchange (cPIE) is a novel technique to improve the data availability using data present within the peers in the network. But the performance degrades gradually with increase in the number of peers in the network. In this paper, we proposed an effective Cooperative Peer to Peer Information Exchange technique using clustering approach for large wireless P2P networks. The proposed technique formulate the cPIE system perform efficiently even at large wireless networks by segregating it into number of clusters. We also elaborated about the various operations performed in the wireless network such as cluster leader selection and cluster formation. Our simulation results justify that the proposed technique is more efficient than the existing PIE techniques for the network which holds large number of peers. Our future works will focus on incorporating network coding technique to improve the efficiency of cPIE technique further and constraint based cluster leader selection.
REFERENCES
[1]. J. Mundinger, R.R. Weber, G. Weiss, Analysis of Peer-To-Peer File Dissemination amongst Users of Different Upload Capacities, in: Performance Evaluation Review, Performance 2005 Issue. [2]. R. Kumar, Y. Liu, K. Ross, Stochastic Fluid Theory for P2P Streaming Systems, in: Proceedings INFOCOM 2007 [3]. J. Li, Philip A. Chou, C. Zhang, Mutualcast: An Efficient Mechanism for Content Distribution in a peerTo-Peer (P2p) Network, in: Proceedings of ACM SIGCOMM Asia Workshop, China, April 2005. [4]. S. Liu, R .Zhang-Shen, W. Jiang, J. Rexford, M. Chiang, Performance Bounds for Peer-Assisted Live Streaming, in: Proceedings of ACM SIGMETRICS, 2008. [5]. D. Qiu, R. Srikant, Modeling and Performance Analysis of Bittorrent like Peer-To-Peer Networks, in: Proceedings of ACM SIGCOMM, 2004. [6]. L. Massoulie, M. Vojnovic, Coupon Replication Systems, in: Proceedings of ACM SIGMETRICS, Banff, Alberta, Canada, 2005. [7]. B. Fan, D.M. Chiu, J.C.S. Lui, The Delicate Tradeoffs in Bittorrent-Like File Sharing Protocol Design, in: Proceedings of ICNP, 2006. [8]. Li J (2008), On Peer-To-Peer (P2P) Content Delivery. Peer-to-Peer Netw Appl 1:4563 [9]. Keong Lua E, Crowcroft J, Pias M, Sharma R, Lim S (2005) A Survey and Comparison of Peer-To-Peer Overlay Network Schemes. IEEE Commun Surveys Tut 7(2):7293, Second Quarter. [10]. A. Iamnitchi, M. Ripeanu, and I. Foster, Small-World File-Sharing Communities, in Proc. IEEE INFOCOM, Mar. 2004, vol. 2, pp. 952- 963.
416
Vol. 3, Issue 1, pp. 405-417
[11]. A. Klemm, C. Lindemann, M. K. Vernon, and O. P. Waldhorst, Characterizing the Query Behavior in Peer-To-Peer File Sharing Systems, in Proc. ACM Internet Measurement Conf. (IMC), 2004. [12]. F. Le Fessant, S. Handurukande, A. M. Kermarrec, and L. Massouli, Clustering in Peer-To-Peer File Sharing Workloads, in Proc. IPTPS, Feb 2004 [13]. Yanfei Fan, Yixin Jiang, Haojin Zhu, Xuemin Shen, Cooperative Peer-to-Peer Information Exchange via Wireless Network Coding, IEEE Global Telecommunications Conference, 2009. GLOBECOM 2009. [14]. K. Akkaya, M. Younis, A Survey on Routing Protocols for Wireless Sensor Networks, ELSEVIER journal of Ad Hoc Networks 3 (3) (2005)325349. [15]. Akbari Torkestani J, Meybodi MR. A Mobility-Based Cluster Formation Algorithm For Wireless Mobile Ad Hoc Networks. J Cluster Comput, 2011, in press. [16]. M. Younis, M. Youssef, K. Arisha, Energy-Aware Management in Cluster-Based Sensor Networks, Computer Networks 43 (5) (2003) 649668. [17]. K. Akkaya, M. Demirbas, R.S. Aygun, The Impact of Data Aggregation on the Performance of Wireless Sensor Networks: A Survey, Wiley Journal of Wireless Communications and Mobile Computing (2006). [18]. Gupta P, Kumar PR. The Capacity of Wireless Networks. IEEE Trans Inf Theory 2000; 46:388404. [19]. Baker D, Ephremides A, Flynn JA.The Design and Simulation of a Mobile Radio Network with Distributed Control. IEEE J Sel Areas Commun 1984; 2:22637. [20]. Basagni S, Mastrogiovanni M, Panconesi A, Petrioli C. Localized Protocols For Ad Hoc Clustering And Backbone Formation: a performance comparison. IEEE Trans Parallel Distributed System 2006; 17:292306. [21]. Nguyen, D.N.; Krunz, M.: A cooperative clustering protocol for energy constrained networks, in: Proceedings of 8th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks (SECON), 2011 [22]. Dang Nguyen; Minet, P.; Kunz, T.; Lamont, L.; New findings on the complexity of cluster head selection algorithms, in: IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM), 2011 [23]. Abboud, Khadige; Zhuang, Weihua: Impact of Node Clustering on Routing Overhead in Wireless Networks,in : IEEE Global Telecommunications Conference (GLOBECOM 2011), 2011 . [24]. Muthuramalingam, S.; Sujatha, M.; Surya, R.; Rajaram, R.:An enhanced sectorized clustering scheme based on transmission range for MANETS, in: International Conference on Recent Trends in Information Technology (ICRTIT), 2011 . [25]. Ka Hung Hui; Tianyi Li; Dongning Guo; Berry, R.A.;Exploiting peer-to-peer state exchange for distributed medium access control ,IEEE International Symposium on Information Theory Proceedings (ISIT), 2011. Authors S Nithya is a Master Degree student, Department of Computer Science, Pondicherry University, Pondicherry, India. She has obtained her B.Tech in the field of Computer Science, Pondicherry University, India. Her research areas include Wireless Systems, Mobile Computing and Web Services.
K. Palanivel received his B.E in Computer Science & Engineering from Bharathiar University, Coimbatore, India in 1994 and M. Tech. in Computer Science & Engineering and Pondicherry University, Pondicherry, India in 1998. He was the member of Indian Society of Technical Education in 2004. He is also a member of International Society of Scientist, Technocrats and Engineers (ISSTE). He is pursuing his Ph.D. in Computer Science and Engineering in Pondicherry University, India since 2008. He is working as Systems Analyst in Computer Centre of Pondicherry University. His field of interest is Software Engineering, Computer Networks and Design Patterns. He is having more than 10 years of experience in teaching and computing.
417
Vol. 3, Issue 1, pp. 405-417
PERFORMANCE EVALUATION OF OBJECT RECOGNITION USING SKELETAL SHOCK GRAPH: CHALLENGES AND FUTURE PROSPECTS
Pankaj Manohar Pandit1, Sudhir Gangadharrao Akojwar2
1
Dept. of Ele. & Tele. Engg., Jawaharlal Darda Institute of Engg. & Technology, Yavatmal 2 Principal, Dr. Bhausaheb Nandurkar College of Engg. & Technology, Yavatmal
A BSTRACT
This paper describes a graphical method of Skeletal Shock Graph, which is based on the shape or geometry of the object. Shock Graph is an abstraction of the skeleton of a shape onto a Directed Acyclic Graph (DAG) in which the skeleton points are labeled according to the local variation of the radius function at each point. A large image data base is created by using suitable image acquisition technique, which is converted into binary images. Skeleton and its labeling of the binary image is obtained by applying Skeletonization Algorithm. Then next steps adopted are formation of Shock Graph and labeled tree, indexing the data base and generation of attribute vectors, pruning the data base and lastly matching the tree of query image with that of database images for recognition. This paper takes a critical review of the earlier contributions in this field and discusses the existing challenges and prospective research areas in Skeletal Shock Graph based object recognition.
K EYW ORDS:
labeled tree.
Directed Acyclic Graph, Skeletonization, Pruning of data base, Skeletal Shock Graph,
I.
INTRODUCTION
One of the most fascinating features of the visual system of human being is ability to quickly recognize a stimulus from a large visual memory. Methods of object recognition and matching are broadly classified as intensity based methods relying on color, texture and geometry based methods, which use shape and structure of the object to match and retrieve from the query images. Most commonly used techniques of managing databases of image are Keywords or text-based technique, which is a common technique to provide information about the contents. It is suitable for texture based image with suitable segmentation. But for satisfactory degree of correctness & details, it requires large & complicated keyword system and manual annotations which are highly timeconsuming, costly and dependant on the subjectivity. The second technique is content based technique which involves generation of image description by analyzing content of image. The process is very common and simple. First, it retrieve image from database similar to the query image, this description store and extract features from query image and finally it estimates the similarity measures resulting image retrieval. The third method is Graphical or Graph based method which is based on the shape or geometry of the object and uses a basic fact that humans show a remarkable ability to recognize objects without effort, despite the fact that objects may vary in color, texture, or size. For recognition purposes, objects are often represented as attributed graphs whose nodes represent their features (or their abstractions) and whose edges represent relations (or constraints) between the features. These graph representations allow us to express many perceptually significant object properties, such as geometric or hierarchical part structures. Skeletal Shock Graph is an effective tool for object recognition, based on its geometry. The concept of Shock Graphs was first launched by Siddiqi and Kimia. Basically, Shock Graph is an abstraction of
418
Vol. 3, Issue 1, pp. 418-426
skeleton of a shape onto a Directed Acyclic Graph (DAG) in which the skeleton points are labeled according to the local variation of the radius function at each point. Since its inception and evolution, it has been effectively used as a significant tool in the area of object recognition, which finds tremendous applications in automated or computer assisted environments. Shape recognition is one of the primary goals of computer vision, allowing an image signal to be semantically labeled according to a priori knowledge of objects in the world. Early shape recognition work in the 60s and 70s focused on the categorization or generic object recognition problem, in which exemplar objects, i.e., specific object instances, were matched to coarse, prototypical models designed to be invariant to within-class shape and appearance deformation. Inspired by Blum's original idea of using directed graphs to define equivalence classes of shapes, Siddiqi and Kimia [1] defined the concept of a shock graph. II. SKELETON AND SKELETONIZATION Abstraction of object information is an important need for any automated or computer supervised or assisted environment for shape representation and modeling. Of-course, the level of abstraction depends on the task for which the shape information is being required. Skeleton or medial axes have been extensively used for characterizing objects satisfactorily using structures that are composed of thin line or arc like patterns. Skeletonization is an image processing operation which reduces input shapes to axial stick-like representations. It has many applications like Industrial inspection, Expert system for Animal census, Intruder detection, Expert Security systems, Image / document processing, Biological part matching, Computer graphics, Robotics etc.
2.1 Medial Axis Transform

The medial axis transform (MAT) is basically described by using a medial axis and a radius function. The skeleton of a shape is like a stick diagram. One of the first formal definitions of the skeleton was proposed by Blum, who defined the medial axis of a shape as the locus of center points of all the maximal circles contained within the shape boundary. Blum called the shape's skeleton as the Medial Axis Transform (MAT). [28, 29, 30]
Fig.1
Skeleton of object
Blum also proposed an alternate definition of MAT in terms of a grassfire frontThe idea is to imagine that the shape is uniformly covered with grass and that fire is set simultaneously to all boundary points. The fire front will start propagating toward the center and will extinguish when meeting an opposing front. The collision points of the fronts form the medial axis of the shape. The distance of any skeleton point to its closest boundary points defines the time of formation of the points, and it corresponds to the radius of the maximal inscribed circle. Skeleton is a representation of a 2D object which preserves the details of the silhouette of an object. Skeletons are converted to shock graph. Shock graph is nothing but shape abstraction which groups skeleton points according to the local variation of a radius function. These groups are called as Shock groups and are labeled as 1, 2, 3 or 4.Such a grouping will provide a decomposition of a skeleton into parts, whose interrelation will confirm to a well-defined grammar [1, 2, 31].The shock graph is a directed acyclic graph and is obtained from these Shock groups which are the constituent primitives of the Shock Graph. The arrangement of shock graph is according to the Shock Graph grammar. Shock Graph is generated for each 2D image in the data base and is compared with the shock graph of the query image very efficiently and leads to an efficient technique for pattern matching. Type 1 shocks form a segment of skeleton points, in which the radius function varies monotonically, as is the case for a protrusion. A type 2 arises at a neck, and is immediately followed by two type-1 branches bowing away from it in opposite directions. Type 3 shocks belong to an interval of skeleton
419
Vol. 3, Issue 1, pp. 418-426
points, in which the radius function is constant. Finally, a type 4 arises when the radius function achieves a strict local maximum i.e. the boundary collapses to a single point.
Fig.2- Types of shock graphs
Fig.3- Shock graphs based matching
III.
STEPS IN CONSTRUCTION OF SHOCK GRAPH
The shock graph construction involves following steps3.1. Labeling

R(s) be the radius function R : S R+ L(s) be a shock labeling function L : S {1,2,3,4} S is the set of continuous medial axis points (i.e., shocks) of shape X Then L(s) for s S is defined as:
3.2. Grouping
Let B1,..Bn be the longest groups of connected shocks in S s.t. s, s Bi, L(s) = l(s) and s Bi, either |N(s) |2 or |N(s) | >2, then s must be a terminal point of Bi ( i.e., Bi \ {s} is connected ), for 1 i n. Let the groups label, l(Bi), be the label of the label of shocks in Bi, and similarly, let the time of formation of the group, t(Bi), be the interval [min (R(s)), max (R(s))] defined by the time of formation R(s) for all s Bi
3.3.Shock Graph
The shock graph of a 2-D shape is a labeled graph G= (V, E, ) such that : Vertices V = {0,..n}, corresponding to the groups B1,..Bn and 0 denoting a root node; V directed from vertex i to vertex j if and only if i j, and i 0 Edges ( i, j) E t(Bi) t (Bj) Bi Bj is a connected set of shocks, or i =0 k 0 (k, j) E; { #,1,2,4} s.t. (i) = # if i = 0 and = l(Bi) otherwise. Labels : V The critical survey of related work indicates various steps involved in object recognition using shock graph are Creating a large image data base by using suitable image acquisition technique. Converting an image of object or a 2-D shape into a binary image. Obtaining the skeletons of the binary images (Skeletonization Algorithm). Labeling the skeleton. Construction of the shock graphs and labeled tree from the skeleton. Indexing the data base and generation of attribute vectors. Pruning the database (if required) by comparing the shock graphs of objects in the data base with the shock graph of the query image and retaining only the closest ones. Matching the tree of query image with that of database images for recognition.
420
Vol. 3, Issue 1, pp. 418-426
Shock Graph offers the advantages that the labeling of skeleton points or shocks is invariant to translation and rotation of the shape and since the labeling is locally determined, it is stable under uniform scaling and within-class part deformation. Shock graphs offer a powerful and challenging framework for representing and matching qualitative shape [3]. But, quite limited efforts have been devoted to their use in realistic scenes in which a silhouette cannot be properly segmented. A careful examination of shock graph based recognition proves that most of the approaches and methodologies are conventionally applied to un-occluded, presegmented counter. Also, the overview of the research work in this field indicates that there is a tremendous scope in the area of shape recognition especially about recognition issues under the condition of image obscured and deformation. Hence it is necessary to focus on these issues and try to evolve a robust recognition based strategy, based on shock graphs.
IV.
RELATED WORK
Although there has been quite extensive work in the area of shape representation and matching, shape recognition is still an open research issue [2]. With the time, many shape-matching approaches have emerged, but high space, time complexity and moderate recognition rate are still the limiting factors for their widespread acceptance as well as popularity. Sebastian et al. [4] use the full classification of the shock graph (MA endowed with a finer classification based on dynamics of flow) [5]. The shape recognition technique has to be robust to visual transformations like articulation and deformation of parts, viewpoint variation and occlusion. Thus, the shape representation has to effectively capture the variations in shape due to these transformations. In previous recognition applications, shapes have been represented as curves [6, 7], point sets or feature sets, and by medial axis [8, 9, 10, 11, 12] among others. 2D shape matching is achieved using hierarchical tree structure extracted from shock graph that in turn extracted from the skeleton of the shape of interest. It attempts to decompose a shape into a set of parts. Xiang Bai et.al.[15]matched skeleton graphs by comparing the geodesic paths between skeleton endpoints. In contrast to typical tree or graph matching methods, it does not consider the topological graph structure. The proposed comparison of geodesic paths between endpoints of skeleton graphs yields correct matching results. This method is able to produce correct results in the presence of articulations, stretching, and contour deformations. But the performance of this method is limited in the presence of large protrusions, since they require skipping a large number of skeleton endpoints. Shahbudin S.et.al.[16] highlights the efficacy of the simplified shock graph (SSG) derived as a result of pruning from the original shock graph. It tests the feature vectors of the SSG with three different multi-class Support Vector Machines (SVM) classifier models namely the one-against-all (OAA), one-against-one (OAO) and the bias support vector machines (BSVM). As such, the validity of SSG as feature vectors to represent human postures is confirmed based on the findings from all three methods of SVM applied. Macrini et.al. [17] presented an algorithm for identifying and representing the ligature structure, and restoring the non-ligature structures that remain. This leads to a bone graph, a new medial shape abstraction that captures a more intuitive notion of an objects allies parts than a skeleton or a shock graph. It offers improved stability and within-class deformation invariance. Zaboli et. al. [19] presented a modified and optimized shock graph based method by adding branch points as key points to the shock graph and its grammar and consequently it developed a new grammar. But the recognition rate is not much attractive. Also, it had a limited focus on stability issues of the shock graph. Behnam Hosseini et.al. [27] proposed some critical points on skeleton of silhouette are used for recognition and classification of human body movement and uses MLP with back-propagation for recognition of movement. But it does not focus the stability issues under articulation and deformation. Wei Zhao et.al. [34] proposed the use of skeletonization algorithm for removal of unwanted edges in processing of MRI images of skull and the brain tissue. It used the thresholding methods for this and use of skeletonization is limited to this purpose only. Aaron D. Ward et.al. [25] proposed the G-MAT( Group-wise Medial Axis Transform), which is a group-wise skeletonization framework that yields a fuzzy significance measure for each branch, derived from information provided by the group of shapes for the determination of the pruning order
421
Vol. 3, Issue 1, pp. 418-426
of skeleton branches, using geometric and topological branch feature information provided by the group as a whole. Although the geometric and topological branch features used in this paper provided superior performance, there remain substantial opportunities for improvement using more sophisticated, pose-invariant features. Jianhao Ding et.al. [13] discussed skeletonization method for the skeleton feature points extracted from human body based on silhouette images, which uses gradient of distance transform to detect critical points inside the foreground. Then, it converge and simplify critical points in order to generate the most important and elegant skeleton feature points and finally, presents an algorithm which connects the skeleton feature points and estimates the position of skeleton joints. But the results are unsatisfied and even encounter some mistakes when the joints are hidden or kept out. That is because this approach uses the uniform scales of the human body parts to determine the position of the inner joints such as elbow and knee joints, there might cause some errors between the scales of the different bodies. Behnam Hosseini et.al. [14], presented skeleton of silhouette, being used for recognition and classification of human body movement and MLP with back-propagation is being used for recognition of movement. It has resulted average recognition rate of 92 %. But, it did not focus the stability issues under articulation and deformation. An object recognition by comparing the sub-graphs by inserting a node in Shock branch whenever there is a sharp direction change has also been proposed by S. P. Hingway et.al. [32]. Addition of node improves the object recognition results. This changes the Shock graph topology. The object identification strategy called as Sub Graph matching method is used for Object recognition. But the improvement in result is not significant when the Skeleton does not have many curved segments.
V.
LIMITATIONS AND SCOPE
Thus the critical review of the previous work gives rise to many unaddressed challenging research issues that need to be tackled effectively for the efficient object recognition. Scale and viewpoint changes may often originate new branches in a skeleton due to small details on the boundary that were not identified at a lower scale or that were not visible from a different viewpoint. These new branches can affect, to a certain extent, the structure a shock graph. The shock labels depend on the pair-wise difference of the radius function at adjacent shocks which requires, in practice, to define some threshold to increase the tolerance to differences in radii. Concave corners in the shape boundary may produce skeleton segments, where all the points are related only to the corner points at the boundary. Thus, small changes in the positions of these corner points will cause rather big changes in the ligature segments, which will ultimately give rise to differences in the corresponding shock graph. Noise may also have effect on the structure of shock graph. The medial axis is sensitive to boundary perturbations (Boundary Noise), which introduce spurious edges in the graph. This may be typically tackled by regularization during the detection process or post-detection. Here regularization is an integral part of the recognition process. Articulation and Deformation of Parts is also the issue of focus. The shock graph inherently segments a shape into parts and captures the hierarchical relationship between them. Hence it is robust to changes which may occur in some of the parts. The viewpoint variations changes gradually, the spatial location and the shape of parts and it may be handled by deform edit. At certain views, there is a sudden appearance or disappearance of a part, which may be handled by the remaining edits. Appearance of a part may be handled by the splice edit. Shadows and highlights often lead to segmentation errors in figure-ground segregation. Highlights tend to cause small changes in the boundary, but may typically affect the shock-graph topology. Shadows are often more global in nature, but tend not to affect the shock-graph topology. Increasing Partial occlusion i.e. occluded blends with the shape or occluded blends with the background increases Shock Graph edit cost. Hence, study of stability issues of Shock Graphs under various challenging situations may be the matter of focus. To accommodate this, robust method for labeling the skeleton points that will also accommodate small differences in scale may be one of the steps. Shock graphs define meaningful equivalence classes for shapes i.e. two shapes abstracted by the same graph are expected to be
422
Vol. 3, Issue 1, pp. 418-426
perceptually similar at a coarse level. This property will allow us to index a model database to recover a set of candidate models with similar shock graphs. The next step may then be to verify each of these candidates by evaluating the information that we have ignored in the abstraction. With this goal in mind, a list of attribute vectors is associated with every point s in every node in the graph. Such vectors are 4-tuples (x, y, t, ) representing the attributes of each shock point in the node: Euclidean coordinates (x, y), time of formation t, and direction . The set of attributes of each node may be used to define a local measure of similarity between two nodes in different graphs. Use of suitable AI technique like Fuzzy Logic, Neural Network or Hybrid tools for attribute vectors associated with every point may result into effective object recognition.
VI.
Figure 4 : Snapshot of the MATLAB window showing query image
Figure 5 : Snapshot showing skeleton & labeling for query image
Figure 7: Snapshot showing the result of tree matching with database for query image. Figure 6: Snapshot of the MATLAB window showing shock-tree of query image.
Figure 8: Snapshot of the complete GUI.
423
Vol. 3, Issue 1, pp. 418-426
VII.
IMPLICATIONS
The proposed research outcome will definitely be a valuable contribution in the vast and ever open field of shape recognition. The outcome of the proposed research work may have several applications in the fields of Automated Industrial inspection, Expert system for Animal census, Intruder detection, Expert Security systems, Image / document processing, Biological part matching, Computer graphics, Robotics etc. The labeling of skeleton points or shocks is invariant to translation and rotation of the shape. Since the labeling is locally determined, it is stable under uniform scaling and within-class part deformation. But all these issues are always posed to the challenging situations like boundary noise, articulation and deformation of parts, view point variations and partial occlusion. So efforts will be concentrated on improving the stability of shock graph for object recognition under these challenging issues. Further, it is expected that the use of suitable AI technique will make the recognition process fast and accurate.
VIII.
CONCLUSION
Skeletal Shock Graph is an effective tool for object recognition, based on its geometry Object recognition is one of the primary goals of computer vision, allowing an image signal to be semantically labeled according to a priori knowledge of objects in the world. Scale and viewpoint changes, concave corners in the shape boundary, boundary perturbations, viewpoint variations, segmentation errors due to shadows and highlights, increasing partial occlusion are still some of the challenging situations in object recognition, may be the matter of focus for the researchers. To tackle this, robust method for labeling the skeleton points that will also accommodate small differences in scale may be one of the steps. Shock graphs define meaningful equivalence classes for shapes which allow us to index a model database to recover a set of candidate models with similar shock graphs. Thus, an improved skeletonization, improved indexer, improved matcher and use of suitable AI tool may further enhance the recognition efficiency of the system, especially under the challenging situations for a large images data base of varieties of shapes concerned for varieties of object recognition applications in automated or computer assisted environments.
REFERENCES
[1] K. Siddiqi and B. B. Kimia. A shock grammar for recognition IEEE Conference on Computer Vision and Pattern Recognition, San Francisco, USA, 1996. [2] K. Siddiqi, A. Shokoufandeh, Sven J. Dickinson, and Steven W. Zucker, Shock graphs and shape matching, IEEE International Conference on Computer Vision, pages 222-229, 1998. [3] Aurelie Bataille, Sven Dickinson. Coarse-to-Fine Object Recognition Using Shock Graphs, SpringerVerlag Berlin Heidelberg, LNCS 3434, pp. 203212, 2005. [4] T.B. Sebastian, P.N. Klein, and B.B. Kimia, Recognition of Shapes by Editing Their Shock Graphs, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, no. 5, pp. 550-571, May 2004. [5] P.J. Giblin and B.B. Kimia, On the Local Form and Transitions of Symmetry Sets, Medial Axes, and Shocks, International Journal on Computer Vision , vol. 54, nos. 1-3, pp. 143-157, Aug. 2003. [6] Yoram Gdalyahu and Daphna Weinshall, Flexible syntactic matching of curves and its application to automatic hierarchical classification of silhouettes, PAMI, vol. 21, no. 12, pp. 13121328, 1999. [7] Laurent Younes, Computable elastic distance between shapes, SIAM J. Appl. Math., vol. 58, pp. 565 586, 1998. [8] S. C. Zhu and A. L. Yuille, FORMS: A flexible object recognition and modeling system, International Journal on Computer Vision, vol. 20, no. 3, 1996. [9] Daniel Sharvit, Jacky Chan, Huseyin Tek, and Benjamin B. Kimia, Symmetry-based indexing of image databases, JVCIR, vol. 9, no. 4, pp. 366380, 1998. [10] K. Siddiqi, A. Shokoufandeh, S.J. Dickinson, and S.W. Zucker, Shock graphs and shape matching, International Journal on Computer Vision, vol. 35, no. 1, pp. 1332, 1999. [11] M. Pelillo, K. Siddiqi, and S.W. Zucker, Matching hierarchical structures using association graphs, PAMI, vol. 21, no. 11, pp. 11051120, 1999. [12] T. Liu and D. Geiger, Approximate tree matching and shape similarity, ICCV, pp. 456462, 1999. [13] Jianhao Ding, Yigang Wang, Lingyun Yu, Extraction of Human Body Skeleton Based on Silhouette Images, Second International Workshop on Education Technology and Computer Science, 2010.
424
Vol. 3, Issue 1, pp. 418-426
[14] Behnam Hosseini, Saeed Vahabi Mashak, S.A.R. Abu-Bakar, Categorization of Human Movement Based on Rule-Based Classifier, Fourth Asia International Conference on Mathematical/Analytical Modelling and Computer Simulation, 2010. [15] Xiang Bai and Longin Jan Latecki, Path Similarity Skeleton Graph Matching,IEEE Transactions on pattern analysis and machine intelligence, VOL. 30, NO. 7, JULY 2008. [16] Shahbudin S;Hussain A;Tahir, N.M.; Samad, S.A.; Multi-class Support Vector Machine for human posture classification using a simplified shock graph International Symposium on Information Theory and Its Applications, 2008. ISITA 2008. Publication Year: 2008 , Page(s): 1 4. [17] Macrini, D.; Siddiqi, K.; Dickinson, S.; From skeletons to bone graphs: Medial abstraction for object recognition, IEEE Conference on Computer Vision and Pattern Recognition, 2008. CVPR 2008. Page(s): 1 8. [18] Gang Xu; Yuqing Lei;A new image recognition algorithm based on skeleton, IEEE International Joint Conference on Neural Networks, 2008. IJCNN 2008. (IEEE World Congress on Computational Intelligence). Publication Year: 2008 , Page(s): 777 782. [19] Zaboli, H.; Rahmati, M.; An Improved Shock Graph Approach for Shape Recognition and Retrieval, First Asia International Conference on Modelling & Simulation, 2007. AMS '07. Publication Year: 2007 , Page(s): 438 443. [20] Tahir, N.M.; Hussain, A.; Samad, S.A.; Shock Graph for Shape Recognition, 4th Student Conference on Research and Development, 2006. SCOReD 2006. Publication Year: 2006 , Page(s): 103 107. [21] Tahir, N.M.; Hussain, A.; Samad, S.A.; Husain, H.; Jamaluddin, M.Y.; Shock Posture for Shape Matching 2nd International Conference on Information and Communication Technologies, 2006. ICTTA '06. Volume: 1,Publication Year: 2006 , Page(s): 169 - 172 . [22] Van Eede, M.; Macrini, D.; Telea, A.; Sminchisescu, C.; Dickinson, S.; Canonical Skeletons for Shape Matching 18th International Conference on Pattern Recognition, 2006. ICPR 2006. Volume: 2 Publication Year: 2006 , Page(s): 64 69. [23] Giblin, P.J.; Kimia, B.B.; On the intrinsic reconstruction of shape from its symmetries IEEE Transactions on pattern analysis and machine intelligence, Volume: 25 , Issue: 7 Publication Year: 2003 , Page(s): 895 911. [24] Macrini, D.; Shokoufandeh, A.; Dickinson, S.; Siddiqi, K.; Zucker, S.; View-based 3-D object recognition using shock graphs 16th International Conference on Pattern Recognition, 2002. Proceedings. Volume: 3, Publication Year: 2002 , Page(s): 24 - 28 vol.3. [25] Aaron D. Ward and Ghassan Hamarneh, The Group wise Medial Axis Transform for Fuzzy Skeletonization and Pruning, IEEE Transactions on pattern analysis and machine intelligence, VOL. 32, NO. 6, JUNE 2010. [26] Tao Wang, Irene Cheng, Victor Lopez, Ernesto Bribiesca and Anup Basu, Valence Normalized Spatial Median for Skeletonization and Matching, IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops, 2009. [27] Behnam Hosseini, Saeed Vahabi Mashak, S.A.R. Abu-Bakar, Categorization of Human Movement Based on Rule-Based Classifier, Fourth Asia International Conference on Mathematical/Analytical Modelling and Computer Simulation, 2010. [28] H. Blum, Biological shape and visual science, J Thror. B iol, 38: pp 205-287. 1973. [29] E. Remya, E. Thielb, Exact medial axis with Euclidean distance. Image and Vision Computing 23(2005) pp 67-175,2005. [30] Alexandru Telea, Jarke J. van Wijk, An Augmented Fast Marching Method for Computing Skeletons and Centerlines, Joint Eurographics IEEE TCVG Symposium on Visualization(2002). [31] P. Dimitrov, C.Phillips, and K. Siddiqi, Robust and efficient skeletal graphs, IEEE Conference on Computer Vision and Pattern Recognition, Hilton Head, SC, June 2000. [32] Hingway S.P., K.M. Bhurchandi, An Improved Shock Graph for an Improved Object Recognition, Cyber Journals: Multidisciplinary Journals in Science and Technology, Journal of Selected Areas in Software Engineering (JSSE), May Edition, 2011 [33] Svetlana Stolpner, Sue Whitesides, Medial Axis Approximation with Bounded Error IEEE Sixth International Symposium on Voronoi Diagrams, 978-0-7695-3781-8/09, 2009 [34] Wei Zhao, Mei Xie, Jingjing Gao and Tao Li, A modified skull-stripping method based on Morphological Processing, IEEE Second International Conference on Computer Modeling and Simulation, 978-0-7695-39416/10, 2010.
425
Vol. 3, Issue 1, pp. 418-426
Authors
Pankaj M. Pandit is working as Associate Professor in Dept. of Electronics & Tele. Engg. of Jawaharlal Darda Institute of Engg. & Technology, Yavatmal (M.S.) He completed B.E. in Electronics & Telecomm. Engg. from Govt. College of Engg., Amravati in 1994 and then M.E. with specialization of Digital Electronics from P.R.M.I.T.& R., Badnera. He also completed M.B.A. with marketing as specialization from YCMOU, Nashik in 2007. Currently he is pursuing Ph.D. in Amravati University. He has 15 years of teaching experience and contributed 8 research publications at National and International Journals and Conferences. His areas of expertise and interest are Analog & Digital Electronics, Artificial Intelligence, Digital Image Processing, Machine Intelligence, Computer Vision, Entrepreneurship Development, Consumer Behavior, Lifestyle Marketing, and Consumer Psychology. He is Life Member of I.S.T.E. and IETE. He is also contributing as Executive Member of IETE Center, Amravati and ISF Faculty Advisor of the present Institution. Sudhir G. Akojwar, is working as Professor of Electronics and Tele-Communication Engineering and Principal at Dr Bhausaheb Nandurkar College of Engineering & Technology, Yavatmal(M.S.). He had 22 Years of teaching experience and published 27 research publications at National and International conferences and journals. He completed Ph.D. from VNIT, Nagpur. His area of research is WSN, Image processing, Neural Networks and Embedded Systems. He completed his B.E.in Instrumentation Engineering form Shri. Guru Gobindsingh College of Engg. and Tech., Nanded, in 1990. He joined as lecturer in the Department of Electronics Engg. at RCERT in 1990. He completed his M.Tech. in Electronic Instrumentation from REC, Warangal [presently NIT, Wrangal]. He was promoted as Asst. Prof. and then as Professor in the same college. He worked as Secretary and now as a Chairman of the local ISTE-Chapter. He worked as Chairman of the board of studies in Information Tech. at R.T.M. Nagpur university. He is member of IEEE and IEEE computer society. He is also member of ISTE, IETE, IE and BMESI.
426
Vol. 3, Issue 1, pp. 418-426
AN INTEGRATED COLOR AND HAND GESTURE RECOGNITION CONTROL FOR WIRELESS ROBOT
Deepali Suryawanshi and C. S. Khandelwal
Department of Electronics Engineering, J.N.E.C., Aurangabad, Maharashtra, India
A BSTRACT
Recently, the interaction between humans and robots has become an important issue for extending the use of robots. The purpose of this project is to show how a real-time human-robot interaction with hand gesture recognition which can be constructed combining color and shape cues. Hence, it is required that the following two major questions be resolved in this study. Firstly, either staying in indoor or outdoor lighting conditions, the robot must be able to detect and spot hand gestures with a specific color from data of robot vision system. Secondly, it is necessary that the robot also can interpret hand gestures performed by humans.
K EYW ORDS:
Hand Gesture, Gesture recognition, wireless robot
I.
INTRODUCTION
Several successful approaches to spatio-temporal signal processing such as speech recognition and hand gesture recognition have been proposed. A gesture is a form of non-verbal communication made with a part of the body and used instead of verbal communication (or in combination with it). Most people use gestures and body language in addition to words when they speak. A sign language is a language which uses gestures instead of sound to convey meaning combining hand-shapes, orientation and movement of the hands, arms or body, facial expressions and lip-patterns. Similar to automatic speech recognition (ASR), we focus in gesture recognition which can be later translated to a certain machine movement. Hand gestures can be classified into two classes: (1) Static hand gestures: This relies only the information about the angles of the lingers and (2) Dynamic hand gestures: This relies not only the fingers flex angles but also the hand trajectories and orientations. The goal of this project is to develop a program implementing real time gesture recognition. At any time, a user can exhibit his hand doing a specific gesture in front of a video camera linked to a computer. However, the user is not supposed to be exactly at the same place when showing his hand. The program has to collect pictures of this gesture thanks to the video camera, to analyze it and to identify the sign. It has to do it as fast as possible, given that real time processing is required. We propose a fast algorithm for automatically recognizing a limited set of gestures from hand images for a robot control application. Hand gesture recognition is a challenging problem in its general form. We consider a fixed set of manual commands and a reasonably structured environment, and develop a simple, yet effective, procedure for gesture recognition. This paper is consisting of the two important parts software development and hardware development. The software consists of the different phases of conversion of the different hand gestures in to the signals or commands for the robot and hardware consist of details of the transmitter and receiver of the project.
II.
RELATED WORK
Color-based target recognition is inherently difficult, due to variation in the apparent color of targets under varying imaging conditions. A number of factors might lead to the problem, namely, the color
427
Vol. 3, Issue 1, pp. 427-435
of incident daylight, surface reflectance properties of the target, illumination geometry, and viewing geometry [1]. There have been numerous studies in the literature dealing with varying color in highly constrained environments, such as color constancy algorithms [2], but are generally not applicable to outdoor images. Buluswar and Draper [3] developed color models for illumination and surface reflectance for use in outdoor machine vision. Stachowicz and Lemke [4] presented a color recognition method to compress the color information in an image into a small feature space that still contains enough information to separate one image from another. Liu et al. [5] provided an approach combining multiple color models to make color-based object recognition stable under varying imaging conditions. Another idea is to establish a supervised learning mechanism based on neural network for color measurement [6].This approach offers good performance and is more convenient to be implemented in human-robot interaction. The ability to quickly and correctly recognize hand gestures plays an important role in human-robot interaction applications. Numerous approaches have been applied to interpreting hand gestures. More recently, combining color space transformation and color thresholding method to classify each pixel in an image into one of a discrete number of color classes has increased noticeably [7]. This approach offers good performance and is more convenient to be implemented in human-robot interaction. The ability to quickly and correctly recognize hand gestures plays an important role in human-robot interaction applications. Numerous approaches have been applied to interpreting hand gestures. In literature [8]-[10], several hand geometry based systems are proposed for measuring a hand shape to extract its features, including lengths and widths of fingers, and hand contour, for recognition. Other methods are developed for improving recognition efficiency by machine learning and supervised training algorithms, including fuzzy neural network and modified Fourier descriptors (FDs) [11], [12]. More recent studies in gestures recognition have focused on support vector machines (SVM). These approaches have produced highly accurate systems capable of recognizing gestures [13], [14]. The main objective of this work was the developing of a control system for a robot, based on gesture recognition. A web camera will be attached to a computer which will acquire live feed of gestures done by the operator. The operators hand will have two different colors on it, one located on the thumb & the other on nuckel. The operator will give a gesture of thumbs up. Now when the thumb is pointing in upward direction, the command is move forward, when to left move left etc.
III.
SOFTWARE DEVELOPMENT
In this project, we propose a gestures recognition method to recognize a number of well-defined hand gestures representing a limited set of commands that the humans can give to the mobile robot 1. Firstly, by analyzing the distribution of the specific color in hue saturation intensity (HSV) color space, a color multithresholding method was developed for detecting hand gestures in video sequence under varying lighting conditions. Secondly, the feature extraction of hand gestures was performed by the detection algorithm and and by applying threshold value. Thirdly we can obtain the particular signal for robot movement subtracting the coordinates of the two colors. Finally, a real-time vision system on hand gesture recognition for a human-robot interaction is presented. The frames captured by the camera are in HSV format. We convert it to grayscale image. Then applying threshold value for one color the gray image is converted to binary image. In binary image the thresholded color turns white & the background becomes black. Now using moments we can find the co-ordinates of this blob. Same thing has to be done for the other color. Now we have the co-ordinates of the two colors. On subtraction of x & y co-ordinates we get the direction of the vector. Thus then we give command to the robot in this way. Flow chart for recognizing gesture & give commands to the robot
428
Vol. 3, Issue 1, pp. 427-435
Fig.1 Flowchart of gesture recognition
Algorithm for the program which converts the gestures to commands for the robot is as follows: 1) Capture a frame from camera. Its in BGR format. 2) Convert this BGR image to HSV OR RGB. (HSV is preferred since it is more noise free than RGB). 3) Convert this image to binary image by applying lower range & upper range of desired color (i.e. to make desired color white & rest black). 4) Find out the moment of each color on thumb and nuckel. 5) From the moments find out the coordinates of the each color viz X coordinate and Y coordinate 6) Subtract the coordinates to find out the direction of the movement of the robot.
3.1 BGR TO HSV Conversion

This module consists of two phases namely the image acquisition and image processing phases. The acquired images using a web cam for each hand gesture are in BGR format. We will convert it to HSV or RGB image. We prefer HSV since it is more noise free than RGB. The image is then subjected to an image enhancement process to adjust the intensity values in HSV image to increase the contrast of the output image. Figure 2 shows an acquired image of a sample hand gesture after enhancement where the captured green portion is marked by a green circle & yellow portion by yellow circle. This gesture represents a vector in forward direction.
Fig.2 : Original HSV image
429
Vol. 3, Issue 1, pp. 427-435
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 3.2 HSV TO Binary conversion (Thresholding)
Since we deal with the segmentation of color on thumb and nuckel color, a threshold value of the particular color level is chosen to convert the image to binary. The threshold value for the two colors must be different. This threshold value should be chosen carefully to avoid effect noise or any other color detection in the territory. Figure 3 shows the resulting black and white image called binary image after the thresholding process. We can convert this image to binary image by applying lower range & upper range of desired color (i.e. to make desired color white & rest black).
Fig.4 Binarized image showing the yellow marker. Here the original image is binarized with values to make yellow color highlight & background black.
We are going to use software opencv because opencv has a function to convert the HSV OR RGB image (step 2) into binary image directly. i.e. no need to convert image to grayscale & then to binary. The removal of these extra results in reduction in time complexity & space complexity.
3.3. FILTERING THE BINARY IMAGE

The binary image obtained as shown in fig.4 can be filtered with help of median smoothing algorithm. This filtering is required to remove the noise if any introduce by thresholding the grayscale image. Median smoothing algorithm is the simple algorithm used for removal of the noise from the signal or an image.
3.4 FINDING THE COORDINATES

Once we obtain the binary image from thresholding the colors we get two white blobs of two colors as shown in figure. By applying logic in physics we can find out the moment of the white blob. An Image moment is a number calculated using a certain formula. With help of the moment of the blob we can find out the coordinates of the color on the thumb with equation 1 and 2 X coordinate = Moment of X Total area of the white blob Y coordinate=Moment of Y Total area of the white blob (2) (1)
Once we obtain the coordinates (X1, Y1) for the color on thumb, we are able to find out the coordinates of the other color on the nuckel(X2, Y2). Fig5 shows the coordinates obtained for the green color and yellow colored ring in the fig1 for forward direction.
430
Vol. 3, Issue 1, pp. 427-435
Fig. 5: Original images captured co-ordinates.
3.5 Coordinate Subtraction

Find out which movement should be there (X, Y), we will need to subtract the coordinates of the two colors. X= X1- X2 , Y= Y1-Y2 After subtracting the coordinates for two colors we are able to find direction of movement of the robot. According to the decisions made in the program the robot will move. Same procedure is followed for all the robot movements.
IV.
HARDWARE ARCHITECTURE
The hardware architecture of this project is consist of two basic parts namely Transmitter unit, Receiver unit. The block diagram of these parts is as shown in figure 3.5 and figure 3.6 respectively. On the transmitter side it is consist of PC with gesture recognition software connected to web camera which in turn connected to RF transceiver through RS232 whereas the receiver unit is consist of blocks like RF transceiver module, PIC16F877 , L293 motor driver along with DC geared motor ,5V regulated power supply.
Fig.5 Block Diagram of Transmitter unit
431
Vol. 3, Issue 1, pp. 427-435
Fig.6 Block diagram of Receiver Unit
4.1 Microcontroller (PIC16F877)

PIC16F873A/876A devices are available only in 28-pin packages, while PIC16F874A/877A devices are available in 40-pin and 44-pin packages. All devices in the PIC16F87XA family share common architecture. Following are some of the properties of the PIC16F877. High-Performance RISC CPU Only 35 single-word instructions to learn All single-cycle instructions except for program branches, which are two-cycle Operating speed: DC 20 MHz clock input DC 200 ns instruction cycle Up to 8K x 14 words of Flash Program Memory, Up to 368 x 8 bytes of Data Memory (RAM), Up to 256 x 8 byte of EEPROM Data Memory.
4.2 Motor Driver L293

The L293 is an integrated circuit motor driver that can be used for simultaneous, bidirectional control of two small motors. The L293 is limited to 600 mA, but in reality can only handle much small currents unless you have done some serious heat sinking to keep the case temperature down. Unsure about whether the L293 will work with your motor? Hook up the circuit and run your motor while keeping your finger on the chip. If it gets too hot to touch, you can't use it with your motor. The L293 comes in a standard 16-pin, dual-in line integrated circuit package.
4.3 DC Geared Motor

Carbon brush of DC motor for the main role to play for, for referring to the rotating armature winding components from one slip to another slip, the brush in the process of short-circuit current generated by components within Change of direction. For the poor performance at run-time is the electrical spark happened. Level and sparks more than a certain limit will cause the carbon brush and commutator surface damage, with the result that should not continue to run the motor. When the motor for the situation well, sparks larger, an increase of the electrical brush wear. Especially when the commutator surface oxide film is damaged, the wear will be increased significantly.
4.4 RF Transceiver
A transceiver is a device comprising both a transmitter and a receiver which are combined and share common circuitry or a single housing. When no circuitry is common between transmit and receive functions, the device is a transmitter-receiver. The term originated in the early 1920s. Technically, transceivers must combine a significant amount of the transmitter and receiver handling circuitry. The RF Transceiver uses RF modules for high speed data transmission. The microelectronic in the
432
Vol. 3, Issue 1, pp. 427-435
digital-RF architecture work at speeds up to 100 GHz. The RF transceiver module which we are using is CC2500 RF module. It is a transceiver module which provides easy to use RF communication at 2.4 GHz. It can be used to transmit and receive data at multiple baud rates from any standard CMOS/TTL source. This module is a direct line in replacement for your serial communication it requires no extra hardware and no extra coding to turn your wired communication into wireless one. It works in Half Duplex mode i.e. it provides communication in both directions, but only one direction at same time (not simultaneously). This switching from receiver to transmitter mode is done automatically.
4.5 Web Camera

A webcam is a video camera which feeds its images in real time to a computer or computer network, often via USB, Ethernet or Wi-Fi. Their most popular use is the establishment of video links, permitting computers to act as videophones or videoconference stations. This common use as a video camera for the World Wide Web gave the webcam its name. Other popular uses include security surveillance and vision. Webcams are known for their low manufacturing cost and flexibility, making them the lowest cost form of video telephony.
4.6 RS232
In telecommunications, RS-232 (Recommended Standard 232) is the traditional name for a series of standards for serial binary single ended data and control signals connecting between a DTE (Data Terminal Equipment) and a DCE (Data Circuit-terminating Equipment). It is commonly used in computer serial ports. The standard defines the electrical characteristics and timing of signals, the meaning of signals, and the physical size and pin out of connectors. However, the limited transmission speed, relatively large voltage swing, and large standard connectors motivated development of the universal serial bus which has displaced RS-232 from most of its peripheral interface roles.
Fig.7 Detailed circuit Diagram
433
Vol. 3, Issue 1, pp. 427-435
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 4.7 Power Supply
This circuit is a small +5V power supply, which is useful when experimenting with digital electronics. Small inexpensive wall transformers with variable output voltage are available from any electronics shop and supermarket. Those transformers are easily available, but usually their voltage regulation is very poor, which makes then not very usable for digital circuit experimenter unless a better regulation can be achieved in some way.
V.
CONCLUSION
This project presented a gesture controlled wireless robot that does not require any special markers or gloves and can operate in real-time on a commodity PC with low-cost camera. Specifically, the system can track the coordinates of each color on the thumb and nuckel for each hand, assuming that a calibrated camera is viewing the hands from above with the palms facing downward. The motivation for this hand tracker was a desktop-based two-handed interaction system in which a user can select and manipulate 2D geometry in real-time using natural hand motions. The algorithm for gesture recognition is implemented in opencv and .net. This project presents a fast, robust and accurate method for hand gestures recognition under unconstrained scenes. The failure of the system to recognize the gesture is mainly due to the very changeable lighting conditions and moving objects (persons) entering the scene, operators failure to move the hand to the proper posture. It must be emphasize that after a short experience operators get used to the system. Future work will be focused on algorithm improvement, by using a combination of segmentation techniques and robot motion control by tracking the arm movement and its speed.
REFERENCES
[1] S. D. Buluswar and B. A. Draper, Learning object color: ATR in color images, 8th British Machine Vision Conference, Sept. 1997. [2] D. A. Forsyth, A novel approach to colour constancy, 2nd International Conference on Computer Vision, pp. 9-18, Dec. 1988. [3] S. D. Buluswar and B. A. Draper, Color models for outdoor machine vision, Computer Vision and Image Understanding, vol. 85, no. 2, pp.71-99, 2001. [4] M. S. Stachowicz and D. Lemke, Color recognition, Proc. of the 22nd Int. Conf. on Information Technology Interfaces, pp. 329-334, June 2000. [5] T. Liu, H. Guo, and Y. Wang, A new approach for color-based object recognition with fusion of color models, Congress on Image and Signal Processing, vol. 3, pp. 456-460, May 2008. [6] T. Lalanne and C. Lempereur, Color recognition with a camera: a supervised algorithm for classification, IEEE Southwest Symposium on Image Analysis and Interpretation, pp. 198- 204, Apr. 1998. [7] J. Bruce, T. Balch, and M. Veloso, Fast and Inexpensive color image segmentation for interactive robots, Proc. of the IEEE/RSJ Int. Conf. on Intellignet Robots and Systems, vol. 3, pp. 2061-2066, Japan, Oct. 2000. [8] X. Jiang, W. Xu, L. Sweeney, Y. Li, R. Gross, and D. Yurovsky, New directions in contact free hand recognition, IEEE Int. Conf. on Image Processing, vol. 2, pp. II 389-392, Sept. 2007. [9] S. Selvarajan, V. Palanisamy, and B. Mathivanan, Human identification and recognition system using more signification hand attributes, Int. Conf. on Computer and Communication Engi., pp. 1211-1216, 2008. [10] M. Bennewitz, T. Axenbeck, S. Behnke, and W. Burgard, Robust recognition of complex gestures for natural human-robot interaction, Proc. of RSS Workshop on Interactive Robot Learning, Zrich, 2008. [11] A. Licsr and T. Szirnyi, Supervised training based hand gesture recognition system, 16th Int. Conf. on Pattern Recognition, vol. III, pp.999-1002, Canada, Aug. 2002. [12] N. D. Binh and T. Ejima, Hand guesture recognition using fuzzy neural network, Int. Conf. on Graphics, Vision, and Image Proc., Egypt, 2005. [13] D. Y. Huang, W. C. Hu, and S. H. Chang, Vision-based hand gesture recognition using PCA +Gabor filters and SVM, IEEE Int. Conf. on Intelligent Info. Hiding and Multimedia Signal Proc., pp. 1-4, Sep. 2009. [14] Y. Ren and F. Zhang, Hand gesture recognition based on MEB-SVM, Int. Conf. on Embedded Software and Systems, pp. 344-349, May 2009. [15] A. Hanbury, Constructing cylindrical coordinate colour spaces, Pattern Recognition Letters, vol. 29, no. 4, pp. 494-500, 2008. [16] N. Otsu, A thresholding selection method from gray-level histogram, IEEE Trans. Syst. Man Cybernet., vol. 9, no. 1, pp. 62-66, 1979.
434
Vol. 3, Issue 1, pp. 427-435
[17] Y. H. Liu, Feature Analysis and Classifier Design and Their Applications to Pattern Recognition and Data Mining, Ph.D. Disseration, National Taiwan Univ., R.O.C., 2003. [18] R. L. Cosgriff, Identification of shape, Technical Report 802-11, Ohio State University, Columbus, OH, Dec. 1960. [19] C. T. Zahn and R. Z. Roskies, Fourier descriptors for plane closed curves, IEEE Trans. On Computers, vol. c-21, no. 3, pp. 269-281, 1972. [20] V. N. Vapnik, The Nature of Statistical Learning Theory, New York, Springer, 1995.
Authors Deepali Suryawanshi received B.E.(Electronics) degree in 2008 from Pune University Maharashtra, India. Doing M.E. (Electronics) from J.N.E.C., Aurangabad, Dr. B.M.A.U, Maharashtra, India. Doing a PG project on Gesture Controlled wireless Robot, Presented paper international conference at Mumbai. Also presented paper in national conference at Nasik. She is currently working as lecturer. She is fellow member of IEEE.
Chhaya S. Kandelwal completed her U.G. and P.G. in Electronics with first Class. She is having 2 years of industry experience and 10 years of teaching experience. She had presented 18 national and 4 international paper in conferences and published 1 national and international paper in journal. She is having a fellow membership of IETE and life membership of QCFI. She is currant working as Assistant professor in J.N.E.C. Aurangabad, Maharashtra, India since 2006. She had guided almost 10 master level Projects.
435
Vol. 3, Issue 1, pp. 427-435
CONVENTIONAL ETHANOL REFORMING TECHNOLOGY DEVELOPMENTS FOR THE PRODUCTION OF HYDROGEN

1
Prashant Tayade1, Vilas Sapkal2, Chandrasekhar Rode3, Rajendra Sapkal4

Chem. Engg. Dept., Jawaharlal Darda Institute of Engg. & Tech., Yavatmal, M.S., India 2 Rashtrasant Tukadoji Maharaj Nagpur University, Nagpur, M.S., India 3 Chemical Engg. Division, National Chemical Laboratory, Pune, M.S., India 4 U.D.C.T., Sant Gadge Baba Amravati University, Amravati, M.S., India
ABSTRACT
In the unsafe horizon of pollution, global warming, and energy crises, all resulted owing to the fossil fuels, hydrogen fuel obtained by steam reforming of bio-derived ethanol (bioethanol), can be a strong candidate as a clean energy carrier based on the renewable sources of energy. In this paper, the key issues related to the steam reforming of bioethanol for the production hydrogen are discussed in the view of potential application of hydrogen for polymer electrolyte membrane (PEM) fuel cell. The need for the development of catalyst for CO free production of hydrogen is emphasized to obtain a commercially viable bioethanol steam reforming system; also the commercial feasibility of hydrogen production is discussed. The proper distinguish has been done in the existing conventional reforming technologies and the new developed technologies along with catalyst developments. The commercial feasibility of reforming of ethanol or bioderived liquids (Bio-ethanol) is criticized by taking consideration of existing technologies of power production.
KEYWORDS: Hydrogen Production, Bioethanol, Steam Reforming, Reforming Catalyst, Membrane Reactor
I.
INTRODUCTION
Modern society is profoundly tied to the fossil fuel energy and its sources which powers the society. From the commencement of the Industrial Revolution, fossil fuels have supplied this energy, enabled a modern and mobile society. But, now fossil fuel energy and its sources, evolving dangerous situations like environmental pollution, global warming and energy crises etc. To avoid this dangerous situation, it is inevitable to think on the alternative clean energy and its resources like hydrogen. Hydrogen is the potential source of clean energy and possesses the highest energy content per unit weight (i.e. 120 KJ/g), as compare to other fuels. The benefits of hydrogen as a fuel are (1) Reduce dependence on fossil fuels (2) Environmentally benign (3) Reduce air pollution (4) Renewable in nature (5) Obtained from diverse sources (6) Carbonless Fuel (7) High energy content on weight basis (8) Higher octane rating. By keeping the view of global cycle of clean and sustainable production of energy, new eco-friendly reservoirs of hydrogen are needed and this can be fulfill by the ethanol or bio-derived ethanol (bioethanol). Hydrogen can be efficiently produced from bioethanol, represented by following simplest endothermic steam reforming stoichiometry equation. CH3CH2OH + 3H2O 6H2 + 2CO2 H0 = + 174 KJ mol-1 (1) Although, most of the used technologies to produce hydrogen are based on natural gas steam reforming, the hydrogen production from biomass (bioliquids) is a promising technology for the future, because it appears as the more attractive alternative due to its renewable characteristics (it is a part of CO2 neutral cycle), high yield and low CO2 emissions [1 2 3]. Ethanol is the most heavily researched renewable liquid because it is easy to store, handle, and transport due to its low toxicity and volatility. The ethanol production and transportation infrastructure already exists [4 5 6], and its forecasted worldwide production is given in the Fig.1 (Source: Market Research Analyst). Therefore reforming of ethanol (bioethanol) for the production of hydrogen appears to be the most viable mid-
436
Vol. 3, Issue 1, pp. 436-450
term approach than other reforming fuels which gives hydrogen. Other renewable reforming fuel (bioliquids) options include sorbitol, glucose, glycerol, bio-oil, methanol, propylene glycol, and less refined sugar streams (cellulose, hemicelluloses). Table 1 list the potential reforming bio-liquid feed stocks, its theoretical hydrogen yield, and advantages and disadvantages.
A A
8000
Millions of Gallons of Ethanol
6000
A B C D E B F G H
U SA Brazil C hina India France Spain Germ any C anada
4000
2000
C D
0 2010
C E F GH D E FG H
C D E FGH
2011
2012
Year
Fig.1 World Ethanol Production Forecast (Source: Market Research Analyst)
Now a days membrane technology has created the renewed interest in the development of small-scale hydrogen production technology for the reforming of ethanol or bio-derived ethanol (bioethanol) by introduction of new conceptual catalytic membrane reactor (CMR) (see Fig. 2) and has given some extent of economical breakthrough. The conventional reforming processes for the production of 99 % pure H2 involves the use of reforming reactor followed by CO and CO2 separation units shown in Fig. 3, i.e. (1) Vaporizer to vaporize inlet liquid fuel and water (2) Reformer to carry out the reforming reaction in presence of catalyst (3) High Temperature Shift Reactor (HTS) to convert the dangerous CO into CO2 and H2 in presence of catalyst (4) Low Temperature Shift Reactor (LTS) for the complete conversion of CO to CO2 in presence of catalyst (5) if any stresses of CO remains in the product stream, can be removed with Preferential Oxidation (PrOX) of CO into CO2. (6) Pressure Sewing Adsorption of CO2. These processes are usually complex, energy intensive and costly [6]. By using catalytic membrane reactor (Fig. 2), it is possible to reduce above mentioned number of unit operations in conventional reforming processes (i.e. HTS, LTS, PrOX, Pressure sewing adsorption), as shown in Fig. 3[11]. The catalysts are the center point for the research in the field of ethanol reforming technologies, over the period of last two decades, there has been a continuous improvement in the ethanol reforming catalyst and sustained a crucial role in the production of hydrogen. Since, reforming of ethanol is accompanied by many by-products (CO, CO2, CH4, Aldehydes); the catalyst has to optimize the hydrogen production and to discourage the by-product formation. The selection of proper support for catalyst and the methods of catalyst preparation significantly affect the overall performance of catalysts. The MgO, ZnO, CeO2, and La2O3 are suitable supports for Rh and Ni due to their basic characteristics, while oxide supports plays a crucial role in the performance of catalyst. The development of bimetallic catalysts, and alloy catalysts is to be a promising answer to enhance hydrogen production and long-term catalyst stability [10]. But, still all these catalyst gives some amount of CO in the reforming reaction which is the main component responsible to increase the number of steps (HTS, LTS and PrOX) for the production of hydrogen in conventional reforming process (Fig. 3). Further research in catalyst development has been done to eliminate CO in the reforming reaction. The deployment of ZnO supported cobalt catalyst with metallic precursor was done in the view of elimination of CO in the reforming reaction and Narcis Homes and Jordi Llorca got the success to eliminate CO in the reforming reaction [62, 50, 52, 53]
437
Vol. 3, Issue 1, pp. 436-450
Table 1: Potential Bio-Liquid Reforming Pathway (Source: U.S. Department of Energy, HFCIT Program Bio-Liquid Ethanol Theoretical H2 Yield 0.78 kg H2/ gallon ethanol 0.26 kg H2/kg ethanol 0.64 kg H2/gallon Methanol 0.22 kg H2/kg methanol 0.13 kg H2/kg Glucose 0.15 kg H2/kg pure glycerol 13.8 g H2/100 g Bio-oil, 0.06 kg H 2/ lb bio-oil 0.13 kg H2/kg Sorbitol 0.13 kg H2/kg Cellulose hemicelluloses Feedstock Advantages/ Disadvantages Low toxicity, Low sulfur content, Use of dilute ethanol would reduce reforming feedstock costs More easily reformed to hydrogen than ethanol. High toxicity. Higher corrosively, volatility than ethanol Low volatility. Non-toxic, non-flammable Utilizes low-value glycerol by-product from biodiesel production, Low volatility, Non-toxic, non-flammable Low volatility. High reactivity; potential of forming carbonaceous deposits or converting to aromatics that are more difficult to reform to H2 Low volatility. Non-toxic, non-flammable Low volatility. Non-toxic, non-flammable
Methanol Glucose Glycerol Bio-oil
Sorbitol Cellulose/ Hemicelluloses
Conventional hydrogen gas production methods are all energy intensive processes requiring high temperature (> 850 0C). The rigorous research in last decade has been given the improved technologies i.e. membrane processes, selective oxidation of methane and oxidative dehydrogenation [7, 8]. Currently methanol-water mixture reforming is the most common technology for hydrogen production on-board. However methanol as a fuel has some disadvantages connected with its high toxicity, low availability on ordinary basis and coal-natural gas based production technology, which leads to additional CO2 emissions [9].
Figure 2. Catalytic Membrane Reactor Concept: Recent Development in Technology for Production of Hydrogen
The steam reforming of ethanol (bioethanol) can be carried out in different type of reactors mainly it is in fixed bed reactor, fluidized bed reactor and recently number of authors putting their efforts for studying this reaction in catalytic membrane reactor [12, 13, 14].
II.
CONVENTIONAL REFORMING PROCESSES
Hydrogen market is large and growing rapidly, 95% of hydrogen is made from natural gas via steam methane reforming (SMR) [15]. The ethanol steam reforming process reactions are represented by
438
Vol. 3, Issue 1, pp. 436-450
following reactions, other reaction that can occur in the process depending on the type catalyst used [16]. C2H5OH + H2O CH4 + CO2 + 2H2 H0298 = -8.0 KJ/mol (2) CH4 + H2O CO + 3H2 H0298 = 206.2 KJ/mol (3) CO2 + H2 H0298 = -41.2 KJ/mol (4) CO + H2O Methods of hydrogen production can be broadly classified as (a) Thermo-chemical, (b) Electrolysis of water, (c) Photochemical and (d) Biological.
Figure 3. Conventional Reforming Process for the Production of Hydrogen
2.1
Thermo-chemical processes
2.1.1 Steam Reforming Processes (SR) Steam reforming refers to the reaction between steam and a hydrocarbon mainly methane, ethanol and natural gas, typically producing hydrogen in the process. Use of natural gas or naphtha as a raw material is an industrial process. Typical reactions are highly endothermic in nature. Depending on the purity of hydrogen required, WGS and PrOX reactors are also appended to this process. The industrial catalyst used for this reaction is Ni/- Al2O3. While the steam/carbon ratio required for this reaction is 2, generally 2.5-3.0 molar ratios are used to avoid coke formation. Since this is an endothermic and kinetically controlled reaction, temperatures in the range of 1073-1173 K are used to obtain desirable reaction rates. Main disadvantages of this process include the formation of CO2, a global warming gas, as a byproduct, and the use of non-renewable natural gas feedstock. 2.1.2 Partial Oxidation (PO) Partial oxidation is another industrially used process to produce H2 or syngas from hydrocarbons, with following stoichiometry equation: CxHy + 0.5x O2 xCO + 0.5y H2 (5) PO compares better than SR, as it is not only non-catalytic but also does not require external thermal energy input. The process requires temperatures of 1400-1700 K and pressures ranging from 30-100 bars. Pure oxygen is used instead of air to avoid the N2 separation step. Catalyst may be used to appreciably reduce the reaction temperature. The primary disadvantage of this process as compared to SR is low hydrogen yield. 2.1.3 Auto-Thermal Reforming (ATR) Combining the features of SR and PO, ATR has been developed, utilizing both steam and oxygen along with the feedstock [17]. The oxygen to fuel ratio is sub-stoichiometry and can be altered to adjust the process temperature. The hydrocarbon feed stock first reacts with oxygen and the released heat is used for steam reforming reaction. This takes place on reforming catalyst placed after the burner in the reactor. The autothermal reforming of methane process was designed to save energy, because the thermal energy required is generated in the partial oxidation of methane. As this process consumes the thermal energy that it produces, it is called autothermal [18]. In this process the hydrogen yield is more than that in PO but less than that in SR. Moreover, this process does not require any external heat to be supplied.
439
Vol. 3, Issue 1, pp. 436-450
2.1.4 Biomass Gasification and Pyrolysis Hydrocarbons, when heated to ~1600C, are converted to pure carbon and H2, with no other significant side products. This, therefore, is quite a viable process for H2 production; of course, till fossil fuel based feedstock is economically available. In biomass conversion processes, a hydrogencontaining gas is normally produced in a manner similar to the gasification of coal. However, no commercial plants exist to produce hydrogen from biomass. Currently, the pathways followed are steam gasification (direct or indirect), entrained flow gasification, and more advanced concepts such as gasification in supercritical water, application of thermo-chemical cycles, or the conversion of intermediates (e.g. ethanol, bio-oil or torrified wood). None of the concepts have reached a demonstration phase for hydrogen production. [19]
2.2
Electrolysis
Electrolysis of water, i.e. applying electricity to split it, is the process by which almost 100% pure hydrogen can be obtained. This is the next commonly used industrial method, after steam reforming of methane. At present, conventional alkaline electrolyte based electrolysis is economical, only if cheap electricity is available. Research is primarily focused on the feasibility of using solid oxide steam electrolyzer, PEM electrolyzer etc. [20, 21, 22, 23].
2.3
Photochemical decomposition
Photo-catalytic splitting of water, using nano-sized TiO2 catalyst, is a potentially feasible process for production of hydrogen from water. Very low process efficiency is the main hurdle. This low efficiency is attributed to (i) rapid recombination of photo generated electron/hole pair and, (ii) backward reaction. Some sacrificial reagents such as 2- 2- 3 S /SO Ce4+/Ce3+ [24, 5, 25] and IO3 -/I[26] and electron donors such as methanol [27] are under study for enhancing production. Catalyst modification techniques, such as noble metal loading [28, 29], ion doping [30], dye sensitization [31] and metal ion implantation are also being studied [32].
2.4
Bio-hydrogen
Biological hydrogen production is carried out by several methods including direct and indirect biophotolysis, photo-fermentation and dark-fermentation, the latter being the most efficient [33]. Dark fermentation involves the use of anaerobic bacteria (Clostridium pasteurianum, C butyricum, and C. beijerinkii e) grown in the dark on carbohydrate rich substrates to produce H2 and CO2 and traces of CH4, CO, and H2S. By using Rhodospirillum rubrum, H2 can be produced from WGS reaction when only CO is available. The main disadvantage of this method is the very slow rate of production.
III.
CURRENT STATUS OF REFORMING PROCESS
The reforming process is a thermodynamically efficient way to produce high purity hydrogen from bioethanol and studies has shown that, this process is feasible for temperature higher than 230 C, being methane, carbon oxides and H2 the main products [34, 35]. From 1996 onwards, lot of research and modifications of this environmentally benign process has been done around the World. Some of technologies to produce hydrogen, their H2 yield, operating temperature, and production cost/Kg of H2 and energy efficiency are given in Table 2, and it clears that ethanol steam reforming and methane steam reforming technology are the most economical hydrogen production technologies till date. Also, these technologies are compared on the basis of Feedstock and conversion efficiencies given in Table 3. The production of hydrogen from natural gas is an integral part of the strategy to introduce hydrogen into the transportation and utility energy sectors, by reducing the cost of conventional and developing innovative hydrogen production processes that rely on cheap fossil feedstocks.
Table 2: Hydrogen Production Technologies Technologies to Produce Hydrogen Ethanol Steam Reforming methane steam H2 Yield Operating Temperature cost ($/kg H2)a 2.69-4.27 3.94 energy Efficiency (%) 67 70
85% 5 (H2/CO Ratio)
573-723 7001000 C
440
Vol. 3, Issue 1, pp. 436-450
reforming coal gasification biomass gasification water electrolysis 39-42% 62-100 0.09 Kg H2/Kg H 2O 1100-1600 100-500 Solid oxide electrolyzers 500-800 0C PEM electrolyzers 80100C Alkaline electrolyzers 100C150C 2.17 7.07 6.82 60 45-50 56-73
Modification of the conventional steam methane reforming (SMR) process to incorporate an adsorbent in the reformer to remove CO2 from the product stream may offer a number of advantages over conventional processes. Upsetting the reaction equilibrium in this way drives the reaction to produce additional hydrogen at lower temperatures than conventional SMR reactors. Although still in the research stage, the cost of hydrogen from this modified process is expected to be 25%-30% lower, primarily because of reduced capital and operating costs [36].
Table 3: Comparison for Hydrogen production technologies in terms of Feedstock and conversion Efficiencies [37] Process Feedstock Conversion References efficiencies Steam reforming Aqueous reforming Electrolysis Partial oxidation Biomass gasification Methane, glycerol, alcohols, polyols, sugars, organic acids Glycerol, alcohols, polyols, sugars, organic acids H2O + electricity Hydrocarbons ( Glycerol, alcohols, polyols, sugars, organic acids) Biomass ( Lignocellulosic biomass) 7085% 35100% 5060% 6075% 3550% [38, 39] [40, 41, 42] [43] [38] [44]
Fuel Cells and Infrastructure Technologies Program (HFCIT), U. S. department of energy is currently supporting several distributed bio-liquid reforming research efforts [45]. These technologies target a range of bio-liquid feedstock, including ethanol, glucose, sorbitol, glycerol, and bio-oil. In the near term, distributed hydrogen production technologies may be the most viable renewable hydrogen pathway due to their lower capital investment requirements (including hydrogen transport and delivery infrastructure). In India the production of hydrogen using reforming as energy source is in the initial stages, but is used for the other purposes like fertilizers, refineries and chlor-alkalies as shown in Fig. 4.
Fig. 4 H2 Consumption worldwide and in India
441
Vol. 3, Issue 1, pp. 436-450
Hydrogen purity is another important issue that has to be discussed, as the carbon monoxide (CO) must be less than 0.2 ppm in the hydrogen supplied to the fuel cell to avoid the degradation of catalyst in the fuel cell. Pressure swing adsorption (PSA), membrane separation, cryogenic distillation, or methanation can be used to lower the CO content of the hydrogen product [16, 46, 47] which leads to the number of unit operations (HTS, LTS, PrOX) in the production of hydrogen process except membrane separation. The characteristics of these technologies are given in Table 4 [46]: Of the four technologies, membrane technology is the recent concept (see Fig. 3) in the production of hydrogen by steam reforming of ethanol and also, methane, having the greatest potential for hydrogen purification using membrane rather than CO purification steps (HTS, LTS, PrOX) which minimizes the capital investment. Catalysts for reforming and water-gas shift reactions could be incorporated into the membrane so that the reversible reforming/water-gas shift reactions shift to the right, favoring hydrogen production. This would enable the reforming and separation to be performed in one unit, reducing capital costs. Although the use of both inorganic (e.g., metal, ceramics) and organic (polymers) membranes has been investigated, much of current research focuses on metal and ceramic membrane technologies due to their ability to tolerate harsher conditions than organic membranes [46].
Table 4: Separation Technologies used in the production of H2 Separation Technology Availability Hydrogen Purity Achieved Pressure Suing Adsorption Commercially available up to 99.99% Membrane Separation Commercially available up to 99.99% Cryogenic Distillation Commercially available up to 95 %
Followings are some points which highlight the scope for the developments in the reforming system, 1) what are the pros and cons of fuel flexible reformers? 2) What are the impacts of bio-derived liquid feedstock impurities on pressure swing adsorption and hydrogen membranes? 3) What is the optimal Water-Gas-Shift technology for bioderived liquid reforming? 4) What are the gaps in catalyst development for bio-derived liquid reforming? 5) Are lower system efficiencies for aqueous phase reformers (compared to distributed natural gas reformers) acceptable given lower capital costs?
IV.
POTENTIAL CATALYST USED FOR REFORMING OF BIOETHANOL
The use of catalysts plays a crucial role in hydrogen production through ethanol reforming. Rh and Ni are so far the best and the most commonly used catalysts for ethanol steam reforming towards hydrogen production. In terms of hydrogen production and long-term stability, MgO, ZnO, CeO2, and La2O3 are suitable supports for Rh and Ni due to their basic characteristics, which favor ethanol dehydrogenation but inhibit dehydration. As Rh and Ni are inactive for water gas shift reaction (WGSR), the development of bimetallic catalysts, alloy catalysts, and double-bed reactors is promising to enhance hydrogen production and long-term catalyst stability. The steam reforming of ethanol over Ni, Co, Ni/Cu and noble metals (Pd, Pt, Rh) has been extensively studied. The greatest concern lies in developing an active catalyst that inhibits coke formation and CO production [48]. The catalyst performance characteristics suggest strong metalsupport interaction. The reaction pathway is complex and a number of undesirable side reactions occur thereby affecting the selectivity to H2. Catalyst coking is mainly due to the formation of ethylene by ethanol dehydration. The performance of various catalyst reported in the literature is summarized in Table 5.
Table 5: Summary of different catalyst used for the steam reforming of bioethanol Catalyst 0.06 NaCoZn 0.23 NaCoZn 0.78NaCoZn 0.98 NaCoZn Co(N)/ZnOred Co(CO)ZnO T (K) 723 723 673 723 623 723 EtOH Conv (%) H2 100 100 100 100 100 100 73.1 73.7 74.2 74.4 3.2 72.6 % Selectivity CO2 CO CH4 24.2 24.9 25.0 24.9 25.0 25.0 0.7 0.1 1.6 1.2 0.9 0.7 1.8 2.4 Ref [49] [49] [49] [49] [50] [50]
442
Vol. 3, Issue 1, pp. 436-450
Ni/Zr Co/Zr Cu/Zr 10Ni/ZnO 10Cu/ZnO Ce2Zr1.5Co0.5O8- Rh/SiO2 Rh/CeO2 Ni/CeO2 Ni-Rh/CeO2 Cu/Ni/K/-Al2O3 973 973 973 723 723 713 873 668 668 668 573 100 100 100 100 84.1 100 100 100 100 100 90 75.0 74.0 45.0 66.5 58.6 70.0 70.0 38.8 29.7 34.5 18.0 18.5 2.5 23.8 9.0 24.0 15.0 62.3 56.0 55.0 5.5 3.5 15.0 1.3 4.0 5.0 8.2 4.9 8.6 1.5 4.0 13.0 8.4 1.2 2.0 5.0 29.5 39.0 33.5 [51] [51] [51] [52] [52] [53] [54] [55] [55] [55] [56]
The focus of current research has shifted towards finding a transition metal alternative that can perform as well as the noble metal catalysts. Cavallaro and coworkers [57] developed a Co/MgO catalyst that showed comparable catalytic performance to Rh/Al2O3. It was determined that the acidic nature of the Al2O3 support caused an increase in coke formation on the Co/Al2O3 catalyst. Studies by Haga and coworkers [48, 58] examined cobalt catalysts on various supports. In contrast to the findings of Cavallaro, Haga determined that Co/Al2O3 had a greater activity and selectivity than Co/MgO in the steam reforming reaction. Recently, Haryanto and coworkers [16] have reviewed steam reforming of ethanol, examined the various catalysts reported till date and presented a comparative analysis. They concluded that the ethanol conversion and H2 production varies greatly with the reaction conditions, the type of catalyst and the method of catalyst preparation. The importance of process engineering related aspects is evident and these need to be discussed at length. This article is aimed at fulfilling this need. It reviews the available literature on catalytic steam reforming of ethanol. The role of the catalyst composition and the process conditions in determining product distribution is elucidated. The possible reaction pathways and the kinetic and thermodynamic considerations have also been discussed. The coupling of ethanol steam reforming with selective removal of CO2 by chemisorptions to produce high-purity H2 at low temperatures has been discussed. It is expected that this will provide an insight into steam reforming of ethanol. Haga, Sahoo and coworkers [58, 59] studied the catalytic properties of Co among other metals and found that selectivity to H2 was in the order Co > Ni > Rh > Pt, Ru, Cu. Metals alone may not assist hydrogen production significantly. So performance of metal catalysts could be improved using supports. The nature of the support also plays a key role in determining the selectivity to H2. Choice of the support is hence crucial. Al2O3 is commonly used as a support in the steam reforming reaction. However, it is acidic and promotes dehydration of ethanol to ethylene, which in turn polymerizes to form coke on the catalyst surface. In contrast, MgO is basic. ZnO also has basic characteristics. CeO2 is also basic and has redox properties [59, 60, 61]. Upon reviewing existing studies over the various reforming catalysts it is evident that the role of the support is influential in the overall performance of a reforming catalyst. Llorca and coworkers [62] investigated the performance of 1% cobalt reforming catalysts using various supports. MgO, g-Al2O3, SiO2, TiO2, V2O5, ZnO, La2O3, CeO2, and SmO2 prepared by impregnation. The catalysts were tested for the steam reforming of ethanol with water to ethanol ratio of 13 at a reaction temperature of 500C and a gas hour space velocity of 5000 h-1. The group determined that Co/ZnO was selective only for H2 and CO2 while the other oxygenated supports produced unwanted byproducts such as acetaldehyde and dimethyl ether [62, 50]. In a study by Aupretre [63] the influence of metal used as well as the role of the support was examined. The catalytic activity for the steam reforming catalysts increased with increasing hydroxyl mobility but decreased for catalysts that promoted the water gas shift reaction.
V.
CHALLENGES FOR REFORMING AND SEPARATION
The ability of ethanol or bio-liquid (bioethanol) reforming to meet the demand for clean, costcompetitive renewable fuel will require advances in catalyst systems, purification technologies, and integration of these technologies into a single process. A multitude of catalyst systems have been
443
Vol. 3, Issue 1, pp. 436-450
investigated for the steam reforming of ethanol, bio-oil, sugar alcohols, and other bio-liquids. A common problem with the catalysts reviewed is deactivation due to coking which occurs when side reaction products (e.g., acetaldehyde, ethylene) deposit on the catalyst [64, 65, 66, 67]. To a certain degree, process parameters such as the steam-to-carbon (water-to-ethanol) ratio can be modified, again if it increased to limit the carbon deposits, the cost of greater process energy will require. The water-to-ethanol feed ratio and operating temperature also influences the selectivity to hydrogen. Water to ethanol molar ratios of three or greater, and temperatures above 500C have been shown to favor the production of hydrogen over methane and other reaction intermediates [65, 67, 68]. However, higher water to ethanol ratios will require increased energy inputs. Low-temperature (<500C) reforming technologies are also under investigation. The advantages of low-temperature technologies are reduced energy intensity, compatibility with membrane separation, favorable conditions for water-gas shift reaction, and minimization of undesirable decomposition reactions typically encountered when carbohydrates are heated to high temperatures [2, 54]. Aqueousphase reforming is a promising technology that has been applied to glucose, ethylene glycol, sorbitol, glycerol, and methanol. Studies have shown that the following factors promote selectivity to hydrogen rather than alkanes [65, 51] 1) Catalysts made of platinum, palladium, and nickel-tin (nickel catalysts favor alkane production). 2) More basic catalyst support materials (e.g., alumina). 3) Neutral and basic aqueous. 4) Feedstock type (in descending order of hydrogen selectivity) polyols (selectivity decreases with increasing carbon number), glucose (selectivity decreases as weight % increases from 1 to 10). Catalyst coking and deactivation are not significant problems as they are in steam reforming, but that may be the result of differences in feedstock reaction pathways. While hydrogen yields are highest from the aqueous phase reforming of sorbitol, glycerol, and ethylene glycol, glucose reforming which has low hydrogen yields may be more practical. Improvements in catalyst performance, reactor design, and reaction conditions may help increase hydrogen selectivity. Low-temperature gas phase reforming of ethanol is also being investigated, but there is a tradeoff between catalyst activity and resistance to deactivation (due to coking) and research is ongoing. Hydrogen purity is critical as the carbon monoxide (CO) must be less than 0.2 ppm to avoid fuel cell degradation [67]. Pressure swing adsorption (PSA), membrane separation, cryogenic distillation, or methanation can be used to lower the CO content of the hydrogen product. The characteristics of these technologies are as follows [10, 68]: PSA: commercially available, able to produce high-purity hydrogen (up to 99.99%); DOE-funded research has achieved the DOE 2005 target for hydrogen purification/separation efficiency of 82%.
VI.
HYDROGEN ECONOMY
The transition of the use of hydrogen in industrial markets to a broader hydrogen economy would occur, if devices that use hydrogen (e.g., fuel cells) must compete successfully with devices that use competing fuels (e.g., hybrid propulsion systems). Equally important, hydrogen must compete successfully with electricity and secondary fuels (e.g., gasoline, diesel fuel, and methanol), at the same time to make the hydrogen commercial feasible, requires a market and a mechanically viable technology. The world economy currently consumes about 42 million tons of hydrogen per year. About 60 percent of this becomes feedstock for ammonia production and subsequent use in fertilizer. Petroleum refining consumes another 23 percent, chiefly to remove sulfur and to upgrade the heavier fractions into more valuable products. Another 9 percent is used to manufacture methanol, and the remainder goes for chemical, metallurgical, and space purposes (Fig. 5). The United States produces about 9 million tons of hydrogen per year, 7.5 million tons of which are consumed at the place of manufacture. The remaining 1.5 million tons are considered merchant hydrogen. Followings are the key challenges to pave the way for commercialization of fuel cell and hydrogen infrastructure technologies:
6.1 Fuel Cell Cost and Durability

Fuel cells are efficient; they convert hydrogen and oxygen directly into electricity and water, with no combustion in the process. The resulting efficiency is between 50 and 60% about double that of an internal combustion engine [69].
444
Vol. 3, Issue 1, pp. 436-450
Hydrogen-powered fuel cells are not only pollution free, but also can be more than two times more efficient than internal combustion technologies. A conventional combustion-based power plant typically generates electricity at efficiencies of 33 to 35 percent, while fuel cell systems can generate electricity at efficiencies of up to 60 percent (and even higher with cogeneration) [70]. The gasoline engine in a conventional car is less than 20% efficient in converting the chemical energy in gasoline into power that moves the vehicle in normal driving conditions. Hydrogen fuel cell vehicles, which use electric motors, are much more energy efficient and use 40-60 percent of the fuels energy corresponding to more than a 50% reduction in fuel consumption, compared to a conventional vehicle with a gasoline internal combustion engine. Long-term operation of the PEMFC has been proven, but mostly for fuel cells containing high loadings of noble metals, and under relatively ideal conditions. However, especially for automotive application, PEMFCs need to operate in a wide range of conditions, which may negatively affect durability and lifetime. The issues of cost and durability are closely related since reduction of cost is associated with the use of less or cheaper materials, e.g. reduction of Pt content or use of thinner perfluorinated, partially fluorinated or nonfluorinated membranes [71]. At present there are many uncertainties to the success of fuel cells and the development of a hydrogen economy: Fuel cells must obtain mass-market acceptance to succeed. This acceptance depends largely on price, reliability, longevity of fuel cells and the accessibility and cost of fuel. Compared to the price of present day alternatives e.g. diesel-engine generators and batteries, fuel cells are comparatively expensive. In order to be competitive, fuel cells need to be mass produced less expensive materials developed.
Worldwide Hydrogen Consumption
23%
9%
8% 60%
For ammonia production in fertilizer industry For petroleum refining For manufacture of methanol For chemical, metallurgical, and space purposes
Fig. 5 Worldwide H2 consumption in different industries
6.2 Hydrogen Storage

Today, hydrogen for transportation applications is compressed and stored in high-pressure metal and composite storage tanks. Hydrogen is also stored by cooling it to its liquid form and containing it in super-insu lated tanks. Table 6 shows the most probable (state of the art) H2-storage methods today [72]. Hydrogen is a molecular gas. At room temperature and atmospheric pressure, 4 kg of hydrogen occupies a volume of 45 m3. Comparative properties of hydrogen are given in Table 7.
Table 6: Physical and chemical properties of hydrogen, methane and petrol Properties Hydrogen (H2) Methane (CH4) Petrol (CH2) Lower heating value (kWh kg1) 33.33 13.9 12.4 Self-ignition temperature (7C) 585 540 228501 Flame temperature (7C) 2,045 1875 2,200 Ignition limits in air (Vol %) 475 5.315 1.07.6 Minimal ignition energy (mW s) 0.02 0.29 0.24 Flame propagation in air (m s1) 2.65 0.4 0.4
445
Vol. 3, Issue 1, pp. 436-450
Diffusion coefficient in air (cm2 s) Toxicity 0.61 No 0.16 No 0.05 High
Table 7: State of the art technology and estimates for system gas, cryogenic liquid, and hydride storage of 3 kg H2 [72] Technology Volume [liters] 35 MPa (350 bar) compressed H2 145 70 MPa (700 bar) compressed H2 100 Cryogenic liquid H2 90 Low-temperature metal hydride 55
volumes and weights for vehicular compressed Weight [kg] 45 50 40 215 Density [Wt. % H2] 6.7 6.0 7.5 1.4
High-pressure tanks can be used to store hydrogen, but present a considerable risk the compression itself is the most dangerous and complicated part. In Japan such vessels are prohibited on the roads in ordinary cars [73]. The hydrogen storing pressure ranges for compressed tanks remain inadequate and the energy consumed to compress the hydrogen reduces the efficiency of this storage media. Current research on storage technology for hydrogen includes: Metal hydride technology, Chemical hydride slurries or solutions and Carbon nanotubes. The principal challenges to improving hydrogen storage technologies relate to increasing their efficiency, size, weight, capacity and, ultimately, their cost. The cost of on-board hydrogen storage systems is currently too high, particularly in comparison with conventional storage systems for petroleum fuels. Low-cost materials and components for hydrogen storage systems are needed, as well as low-cost, high-volume manufacturing methods.
6.3 Hydrogen Delivery

Hydrogen is currently transported by pipeline or by road via cylinders, tube trailers, and cryogenic tankers, with a small amount shipped by rail or barge. Due to the energy intensive nature and the cost associated with hydrogen distribution via high-pressure cylinders and tube trailers, this method of distribution has a range limited to approximately 200 km. For longer distances of up to 1,600 km, hydrogen is usually transported as a liquid in super-insulated, cryogenic, over-the-road tankers, railcars, or barges, and then vaporized for use at the customer site [74]. This is also an energy intensive and costly process. Therefore, breakthroughs are necessary in material science to reduce the cost of transporting hydrogen over long distances. Two fundamental questions are how much energy is required to extract hydrogen from naturally occurring, stable hydrogen-rich compounds and whether hydrogen should be produced at large-scale central locations that will require the development of a dedicated infrastructure to store and transport it to end use destinations. Both of these questions demand close evaluation of the related social, economic, and environmental costs and benefits associated with developing a hydrogen production and transportation infrastructure.
6.4 Public Acceptance

The hydrogen economy will be a revolutionary change from the world we know today. Education of the general public, training personnel in the handling and maintenance of hydrogen system components, adoption of codes and standards, and development of certified procedures and training manuals for fuel cells and safety will foster hydrogen's acceptance as a fuel.
VII.
CONCLUSION
The major benefit of demonstrating the processes for H2 production from a renewable energy source (biomass) is the high overall efficiency and zero emissions to the environment. Fundamental advances in catalysis, membranes, and gas separation could enable more efficient, lower-cost hydrogen technologies. The processes that combine hydrogen generation and separation in a single reactor (for example, membrane reactors for methane steam reforming) could improve conversion efficiency and
.
reduce the initial cost of production of hydrogen by steam reforming of ethanol or bio-derived liquids The overall challenge to hydrogen production is the cost reduction and storage safety. In order to succeed in the commercial marketplace hydrogen must be comparable to conventional fuels and technologies on a per-mile basis.
446
Vol. 3, Issue 1, pp. 436-450
ACKNOWLEDGEMENT
The authors are grateful to Dr. Kamal Singh, former Vice Chancellor, Sant Gadge Baba Amravati University Amravati, Maharashtra, India, for her inputs to the IMSTR (Innovative Membranes & Separation Technologies Research) group, SGBAU, Amravati, M.S., India. Also, Mr. Kishor Darda, Chairman, Darda Education Society, Yavatmal, M.S., India and Dr. Avinash Kolhatkar, Principal, Jawaharlal Darda Institute of Engineering and Technology, Yavatmal, (MS), India, supported this research work.
REFERENCES
[1] [2] [3] [4] Comas, J., F. Marino, M. Laborde, N. Amadeo, Bio-ethanol steam reforming on Ni/Al2O3 catalyst. Chemical Engineering Journal 98: 61-68, 2004. Athanasio, N. F., and X. E. Verykios, Reaction network of steam reforming of ethanol over Ni-based catalysts. Journal of Catalysis 225: 439-452, 2004. Cavallaro, S. and S. Freni, Ethanol steam reforming in a molten carbonate fuel cell. A preliminary kinetic investigation. International Journal of Hydrogen Energy 21; 465-469, 1996 Yakimova1, M.S., A.S. Lermontov1, O.S. Polezhaeva, V.K. Ivanov, V.F. Tretyakov, Hydrogen Production via Bio-Ethanol Steam Reforming on CeO2-Based Catalysis. 2nd International IUPAC Conference on Green Chemistry, Russia, 2008. Meng, Ni, Dennis Y.C. Leung, Michael K.H. Leung, A review on reforming bioethanol for hydrogen production, International Journal of Hydrogen Energy 32: 3238-3247, 2007. Ilias Shamsuddin, Reforming of Liquid Hydrocarbons in a Novel hydrogen selective Membrane based Fuel Processor UCR/HBCUs/OMIs Contractors Meetings, Pittsburgh. 2003(Available from: www.osti.gov/bridge/servlets/purl/825381-Dvc4sV/native/825381.pdf.) Ilgi, K. Kapdan, F. Kargi, Bio-hydrogen production from waste materials. Enzyme and Microbial Technology 38: 569-582, 2006. Armor, J. N., The multiple roles for catalysis in the production of H2. Applied Catalysis A Gen 176: 159-76, 1999. Yakimova1, M.S., A.S. Lermontov1, O.S. Polezhaeva, V.K. Ivanov, V.F. Tretyakov, Hydrogen Production via Bio-Ethanol Steam Reforming on CeO2-Based Catalysis. 2nd International IUPAC Conference on Green Chemistry, Russia, 2008. Meng, Ni, M.K.H. Leung, D.Y.C. Leung, Energy and exergy analysis of hydrogen production by a proton exchange membrane (PEM) electrolyzer plant. Energy Conversion and Management 49: 27482756, 2008a Gernot, E., F. Aupretre, A. Deschamps, F. Epron, P. Marecot, D. Duprez, C. Etievant. Epron, Production of hydrogen from bioethanol in catalytic membrane reactor. WHEC 16. Lyon France, 2006. Basile, A., F. Gallucci, A. Iulianelli, M. De Falco, S. Liguori, Hydrogen Production by ethanol steam reforming: Experimental study of a Pd-Ag membrane reactor and traditional reactor behavior, International Journal of Chemical Reactor Engineering 6 Article A30: 1-17. 2008. Chen, Y., H. Xu, Y. Wang, G. Xiong, Hydrogen production from the steam reforming of liquid hydrocarbons in membrane reactor. Catalysis Today 118: 136-143, 2006. Ayturk, M. E., N. K. Kazantzis, Y. H. Ma, Modeling and performance assessment of Pd- and Pd/Aubased catalytic membrane reactors for hydrogen production. Energy & Environmental Sciences 2: 430438, 2009. Jeff McDaniel. The Hydrogen Economy Myths & Realities. Ohio Fuel Cell Symposium, May 24 2007. Haryanto, A., S. Fernando, N. Murali. Current status of hydrogen production techniques by steam reforming of ethanol: A Review. Energy Fuels 19: 2098-2106, 2005. Freni, S., G. Calogero, S. Cavallaro. Hydrogen production from methane through catalytic partial oxidation reactions. Journal of Power Sources 87: 28-38, 2000. Ayabe, S.; Omoto, H.; Utaka T.; Kikuchi, R.; Sasaki, K.; Teraoka, Y.; Eguchi, K. Catalytic autothermal reforming of methane and propane over supported metal catalysts, Applied Catalysis A: General, No. 241, p. 261-269. 2003. Trygve Riis, Elisabet F. Hagen, Preben J. S. Vie, ystein Ulleberg HYDROGEN RODUCTION R&D:PRIORITIES AND GAPS, HYDROGEN PRODUCTION R&D: PRIORITIES AND GAPS Barbir, F. 2005. PEM electrolysis for production of hydrogen from renewable energy sources. Solar Energy 78: 661-669. Grigoriev, S.A., V. I. Porembsky, V.N. Fateev. 2006. Pure hydrogen production by PEM electrolysis for hydrogen energy. International Journal of Hydrogen Energy 31: 171-175.
[5] [6]
[7] [8] [9]
[10]
[11] [12]
[13] [14]
[15] [16] [17] [18]
[19] [20] [21]
447
Vol. 3, Issue 1, pp. 436-450
[22] [23] [24] Ouweltjes, J. P. P., M.V. Tuel, F. P. V. Berkel, B. Rietveld. 2007. Solid Oxide Electrolyzers for Efficient Hydrogen Production. ECS Transactions 7: 933-940. Wang, W., J. Vohs, R. Gorte, 2007. Hydrogen Production via CH4 and CO Assisted Steam Electrolysis. Topics in Catalysis 46: 380-385. Koca, A., M. Sahin. 2002. Photocatalytic hydrogen production by direct sunlight from sulfide/ sulfite solution. International Journal of Hydrogen Energy 27, 363-367. Kerby, R., P. Ludden, G. Roberts. 1995. Carbon monoxide-dependent growth of Rhodospirillum rubrum. Journal of Bacteriology 177: 2241-2244. Bamwenda, G.R., S. Tsubota, T. Nakamura, M. Haruta. 1995. Photoassisted hydrogen production from a water-ethanol solution: a comparison of activities of Au/TiO2 and Pt/TiO2. Journal Photochemistry and Photobiology A: Chemistry 89: 177-189. Abe, R., K. Sayama, K. Domen, H. Arakawa. 2001. A new type of water splitting system composed of two different TiO2 photocatalysts (anatase, rutile) and a IO3-/Ishuttle redox mediator. Chemical Physics Letters 344: 339-344. Ashokkumar, M. 1998. An overview on semiconductor particulate systems for photoproduction of hydrogen. International Journal of Hydrogen Energy 23: 427-438. Anpo, M., M. Takeuchi. 2003. The design and development of highly reactive titanium oxide photocatalysts operating under visible light irradiation. Journal of Catalysis 216: 505-516. Bamwenda, G.R., S. Tsubota, T. Nakamura, M. Haruta. 1995. Photoassisted hydrogen production from a water-ethanol solution: a comparison of activities of Au/TiO2 and Pt/TiO2. Journal Photochemistry and Photobiology A: Chemistry 89: 177-189. Choi, W., A. Termin, M. R. Hoffmann. 1994. The Role of Metal Ion Dopants in Quantum-Sized TiO2: Correlation between Photoreactivity and Charge Carrier Recombination Dynamics. Journa1 of Physical Chemistry 98: 13669-13679 Choi, W., 2007. Photocatalytic hydrogen production using surface-modified Titania nanoparticles. Solar Hydrogen and Nanotechnology II. Edited by Guo, Jinghua. Proceedings of the SPIE, Volume 6650. 66500L (Available From: http://adsabs.harvard.edu/abs/2007SPIE.6650E..17C) Li, Y., H. Zhang, Z. Guo, J. Han, X. Zhao, Q. Zhao, S. J. Kim. 2008. Highly Efficient Visible-LightInduced Photocatalytic Activity of Nanostructured AgI/TiO2 Photocatalyst. Langmuir 24: 8351-8357. Levin, D. B., L. Pitt, M. Love. 2004. Biohydrogen production: prospects and limitations to practical application. International Journal of Hydrogen Energy 29: 173-185. Garca, E. Y., M. A. Laborde. Hydrogen production by the steam reforming of ethanol: Thermodynamic analysis. International Journal of Hydrogen Energy 16 (5): 307-312, 1991. Vasudeva, K., N. Mitra, P. Umasankar, S. C. Dhingra. Steam reforming of ethanol for hydrogen production: Thermodynamic analysis. International Journal of Hydrogen Energy 21: 13-18, 1996. http://www.fuelfromthewater.com/production.htm David B. Levin, Richard Chahine, Challenges for renewable hydrogen production from biomass, International Journal of Hydrogen Energy xxx , 1 8, (2009). McHugh K. Hydrogen production methods. MPR Associates Inc.; 2005. p. 41. Muradov N, Smith F. Thermocatalytic conversion of landfill gas and biogas to alternative ransportation fuels. Energy and Fuels 2008;22:205360. Milliken J. Hydrogen generation from biomass-derived carbohydrates via aqueous phase reforming process. Washington, DC: U.S. Department of Energy; 2008; PD-6. Reichman B, Mays W, Strebe J, Fetcenko M. Fuel reformation using alkaline enhancement. In: Proceedings International Hydrogen Energy Congress and Exhibition. Istanbul, Turkey; 1315 July 2007. Wen G, Xu Y, Ma H, Xu Z, Tian Z. Production of hydrogen by aqueous-phase reforming of glycerol. International Journal of Hydrogen Energy 2008;33:665766. Srensen B. Hydrogen and fuel cells emerging technologies and applications. New York: Elsevier Academic Press; 2005; p. 450. Guoxin H, Hao H. Hydrogen rich fuel gas production by gasification of wet biomass using a CO2 sorbent. Biomass and Bioenergy 2009;33:899906. Anderson, A., Bio-Derived Liquids to Hydrogen Distributed Reforming Working Group Background Paper. U.S. Department of Energy, HFCIT Program Tracy Carole, and Energetics Incorporated, 2006. (Available from: www1.eere.energy.gov/hydrogenandfuelcells/bio_derived_liquids_meeting.html) Adhikari, S., S. Fernando. Hydrogen membrane separation techniques. Industrial and Engineering Chemistry Research 45: 875-881, 2006. Voss, C., Applications of pressure swing adsorption technology. Adsorption 11: 527-529, 2005.
[25]
[26]
[27] [28]
[29]
[30]
[31]
[32] [33] [34] [35] [36] [37] [38] [39] [40] [41]
[42] [43] [44] [45]
[46] [47]
448
Vol. 3, Issue 1, pp. 436-450
[48] [49] [50] Haga, F., T. Nakajima, H. Miya, S. Mishima. Catalytic properties of supported cobalt catalysts for steam reforming of ethanol. Catalysis Letters 48: 223-227, 1997. Pilar, R. P., N. Homs, S. Joaquim, J. L. Fierro., Effect of sodium addition on performance of Co-ZnO based catalysts for hydrogen production from bioethanol. Journal of catalysis 222: 470-480. 2004. Pilar, R. P., J. Llorca, N. Homs, S. Joaquim, J. A. Dalmon., CO-free hydrogen from steam reforming of bioethanol over ZnO-supported cobalt catalysts: Effect of the Mettalic precursor. Applied catalysis B: Environmental 43: 355-369, 2003. Benito, M., R. Padilla, L. Rodriguez, J. L. Sanz, L. Daza., Zirconia supported catalyst for bioethanol steam reforming: Effect of Active Phase and Zirconia Structure, Journal of Power sciences 169: 167176, 2007. Homs, N., J. Llorca, R. P. Pilar., Low temperature steam-reforming of ethanol over ZnO-supported catalyst: The effect of nickel and copper addition to ZnO supported cobalt-based catalyst. Catalysis Today 116: 361-366, 2006. Vargas, J. C., S. Libs, A. C. Roger, A. Kiennemann., Study of Ce-Zr-Co fluoritetype oxide as catalyst for hydrogen production by steam reforming of bioethanol. Catalysis Today 107-108: 417-425, 2005. Chang, Y. U., D. W. Lee, S. J. Park, K. Y. Lee., Study on catalytic membrane reactor for hydrogen production from ethanol steam reforming. International Journal of Hydrogen Energy 34: 2947-2954, 2009. Kugai, J., S. Velu, C. Song., Novel CeO2 supported Ni-Rh bimetallic catalysts for reforming of bioethanol to produce hydrogen for fuel cells. Preprint Papers -American Chemical Society, Division of Fuel Chemistry 48(2): 754-755, 2003. Marino, F., M. Boveri, G. Baronetti, M. Laborde, Hydrogen production from steam reforming of bioethanol using Cu/Ni/K/-Al2O3 catalyst. Effect of Ni. International Journal of Hydrogen Energy, 26(7): 665-668, 2001. Cavallaro, S., N. Mondello, and S. Freni., Hydrogen production from ethanol for internal reforming molten carbonate fuel cell. Journal of Power Sources 102: 198-204, 2001. Haga, F., T. Nakajima, K. Yamashita, S. Mishima., Effect of crystallite size on the catalysis of alumina-supported cobalt catalyst for steam reforming of ethanol. Reaction Kinetics and Catalysis Letters 63: 253-259, 1998. Sahoo, D. R., S. Vajpai, S. Patel, K. K. Pant. Chem. Eng. J. 125, 139, 2007. Erdhelyi, A., J. Rask, T. Kecsks, M. Tth, M. Dmk, K. Ban, Catalysis Today 116: 367, 2006. Montini, T. L. D. Rogatis, V. Gombac, P. Fornasiero, M. Graziani. Rh (1%)@CexZr1xO2Al2O3 nanocomposites: Active and stable catalysts for ethanol steam reforming. Applied Catalysis B: Environmental 71: 125-134, 2007. Llorca, J., N. Homs, J. Sales, and de la P.R. Piscina., Efficient production of hydrogen over supported cobalt catalysts from ethanol reforming. Journal of Catalysis 209: 306, 2002. Aupretre, F., C. Descorme, and D. Duprez., Bio-ethanol catalytic steam reforming over supported metal catalysts. Catalysis Communications 3: 263-267, 2002. Delsman, E. R., B. J. P. F. Laarhoven, M. H. J. M. De Croon, G. J. Kramer, J. C. Schouten, Comparison between conventional fixed-bed and microreactor technology for portable hydrogen production case. Trans IChemE, Part A, chemical engineering Research and Design 83(A9): 10631075, 2005. Vaidya, P. D., A. E. Rodrigues., Insight into steam reforming of ethanol to produce hydrogen for fuel cells. Chemical Engineering Journal 117: 39-49, 2006. Davda, R.R., J. W. Shabaker, G. W. Huber, R. D. Cortright, J. A. Dumesic., A review of catalytic issues and process conditions for renewable hydrogen and alkanes by aqueous-phase reforming of oxygenated hydrocarbons over supported metal catalysts. Applied Catalysis B: Environmental 56: 171186, 2005. Benito, M., J.L. Sanz, R. Isabel, R. Padilla, R. Arjona, L. Daza., Bio-ethanol steam reforming: insights on the mechanism for hydrogen production. Journal of PowerSources 15: 11-17, 2005. Yang, Y., J. Ma, F. Wu., Production of hydrogen by steam reforming of ethanol over a Ni/ZnO catalyst. International Journal of Hydrogen Energy 31: 877-882, 2006. http://fuelcellstore.com/products/heliocentris/INTRO.pdf http://www.h2technology.com/hydrogen-fuel-cell/#more-17 F. A. Bruijn1, V. A. T. Dam, G. J. M. Janssen, Review: Durability and Degradation Issues of PEM Fuel Cell Component, Fuel Cell, 8, 1, 3-22, 2008. Trygve Riis, Gary Sandrock , ystein Ulleberg, Preben J.S., Hydrogen Storage Gaps and Priorities, HIA HCG Storage paper, http://ieahia.org/pdfs/HIA_Storage_G&P_Final_with_Rev.pdf Louis Schlapbach, Andreas Zttel, Hydrogen-storage materials for mobile applications, Nature, 414, 353-358, 2001
[51]
[52]
[53] [54]
[55]
[56]
[57] [58]
[59] [60] [61]
[62] [63] [64]
[65] [66]
[67] [68] [69] [70] [71] [72] [73]
449
Vol. 3, Issue 1, pp. 436-450
[74] http://www.iphe.net/docs/Fact_Sheets/Hydrogen_Production_and_Delivery.pdf
Authors Biography
Prashant R. Tayade currently working as Assistant Professor in Chem Engg Department, JDIET, Yavatmal, M.S., India, having 05 years of teaching experience. He has completed his M.Tech. in Chemical Engg. from Dr. BATU, Lonere, Raigad in 2005 and now pursuing Ph.D. from SGBAU, Amravati M.S., India. He is involved in research activity by publishing 02 papers in international journal and presented 06 papers in national and international conferences. He is currently working on the reforming of bioethanol for the production of hydrogen, CO2 sequestration using biochar, pyrolysis of biomass for the production of biofuels and biochar Vilas. S. Sapkal is currently the Vice-Chancellor of Rashtrasant Tukadoji Maharaj Nagpur University, Nagpur, M.S., India. Formerly, he was Professor & Head for 15 years in University Department of Chemical Technology, SGBAU, Amravati, M.S., India. He has completed his M.Tech. & Ph.D. from IIT Pawai, Mumbai in 1987. He has 120 Publications in his credit (Presentations in National and International Journals and Conferences) including invited talks and Key-Note addresses in the field of Energy Conservation, productivity improvement, quality enhancement and environmental protection, application development etc. Honored to be Session Chairman and Chief Guest. He has Worked on large number of academic bodies and contributed significantly for academic excellence, quality research and professional administration. He has worked as expert for National Board of Accreditation of AICTE, New Delhi, for UGC for granting Deemed University Status for Director, Technical Education for Approval of new colleges, For Industrial Health and Safety Pollution Control Board, Non-conventional Energy Sources. He is recognized guide from Sant Gadge Baba Amravati University, Amravati, Dr. Babasaheb Ambedkar Maratwada University, Aurangabad, Yashwantarao Chauhan Open University, Nashik. Under his guidance 05 Ph.D. are awarded, 10 ongoing, 27 projects guided for M. Tech. He has handled research Project funded by DST, AICTE, UGC, MNES for Rs.1.7crores in the field of product and technology development. He has got the best Service Award among all University Teachers of Amravati University in 2000, Rashtriya Nirman Award for Outstanding achievements in the chosen field of activity by international Business council, New Delhi in 2001, Life Time Achievement Award by National And International Compendium, New Delhi In 2001. Chandrashekhar V. Rode currently working as Senior Principal Scientist Chemical Engineering & Process Development Division National Chemical Laboratory, Pune. He has completed his Ph.D. (Chemistry) 1990, University of Pune Title, Studies in Multiphase Reactors. He has a vast research experience having No. of Peer-reviewed Publications: 100, No. of Granted US / International Patents: 18. He is Recognized Guide of University of Pune, Pune, SRTM University, Nanded; ICT Mumbai; Shivaji University, Kolhapur. No. of Ph.D. Students Guided: 6 No. of Ph.D. Students Working: 8. He has Excellent Professional Records: 1) AIST Sendai, Japan, Oct. to Dec 2006, Oct-Nov. 2007 and 2008. 2) Visiting scientist at National Institute of Advanced Industrial Science and Technology, Sendai Japan under JSPS Invitation programme during February-November 2002. 3) Awarded JSPS (Japan Society for Promotion of Science) fellowship at Institute for Chemical Reaction Science, Tohoku University, at Sendai, Japan from March 95 to March 96. 4) Visiting scientist at Zelinski Inst. of Organic Chemistry, Moscow, July1990 June 1991. His Highlights of Achievements: 1. Fellow of Maharashtra Academy of Sciences 2. A pilot plant of 30 TPA for hydrogenation of nitrobenzene to p-aminophenol has been commissioned, at the clients site Oct-Nov 2009. 3. Five silver medals by NCL Research Foundation, for the grant of US patents 4. DuPonts (UK) citation in 2000: In recognition of your creative insight and contributions to the advancement of liquid phase oxidation of terephthalic acid Rajive S. Sapkal currently working as Associate Professor & Head, University Department of Chemical Technology, SGBAU, Amravati, M.S., India. He has completed his Ph.D. and having vast teaching experience of 17 years and industry research experience of 12 years. He has published and presented papers in 52 International & National Conferences and Seminars. He is the Section-In-Charge of Project Sanctioned by D.S.T. Govt. of India, New Delhi, Project -In-Charge of Cellulose Technologist, Project sanctioned by CAPART, New Delhi for Research & Training. Also, he is steering the post of Assistant Director, Khadi & Village Industries Commission, Ministry of Industries, Govt. of India, New Delhi.
450
Vol. 3, Issue 1, pp. 436-450
AN EFFICIENT BIT RATE PERFORMANCE OF SERIAL-SERIAL MULTIPLIER WITH 1S ASYNCHRONOUS COUNTER

P.Rajesh1 and B.Gopinath2
1
Department of ECE, SNS Tech, Coimbatore City, T.N., India 2 Assistant professor, SNS Tech, Coimbatore, T.N., India
A BSTRACT
Traditional Serial-Serial multiplier addresses the high data sampling rate. It is effectively considered as the entire partial product matrix with n data sampling cycle for nn multiplication function instead of 2n cycles in the conventional multipliers. The existing Serial-Serial multiplier is the first bit serial structure. Newly developed serial-serial multiplier design is capable of processing input data at (GBs) without buffering and with reduced total number of computational cycle. This multiplication of partial products by considering two series inputs among which one is starting from LSB the other from MSB. Using this feed sequence and accumulation technique it takes only n cycle to complete the partial products. It achieves high bit sampling rate by replacing conventional full adder and highest 5:3 counters. Here asynchronous 1s counter is presented. This counter takes critical path is limited to only an AND gate and D flip-flops. Accumulation is integral part of serial multiplier design. 1s counter is used to count the number of ones at the end of the nth iteration in each counter produces. The proposed multiplier consists of a serial-serial data accumulator module and carries save adder that occupies less silicon area than the full carry save adder. In this paper we proposed model address for the 8bit 2s complement implementing the Baugh-wooley algorithm and unsigned multiplication implementing the proposed architecture for 88 Serial-Serial unsigned multiplication. We can able to extend the 16bit multiplication.
K EYW ORDS:
Binary multiplication, Serial-Serial Asynchronous counter; Serial-Link bus architecture.
multiplication,
Serial-Parallel
multiplication;
I.
INTRODUCTION
Serial-serial multiplication techniques have been in use for many years . The proposed a structure based on 1s counters , which computed the N bits of the product of an NxN multiplication in N bits clock cycles, using N processing cells. One cell was dedicated to the computation of each output bit, In this paper, the design of a bit serial-serial multiplier which utilises a bit serial, least significant bit first format for both input and output based on particular methods. The design and implementation approaches of multipliers contribute substantially to the area, speed and power consumption of computational intensive VLSI systems. Often the delay of multipliers dominates the critical path of this System and due to issues reliability and portability. Power consumption is a critical criterion for applications that demand low-power as its primary metric. While low power and high speed multiplier circuits are highly demanded, it is not always possible to achieve both criteria simultaneously. Therefore, a good multiplier design requires some tradeoffs between speed and power consumption. The main objective of designing bit-parallel multipliers which makes them different from bit-serial and digit-serial multipliers. Bit-serial multipliers provide the lowest possible area complexity. In the literature, the bit-serial algorithms have been studied for the polynomials basis and two different algorithms have been proposed, namely the Least Significant Bit (LSB) first and the Most Significant Bit (MSB) first bit-serial Algorithms. Fast arithmetic circuits are key elements of high performance computers and data processing systems. In the majority of these applications, multipliers have been a critical and obligatory component in dictating the overall circuit performance when constrained by power consumption and computation speed. Compressors are a critical component of
451
Vol. 3, Issue 1, pp. 451-459
the multiplier circuit, which greatly influence the overall multiplier speed. The authors propose two novel high performance 5:2 compressor architectures. The main objective of their designs is to limit the carry propagation to a single stage, thereby reducing the overall propagation delay. Based on the partial product formation multipliers are classified into two types. They are serial and parallel multiplier. Parallel multipliers are popular for their high speed operation but for long word length they are limited by hardware cost and power consumption of the applications. Serial multipliers are useful in low cost and low power applications. Several architectures have been reported in the literature for the implementation of serial multipliers but do not find any significant work in the reduction of partial product formation. Each architecture produce the partial product in 2N cycles for an N x N multiplication. A new approach to form the entire partial product matrix in just N sampling cycles for an N x N multiplication instead of at least 2N cycles in the conventional serial-serial multipliers. A new approach to serial accumulation of data by using asynchronous counters is suggested here which essentially count the number of 1s in respective input sequences. The counters effectively replace the full adders in the accumulation circuit. In this project the partial products are formed in serial manner and they are accumulated by the counters and finally the product can be obtained by adding the accumulated output by the ripple carry adder. Serial multipliers are popular for their low area and power. They are broadly classified into two categories, namely serial-serial and serial-parallel multiplier. In a serial-serial multiplier both the operands are loaded in a bit-serial fashion, reducing the data input pads. To reduce the number of computational cycles from 2n to n in an nn serial multiplier, several serial-parallel multipliers.
II.
RELATED WORK
A fast serial-parallel multiplier based on the Baugh-Wooley algorithm is proposed. It is shown that sign extension of the sum or carry bit, produced during the addition of the bit product rows, can be avoided, and a more efficient multiplier can be obtained. We present a twos-complement FSP multiplier that is based on the well known Baugh-Wooley algorithm. In the proposed multiplier, there is no need to sign extend the sum or carry bits of the partial product sums. The technique described can be extended to twos complement multiplication. The sign of the twos complement number is extended either directly, as the extension of the sign bit of the bit-product sum, or created as a result of a carry generated by previous row addition. Main contribution of our wok will follow, 2.1. Major participation of our work An FSP multiplier based on the Baugh-Wooley algorithm has been developed. It has been shown that there are 2s complement sign extensions of the sum or the carry bits produced during the addition of two-bit product row. 1s counter is used to count the number of ones at the end of the nth iteration in each counter produces. The proposed multiplier consists of a Serial-Serial data accumulator module and carries save adder that occupies less silicon area than the full carry save adder. Unsigned multiplier is based on the proposed architecture for 8 8 serial-serial unsigned multiplication. It is effective multiplier for a time, area and delay analysis technique. 2.2. Compressor architecture In conventional method consist of the several architectures and designs of low-power 4-2 and 5-2 compressors capable of operating at ultra low supply voltages. These compressor architectures are anatomized into their constituent modules and different static logic styles based on the same deep sub micrometer CMOS process model are used to realize them. Different configurations of each architecture, which include a number of novel 4-2 and 5-2 compressor designs, are prototyped and simulated to evaluate their performance in speed, power dissipation and power-delay product. The newly developed circuits are based on various configurations of the novel 5-2 compressor architecture with the new carry generator circuit, or existing architectures configured with the proposed circuit for the exclusive OR (XOR) and exclusive NOR (XNOR) [XORXNOR] module. The proposed new
452
Vol. 3, Issue 1, pp. 451-459
circuit for the XORXNOR module eliminates the weak logic on the internal nodes of pass transistors with a pair of feedback PMOSNMOS transistors. Driving capability has been considered in the design as well as in the simulation setup so that these 4-2 and 5-2 compressor cells can operate reliably in any tree structured parallel multiplier at very low supply voltages. Two new simulation environments are created to ensure that the performances reflect the realistic circuit operation in the system to which these cells are integrated. Simulation results show that the 4-2 compressor with the proposed XORXNOR module and the new fast 5-2 compressor architecture are able to function at supply voltage as low voltage. 2.2.1. 4-2 Compressor architectures A 4-2 compressor has five inputs and three outputs, as shown in Fig. 1, The four inputs x1,x2 ,x3 , and x4 and the output sum have the same weight. The output carry is weighted one binary bit order higher.
Figure.1 4-2 compressor
The conventional implementation of a 4-2 compressor is composed of two serially connected full adders, at gate level, high input compressors are anatomized into XOR gates and carry generators normally implemented by multiplexers (MUX). Therefore, different designs can be classified based on the critical path delay in terms of the number of primitive gates. Since the difference between the delays of widely used XOR gate and carry generator is trivial in an optimized design, the delay of the compressor is more commonly specified as (m+n) . The throughput rate has been increased by judicially changing the way the partial products are accumulated. Main benefit of this project is it occupies less area and fast operation. It considers the 16 bit operation also it must be very efficient. The bit consideration is very important for any multiplication. Each iteration required single n cycle. The two carry signals carry and Cout is generated from both the XOR and XNOR functions of the input signals. The Sum output is generated by several two-input XOR circuits, some internal signals of which can be used to generate the two carries. Fig.2, shows the logic decomposition of this 4-2 compressor architecture. It is mainly composed of six modules, four of which are XOR circuits and the other two are 2-1 MUX. Three special XORXNOR modules marked with X-OR generate both the XOR and XNOR signals simultaneously to other modules driven by them.
Figure.2 Logic decomposition of 4-2 compressor
453
Vol. 3, Issue 1, pp. 451-459
2.2.2. 5-2 Compressor architectures The 5-2 compressor is another widely used building block for high precision and high speed multipliers. The block diagram of a 5-2 compressor is shown in Fig.3, which has seven inputs and four outputs.
Figure.3 5-2 compressor
Five of the inputs are the primary inputs x1, x2, x3, x4 and x5, and the other two inputs, Cin1 and Cin2 receive their values from the neighbouring compressor of one binary bit order lower in significance. All the seven inputs have the same weight. The 5-2 compressor generates an output of the same weight as the inputs, and three outputs Carry, Cout1, Cout2 and weighted one binary bit order higher. A third widely used compressor of significant importance is the 5-2 compressor. Its block diagram is shown in Fig.3, it has seven inputs of which five are direct inputs and two are carry-in bits from a previous stage. Similarly, there are four outputs of which two are carry-out bits to the next stage and the other two are sum and carry bits. A simple implementation of the 5-2 compressor is to cascade three full adders in a hierarchical structure, as shown in Fig.4, which has a critical path delay of 6 .
Figure.4 implementing using 3:2 compressors
In this simplest form, a 5:2 compressor can be designed by cascading three 3:2 compressors as shown in Fig. 5. This structure has a delay of 6 XORs and is slower than the 6:2 compressor presented in delayed convention which has a delay of only five XORs. A faster implementation of the 5:2 compressors with 5 XOR delays is presented in [7]. The main objective in our designs is to limit the carry propagation. In the conventional implementation of the 5:2 compressors using 3:2 compressors shown in Fig.5, the first compressor generates Co1, the second generates Co2 and the third generates the Sum and Carry. The second compressor receives inputs x4, ci1 and the output from the first compressor. Hence, Co2 is a function of ci1, thereby propagating ci1 across the compressor. These types of compressors require many numbers of inputs and outputs. So it is complex structured than 1s counter. It can possible to count the number of 1s.
454
Vol. 3, Issue 1, pp. 451-459
2.3. Review of serial multiplier In a serial-serial multiplier both the operands are loaded in a bit-serial fashion, reducing the data input pads to two serial multipliers are popular for their low area and power. Bit-serial processing can result in efficient communications, both within and between VLSI chips, because of the reduced number of interconnections required. Serial multiplier designs which are particularly suitable for applications where input data are sequentially presented .The operating speeds are determined mainly by the propagation delays along the critical path within the processing elements. A major advantage offered by bit serial processors is when the operands are available only one bit at a time, the processing speed can be improved using bit-serial arithmetic elements. The structures that use this approach can achieve moderate speeds with comparatively small area. 2.3.1. Implementation stages for serial- serial multiplier This operation consists of three stages i) The generation of partial products. ii) The reduction of partial products. iii) Final carry-propagation addition. The partial products can be generated either in parallel or serially, depending on the target application and the availability of input data. The partial products are generally reduced by carry-save adders using an array or a tree structure. Carry propagation addition is inevitable when the number of partial products is reduced to two rows. This final adder can be a simple ripple carry adder for low power or a carry look-ahead adder for high speed. It is highly desirable to reduce the number of partial products before the carry-save adders stage. The drawback is that and higher order compressors are slower and consumes more power than the full adders. An accumulator is an adder which successively adds the current input with the value stored in its internal register. 2.3.2. Serial multiplier with counter based accumulator In this type of multipliers the partial products are formed in a serial manner. Instead of using full adders counters are used to count the number of 1s in a column in this structure. This method provides power minimization from the fact that the counter output will not toggle unless a 1 is present at its input. The counter output is applied to a ripple carry adder to obtain the final product.By using this counter based accumulator the partial products are formed in N cycles for an NXN multiplication. A new approach to serial/parallel multiplier design by using parallel 1s counters to accumulate the binary partial product bits. The 1s in each column of the partial product matrix due to the serially input operands are accumulated using a serial T-flip flop (TFF) counter. 2.4. Proposed serial-serial multiplier Accumulation is an integral part of serial multiplier design. A typical accumulator is simply an added that successively adds the current input with the value stored in its internal register. Generally, the adder can be a simple RCA but the speed of accumulation is limited by the carry propagation chain. The accumulation can be speed up by using a CSA with two registers to store the intermediate sum and carry vectors, but a more complex fast vector merged adder is needed to add the final outputs of these registers. A new approach to serial accumulation of data by using asynchronous counters is suggested here which essentially count the number of 1s in respective input sequences. 2.4.1. 8 and 16 bit word length for unsigned multiplier A new technique of generating the individual row of partial products by considering two serial inputs, one starting from the LSB and the other from MSB. It takes only n cycles to complete the entire partial product generation.
455
Vol. 3, Issue 1, pp. 451-459
Figure.5 conventional partial product formation
The PP generation of an 88 multiplier for two unsigned numbers X and Y. Fig.5, shows the conventional partial product formation and Fig. 6, shows the generation sequence of the PPs. Row r generated in cycle r , for r=0,1..n-1. The PPs in Fig.7, are generated in such an unconventional way in order to facilitate their accumulation on-the-fly by the proposed counter-based accumulation technique. A PP bit corresponding to the middle column of the PP is produced by the centred AND gate.
Figure.6 Proposed partial product generation
Figure.7 Proposed architecture for 8 8 serial-serial unsigned multiplication
456
Vol. 3, Issue 1, pp. 451-459
The latched outputs are wired to the correct FAs and HAs (half adders) according to the positional weights of the output bits produced by the counters. From Fig.7, it is observed that the column height has been reduced from 8 to 4 and the final product, can be obtained with two stages of CSA tree and a final RCA. Similarly, for 1616, 3232 and 64 64 multipliers the column heights are reduced logarithmically from 16, 32, and 64 to 5, 6, and 7, respectively. This drastic reduction in column height leads to a much simpler CSA tree, and hence reducing the overall hardware complexity and power consumption. The latching register between the counter and the adder stages not only makes it possible to pipeline the serial data accumulation and the CSA tree reduction, but also prevents the spurious transitions from propagating into the adder tree. 2.4.2. 8 and 16 bit word length for 2s complement numbers The Most digital systems operate on signed numbers commonly represented in 2s complement. 2s complement numbers using the BaughWooley algorithm. The architecture of the proposed 2s complement serial-serial multiplier is depicted below structure. The addition of the term raises the height of CSA tree by only two bits regardless of the word length of the operands.
Figure.8 Proposed architecture for 88 serial-serial 2s complement multiplication
2.5. Simulation result.

Table 1: Proposed delay cycle of the serial-serial multiplication Method Bit rate Operating Mode Unsigned 2s complement Unsigned 2s complement Cycles
Proposed Proposed Proposed Proposed
88 88 1616 1616
N N N N
The both 8 bit and 16 bits are proposed based on their respective algorithms and architecture. 2s complement is using to implement in baugh-wooley algorithm. Unsigned multipliers are using to implement in proposed architecture. These different word lengths are support for the 1s counter. Both 8 bit and 16 bit outputs are getting 8 clock pulses(n cycles) so required time is vey minimum. In previous session did only 8bit multiplication here we completed 16 bit also.
457
Vol. 3, Issue 1, pp. 451-459
Figure.9 Snapshot of the output sequence of the proposed multiplier for 8 bit unsigned numbers
Figure.10 Snapshot of the output sequence of the proposed multiplier for 16 bit multiplier
III.
CONCLUSION
In this paper, we have studied for computing serial-serial multiplication is introduced by using low complexity asynchronous counters. By exploiting the relationship among the bits of a partial product matrix, it is possible to generate all the rows serially in just n cycles for an nn multiplication. Employing counters to count the number of 1s in each column allows the partial product bits to be generated on-the-fly and partially accumulated in place with a critical path delay of only an AND gate and a DFF. The counter-based accumulation reduces the PP height logarithmically and makes it possible to achieve an effective reduction rate. The proposed counter-based multiplier performs many
458
Vol. 3, Issue 1, pp. 451-459
serial-serial and serial-parallel multipliers in speed but its hybrid architecture does carry an area overhead.
REFERENCES
[1] A High Bit Rate Serial-Serial Multiplier with On-the-Fly Accumulation by Asynchronous Counters, in proc. IEEE Conf. 2010. [2] Aggoun.A, Ashur.A, and Ibrahim,M.K (2000),Area-time efficient serial-serial multipliers, in Proc. IEEE Conf. Circuits Syst. (ISCAS), Geneva, Switzerland, pp. 585588. [3] Almiladi.A, Ibrahim.M.K,Al-Akidi.M, and Aggoun.A (2007),High performance scalable bidirectional mixed radix-serial serial multipliers,IET Proc-Comput. Digit. Tech., vol.1, no. 5, pp. 632639. [4] Bi.G and Jones.E.V, High-performance bit-serial adders and Multipliers (1992), IEEE Proc G-Circuits Devices Systs, vol. 139, no. 1, pp.109113. [5] Dobkin.R, Moyal.M, Kolodny.A, and Ginosar.R (2010), Asynchronous current mode serial communication, IEEE Trans. Very Large Scale Integer. (VLSI) Syst., vol. 18, no. 7, pp. 11071117. [6] Ghoneima.M, Ismail.Y, Khellah.M, Tschanz.J, and De.V (2009),Serial link bus: A low-power on-chip bus architecture, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 56, no. 9, pp. 20202032. [7] Gnanasekaran.R (2006), On a bit-serial input and bit-serial output multiplier, IEEE Trans. Compute., vol. C-32, no. 9, pp. 878880. [8] Menon.R and Radhakrishnan.D (2006), High performance 5:2 compressor architectures, IEEE ProcCircuits Devices Syst., vol. 153, no. 5, pp. 447452. [9] Nibouche.O, Bouridarie.A, and Nibouche.M (2001), New architectures for serial-serial multiplication, in Proc. IEEE Conf. Circuits Syst. (ISCAS), Sydney, Australia, vol. 2, pp. 705708. [10] Saleh.H.I, Khalil.A.H, Ashour.M.A, and Salama.A.E (2001),Novel serial-parallel multipliers, IEEE ProcCircuits Devices Syst., vol. 148, no. 4, pp. 183189. [11] Sunder.S, El-Guibaly.F, and Antoniou.A (1995), Twos-complement fast serial-parallel multiplier, IEEE Proc.-Circuits Devices Syst, vol. 142, no. 1, pp. 4144. Authors Rajesh received the B.E degree from P.S.R College of technology, Tamilnadu, India, in 2009. Now doing M.E degree for VLSI design in S.N.S College of technology, Tamilnadu, India. Am atttened many conferences for both national and international journals.Now doing my research work.
459
Vol. 3, Issue 1, pp. 451-459
CONTROLLER AREA NETWORK DATA EXTRACTION FOR AUTOMOBILE

Tejal Farkande and S. N. Pawar
Department of Electronics and Telecommunication, J.N.E.C, Aurangabad, India
ABSTRACT
Nowadays networked Electronic Control Units (ECUs) are increasingly being deployed in automobiles to realize various functions and Controller Area Network (CAN) is deployed for the communications among ECUs. Our primary objective is to build both hardware and software that interface and communicate directly with CAN network and extract CAN messages for reliable communication between automobile ECUs. The hardware is a circuit board that is capable of extracting CAN messages released from an automobile ECUs which can be used for reliable automobile ECUs communications, for example, as operating engine temperature crossed threshold limit a cooling system can turn on, or cabin temperature increases beyond comfort level, AC can turn on with interface and indication to user. Tests and trials of the completed board are also carried out to ensure that the system is performing decently well.
KEYWORDS: CAN, PIC microcontroller, CAN transceiver, LM 35 temperature sensor
I.
INTRODUCTION
In recent years, control systems of cars have moved from the analog to the digital domain. In particular, x-by-wire systems are appearing and drive research efforts of the whole automotive industry for the recent decade. Electronic Control Units (ECUs) are increasingly being deployed in automobiles to controls one or more electrical subsystems to realize the various functions. When a driver drives a car, there are many signals that are passed between the various Electronic Control units embedded inside the car. Output signals from an ECU contain information about the current state of the car as the driver interacts continuously with the car. A modern automobile can consists of up to 70 ECUs [1, 2] sensing and taking tabs of the various parameters of the automobile. This rapid and complex exchange of signals ensures the proper functioning of the car. The Controller Area Network (CAN) protocol was thus introduced to alleviate the problem of sending these signals efficiently in the form of structured message frames called CAN messages. In this we are construction the solutions of extracting needed CAN messages from automobile ECUs to make reliable communication among ECUs and this will also helps to reduce wiring harness in the car. The remainder of this paper is organized as follows: the CAN related information is given in the next subsections. We present our system design and operation in section 2, while results are presented in section 3. Finally we conclude the results in section 4.
1.1. Controller Area Network

In the early 1980s, engineers at Bosch evaluated existing serial bus systems regarding their possible use in passenger cars and found that none of the available network protocols were able to fulfill the requirements of the automotive. In February of 1986, CAN was born at the SAE (Society of Automotive Engineers) congress in Detroit, the new bus system developed by Bosch was introduced as 'Automotive Serial Controller Area Network'. Historical facts and rationales for the CAN protocol are given in [3, 4, 5]. Controller Area Network (CAN) is a serial data communications bus for realtime applications operates at data rates of up to 1 Mbps and has excellent error detection and
460
Vol. 3, Issue 1, pp. 460-465
confinement capabilities. CAN was originally developed for use in cars to provide a cost-effective communications bus for in-car electronics and as alternative to expensive, cumbersome and unreliable wiring looms and connectors. The car industry continues to use CAN for an increasing number of applications, but because of its proven reliability and robustness, CAN is now also being used in many other control applications. CAN is an international standard and is documented in ISO 11898 (for high-speed applications) and ISO 11519 (for lower-speed applications). Low-cost CAN controllers and interface devices are available as off-the-shelf components from several of the leading semiconductor manufacturers. Custom built devices and popular microcontrollers with embedded CAN controllers are also available. There are many CAN-related system development packages, hardware interface cards and easy-to-use software packages that provide system designers, builders and maintainers with a wide range of design, monitoring, analysis, and test tools. CAN network topology is given in Figure1, which gives it the advantage of easily adding new, CAN nodes to an existing network. Furthermore, the standardization of the protocol means all ECUs will conform to the CAN standards while transmitting data. All CAN nodes are fitted with a mandatory CAN transceiver chip that connects it to the CAN bus.
Figure 1.CAN network topology
CAN is based on the "broadcast communication mechanism". This means that all nodes can "hear" all Transmissions. There is no way to send a message to just a specific node; all nodes will invariably pick up all traffic. The CAN hardware, however, provides local filtering so that each node may react only on the interesting messages. Therefore, CAN is based on a message-oriented transmission protocol. Every message has a message identifier, which is unique within the whole network since it defines the priority of the messages.
1.2. CAN in Cars

To satisfy customer requirements for greater safety, comfort, and convenience, and to comply with increasingly stringent government legislation for improved pollution control and reduced fuel consumption, the car industry has developed many electronic systems. Anti-lock Braking, Engine Management, Traction Control, Air Conditioning Control, central door locking, and powered seat and mirror controls are just some examples. The complexities of these control systems, and the need to exchange data between them meant that more and more hard-wired, dedicated signal lines had to be provided. Sensors had to be duplicated if measured parameters were needed by different controllers. Apart from the cost of the wiring looms needed to connect all these components together, the physical size of the wiring looms sometimes made it impossible to thread them around the vehicle (to control panels in the doors, for example). In addition to the cost, the increased number of connections posed serious reliability, fault diagnosis, and repair problems during both manufacture and in service.
1.3. Working Principle of CAN

Data messages transmitted from any node on a CAN bus do not contain addresses of either the transmitting node, or of any intended receiving node. Instead, the content of the message (e.g. Revolutions per Minute, Hopper Full, X-ray Dosage, etc.) is labeled by an identifier that is unique throughout the network. All other nodes on the network receive the message and each performs an acceptance test on the identifier to determine if the message, and thus its content, is relevant to that
461
Vol. 3, Issue 1, pp. 460-465
particular node. If the message is relevant, it will be processed; otherwise it is ignored. The unique identifier also determines the priority of the message. The lower the numerical value of the identifier, the higher the priority. In situations where two or more nodes attempt to transmit at the same time, a non-destructive arbitration technique guarantees that messages are sent in priority order to avoid the loss of messages.
II.
SYSTEM DEPLOYMENT
The project can be largely divided into three parts: the design and fabrication of hardware main board, Programming of firmware. Figure 2 gives a block diagram of the overall design of the system.
Figure 2. System block diagram
Main board consist of consist of PIC18F4680 flash microcontroller, CAN Transceiver (MCP2551), temperature sensor LM35 and CAN. Essentially, the main board is the one that is receiving input signals from the controller via the CAN interface. These signals are passed via the RS-232 port to the PC and LCD display.
2.1. Temperature Sensor

Various types of temperature sensors are available in the market and sensor depending upon the application can be used. The LM35 has temperature range of 55 to +150C. It can be used with single power supply or with plus and minus supply. The LM35 output voltage is linearly proportional to the Celsius (Centigrade) temperature. The LM35 thus has an advantage over linear temperature sensors calibrated in Kelvin, as the user is not required to subtract a large constant voltage from its output to obtain convenient centigrade scaling [6]. The LM35s low output impedance, linear output, and precise inherent calibration make interfacing to readout or control circuitry especially easy.
2.2. Microcontroller
It is proposed to use PIC18F4680 is the enhanced flash microcontroller with ECAN Technology, 10-Bit A/D and nanowatt Technology. Other family of microcontroller units can also be used depending upon the use. The work of the microcontroller is to control the flow of information according to CAN messages from temperature sensor. The signals sensed by sensor and transmitted through CAN transceiver to LCD [7].
462
Vol. 3, Issue 1, pp. 460-465
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 2.3. CAN Transceiver (MCP 2551)
The MCP2551 is a high-speed CAN, fault-tolerant device that serves as the interface between a CAN protocol controller and the physical bus [8]. The MCP2551 provides differential transmit and receive capability for the CAN protocol controller and is fully compatible with the ISO-11898 standard, including 24V requirements. It will operate at speeds of up to 1 Mb/s. typically; each node in a CAN system must have a device to convert the digital signals generated by a CAN controller to signals suitable for transmission over the bus cabling (differential output). It also provides a buffer between the CAN controller and the high-voltage spikes that can be generated on the CAN bus by outside sources (EMI, ESD, electrical transients, etc.).The schematic of CAN transceiver is shown in Figure3.
Figure 3. Schematic of CAN transceiver
The CAN bus has two states: Dominant and Recessive. A dominant state occurs when the differential voltage between CANH and CANL is greater than a defined voltage (e.g., 1.2V). A recessive state occurs when the differential voltage is less than a defined voltage (typically 0V). TXD is the transmitter data input pin usually connected to the transmitter data output of the microcontroller to receive the data from the nodes on the CAN bus. And RXD is receiver data output usually connected to the receiver data input of microcontroller.
2.4. Display Unit

All the activities which are taking place with CAN messages, sensors and controller will be displayed on the LCD panel connected to Node2.
2.5. System Operation

In this there are three nodes referring to Figure 2. The simplified system flow chart of algorithm in 18F4680 PIC microcontroller is shown in Figure 3.
Figure 4. System flowchart for 18F4680 at node 1
463
Vol. 3, Issue 1, pp. 460-465
The Node-1 is connected to the temperature sensor LM35. The temperature data will be capture by the temperature sensor which gives the analog output. This analog output is given to the ADC which is inbuilt in the PIC microcontroller. This analog output is processed and passed on the CAN bus through CAN transceiver (MCP 2551) to node 2 and 3.This CAN transceiver would be used for TTL/CMOS to CAN Voltage levels and vice versa. The temperature sensor will resemble the engine or car cabin temperature. When the temperature data goes above set value the user get indication by red LED on Node 2 to take necessary action. Node -2 will consist of two LED which would indicate the status of the left indication and right indication done by manually and send the CAN message related to this on Node 3.And LCD display which displays the temperature value received from node 1.The node 3 would be connected PC and display the same temperature value. The PC communication is through RS232. Figure .5 shows the complete main board of the system consisting of three nodes.
Figure 5.Complete main board
III.
Three nodes has been implemented, the temperature data sent by node 1 will be extracted at node 2 and node 3. The LCD display output and the screen output at node 2 and node 3 respectively are shown in Figure 6. When this temperature exceeds beyond the threshold value the red LED on node 2 will glow and then user can take necessary action. Means that using the extracted CAN message us can make reliable communication between ECUs without wiring harness. This information will guide the user to take necessary action.
Figure 6.LCD display output at node2 and screen output at node 3
464
Vol. 3, Issue 1, pp. 460-465
IV.
CONCLUSIONS
This paper presents our project that successfully extract CAN messages from automobile ECUs. It presents the details of building both hardware and software that interface and communicate directly with CAN network embedded in the automobile. This extracted CAN Messages can be used for communication among automobile ECUs in order to take necessary action and this data will helps user to keep current status of automobile. Use of CAN to network controllers, actuators, sensors, and transducers we can reduce design time, connection costs because of lighter, smaller cables and connectors and improve reliability due to fewer connections so wiring harness reduces too. In the future works, we will see towards the prospects of integrating the system with an embedded info-security system for vehicular networks and forward CAN data for reliable car communications. And with the advent of Vehicular Ad-Hoc Network that allows vehicles within reasonable proximity to connect wirelessly to one another via an ad-hoc manner, vehicles are able to share information from extracted CAN messages about their current States to other.
REFERENCES
[1]. Heffernan, D., & Leen, G. (2008, September). ICT based research at Limerick contributes to automotive 'drive-by-wire' technology. Available:http://www.irishscientist.ie/2002/contents.asp?contentxml=02p237b.xml&contentxsl=is02pa ges.xsl [2]. Vasilash, G. S. (2005, May). Hardware-in-the-loop: assuring performance & quality by combining the real & the virtual. Testing. Automotive Design & Production. Gardner Publications, Inc. Available: http://findarticles.comlp/articles/mimOKJIIis /ai nl3784131 [3]. Texas Instruments. 2008, July). Introduction to the Controller Area Network (CAN). Available: http://focus.tij.co.jp/jp/lit/an/sloaIOla/sloaIOla.pdf [4]. CAN in Automation (CiA). Controller Area Network (CAN). Available: http://www.can-cia.orgl [5]. Robert Bosch Gmbh. Controller Area Network (CAN). Available: http://www.semiconductors.bosch.de/en/20/can/index.asp [6]. LM 35 data sheet Available: http://www.national.com/ds/LM/LM35.pdf [7]. PIC 18F4680 data sheet Available: http://www.datasheetsite.com/datasheet/PIC18F4680 [8]. MCP2551 High-speed CAN Transceiver Data sheet, Microchip Technology Inc. 2003.
Authors Biography Tejal Farkande: Student of Master of Engineering at Jawaharlal Neharu Engineering College, Aurangabad (India). She has obtained Bachelors Degree in 2008.Now working as lecturer in Sri. Venkateshwara College of Engineering, Bangalore, India.
Sunil Pawar: is working as Associate Professor in Electronics and Telecommunication Engineering Dept. of Jawaharlal Nehru Engineering College, Aurangabad, Maharashtra State, India. He has about 18 years of teaching experience. Now, he is working on Network Intrusion detection using Genetic Algorithm. He has about 30 research papers on his credit. His area of interest is network security and its related aspects.
465
Vol. 3, Issue 1, pp. 460-465
COMPARATIVE STUDY OF EXPERIMENTAL AND THEORETICAL LOAD CARRYING CAPACITY OF STONE COLUMN WITH AND WITHOUT ENCASEMENT OF GEOSYNTHETICS
Kameshwar Rao Tallapragada & Golait Y. S.
Professor, Civil Engineering Department, Shri Ramdeobaba College of Engineering and Management, Katol Road, Nagpur, India
ABSTRACT
The techniques for soil improvement have been changing and during the last three decades the concept of soil improvement by reinforcing it with tension resistant elements in various forms has received the attention. The reinforcement can be provided in vertical as well as in horizontal direction. The concept of introducing Geosynthetic Encased Column is found worth considering in Improvement of Load carrying capacity of soft clay ground intended to be treated by installation of Stone Column. In the present study a simple theoretical analysis is attempted to analyze a modified stone column treated ground and developed expressions to calculate the Ultimate Load Carrying capacities of Stone Column with and without encasement of geosynthetics. The correctness of these expressions is ratified by comparing the values obtained from the expressions with the observed results from tests on modeled stone column treated soil on a limited scale with and without encasement of geosynthetics.
KEYWORDS:
Soil improvement, Stone column, Failure load, Geosynthetics, Encasement.
I.
INTRODUCTION
Nowadays there are several techniques for soil improvement. Out of these techniques available for improving the weak strata, stone columns have been used to a large extent for several applications. The concept of soil improvement by reinforcing it with tension resistant elements in various forms has received the attention. This reinforcement can be provided in Vertical as well as in horizontal direction. In the present study modifications to existing stone column technique are proposed. The modification is in the form of introducing marginally small percentage of Sand, Stone Dust and lime to the conventional granular material in addition to this the Encasement of Geosynthetic is suggested for enhancing the load carrying capacity of the Stone Column in treated ground. In the conventional stone Column, the Bearing Capacity of the treated ground is mainly derived from the passive resistance offered by the surrounding soil due to lateral bulging of Stone Column material under axial load. Thus, logically if the Stone Column material is in the form adequately granular material, with small quantity of Lime, sand and Stone Dust, it is obvious that the effectiveness of the Stone Column treated ground is going to be enhanced. Also the introduction of Geosynthetic Encasement results in further enhancement of Load Carrying Capacity of Stone Column treated ground. In order to improve the stiffness of the column material, it is necessary to provide some adhesion in the material matrix. This will prevent easy migration of particles readily outwards. Hence mixing of Sand and Stone Dust with combination of Lime of very insignificant quantity supplemented by filler material is used to induce certain bonding. The concept of Load transfer mechanism as it occurs in the pile is also considered for simplifying the analytical approach.
466
Vol. 3, Issue 1, pp. 466-476
With the aforesaid considerations in view, the present study has been planned and carried out to develop the idealized load transfer mechanism. A simple theoretical analysis is attempted to analyze firstly, with a unit cell for untreated soft clay, secondly, unit cell for straight shafted stone column without encasement of geosynthetic, lastly, unit cell for straight shafted stone column with encasement of geosynthetic. Then mathematical expressions are developed for computing the ultimate load carrying capacities of stone column. The correctness of these expressions is ratified by comparing the values obtained from the expressions with the observed results from tests on modeled stone column treated soil on a limited scale. Different parameters length to diameter ratio rl (L/d), shear strength of the surrounding soil and, the area replacement ratio ra (a/A) and others were studied to show their effect on the bearing improvement and settlement reduction of the stone column.
II.
LITERATURE REVIEW
Soft clays extending over large depths, stone columns could be the ideal choice from the point of view of technical feasibility and suitability. Improvement of strength and compressibility characteristics of soft and /or loose soil or heterogeneous fill by installation of stone column as load bearing member has been identified as an effective means of ground improvement technique. The beneficial effects of installation in weak or difficult subsoil deposits is manifested in the form of increased load carrying capacity, significant reduction in total and differential settlements, minimization of post construction settlements. Ranjan & Rao [19][20] utilized the analogy of expansion of cylindrical cavity in a homogeneous, infinite soil mass, a simple method of estimating the ultimate capacity of these piles both plane skirted has been developed. The analysis has been extended to pile group also. Mokashi et al [17] found that the passive restrained on the peripheral granular piles can also be increased by providing surcharge around the foundation. di Prisco, C et al [5] studied the mechanical behavior of sand columns reinforced by means of geosynthetics by performing small scale tests on single elements. Kameshwar Rao Tallapragada et al [11]fount that the percentage reduction in settlement due to geosynthetic encasement decreased from 11.88% to 41.47% and the contribution of load carrying capacity due to geosynthetic encasement increased from 21.62% to 45.0% with smaller diameter and shorter length to larger diameter and longer length of stone column respectively. Review of literature presented indicates, considerable attempts have been made to evolve a rational analysis and design procedure but most of them relied upon the basis that the Stone Column derives its capacity through the passive resistance offered by the surrounding soil. While many attempts were made to enhance the capacity by externally supporting the Stone Column in top portion or tried to improve the stiffness of the column material by additions. However some of them posed cost impediments, some constructional difficulties. However it is kept in view that, in addition to improve the Load Carrying Capacity, reduction in settlements of soft clays; the Stone Column serves the purpose of enhancement of time rate to consolidation. This functionality of Stone Column is most important as this can considerably reduce long term post construction settlements, which is a natural occurrence in soft clayey soils.
III.
MATHEMATICAL FORMULATION COLUMN
FOR
LOAD CAPACITY
OF
STONE
3.1 Basic Assumptions

i) ii) iii) iv) An approach of Static load equilibrium holds well. Unit cell concept is valid for estimating the performance of large layout of stone column. The soil to be treated is soft saturated clay. The effective length of the stone column for development of shaft surface resistance and the tip resistance is pertaining to length ratio (rL) of 23. v) The reduction in the ultimate tip bearing is assumed to vary linearly from zero percent at depth of 3ds to 100 percent at depth of 23ds. vi) Full mobilization of shaft surface resistance is assumed over the entire length of column (Le) and Le = 23ds.
467
Vol. 3, Issue 1, pp. 466-476
Zone of depth of 3ds is the zone of possible bulging. On reaching the failure of column under axial action, there cannot be possibility of bulging thereafter. If column of length 3ds is constructed, it will act as a pile with full development of ultimate bearing resistance. Hence bearing capacity reduction factor rd is zero at this depth. In view of semi rigid nature of column material, and considering the vertical downward displacement of column, it is assumed that this vertical movement will cease to exist at a depth of 20ds below the 3ds depth level. At this level there will be no end bearing. Thus the value of rd = 100%. It can be expressed as rd = 0.05 (z/ds) - 0.15 . Eqn. (1)
3.2 Notations used in the analysis

ds = Diameter of Column Shaft L = Length of Stone Column a = Cross sectional area of column at ground surface A = Area of Unit Cell Z = Any depth below surface of ground Du = Diameter of Unit Cell Cu = Undrained cohesion of untreated soil Cu = Undrained Cohesion for local shear ra = Area of Replacement ratio (a/A) rL = Length Ratio (L/ds) rd = bearing capacity reduction factor at any depth z Qf = Load Capacity at failure of Unit Cell qf = Failure Load carrying Capacity of Unit Cell (qf/A) Nc = Bearing Capacity Factor for circular rigid load base untreated soft clay Nct = Bearing Capacity Factor for Stone Column treated soft clay Fb = Bearing Capacity Improvement Factor. Rt Column tip resistance. Rs = Soil resistance offered by tributary soil under (A-a) area at local failure of soft clay Rf = Shaft resistance over length L of stone column Rg1 = Resistance developed between Stone Column material and geosynthetic Rg2 = Resistance developed between geosynthetic and surrounding soil.
3.3 Unit Cell Analysis For Untreated Soft Clay

A Rigid Circular base of diameter Du (i.e. diameter of unit cell with area A) when loaded will cause local shear failure in soil. Shear Failure in soil occur up to a depth of Du below ground surface. This case is equivalent to rigid circular surface footing resting on soft clay for which Ultimate bearing capacity is expressed as, qu = cu * Nc . Eqn. (2) This equation is modified for circular area and local shear failure. Thus using the terminology considered in stone column system, qf = 1.2 * [2/3(Cu)] *Nc = [1.2 * (2/3)*0.9]Cu * Nc = 0.72 Cu Nc ....Eqn. (3) Different analysis viz. Prandtl, Terzaghi, Mayerhoff, Skempton have recommended different values Nc for = 0 case. In order to give appropriate weightage to all such values, the average of four theories is considered in the present analysis. Thus, Nc = 5.61 is decided as Bearing capacity factor. Hence Eqn.(3) of failure load carrying capacity of untreated clay becomes, qf = 0.72 * 5.61 Cu = 4.04 Cu ... Eqn. (4) Nc = (qf / Cu) = 4.04 ... Eqn. (5) The failure load on the unit cell for untreated ground is obtained multiplying the failure load capacity with the bearing area. Qf = 4.04 Cu A .. Eqn. (6)
3.4 Unit Cell Analysis For Straight Shafted Stone Column Without Geosynthetic
Fig.1 shows various resistances offered by the system under superimposed loads a single unit cell provided with a straight shafted stone column.
468
Vol. 3, Issue 1, pp. 466-476
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 Du Q
Rs
Rf
Rt
FIG. 1 : static load equilibrium for single unit cell of straight shafted stone column without geosynthetic
Static load equilibrium for this case is given by: Qf = RS + Rf + Rt -------- Eqn. (7) Now, Rs = [1.2*(2/3)*0.9* Nc] Cu* (A-a)] = 4.0392 Cu (A-a) = 4.04 Cu (A-a) ---- Eqn. (8) ----- Eqn. (9) Rf = ( *Cu ) ( *ds * L) = ( ds L) Cu For developing equation for Rt, the Bearing Capacity, reduction factor for column tip level is required to be determined, hence, substituting z = L in Eqn.(1) We get rdt for tip as: rdt = [0.05 (L/ ds) - 0.15] Hence, mobilized tip resistance, Rt Rt =(9Cu)(1- rdt)a =[9(1- 0.05(L/ds)+0.15)] Cu a ---- Eqn. (10) = [10.35 - 0.45(L/ ds)]Cu a Therefore, Qf = [4.04(A-a) Cu] + [( ds L) Cu] + [(10.35-0.45(L/ds)] Cu a --- Eqn. (11) Thus by dividing Qf with area of unit cell A, we get expression for failure load carrying capacity qf as under, qf = Q f / A qf = [4.04(A-a)/A] Cu + [( ds L)/A] Cu + [(10.35-0.45(L/ ds ) (a/A))] Cu ---- Eqn. (12) Incorporating the following non dimensional Parameters ra, rL in Eqn.12 L= rL ds ra = (a/A) and rL = (L/ds), ds/A = (ds/a) * (a/A) = [(ds 4 ra)/ ( ds2)] = (4 ra)/ ( ds) qf = [4.04(1-ra )] Cu + [( 4ra / ds) rL ds] Cu +[(10.35-0.45rL) ra] Cu qf = [4.04(1-ra )] Cu + [(4 ra rL] Cu +[(10.35-0.45rL) ra ] Cu qf = Cu {[4.04(1-ra )] + [(4 ra rL] +[(10.35-0.45rL) ra ] } ----Eqn. (13) Now the failure load carrying capacity of soft clay treated material with straight shafted stone column can be expressed as qf = Cu Nct ----- Eqn. (14) Where Nct = Bearing Capacity Factor of treated soft clay with straight shafted column system. From Eqn. 12 & Eqn. 13 Nct = [4.04(1-ra)]+[(4 ra rL]+[(10.35-0.45rL) ra ] ----- Eqn. (15)
3.5 Bearing Capacity Improvement Factor (Fb)
469
Vol. 3, Issue 1, pp. 466-476
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 For identifying the extent of improvement occurring due to the treatments given by means of installation of stone column, the Bearing Capacity Improvement Factor (Fb) came into existence. The bearing capacity factor, Fb = (Nct / Nc) ------ Eqn. (16) 3.6 Unit Cell Analysis For Straight Shafted Stone Column With Encasement Of Geosynthetic.
Fig.2 shows various resistances offered by the system under superimposed loads a single unit cell provided with a straight shafted stone column.
RS Rg2 Rg1 Rt
FIG 2 : Static Load Equilibrium For Single Unit Cell of Straight Shafted Stone Column With Geosynthetic
Static load equilibrium for this case is given by: Qf = Rs + Rt + Rg1 + Rg2 ----- Eqn. (17) qf = Cu {[4.04(1- ra )] +[(10.35-0.45rL) ra ] }+ Rg1 + Rg2 ----- Eqn. (18) Where, Rg1 = (4 1 ra rL) Cu & Rg2 = (4 2 ra rL) Cu 1 = Adhesion Factor of Stone Column Material and Geosynthetic = (Cug1/ Cu) 2 = Adhesion Factor of Geosynthetic and Surrounding Soil =(Cug2/ Cu) Cug1 =Undrained Cohesion of Stone Column Material and Geosynthetic Cug2 = Undrained Cohesion of Geosynthetic and Surrounding Soil. Hence Eqn.16 is modified as, qf = {Cu{[4.04(1-ra)]+[(10.35-0.45rL) ra]}}+{(4 1 ra rL)Cu+ (4 2 ra rL)Cu} ----- Eqn. (19) Nct = [4.04(1-ra)]+[(10.35-0.45rL) ra]+(4 1 ra rL)+ (4 2 ra rL) ----- Eqn. (20)
IV.
PARAMETRIC STUDY
The purpose is to study the effect of various parameters involved in the analytical treatment of systems considered in optimizing the benefits to be derived for stone column of soft clay treated soil. In addition to establishment of the effect of their variations on the performance of the system, they serve as general guiding tool. With the aforesaid considerations in view, the parametric study is undertaken for the different cases.
For a given soil state and its condition, the failure load carrying capacity depends on the non dimensional bearing capacity factor for treated ground. The non dimensional factor Nct is given by, Nct = [4.04(1-ra)] + [(4 ra rL] + [(10.35-0.45rL) ra] ----- Eqn. (21) 470 Vol. 3, Issue 1, pp. 466-476
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 From above Eqn. 20 it can be seen that the magnitude of Nct depends on Area Replacement Ratio (ra) and Length ratio (rL). Hence, Nct = f (ra, rL) Therefore the parametric study is carried out by considering the different values of ra & rL and considering possible permutations and combinations. The values of Area Replacement Ratio considered for this purpose are 0.05, 0.1, 0.15, 0.2, 0.25, and 0.3. Considering floating type of stone column and from practical installation consideration, the length ratios considered are, 10, 12, 14, 16, 18 and 20. The effect of these values of ra & rL on the non dimensional coefficients Nct and Fb are presented in Table-1, Table-2 and Table-3.
Table-1: theoretical load carrying capacity, Q (Kg) without encasement of geosynthetic Length Dia. Of of stone stone Nct= qf Q ra rL Rs Rf Rt column column (qf/Cu) (kg/cm2) (kg) (L) (ds) 30.0 2.5 0.062 12 3.79 2.62 0.31 6.72 1.24 97.57 30.0 2.0 0.034 15 3.90 1.80 0.12 5.82 1.08 84.57 30.0 1.5 0.022 20 3.95 1.55 0.03 5.53 1.02 80.34 22.5 2.5 0.062 9 3.79 1.96 0.39 6.14 1.14 89.28 22.5 2.0 0.034 11.25 3.90 1.35 0.18 5.43 1.00 78.88 22.5 1.5 0.022 15 3.95 1.16 0.08 5.19 0.96 75.44 15.0 2.5 0.062 6 3.79 1.31 0.47 5.57 1.03 80.98 15.0 2.0 0.034 7.5 3.90 0.90 0.24 5.04 0.93 73.19 15.0 1.5 0.022 10 3.95 0.77 0.13 4.85 0.90 70.53 Table-2: Theoretical load carrying capacity, Q (Kg) with encasement of geosynthetic Dia. Of Rf stone Nct = qf Q rL Rs Rt ra (Rs+Rf+Rt) (kg/cm2) (kg) column Rg1 Rg2 (ds) 2.5 0.062 12 3.79 2.77 2.83 0.31 9.69 1.84 144.62 2.0 0.034 15 3.90 1.90 1.94 0.12 7.86 1.49 117.30 1.5 0.022 20 3.95 1.64 1.67 0.03 7.29 1.39 108.78 2.5 0.062 9 3.79 2.08 2.12 0.39 8.38 1.59 125.00 2.0 0.034 11.25 3.90 1.42 1.45 0.18 6.96 1.32 103.84 1.5 0.022 15 3.95 1.23 1.25 0.08 6.51 1.24 97.17 2.5 0.062 6 3.79 1.38 1.41 0.47 7.06 1.34 105.37 2.0 0.034 7.5 3.90 0.95 0.97 0.24 6.06 1.15 90.39 1.5 0.022 10 3.95 0.82 0.84 0.13 5.73 1.09 85.57 Table-3: Direct Shear Test With Geosynthetic Material B.C SOIL and Geosynthetic Stone column material and Geosynthetic Cohesion Cu (Kg/cm2) 0.2 2 Friction angle () 55 60
Fb = (Nct/Nc) 1.66 1.44 1.37 1.52 1.34 1.29 1.38 1.25 1.20
Length of stone column (L) 30.0 30.0 30.0 22.5 22.5 22.5 15.0 15.0 15.0
Fb= (Nct/Nc) 2.40 1.95 1.80 2.07 1.72 1.61 1.75 1.50 1.42
471
Vol. 3, Issue 1, pp. 466-476
V.
EXPERIMENTAL OBSERVATIONS
With the help of recorded load - settlement observations obtained while conducted the tests, curves were obtained for depicting the behavior of the unit cell under superimposed loads. The load settlement curves are developed for all the considered cases of: Untreated Soil Treated Soil : Straight Shafted stone column without encasement of geotextile Straight Shafted stone column with encasement of geotextile Thus the introduction of geotextile confirms that the bulging phenomenon is the only governing criteria for failure of stone column and geotextile plays a vital role in its resistance to failure. The increased frictional resistance increases the load carrying capacity. The failure load is obtained from the curves of load-settlement in the following manner. Tangents are drawn from the approximated initial straight line portion and from final straight portion of continuous yielding. The abscissa of point of intersection will represent the Experimental failure load. This is conventional and universally accepted method for evaluation of failure load for a continuously yielding type of plastic material. A sample of load-settlement curves are shown in Figure 5 and Figure 6 for the case of stone column having length L = 15cm and diameter d = 1.5cm and failure load is obtained. Failure loads were obtained for all the cases of study and the findings are presented in Table-4 and Table-5. The theoretical failure loads for all the cases which were computed and shown in the Table-1and Table-2 are compared with experimental failure load in the Table-4 and Table-5.
Table-4: VARIATION OF LOAD CARRYING CAPACITY, Q (Kg) WITHOUT ENCASEMENT OF GEOSYNTHETIC Dia. of Length of Load Q Load Q stone stone column (kg) (kg) % Error column (L) (Theoretical) (Experimental) (ds) 30.0 2.5 97.57 100.00 -2.49 30.0 2.0 84.57 92.00 -8.79 30.0 1.5 80.34 86.00 -7.05 22.5 2.5 89.28 94.00 -5.29 22.5 2.0 78.88 85.00 -7.76 22.5 1.5 75.44 80.00 -6.04 15.0 2.5 80.98 85.00 -4.96 15.0 2.0 73.19 75.00 -2.47 15.0 1.5 70.53 74.00 -4.92
472
Vol. 3, Issue 1, pp. 466-476
Table-5: VARIATION OF LOAD CARRYING CAPACITY, Q (Kg) WITH ENCASEMENT OF GEOSYNTHETIC Dia. of Length of Load Q Load Q stone stone column (kg) (kg) column (L) (Theoretical) (Experimental) (ds) 30.0 2.5 144.62 148.50 30.0 2.0 117.30 125.00 30.0 1.5 108.78 110.00 22.5 2.5 125.00 130.00 22.5 2.0 103.84 110.00 22.5 1.5 97.17 100.00 15.0 2.5 105.37 110.00 15.0 2.0 90.39 95.00 15.0 1.5 85.57 90.00
% Error -2.68 -6.56 -1.12 -4.00 -5.93 -2.91 -4.39 -5.10 -5.18
Table-6: THEORETICAL STRENGTH CONTRIBUTION OF GEOSYNTHETIC IN THE STONE COLUMN ENCASEMENT Strength Q (kg) Length % Dia. of Contribution Stone Column Material of Contribution stone of of Strength stone column Geosynthetic Without With column due to (ds) (Kg) Geosynthetic Geosynthetic (L) Geosynthetic 30.0 30.0 30.0 22.5 22.5 22.5 15.0 15.0 15.0 2.5 2.0 1.5 2.5 2.0 1.5 2.5 2.0 1.5 97.57 84.57 80.34 89.28 78.88 75.44 80.98 73.19 70.53 144.62 117.30 108.78 125.00 103.84 97.17 105.37 90.39 85.57 47.05 32.73 28.44 35.72 24.96 21.74 24.39 17.20 15.04 48.22 38.70 35.39 40.01 31.65 28.81 30.12 23.50 21.32
Table-7: EXPERIMENTAL STRENGTH CONTRIBUTION OF GEOSYNTHETIC IN THE STONE COLUMN ENCASEMENT Length Strength % Q (kg) Dia. of of Contribution Contribution Stone Column Material stone stone of of Strength column column Geosynthetic due to Without With (ds) (L) (Kg) Geosynthetic Geosynthetic Geosynthetic 30.0 30.0 30.0 22.5 22.5 22.5 15.0 15.0 15.0 2.5 2.0 1.5 2.5 2.0 1.5 2.5 2.0 1.5 100 92 86 94 85 80 85 75 74 145 125 110 130 110 100 110 95 90 45 33 24 36 25 20 25 20 16 45.00 35.87 27.91 38.30 29.41 25.00 29.41 26.67 21.62
473
Vol. 3, Issue 1, pp. 466-476
474
Vol. 3, Issue 1, pp. 466-476
VI.
DISCUSSION
By introducing Lime and Sand in the material Matrix for the column, the Rigidity of the Stone Column can be very easily improved. The semi rigid nature of Stone Column resulted in development of additional resistances like, Shaft Resistance and Tip Resistance in addition to the surface Resistance. As compared to Conventional Stone Column, the Load Carrying Capacity of Stone Column by Treated material can be enhanced due to the development of these additional resistances. Additional resistance can be created by providing the Encasement of Geosynthetic to the Stone Column. The dual benefit of Treated material of Stone Column and Encasement of Geosynthetic are useful in deriving additional load carrying capacity as compared to Conventional Stone Column. Experimental tests showed how the confining effect given by the geosynthetic improves the Load Carrying Capacity of Stone Column. The analysis for all cases of treated, untreated Stone Column Material with and without Encasement of Geosynthetic were carried out based on Unit Cell Concept.
VII.
CONCLUSION
Load carrying strength of stone column improved by 20% to 66% and 42% to 124% because of treated material without and with encasement respectively as compared to untreated soil. The percentage difference in theoretical and experimental load carrying capacity of stone column without and with encasement respectively, varied from 2.47% to 8.79% and 1.12% to 6.56%. Results for the Experimental Investigations were found in close agreement with the analytical predictions. The theoretical values are lesser when compared to experimental values. For design of stone column the theoretical expressions can be used by multiplying them with 1.1. From the parametric study it can be concluded that with very low area replacement ratios and length ratios, considerable improvement in the load carrying capacity can be achieved. The parametric studies also reveals longer lengths of columns can be replaced by shorter length of geotextile encased Stone column.
REFERENCES
[1]. Ambily A.P., Gandhi S.R. (2004), Experimental and Theoretical Evaluation of Stone Column in Soft Clay, ICGGE, pp 201-206. [2]. Barksdale R D and Bachus R. C., Design and Construction of Stone Column, Vol-I (Dec.- 1983). [3]. Black,A.J., Shivkumar,M.R., Madhav, M.R. & Hamill,G.A.(2007), Reinforced Stone Column In Weak Deposits : Laboratory Model Study, Journal Of Geotechnical And Geo-environmental Engineering , Vol.133,pp.1154-1161. [4]. DaeYoung Lee., Chungsik Yoo., Sisam Park., Suntae Jung. (2008), Field load Tests of Geogrid Encased stone columns in soft ground, Proceedings of Eighteenth International Offshore and Polar Engineering Conference, Vancouver, BC, Canada, July 6-11.
475
Vol. 3, Issue 1, pp. 466-476
[5]. di Prisco,C., Gaslli, A., Cantarelli, E.,Bongiorno, D. Geo Reinforced Sand Columns: small Scale Experimental Tests and Theoretical modeling [6]. Geopier Foundation Co. INC, Technical Paper, Case Histories of Rammed Aggregate Pier Soil Reinforcement Construction over Peat and Highly Organic Soil. [7]. Golait Y. S. (1998), A practical approach of Composite system for problems of black cotton soil as subgrade material in road construction, Proceeding of state of the art workshop on Expansive soil technology (EST-98), March-1998- RKNEC, Nagpur. [8]. Golait Y. S., V. Satyanarayana., S.S.V. Raju. (2009), Concept of Under Reamed Cemented Stone Columns for Soft Clay Ground Improvement, Indian Geotechnical Conference (GEOTIDE), Guntur, India, December 17-19, pp 356-360. [9]. Guetif,Z., Bouassida, M., Debats, J.M. (2007), Improved Soft Clay Characteristics Due To Stone Column Installation, Computers And Geotechniques, Vol.34, pp.104-111. [10]. Jha J.N. (2007), Effect of Vertical Reinforcement on Bearing Capacity of Footing on Sand, Indian Geotechnical Journal, 37(1), 64-78. [11]. Kameshwar Rao Tallapragada., Golait Y.S and Ashwini S Zade (2011), Improvement of Bearing Capacity of Soft Soil Using Stone Column with and without Encasement of Geosynthetics, by at International Journal of Science and Advanced Technology (ISSN 2221 8386), Volume I, No. 7 September 2011, pp 50-59. [12]. Kameshwar Rao Tallapragada and Golait Y.S (2010), Use Of Geogrid As Horizontal Reinforcement In The Swelling Subgrades For Pavement Construction, International Conference DRT 2010, National Institute Of Technology, Rourkela, October 8-10, 2010. [13]. Kemfert H.G., Ground Improvement methods with special emphasis on column type technique, International workshop on Geomechanics of soft soil theory and practice. [14]. Malarvizhi N., Ilamparuthi K. (2004), Load versus Settlement of Clay Bed stabilized with Stone and Reinforced Stone Column, 3rd Asian Reg. Conf. on Geosynthetic, pp.322-329. [15]. Mitra, S. and Chathpadhyay, B.C. (1999). Stone Columns and Design Limitations. Proc. of Indian Geotechnical Conference held at Culcutta , 201 205. [16]. Mohammed Y Fattah., Qutaiba G Majaed. (2009), Behaviour of Encased Floating stone Columns, Engineering & Technological Journal, Vol. 27. [17]. Mokashi, S.L., Paliwal, S.T. and Bapaye, D.R. (1976), Use of stone columns for strengthening soft foundation clay, Proc. Central Board of Irrigation and Power, 45th Annual Research Session, Vol. III-Soil and Concrete 8-11 June. [18]. Radhey S. Sharma, Bhyravajjula R. Phanikumar (2005), Laboratory Study of Heave Behavior of Expansive Clay Reinforced with Geopiles, Journal of Geotechnical and Geoenvironmental Engineering, April. [19]. Ranjan G., and Rao, B. Govind (1983), Skirted granular piles for ground improvement, Proc. VIII European Conference on Soil Mechanics and Foundation Engineering, Helsinki. [20]. Ranjan G., and Rao, B. Govind (1987), Soil improvement by granular piles, Journal of Indian Geotechnical Society, New Delhi, Vol. 17, No. 2.
Authors
Kameshwar Rao Tallapragada, B.Tech (Civil Engg.), M.E. (Transportation Engg. and Management), M.B.A. 26 years teaching experience. Presently teaching to U.G. and P.G. students of Civil Engg. and Geotechnical Engineering. Guided 11 students of M.Tech (Geotechnical Engg.), 2 students of M.Tech (Structural Engg.). Published 1 paper in International Journal, 2 papers in International conferences and 10 papers in National conferences.
Golait Y S, B.E. (Civil Engg.), M.E. (Soil Engineering) (IITB), Ph.D. (Geotechnical Engg.) (IITB), 43 years of teaching. Presently teaching P.G. students of Geotechnical Engineering. Guided 12 students of M.Tech (Geotechnical Engg.), Published 2 papers in International Journal, 11 papers in International conferences and7 papers in National conferences. Recipient of many prestigious awards.
476
Vol. 3, Issue 1, pp. 466-476
A NOVEL FACE RECOGNITION APPROACH USING A MULTIMODAL FEATURE VECTOR

Jhilik Bhattacharya1, Nattami Sekhar Lakshmi Prabha2, Somajyoti Majumder3, Gautam Sanyal4
1,2,3
Central Mechanical Engineering Research Institute, Durgapur , West Bengal, India 4 National Institute of Technology, Durgapur, West Bengal, India
ABSTRACT
This paper proposes a face recognition technique using a composite framework consisting of a number of parameters for facial feature extraction and matching. Recognition of a known face often becomes an intricate task due to the presence of lighting variations, viewpoint changes, facial expression variations and presence of occlusions like spectacles or beard. Often the recognition system may be required to function in the presence of a very small training set; even a single training image in many situations. The feature extraction technique should thus have the potential to extract maximum information from the existing training set and also prove to be resilient in any varying conditions. The basic idea of the multimodal framework stems out from the fact that using an integrated framework can reduce the multiple false alarms or wrong identification obtained while depending on a single feature. The framework mainly aims in giving a computationally inexpensive, consistent and reliable detection rate which can be robust against any variations (such as pose, orientation) and environmental conditions (lighting, surroundings) and can be applied for tasks ranging from object matching to face, gesture recognition and medical imaging applications. Experimental results with a set of images from the PIE database demonstrate the effectiveness of the approach in giving a good recognition rate even in the presence of huge variations between the training and test set images. The framework is also tested with the plastic surgery face database where a post surgery face needs to be identified from a single pre-surgery training image .The paper also uses an efficient feature dimension reduction technique which improves the overall performance of the feature extractors in the framework.
KEYWORDS:
face recognition, feature extraction, gabor filter, wavelet, local binary pattern.
I.
INTRODUCTION
The need for secure, safe and accurate authentication system is increasing with the huge demand in online transactions. The biometric system such as face, iris, fingerprint, gait for example, has already replaced the existing manual inspection process and surveillance systems in many disciplines. Amongst all these biometrics, face is more attractive as it provides information such as identity, expression, gender, ethnicity and age of an individual. This led to the demand of a robust face recognition system that is resilient to the face lighting conditions and different variations like pose, facial expression or occlusions like spectacles or beard. Facial aging and plastic surgery [1], [2] are yet two other challenges that have not received substantial attention compared to other facial variations. Figure 1 displays an example of face images subject to lighting, pose or expression variations. An efficient recognition system should have the capability to match two faces even when they are subject to these variations. A lot of work has been carried out in this field over the years, each being an improvement over the other. One of the most common method used for face recognition is Principal Component Analysis (PCA) done by Turk and Pentland in 1991 [3] .The Linear Discriminant Analysis (LDA) [4] ,[5] ,[6] is an improvement over the PCA as it works by maximizing the between class scatter matrix and minimizing the within class scatter matrix instead of finding the variation among the entire data set, irrespective of class membership. Following this there are various other methods used for face recognition such as Independent Component Analysis (ICA)
477
Vol. 3, Issue 1, pp. 477-486
[7] , Neural Network [8], Geometric matching [9], Template matching [10], Hidden Markov model (HMM) [11], Support Vector Machine (SVM) [12] [13] [14].These techniques however gives a poor recognition rate in presence of huge illumination and pose variations. Their performance was further improved by using distinct image features instead of using the intensity images directly. Popular feature extraction techniques like Gabor wavelet , Active Appearance Models (AAM) [15], and Local Binary Pattern were used for this purpose. Shape, size, color, texture, position and orientation parameters form different discriminative features from a face image. LBP and Gabor wavelet features extract local shape and texture information while AAM provides better global shape information. Different color spaces were also used to obtain distinct facial features from various color planes [16].
Figure 1: Images of the same person at varying light, pose and expressions.
This paper uses a multimodal feature vector, which represents an object in a multidimensional feature space for feature extraction. The feature vector is basically a framework with a composite representation of various features like shape, size, color, texture, position and orientation .The basic idea stems out from the fact that if a large number of parameters are used instead of one then the detection rate increases considerably. Limitations of certain parameters in particular situations can then be conveniently addressed by the remaining set thus making the framework resilient to any environmental conditions. Failures of many face recognition algorithms, mainly due to view point variations, lighting conditions, occlusion and influence of other factors like facial expressions can also be successfully resolved by this framework. The framework uses four different features (CGABOR, LSBH, Edge components and Local Energy and Entropy ) to represent the face image .The CGabor and LSBH are variations of the Gabor and LBP operators and are used to extract the salient texture features, the edge component gives the global shape of the face pattern while the local energy-entropy feature provides unique local features. This paper is organized in the following manner. Section I discuss the basic background of the problem. Section II provides details about the feature extraction technique, the experimental results of recognition using the proposed technique and its comparison with other existing methods. Finally, discussion and conclusion is presented in section III.
II.
FEATURE EXTRACTION FRAMEWORK
The basic idea of the framework is to represent the total feature set x in terms of a multimodal probability distribution in a multidimensional feature space, mathematically expressed as
x = {x1, x2 ,...xn } (1) where xi (i=1...n) denote distinguishable features like color, texture, energy, shape, size etc. Some of these features may be orthogonal to the other. Any number of features may be used in the framework depending on the application requirements and available computational resources. For any object to be represented in this framework it should have more than one feature like color, texture, energy, shape etc. The classification of the object of interest (face for this particular case ) is done using the combined distance metrics using all the feature parameters. The combined distance metrics is calculated from the distance metrics of the individual feature parameters obtained using the mahalanobis distance classifier. The parameter estimation ( = {,2}) (required for calculating the mahalanobis distance between the distributions) of the sample's distribution is done using the
478
Vol. 3, Issue 1, pp. 477-486
maximum likelihood estimator (MLE) as it is known to give a consistent, convergent unbiased and minimum variance estimate for gaussian distributions, provided such an estimate can be determined for a particular sample. It can be proved that using a multimodal probability distribution , the distance metrics comb obtained by combining all the variables is less than the individual distance metrics 1,2,3 for each of the variables x1, x2, x3 (Lemma 1).It is known that for a multiparametric search space the combined deviation is given by equation 2.Lemma 1 can be proved easily from this identity using proof by contradiction.
comb
It is assumed that comb > 1. Thus
1 1 1 1
+ .....
n
1 1 1 1
(2)
+ .....
> 1 ,
>
+ .....
(3)
The above identity is wrong for all values of 1 as 1 is a square root of a squared term and hence is always positive. Thus the above assumption is wrong and it can be stated that comb < 1. comb is similarly less than 2, 3,. etc. Hence it is seen that comb is less than the standard deviations of all the variables and also decreases with the increase in the number of variables. The efficiency of the multiparametric method over other single parameter based methods is thus established. In the present work the framework includes texture, edge components, and energy-entropy features. The superiority of the framework over other conventional approaches is two-fold. Firstly it encapsulates all the distinct features of the object; secondly the classification approach allows the intraclass variations to a great extent. The feature extraction procedure is further explained in the subsequent subsections.
2.1. Texture
Among the various texture modeling techniques like Gray-Level Co-occurrence Matrices (GLCM's), Local Binary Pattern (LBP), Law's texture , autocorrelation based, primitive length based and edge feature based methods, the Gabor filters [17] and Local Binary Pattern [18] [19] are two notable texture analysis methods which are mostly used at present. Gabor filter extracts large amount of discriminating local features due to its ability to operate in selective scales and orientations thus capturing the spacial locality and quadrature phase relationships. Two types of texture descriptors named CGABOR and LSBH are used in the framework.
2.1.1. Compressed Gabor: CGABOR

Usually the gabor kernel . is constructed using five scales =0,1,2,3,4 and eight orientations =0,1.7 , hence resulting in a very high dimensional representation.
k , v 2 z 2
, v ( z) =
i
k , v
2 2
[e
ik , v
ze 2 ]
(4)
where k , = k e , = 2 , k = k max f , f = 2 , k max = 2 , = 8 , z = (r , c) and || || denotes the norm operation. Although PCA or LDA is applied for dimension reduction of the gabor convolved image O , ( z ) it still takes a huge time for computation. Also, there is a decrease in the
performance for ignoring some of the principal components. Thus it is essential to reduce the feature dimension in an optimal way without hampering the recognition accuracy. O , ( z ) = 1 [Im( z )] , v ( z )
(5)
479
Vol. 3, Issue 1, pp. 477-486
A possible alternative can be to reduce the image size by down sampling the image. In the present approach the gabor filter is applied on a compressed version Im(r , c) of the original image Im(r,c) .This is basically a low pass filtered image which can be obtained from a wavelet decomposed image as shown in Figure 2.A sampling technique is then used to select the maximum gabor coefficients from a 3 X 3 sliding window for each of the 40 (8 orientations X 5 scales) filtered images as shown in Figure 4 .Thus the dimension is almost reduced to more than 1/100th of the original size. For example if gabor filter is applied over a 150 X 130 image the total vector size prior to PCA will be 7,80,000 (150 X 130 X 40 ).However for the present case it will be 53040 (39 X 34 X 40 ) after applying gabor filter on the compressed image and 5893 after sampling.
Figure 2: The approximate coefficients of the wavelet decomposition gives the compressed image which is used to extract gabor features
The wavelet function used for compression is selected experimentally .The recognition accuracy increases with the increase in the number of components used but at the cost of computation. Thus the selection needs to be done by keeping a balance between the computational complexity and recognition accuracy. Table 1 shows a list of the number of components for a sampled gabor coefficient vector for each wavelet function at different levels. A computation up to 7000 components may be supported. Thus level 1 wavelet decomposition or level 2 wavelet decomposition for db6 and db8 are not used. A comparison between the second, third and fourth level decomposition using db2 and db4 wavelet as shown in Figure 3 shows that the second level wavelet decomposition gives a better result and is hence chosen. In this case, PCA is applied to the sampled vector O , ( z ) for
transforming the vector to the eigen space rather than reducing its dimensionality. It is observed that this approach not only performs faster than the conventional Gabor-PCA or Gabor-LDA,but also gives a steady hike in its recognition rate. The gabor feature vector is formed using the magnitude value of the gabor coefficients as it is seen to give better accuracy when compared to real,complex or angle value of the coefficients as shown in Table 2.The results of Figure 3 and Table 2 uses the PIE face database with a set of 980 images (35 different individuals with 28 images each).For all the cases 15 out of the 28 images were used for training. Fcgabor = O , ( z ) (6)
Table 1: Sampled Gabor dimension for different Wavelet functions at different levels using an image dimension of 150 X 130 pixels Wavelet Type Level 1 Level 2 Level 3 Level 4 db2 22293 5893 1680 533 db4 23573 6906 2346 933 db6 24888 8000 3111 1520 db8 26240 9173 3995 2248
480
Vol. 3, Issue 1, pp. 477-486
Figure 3: Comparison performance of gabor features using different wavelet functions .
Figure 4: The maximum coefficient is selected from each 3 X 3 sliding window for all the 40 gabor filtered images.
Table 2: Performance Comparisons using the complex, real, magnitude, log magnitude and angular value of the gabor coefficients Type No. of Recognized images Complex 907 Real 903 Mag 915 Log Mag 915 Angle 720
2.1.2. Local Sign Binary Histogram: LSBH

The LBP code of a given pixel fc is computed by finding its local difference with its neighbours fp and generating the binary code cs as shown in equation 7.The LSBH operator is a variation of the LBP
481
Vol. 3, Issue 1, pp. 477-486
operator where the histogram of the sign components computed for each 3 X 3 window of a block is updated as shown in equation 8 .From the equation it is clear that only the histogram bins which are a power of 2 are considered. Thus instead of a 255 component code for normal LBP, an 8 component code for each block n is obtained hence resulting in a much shorter feature vector. For example, for an image of size 150 X 130 pixels, LBP feature extraction with a block size of 12 X 12 pixels gives a vector size of approximately 30,000.The LSBH operator on the contrary gives a vector size of 1000 for the same case, thus reducing the computation complexity to a great extent.
cs =
2p sp
p =0
1 where s p = 0
( f p fc ) > 0 ( f p f c ) < 0
(7)
H n (2 p ) = H n (2 p ) + 2 ps , p = 0.....7 The final vector Flsbh is thus formed by concatenating the histogram Hn of all the b blocks .
Flsbh = [H1 H2 H 3 ... H b ]
(8)
(9)
2.2. Edge components

The edge features are obtained using a kirsch operator over the original image. Generally a kirsch operator gives the edge coefficients in eight directions by 8 consecutive 45 degree rotations of the kernel k1. Here the horizontal, vertical and diagonal edge elements are calculated by combining the 2 horizontal and 2 vertical components and the 4 diagonal components as shown in equation 10.The low pass filtered image and the components are shown in Figure 5 .
Figure 5: The lowpass filtered image along with the horizontal, vertical and diagonal edge components.
5 5 5 k1 = 3 0 3 3 3 3
Imn edge_comp = Im kn Im(z)edge = maxn=1:8 Im(z)n edge_comp flpf() = Im- Imedge
482
Vol. 3, Issue 1, pp. 477-486
fh() =
n =1,5
Imn edge_comp , fd() =
n = 3, 7
Imn edge_comp , fv() = fh() fv() fd()]
n = 2, 4,6,8
Imn edge_comp (10)
f()=[ flpf()
It is observed that using both the low pass filtered image along with the edge components gives a better recognition rate than the one achieved using only the edge components. To reduce the dimensionality the component matrix is sampled by selecting the maximum coefficient value in each 3 X 3 sliding window and the sampled vector f () is generated. It can be verified from Table 3 that the above mention procedure gives a better recognition rate than normal sampling. Feature size can be varied as required by altering the sliding window size or using a downsampled image to calculate the edge components.
Table 3: Comparison of performance using normal sampling and maximum selection
Normal Sampling Maximum Selection 897 913
Fedge= f ()
2.3. Local Energy and Entropy

Energy entropy parameters are rich in information and are often used for object pattern identification purposes. Cunjian chen [20] calculated the energy and entropy from the approximate coefficients of a wavelet decomposed image and used it for face recognition. In the present approach the face image is divided into equal block sizes as shown in Figure 6 and the energy and entropy for each block n is calculated as shown in equation 11. The performance of the recognition algorithm using the energy entropy feature depends on the block size. A performance variation with different block sizes is shown in Table 4 .It is observed that a block size of 15 X 15 gives the best performance when experimented on an image of 150 X 130 pixels. This observation was obtained using images of size 150 X 130 pixels from the PIE and ORL databases. The performance variations with the block size using the ORL database with a set of 80 images is given in Table 4.
255 255 energy n = P ( I ) 2 , entropy n = P ( I ) log 2 P ( I ) I =0 I =0
(11)
where P(I) denotes the probability density function of intensity level I. The final vector Fenergy-entropy is thus formed by concatenating the energy and entropy of all the b blocks . Fenergy-entropy =[energy1 entropy1 energy2 entropy2..... energyb entropyb]
Figure 6: Face partitioned in equal sized blocks and energy entropy vector is obtained by computing the energy and entropy parameter of each block. Table 4:Effect of block sizes on the recognition rate
Block Size 40 x 40 35 x 35 25 x 25 No. of Recognized images 65 66 68
483
Vol. 3, Issue 1, pp. 477-486
20 x 20 15 x 15 12 x 12 10 x 10 5x5 74 79 76 71 58
The total feature vector xj of a face image Im(r,c) can thus be represented by
xj ={F cgabor , F lsbh , F edge, F energy-entropy}

The mean vector of each feature is calculated using equation 13.
(12)
Each feature xi in the feature vector is further transformed in the eigen space using equations 13 to 16.
i = m=1 xij j
The covariance Ci of the feature i in the training set is calculated using the mean adjusted data i
(13)
i = xi - i
Ci =1/m i T i
(14)
The feature is then transformed to the eigen space using the eigen vectors i of the covariance matrix Ci as shown in equation 15.
i
= iT i T
(15)
For any test image the transformed vector in the eigen space is calculated from the feature vector xk using equation 16.
k i
= iT (ik )T
where ik = xik - i
(16)
The distance matrix i is calculated by finding the distance between each training image principal component vector ij and test image principal component vector ik . comb is then computed using equation 2 and the best match is mapped to the jth image such that the minimum value of comb corresponds to combj.
2.4. Experimental Results

The performance analysis of the feature extraction technique is carried out using the PIE face database with a set of 980 images (35 different individuals with 28 images each). For all the cases 15 out of the 28 images were used for training. The face images were normalized to a dimension of 150 x 130 pixels before use. The performance comparison of the proposed technique and conventional gaborpca, lbp-pca and wavelet decomposition techniques is presented in Table 5.The gabor features are computed using the image compressed with db4, level 2 wavelets while the edge features are computed using a downsampled image. Block size for energy entropy parameter is taken as 15 as these give the best results as seen from Figure 3 and Table 4.
Table 5: Recognition results with PIE face database
Methods LBP Gabor Wavelet Feature Vector Test + Training (in %) 88.87 80.41 88.27 95.92 Test (in %) 76.04 57.8 74.73 91.43
The results of applying the framework on the plastic surgery face database is given in Table 6.The database consists of a pre-surgery and post-surgery image of 250 individuals. It is seen that the framework gives a better performance when compared to the other existing approaches.
Table 6: Recognition results with Plastic surgery Face Database
Methods LBP Gabor Test (in %) 60 54
484
Vol. 3, Issue 1, pp. 477-486
Wavelet Feature Vector 48 72
III.
DISCUSSIONS AND CONCLUSIONS
The experimental results verify that the proposed feature matching technique gives a better recognition rate than the conventional gabor, lbp and wavelet feature extraction techniques. It is observed that the energy-entropy parameter and edge component (not commonly applied for face recognition applications) used in the framework gives a high recognition rate (about 88%) thus increasing the overall performance. It is also seen that applying gabor filter over a compressed image and using the local maximum coefficients is far more efficient in terms of both accuracy and computational complexity than the general gabor-PCA or gabor-LDA approach used so far as using the entire feature set of a compressed image (even 1/4th of the original size) gives a better recognition rate than considering a certain percentage of the feature set calculated from the original image. Similar technique of compression when applied before LBP feature extraction also shows considerable hike in performance. This performance can also be achieved using the LSBH operator, which gives a reduced feature size on the original image itself. As a matter of fact, it is noted that the LSBH operator outperforms the traditional LBP operator in terms of speed as well as accuracy. Work is being carried out to further test the performance of the feature extractor on facial age modelling.
REFERENCES
[1]. Unsang Park et al., `Àge-Invariant Face Recognition,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 32, No. 5, May 2010. [2]. Richa Singh et al., ``Plastic Surgery: A New Dimension to Face Recognition,'' IEEE Transactions on Information Forensics and Security, pp. 441-448, Vol. 5, No. 3, Sep. 2010.
[3]. M. Turk and A. Pentland, `Èigenfaces for recognition,'' Journal Cognitive Neuroscience, vol. 3, pp.
71-86, 1991. [4]. Martinez, A.M., Kak, A.C., ``PCA versus LDA,'' IEEE Transactions Pattern Analysis and Machine Intellgence Vol. 23 (2), pp. 228-233, 2001. [5]. Francis Galton, ``Personal identification and description,'' pp. 173-177, June 21, 1888. [6]. P. N. Belhumeur, J. P. Hespanha and D. J. Kriegman, Eigenfaces vs. Fisherfaces: Recognition Using Class Spesific Linear Projection, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 19, No. 7, July 1997. [7]. M. S. Bartlett, J. R. Movellan, and T. J. Sejnowski, Face Recognition by Independent Component Analysis, IEEE Transaction on Neural Networks, Vol 13, pp. 1450-1464, 2002. [8]. N. Jamil, S. Iqbal, N. Iqbal, ``Face Recognition Using Neural Networks,'' IEEE Proc. of INMIC technology for the 21st century, pp. 277-281, 2001. [9]. S. Tamura, H. Kawa, and H. Mitsumoto, ``Male/Female identification from 8x6 very low resolution face images by neural network,'' Pattern Recognition, vol. 29, pp. 331-335, 1996.
[10]. R. Bruneli and T. Poggio, ``Face recognition: features versus templates,'' IEEE Transactions Pattern Analysis and Machine Intelligence, vol. 15, pp. 1042-1052, 1993. [11]. F. Samaria and F. Fallside, ``Face identification and feature extraction using hidden markov models,'' Image Processing: Theory and Application, Elsevier, 1993. [12]. C. J. C. Burges, `À tutorial on support vector machines for pattern recognition,'' Data mining and knowledge discovery, Vol.2, pp. 121-167, 1998. [13]. P. J. Phillips, ``Support vector machines applied to face recognition,'' Advances in Neural Information Processing Systems 11, MIT Press, 1999. [14]. Guoqin Cui, Wen Gao, Feng Jiao, and Shiguang Shan, ``Face Recognition Based on Support Vector Method,'' Asian Conference on Computer Vision, January 2002.
485
Vol. 3, Issue 1, pp. 477-486
[15]. Cootes, T.F., Edwards, G.J., Taylor, C.J., `Àctive Appearance Models,'' Proc. European Conf. on Computer Vision, Vol. 2, pp. 484-498, Springers, 1998. [16]. Z. Liu and C. Liu Robust Face Recognition Using Color Information, Advances in Biometrics, Lecture Notes in Computer Science, 2009, Volume 5558/2009, 122-131 [17]. Chengjun Liu, Harry Wechsler, `À Gabor Feature Classifier for Face Recognition,'' Proceedings of ICCV, pp. 270-275, 2001. [18]. T. Ojala, M. Pietikinen, T. Menp, ``Multiresolution gray-scale and rotation invariant texture classification with local binary patterns,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, pp. 971-987, 2002.
[19]. T. Ahonen, A. Hadid, and M. Pietikinen, ``Face Recognition with Local Binary Patterns,'' Proc. Eighth European Conf. Computer Vision, pp. 469-481, 2004.
[20].
Cunjian.chen Jiashu.zhang, Wavelet Energy Entropy as a New Feature Extractor for Face Recognition, Fourth International Conference on Image and Graphics,IEEE ,2007.
Authors
J Bhattacharya is currently Senior Research Fellow, in Central Mechanical Engineering Research Institute (CMERI), Lab under Council of Scientific and Industrial Research (CSIR), Durgapur, West Bengal. currently pursuing her Ph.D. in Computer Vision from NIT, Durgapur. Her interest area includes Image Processing, Computer Vision, Pattern Recognition, Artificial Intelligence, Soft Computing and Robotics.
N S Lakshmiprabha is currently Scientist in Central Mechanical Engineering Research Institute (CMERI), Lab under Council of Scientific and Industrial Research (CSIR). Her interest area include Image Processing, Computer Vision, Pattern Recognition and Robotics.
S. Majumder is currently the Group Head and Chief Scientist, Surface Robotics Lab, Central Mechanical Engineering Research Institute (CMERI), Lab under Council of Scientific and Industrial Research (CSIR).His research areas include Sensor Fusion, Sensor and Signal Processing, Robotics, Decentralised Computing and Multi agent systems, Image Processing, Artificial Intelligence.
G.Sanyal is currently Professor and Dean of Computer Science & Engineering Department in NIT Durgapur. His research areas include computer architecture, Vision , Networking.
486
Vol. 3, Issue 1, pp. 477-486
INFLUENCE OF GEOFABRICS IN THE CONSTRUCTION OF PAVEMENTS ON EXPANSIVE CLAYEY SUBGRADES

A. V. Narasimha Rao1, D. Neeraja2
1
Professor, Department of Civil Engineering, S.V.U.College of Engineering,Tirupati 2 Assistant Professor(Sr) , Department of Civil Engineering, VIT University,Vellore
ABSTRACT
Design and construction of pavements on weak soils pose lot of problems due to its low load carrying capacity. Roads cause rut formation due to the high axle loads of vehicle on expansive clayey subgrades. There are many techniques to modify the properties of such soils. Among them, soil reinforcement with geofabrics is more promising in improving the load carrying capacity of expansive subgrades. In this paper, emphasis is given to investigate the feasibility of using abundantly available indigenous resources, coir and jute in road construction. In addition to the above mentioned natural fabrics, geogrid, a synthetic material is also considered. The influence of varied number of reinforcing layers on CBR values is investigated. CBR method is employed for the design of flexible pavements in most of the design offices even though it is empirical in nature, since the method is simple and convenient to apply and the CBR value of the subgrade indirectly gives the strength of subgrade soil. The results are analyzed effectively by introducing two non-dimensional factors namely, Effective Depth Ratio (EDR) and Strength Benefit Ratio (SBR). It is found that both CBR and SBR values increases with increase in number of reinforcing layers. Based on the favorable results obtained, it can be concluded that expansive soil can be successfully stabilized using geofabrics.
KEYWORDS: California Bearing Ratio, Effective Depth Ratio and Strength Benefit Ratio, Geofabrics, Expansive
clay
I.
INTRODUCTION
Design and construction of roads on expansive soils is a difficult task because of its low load carrying capacity, high compressibility, high swelling and shrinkage. In United States, damage caused by expansive clays exceeds the combined average annual damage from floods, hurricanes, earthquakes, and tornadoes [1]. Documented evidence of the problems associated with expansive clays is worldwide, which occurred in countries like United States, China, Australia, India, Canada, and regions in Europe [2]. Changes in the moisture content of clay soils are generally accompanied by volume changes. On moisture uptake, there is generally volume increase and moisture loss is accompanied by shrinkage [3]. The properties of these soils can be improved by stabilization with admixtures and geofabrics. But, in stabilization using admixtures technique, it is necessary to select proper stabilizer and it is very difficult to stabilize huge quantity of soil in the field. With the advent of synthetic fibres, soil reinforcement is found to be dynamic in improving load bearing capacity of soft soils. The geofabric reinforcement system prevents the failure of the pavement due to penetration of road crust into subgrade soil or intrusion of soft subgrade soils into base course and inadequate drainage of subbase and base course by acting as separator as well as drainer within the pavement apart from reinforcing function. The geofabrics include synthetic fibres like geogrid and natural fibres like jute and coir. The thickness of the pavement is inversely
487
Vol. 3, Issue 1, pp. 487-493
proportional to the California Bearing Ratio (CBR) of a subgrade. The increase in CBR with the introduction of geofabrics results in reduction in pavement thickness and it is obvious that increase in number of layers will yield higher CBR values which results in further reduction in pavement thickness. But, with increase in number of layers, the ultimate cost of the pavement will go up and hence the experimentation is carried out by restricting the number of layers to four only. The effect of reinforcing layers on the CBR values of the subgrade soil is studied herein.
II.
Soil
MATERIALS AND METHODS
The soil used in this study was obtained from Gajulamandyam near Tirupati. Disturbed but representative soil samples were collected from trial pits at a depth of about 2.0 m from ground level. The soil collected from the site was pulverized with wooden mallet to break lumps and then air-dried. Subsequently it was sieved through 2.36 mm IS sieve and then dried in an oven at 105oC for 24 hours .The soil falls under the CH category i.e., clay of high compressibility as per I.S Classification System [4]. The soil passing through I.S 425 sieve has very high Liquid Limit and Plasticity Index. Based on Differential Free Swell Index, Liquid Limit and Plasticity Index, the soil comes under the category of high degree of expansiveness. The properties of soil are presented in Table 1.
Coir
The coir fabric is obtained from door mats which are uniformly woven. Coir fabrics of 15 cm diameter circular pieces were used for the present study.
Jute
Jute is prepared from jute bags like sugar bags, rice bags etc, of 1.5 mm thick. 15cm diameter circular pieces are cut from the jute bag. A jute geofabric of 3mm thick is prepared by joining two pieces of jute bag.
Geogrid
Geogrid used in this study is a product of Netlon. This is immune to corrosive, chemical and biological attack in soils. Geogrid of 15 cm diameter circular pieces were used for the present study.
Table 1 Properties of Soil CHARACTERISTICS Specific gravity Particle Size distribution a) Gravel (%) b) Sand (%) c) Silt+Clay (%) Liquid limit (%) Plastic limit (%) Plasticity index (%) Differential Free Swell Index (%) Classification of soil Maximum dry unit weight (kN/m3) Optimum moisture content (%) Unconfined Compressive Strength (kN/ m2) Unsoaked CBR (%) VALUE 2.69 Nil 12 88 98 13 85 150 CH 17.99 14 145 6.84
488
Vol. 3, Issue 1, pp. 487-493
III.
TESTS CONDUCTED ON UNREINFORCED SOIL
The compaction tests, unconfined compression tests and CBR tests on unreinforced soil were conducted in accordance with Indian standard codes [5, 6 and 7].
IV.
TESTS CONDUCTED ON REINFORCED SOIL
The main emphasis in the present study is the influence of geofabrics on CBR values of expansive clays, since the thickness of pavement depends on CBR value. The thickness of the pavement is inversely proportional to the CBR of a subgrade. To achieve positive contribution of geofabric has to be placed at optimum depth below the surface of weak subgrade [8]. The increase in CBR with the introduction of geofabrics results in pavement thickness and it is obvious that increase in number of layers will yield higher CBR values which results in further reduction in pavement thickness. The effect of reinforcing layers on the CBR values of the subgrade soil is studied herein. The reinforcing materials employed are coir, jute and geogrid. The tests have been carried out for studying layer effect. The numbers of layers selected are one, two, three and four. The locations of fabrics for one layer is H/6, two layers are top and H/6, for three layers at top, H/6 and 2H/6, for four layers at top, H/6, 2H/6 and 3H/6. All the tests have been conducted for unsoaked condition since the studies on soaked and unsoaked soil specimens indicated that the percentage improvement of reinforced system remain almost same irrespective of soaking of the specimen.
V.
The results are analyzed effectively by introducing two non-dimensional factors namely, Effective Depth Ratio (EDR) and Strength Benefit Ratio (SBR). The Strength Benefit Ratio (SBR) is defined as the per cent increase in CBR value of soil due to the presence of reinforcement when compared to the CBR value of unreinforced soil. SBR =
CBR (Reinforced ) CBR (Unreinforced ) CBR (Unreinforced )
X 100
The Effective Depth Ratio (EDR) is defined as the ratio of depth of reinforcing layer from the top to total height of soil specimen. EDR =
Depth of reinforcing layer from the top Total height of soil specimen
Figures 1, 2 and 3 represent load-penetration curves of the soil for single, two, three and four layers with coir, jute and geogrid respectively. The load- penetration curve of unreinforced soil is also shown in each of these figures. The CBR vales are presented in Table 2 for varied number of reinforcing layers with different geofabrics. Table 3 gives the SBR values obtained with different geofabrics for varied number of reinforcing layers for geofabrics like coir, jute and geogrid. Fig. 4 represents the variation of SBR values with number of reinforcing layers for geofabrics like coir, jute and geogrid. The above tables and figures show that both CBR and SBR are increasing with increase in number of reinforcing layers. It is inferred that coir is more effective than jute. The reason is that coir has more tensile strength and friction than jute. The solid structure of the coir fabric resulting in higher tensile strength contributes to the observed behaviour. The jute has fibrous structure of low tensile strength. It is also observed that geogrid results in less CBR values when compared to natural fabrics even though the geogrid has six to seven times more tensile strength natural fabrics. But, to develop full tensile stress within the geogrid, it should interlock with the aggregates. Otherwise, it will slip due to its smooth nature resulting in low CBR value.
489
Vol. 3, Issue 1, pp. 487-493
It is quite obvious that much tensile stress will mobilize as more number of layers intercepts the effective pressure bulb. It is also observed that the rate of CBR value is high from one layer to two layers in case of natural fabrics such as coir and jute, whereas the rate of increase in CBR value is high from three to four layers in case of synthetic fabric i.e., geogrid. It is due to the presence of one of the layers at depth of H/6 from the top. Four layers proved to be more effective in improving the CBR values of the soil irrespective of type of fabric, as per this study. The maximum SBR i.e., at four layers for the soil with coir, jute and geogrid are of the order of 126.61%, 98.9% and 73.95% respectively.
Figure 1 Load Vs penetration curves of soil reinforced with coir for varied number of reinforcing layers
Figure 2 Load Vs penetration curves of soil reinforced with jute for varied number of reinforcing layers
490
Vol. 3, Issue 1, pp. 487-493
Figure 3 Load Vs penetration curves of soil reinforced with geogrid for varied number of reinforcing layers
Figure 4 SBR(%) values for geofabrics with number of reinforcing layers Table 2 CBR Values (%) for variation of reinforcing layers Geofabric at Type of geofabric CBR without geofabric One layer 9.12 8.7 8.8 Two layers 14.14 12.7 9.5 Three layers 14.6 13.22 10 Four layers 15.5 13.7 11.86
COIR JUTE GEOGRID
6.84 6.84 6.84
491
Vol. 3, Issue 1, pp. 487-493
Table 3 SBR Values (%) for variation of reinforcing layers Geofabric at Type of geofabric One layer Two layers Three layers Four layers
COIR JUTE GEOGRID
33.33 27.19 28.65
106.73 85.67 38.88
113.45 93.27 46.2
126.61 98.9 73.95
VI.
1. 2.
CONCLUSIONS
The CBR value increases with increase in number of reinforcing layers due to mobilization of more tensile strength as more number of layers intercept the effective pressure bulb. The rate of increase in CBR value is high from one layer to two layers in case of natural fabrics such as coir and jute, whereas the rate of increase in CBR value is high from three to four layers in case of synthetic fabric i.e., geogrid. The maximum SBR at four layers for soil with coir, jute and geogrid are of the order of 127%, 99% and 74% respectively. The coir is more effective than jute in reinforcing action due to the fact that the coir has more tensile strength, friction and stiffness than jute. Geogrid has less CBR value than natural fabrics in the conventional CBR test procedure. For complete mobilization of tensile stress within the geogrid, it should interlock with aggregate. Hence, the conventional CBR test procedure should be modified by replacing the surcharge weight with the compacted aggregate with geogrid at zero EDR. Geofabrics are more suitable for low load bearing soft soils. This may be due to the presence of higher percentages of clay fraction in soft soils which is responsible for greater mobilization of tensile strength.
3. 4. 5. 6.
7.
REFERENCES
[1]. Jones D.E. and Holtz W.G.(1973), Expansive Soils- the Hidden Disaster, Journal of Civil Engineering, Issue 8, Vol.43, pp 49-51. [2]. Popescu, M.E., (1986), A comparison between the Behavior of Swelling of Clayey Soils, Geotechnical Engineering, Vol. 12, pp. 19-39. [3]. Gillot,J.E. (1968), Clay in Engineering Geology, Elsevier Publishing Company, AmsterdamLondon-New York, 296 pages. [4]. IS 1498-1970, Classification and Identification of Soils for General Engineering Purposes. [5]. IS 2720 (Part VII)-1980, Method of Test for Soils, Determination of Water Content-Dry Density relation using Light Compaction. [6]. IS 2720 (Part X)-1991, Method of Test for Soils, Determination of Unconfined Compressive Strength.
492
Vol. 3, Issue 1, pp. 487-493
[7]. IS 2720 (Part XVI)-1987, Method of Test for Soils, Laboratory Determination of CBR. [8]. Shroff A.V and. Singh G.N. (1989), Load-Deformation characteristics of fabric reinforced on weak soil subgrades, International workshop on geotextiles, pp 144-147. Authors A. V. Narasimha Rao received his Ph.D degree from IIT Madras in 1979. He published more than 105 papers in international and national journals and conferences and 3 text books in Geotechnical Engineering and Engineering mechanics. He has more than 32 years of teaching experience. He is the recipient of Engineer of the year 2007 by Institution of Engineers and Government of Andhra Pradesh D. Neeraja, received her M.Tech from S.V.University, Tirupati. She published 28 papers in international and national journals and conferences. She has more than 10 years of teaching experience.
493
Vol. 3, Issue 1, pp. 487-493
E-LEARNING SYSTEM ANALYSIS USING SMART AIDED TOOLS THROUGH WEB SERVICES
M. Balakrishnan1 and K. Duraiswamy2
Assistant professor, Department of Computer Science Engineering, Selvam College of Technology, Namakkal, Tamilnadu, India 2 Dean(Acadmic), K.S. Rangasamy College of Technology, Thiruchengode, Tamilnadu, India
1
ABSTRACT
A web service is defined as a software system which is designed to interact with the applications over the network in an independent manner. Nowadays, everything has been taken place through online. Some applications which uses web services like e-shopping, e-learning, e-commerce and so on. To improve the efficiency of e-learning many online tutorial services have been developed. So it is easy for the students to learn all the things through internet. An earlier technique has been developed for mediation infrastructure which provides an e-learning system using Smart Space learning system (SSLS). The main concern over SSLS is that it cannot be applied to an IT-mediated learning which has closed information environments. Another technique for e-learning service is that Open Smart Classroom (OSC) on Open Smart Platform to enable intercontinental and intercultural class with advanced features. The issue over OSC is that it provides e-learning services to only intercultural students and it is unreliable to use. To overcome all the issues, a new technique is presented in this work. An effective e-learning system (ELS) is used to improve the online education/ online tutorial service which have to be run under multiple platforms. The proposed ELS use smart tools to provide the e-learning web service to all environments i.e., both open and closed environments. It can also be applied to multicultural students whoever involving in the online tutoring service schemes. This is built on software infrastructure based on multi-agent system architecture using web service technology. It will improve the reliability and scalability of the service with the help of smart class room.
KEYWORDS: Smart Class Room, ELS, Smart tools.
I.
INTRODUCTION
In a traditional method, learning has been done with students and teachers in a face to face manner which assumes as a better learning system. Since the growth of technology has made to develop many new things i.e., everything is based on the internet. So, learning system has also been done through online. This is referred as e-learning. Many web services have been developed for e-learning system which needs some electronic tools to deliver through online. Although electronic tools are needed to deliver tutoring online, a definition of the online process and its best practices may be needed first in order to help select the appropriate technology. A real time interaction has been made in this e-learning service which is to be more helpful for the distance learners where they can clarify their doubt through online. But the tutor, being a good guide, sent back a Socratic answer with more questioning prompts; the student, expecting the answer, became frustrated. Although the student may expect a give and take interaction in a face-to-face tutoring session, the email format suggested to the student that the question should be answered with a direct answer. Since they are learners, it is difficult for students to get the idea soon. Here technology has not been well developed to provide with this type of service. Normally, e-learning models begin with email but there has been an emergence of new models whenever new tools are available, both in asynchronous and synchronous formats. The following models are presented in the context of the specific tool used, for example, Blackboard or NetTutor. A
494
Vol. 3, Issue 1, pp. 494-499
learning service is termed as an event. To attain the educational objective by learning service, it is necessary to create a learning environment consisting of educators, educational material, communication infrastructure, meeting places, etc. Examples of a learning service are the delivery of a course, the provision of a web-based training application or the provision of self-study material. A new learning system is developed with advanced techniques to help the student and teacher for the better outcome. To increase the human-computer interface, mobile devices play an important role in this learning system. The modeling of learning services with web services also opens the possibility of automated integration of educational services into a smart learning space and the automated combination of them. In the smart class system, three successive phases have been developed: They are Individual Smart space, Open Smart Space, and Smart Community. The first phase describes the communication carries between software modules and the second phase will allow other smart devices like mobile phones for enhancing the communication. The last phase describes the inter-space management scheme which sustains the open smart space systems. These techniques have been developed with a limited scenario and can exist under certain applications. With the increasing requirements of intercultural and intercontinental communication for learning purpose, just enabling remote students to participate in the Smart Classroom and to easily communicate with teacher and local students is inadequate. So for this kind of systems, smart classrooms have been developed. To connect these classrooms, we need to tackle the intercommunication problem in the open network, including video, voice, and controlling messages. This type of services has been processed and taken over by smart class system. In this work, we explore literature review in Section 2. Section 3 introduces the architectural diagram for our proposed work Effective E-Learning System using Smart Tools, its features, design and implementation. Section 4 explains about the results and discussions and we draw our conclusion in Section 5.
II.
LITERATURE REVIEW
In peoples daily life, with the development of human-computing interaction, more and more natural human-computing interfaces have been integrated for enhancing work efficiency [1]. A unique disintegration of web applications can be envisaged since September 2005 when official definition of the term Web 2.0 was published by OReilly. During this period, the basic principle of Web 2.0 was proved to be very successful. Similar changes occurred also in the field of e-learning. Researchers in [2] focus on collaborative tagging in learning systems and the characteristics of the social network that would be created by tags. Several recommendations have also been provided how to effectively integrate a social network acquired from tags into learning systems. E-learning solutions not only increase the demand in providing learning objects accessible in the elearning platforms but also across other Web applications too. This need can be solved with the help of semantic technologies where its main goal is to reveal to the computer applications the resources semantics. Researcher in [3] [4] proposes a model that extends the Learning Object Metadata standard integrated with ontology-based semantic annotations for the efficient use of Learning Objects outside Learning Management Systems. The main objective of eAccess2Learn Framework is to provide tools and services that results in the facilitation of design and development of e-training resources that can be exchanged between different e-training platforms and programs that in turn can be reusable between different user groups. Researchers in [6] discusses about Technology Enhanced Learning which being a challenging application area for adaptive and personalized web services and content involves broad range of learning activities, the use of rich multimedia content, as well as a broad range of users with differing expertise. At the same time, mobile devices such as mobile phone, smart phone, PDA, and laptop have been easily accessible for ordinary people. Researchers in [9] and [11] emphasize that mobile devices play an important role in learning. For example, the teacher uses his Smart Phone to bring the presentation file and to control the slideshow, while the students can use a laptop to discuss with others. Some of these features have been incorporated in several projects. Previous Smart Platform enables mobile devices roaming with users to connect into Smart Space by preinstalled modules (eContainer and eADK-based agent of Smart Platform). However, it lacks
495
Vol. 3, Issue 1, pp. 494-499
convenience for the users, especially for those who first come into Smart Space to use their mobile devices. Open Smart Platform applies OSPG as the Web-based mobile interface for mobile devices interaction in Smart Space. OSPG provides the mobile interfaces, such as PPT upload or Turn-toNext-Page, as a Web page [5], [7]. Researcher in [8] discusses about new technique that facilitates trust in online e-learning activities. The privacy of learners is being protected by way of identity management where personal information can be protected through some process anonymity. Subsequently in order to expect learners to trust other pseudonymous participants, we emphasize the reliable mechanism needed for managing participants reputations and assuring that such reputations are legitimately obtained. Researchers in [10] discuss about real-time interactive virtual classroom where tele-education experience is of great significance in distance learning. Subsequently with the development of pervasive computing aimed at enhancing human-computer interface and alternatively integrating pervasive computing technologies into classrooms enhances the experience of both teacher and students, in which case the teacher easily interacts with local and remote students.
III.
EFFECTIVE E-LEARNING SYSTEM USING SMART TOOLS
Tutoring services for e-learning are comprehensive entities considered for a definite purpose and targeted at a specific audience. The providers will clearly analyze the kind of skills to develop for the learners and trained with some more efforts. The objects of learning are common in nature and of a smaller granularity level. The technique presented here for e-learning through web services produce educational resources. Learning services are used mainly in the corporate world and so the services which are available does not free for cost. So the provisions have been made here for the students to offer placement, booking, announcement and payment of educational services without any interruption.
Web Services
E-Learning System
Environment Open
Smart tools
Closed
Multicultural students User 1 User 2 .. User n
Figure 1 Architecture Diagram of E-Learning System Using Smart Tools
The figure 1 shows the architecture diagram of E-Learning system using smart tools. For e-learning process, the student who acts as a user will request the tutor for e-learning process. According to a specific schedule, tutoring services used manual and physical resources to manage it whenever needed by the user. Normally, tutoring system needs a particular time to discuss about their topics and sessions for an appointment. After receiving the message from the user, the tutor will identify the students skill and respond to them based on their performance. By using smart tools, the online tutoring schemes are applied for multicultural students i.e., any students can attend the e-learning system based on their skills across world wide. The proposed elearning system using smart tools is applied to both open and closed environments. It chooses the learning service efficiently based on users profiles. The mediation infrastructure predicts to
496
Vol. 3, Issue 1, pp. 494-499
correspond heterogeneous learning services offered by various information systems and mediate them using learner profiles. Since online tutoring services increases day by day, it is possible to include more academic subjects to enhance the learning services. The proposed ELS using smart tools are very reliable to use and the scalability of the e-learning process is also be improved.
IV.
Web Services provide a perfect modern solution for hyper linking software components over the Internet. During a month of course, 30 students have been participated in an experimental study. Initially, 9 students only participated. The communication rate was 83%. A rating has been taken under students judgments. After that, the two staff-tutors rated all questions, together with the questions started but not yet rated by the students. Generally, the agreement between the tutors on solved versus not-solved questions is high: 86% or 74%. If we combine the judgment of the students and the tutors, by counting a question as solved if at least two of the three ratings are 4 or above, the number of questions solved is approximately the same as the number indicated by the students. So student opinion does not differ much from expert (staff) opinions.
Message communication rate 40 30 20 10 0 0 10 20 30 No. of users ELS using smart tools Extensible learning system using smart space Medis infrastructure for learning 40 50 60
Figure 2 No. of users vs. Message Communication Rate
Figure 2 describes the message communication rate of different techniques like ELS using smart tools, Extensible learning system using smart space and Medis infrastructure for learning. As the number of users increases for e-learning, the communication rate should be efficient. The proposed ELS using smart tools for e-learning provide comparatively good result when compared to the existing ones.
Response time (sec) 40 30 20 10 0 0 10 20 30 No. of web services ELS using smart tools Extensible learning system using smart space Medis infrastructure for learning 40 50 60
Figure 3 No. of web services vs. Response time
497
Vol. 3, Issue 1, pp. 494-499
Figure 3 describes the response time of web services for different techniques used by the e-learning system. The figure 3 also shows that the increase in the number of web services simultaneously results in lowering of response time provided by the tutor for the student by way of using ELS tutoring scheme for e-learning using smart tools.
40 Availability of service 30 20 10 0 0 10 20 30 No. of Users ELS using smart tools Extensible learning system using smart space Medis infrastructure for learning 40 50 60
Figure 4 No. of Users vs. Availability of service
Figure 4 describes the availability of web services for different techniques used for e-learning system. When the number of users increases for e-learning, then the availability rate across world wide in the proposed ELS tutoring scheme for e-learning is better when contrast to an existing ones. The proposed effective e-learning system using smart tools is experimented to show the performance improvement with the existing model in terms of application throughput, delivery report, interaction of more number of people, and scalability.
V.
CONCLUSION
The evolution of online tutoring has showed us that success may not depend so much upon the tool selected, but on the development of an appropriate culture for online tutoring, an understanding of the process and parameters involved. Learning services which make use of physical or human resources are offered according to a specific schedule since the use of those resources needs to be managed. The proposed effective e-learning system using smart tools is easy for education environment which effectively integrated Web Services to exploit the resource sharing for appropriate online tutoring domain and it can be applied effectively to multicultural students. It efficiently achieved Easy installation, Ease of use, Low maintenance efforts, and Integration with other Internet/Intranet based education tools. The experimental results showed an improved performance of communication rate; delivery time and scalability which is proved that the interactive web service tutoring scheme for elearning is good.
REFERENCES
[1] Yue Suo, Naoki Miyata, Hiroki Morikawa, Toru Ishida, and Yuanchun Shi, Open Smart Classroom: Extensible and Scalable Learning System in Smart Space Using Web Service Technology, IEEE transactions on knowledge and data engineering, vol. 21, no. 6, june 2009. [2] Robert Moro, Ivan Srba, Maros UnCik, Maria Bielikova, Marian Simko, Towards Collaborative Metadata Enrichment for Adaptive Web-Based Learning IEEE International Conferences on Web Intelligence and Intelligent Agent Technology, 2011 [3] Mihaela M. Brut, Florence Sedes, Senior Member, IEEE, and Stefan Daniel Dumitrescu, A SemanticOriented Approach for Organizing and Developing Annotation for E-Learning IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 4, NO. 3, JULY-SEPTEMBER 2011 [4] Demetrios G. Sampson, Senior Member, IEEE, and Panagiotis Zervas, Supporting Accessible TechnologyEnhanced Training: The eAccess2Learn Framework IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. 4, NO. 4, OCTOBER-DECEMBER 2011
498
Vol. 3, Issue 1, pp. 494-499
[5] A. Maedche and S. Staab. Services on the move towards p2p-enabled semantic web services. In Proceedings of the Tenth International Conference on Information Technology and Travel & Tourism, ENTER 2003, Helsinki, Jan. 2003. [6] Cormac Hampson, Owen Conlan, Vincent Wade, Challenges in Locating Content and Services for Adaptive eLearning Courses 2011 11th IEEE International Conference on Advanced Learning Technologies [7] Addison Wesley. (2005). The Tutor Center. Retrieved September 20, 2005 from http://www.awbc.com/tutorcenter/index.html [8] Mohd Anwar and Jim Greer, Facilitating Trust in Privacy-preserving E-learning Environments IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES, VOL. X, NO. X, XXX 2010 [9] Doherty, B. & Atkinson, M. (2004, Spring). A pilot study of online tutoring using Smarthinking. PowerPoint presentation. Retrieved June 15, 2004 from http://www.smarthinking.com [10] Yue Suo, Naoki Miyata, Hiroki Morikawa, Toru Ishida, Fellow, IEEE, and Yuanchun Shi, Senior Member, IEEE, Open Smart Classroom: Extensible and Scalable Learning System in Smart Space Using Web Service Technology IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 21, NO. 6, JUNE 2009 [11] L. Stojanovic, S. Staab, and R. Studer. elearning based on the semantic web. In Proceedings of the World Conference on the WWW and Internet WebNet2001, Orlando, Florida, USA, 2001. [12] Mihaela M. Brut, Florence Sedes, and Stefan Daniel Dumitrescu, A Semantic-Oriented Approach for Organizing and Developing Annotation for E-Learning, IEEE transactions on learning technologies, vol. 4, no. 3, July-September 2011 Authors M. Balakrishnan received the M.E. degrees in Computer Science and Engineering from K.S.Rangasamy College of Technology, Tiruchengode, in 2006 respectively. During 20072009, he worked as Lecturer in K.S.Rangasamy College of Technology in Tiruchengode. He now with Selvam College of Technology, Namakkal, Tamilnadu, India as Assistant Professor in Department of Computer Science and Engineering. K. Duraiswamy received the B.E., M.Sc. and Ph.D. degrees, from the University of Madras and Anna University in 1965, 1968 and 1987 respectively. He worked as a Lecturer in the Department of Electrical Engineering in Government College of Engineering, Salem from 1968, as an Assistant professor in Government College of Technology, Coimbatore from 1983 and as the Principal at K.S.Rangasamy College of Technology from 1995. He is currently working as a Dean in the Department of Computer Science and Engineering at K.S.Rangasamy College of Technology (Autonomous Institution).His research interest includes Mobile Computing, Soft Computing, Computer Architecture and Data Mining. He is a senior member of ISTE, IEEE and CSI.
499
Vol. 3, Issue 1, pp. 494-499
THE IMPACT OF ELECTRONIC CUSTOMER RELATIONSHIP MANAGEMENT ON CONSUMERS BEHAVIOR

Usman Musa Zakari Usman, Abdullah Nabeel Jalal, Mahdi Alhaji Musa
Department of Information Systems, Universit Teknologi Malaysia (UTM), Skudai, Johor Bahru 81310, Malaysia.
ABSTRACT
Electronics customer relationship management (E-CRM) is a business and marketing strategy that analyzes about consumers needs and behavior in order to create an effective relationship between anorganization and its consumers. E-CRM is a combination of hardware and software, process, applications and commitment of management activities to develop high quality of customer service, and customers maintenance.However, the slight is known about how various activities may exert differentiated impact on organization the concerned. This papers focus is to discuss the positive and negative impacts of electronics customer relationship management (E-CRM), as a marketing strategy for an organization. The paper further explains the behavior of consumers toward organizations and the emergence of electronic commerce that brought a positive change towards business in a global market.
KEYWORDS: electronics customer relationship management (E-CRM), consumer behavior
I. INTRODUCTION
Managing customer relationship has become one of the most important concerns for an organization. Appropriately, organizations are seeking for supplementary and effective relationship activities to increase organizationalvalue of customer relationship by improving relationship quality or delivering better relationship benefits. [3]Electronics customer relationship management (E-CRM) is an important tool for measuring the performance of an organization in order to help the company to increase its profit and enhance customer loyalty towards the company product. [8]The performance of E-CRM has becoming a rising alarm in marketing and information technology research and practice.[9] Although the important basis of CRM has been around since 1956, it is within the last few years that CRM has generatedan important impact inglobal business industries. It is a new process that comes from the internet and web technology to make the implementation of E-CRM easy and flexible. [10] The electronics customer relationship management concentrates on internet web-based interaction between service providers (organization) and customers.[11], [12] the most crucial advantage of CRM and E-CRM is to help the company to identify and categorize customer for enhancing their loyalty and generating profit (Rudolph 1999).
II. CUSTOMER RELATIONSHIP MANAGEMENT (CRM)

Customer relationship management (CRM) starts off from new business practices that focus on how to manage customers, such as attracting new customers and retaining them. [13] Customer relationship management is a business strategy process that improves organizations competitive ability. It also creates a successful strategy for an organization which helps them to focus on customers behaviors, promotes the emerging speed of firm and maintains their customers in a very highly competitive marker environment. [14] The need to understand the consumers behavior in various marketplaces is very important in electronic commerce. CRM deals with both customer and
500
Vol. 3, Issue 1, pp. 500-504
employees. [15]. although its perspective is based on outside of an organization (i.e.) in marketplace and eyes of the consumers, CRM communication is a market communication, such as communication various types of relationship and consumers behaviors. [16] Example is communicating the relationship that involved in business-to-business, customer lifecycles, various segments of consumers and their nature. It also communicates the consumer time value and profitability of consumers. [17]
III. IMPACT OF ELECTRONICS CUSTOMER RELATIONSHIP MANAGEMENT (E-CRM)

Electronics customer relationship management (E-CRM) is an electronic commerce (marketing) approach used by organizations to enable them identify anddraw customers that are potentially profitable.[18] This is done by developing proper relationship with customers and drawing them in a way that their future requirements will be satisfied. Practicing E-CRMsystem means developing an accurate image of customer requirements and behavior, [19], [30]E-CRMis one of the most developed managerial methods that can be utilized in any organization. It is one of the best approaches that organizations can use to attract and retain their consumers from switching to other companies in order to prevent customer churn. [20] E-CRM is the complete process and techniques that an organization has to gain, increase, and maintain in order to provide consumers with high quality service through the use of internet. As a result E-CRM is a flexible and more secure channel method to get high profitability, customer retention and maintenance of relationship with the customers in a convenient and effective manner. [21]E-CRM technology should be more advanced and erudite to meet the requirement for developing and knowledgeable customer, usage of CRM technology consistently has a strong impact on CRM performance (Greve and Albers, 2006), they propose that the more comprehensive the CRM technology is, and the higher the CRM technology usage, the better the CRM performance will be across the stages of customer life-cycle. The CRM technology creates a significant impact on the performance of customer relationship with an organization.[22]. According to Sheng [1], the primary objectives of customer relationship are the acquisition and retention of consumers, which is leads to the bottom line of financial benefits.The CRM has started the recent years in order to become the central strategies for organizationsthat aim to provide information of understanding and meeting the needs of customers. [23] The rapid growth of the latest developments in internet-based technology encouragesthe development of electronic customer relationship (ECRM), which allows many innovations in a relationship management, supply chain management, and inter-organizational collaboration [2] E-CRM is one of the most effective application of information system (IS).
IV. CONSUMERS BEHAVIORS

The American marketing research foundation defines consumer behavior as the dynamic interaction of affect and cognition, behavior and the environment by which human being conducts the exchange aspects of their lives (Bennett, 1995). [33] This definition indicated that affect and cognition are significant aspect in understanding consumers behavior. Affect and cognition are consumer psychological respond in various types of situation. Aconsumer is the final user of a companys product or service. Consumer behavior involves psychological processes that consumers go through for them to identify their needs. [32] as a result when trying to solve these needs, there is a need for making a purchase-decision (example is whether the consumer will purchase a product or not, if yes then which brand and where), interpret information, make the plans, and implement the plans by comparison on variety of product or purchasing a product.[24]Consumer behavior is crucial factor determines consumers behavior and intention to accept or reject new information about a companys product. The cognitive dissonance theory states [3] that under normal circumstances, an individuals cognitive system is in behavior inconsistent, which will lead to create tensions and conflicts in the individuals mind, making he/she to change attitude or adjust behaviors.
501
Vol. 3, Issue 1, pp. 500-504
V. THE IMPACT OF E-CRM ON CONSUMERS BEHAVIORS

The E-CRM system is built up from co-corporation of the new technology, such as the internet, telemarking and email, with customer relationship management.[25] E-CRM is built for the business and marketing industries. Most organizations with a good E-CRM system implementation are moving towards more positivethan negative impact.[26] An organization using E-CRM software system will be able to communicate with their customers and access customer information and history, and this will lead the organization to enhance its consumers loyalty implicitly. [27] E-CRM provides organization with chances of understanding the consumers future expectations level, which will provide the organization with a deeper look at the level of consumer satisfaction.Kotler [5] explains that satisfaction refers to a persons feeling of pleasure or disappointment that results from comparing an organizations products perceived performance or outcome in relation to his/her expectations. Satisfaction can be associated with feeling of acceptance, and happiness [6]. Factors that affect customers satisfaction include friendly employees, knowledgeable employees, and accuracy of billing, billing timelines, competitive pricing quality, good value and fast service.(Hokason)[7]Other positive impacts of using E-CRM are improving customer service and providing support, efficient and encouraging cost reduction. [4]In addition E-CRM helps in collecting, updating and fulfilling orders remotely and accurately. [28] For any organization, this requires a long dating and maintaining process to retain the existing customers and attract new customers. [29]Organizations also use ECRM as a tool for centralizing the entire customers data intoa single database storage and allow each department within an organization to exchange customer profile [31].
VI. DISCUSSION
Across the surveys conducted by most experts, we conclude that the E-CRM system requires all the necessary consumer information that will describe the consumers daily activities and the leisure activities to the organization. Such information includes the financial status of a customer, occupation, age and personal interest.This research shows that E-CRM has positive effects at the organizational level for their respective products and services, consumers loyalty, efficiency, internal process, management channels and innovative sufficient account for E-CRM impact. The E-CRM technology and organizational support also have positive impact on E-CRM performance. Within the technological context, it is crucial for an organization to consider the E-CRM technology integration and customer information analysis. On the order hand (organizational context), E-CRM personal asset, consumer knowledge management (CKM), learning and sensing consumer market, appropriately account for the organizational support on good E-CRM. Competition intensity and consumer power in an environmental context should be put into consideration as well.Advanced technology and information provide organization with off-the-shelf software to better electronic system. Better understanding of how employees communicate with customers when facing a problem about the organizations product and services should alsobe put into consideration. As a result, qualitative and quantitative research about the organization and its product and services should conduct thoroughly from both the employees and consumers perspective.
VII. CONCLUSION
The E-CRM framework (impact) presented in this paper was addresses the current electronics business needsfor an organization to become successful in a highly competitive electronic commerce environment. The study of consumers behaviors in an electronic market and the factors that control the consumers behavior towards organizational goals are also discussion in this paper. Consumers perspective about organizational business model may have an impact on the result of a succeeding process of E-CRM. Other factors that influence the results of a successful E-CRM implementation are available there for other expects and/or researcher to recognize in the future.
REFERENCE
[1] Y.P. Sheng. A business model and framework for electronic customer relationship management. InProceedings of the 8thAMCIS Conference. 2002.
502
Vol. 3, Issue 1, pp. 500-504
[2] [3] A. Zornes and M. Gotta, Web & collaboration strategies. 2001, META Group Inc. R.A. Feinberg, R. Kadam, L. Hokama, and I. Kim, The state of electronic customer relationship management in retailing. International Journal of Retail & DistributionManagement, 2002. 30(10), pp. 470 - 481. Ahmed T. 2009, Electronic Customer Relationship Management in Online Banking, Master ThesisElectronic Commerce 2009:070. Kennedy, A. 2006, Electronic Customer Relationship Management (eCRM): Opportunities Hoyer, W. D. & MacInnis, D. J. 2001, Consumer Behavior, 2nd ed., Boston, Houghton Mifflin Company. Hokanson, S. January 2, 1995, The Deeper You analyze, The More You Satisfy Customers, Marketing News, p.16 N.C. Romano Jr. and J. Fjermestad, Electronic commerce customer relationship management: An assessment of research.International Journal of Electronic Commerce, 2001. 6(2), pp. 61-113. A. Gaffney 2007, Improving Customer Satisfaction through More Effective ICT Strategies, AlcatelLucent Enriching Communications Vol.1, Issue 2, 2007 E-Customer Relationship Management: Training Module 2008, http://www.mmv.vic.gov.au/Assets/219/1/ElectronicCustomerRelationshipManagement.pdf, [Accessed on September 17, 2010] Ahmed T. 2009, Electronic Customer Relationship Management in Online Banking, Master Thesis Electronic Commerce 2009:070. Samsudin Wahab, Nor Azila Mohd Nor & Khaled AL-Momani 2010, The Relationship between EService Quality and Ease of Use On Electronic Customer Relationship Management (E-CRM) Performance: An Empirical Investigation In Jordan Mobile Phone Services, 2010 International Conference on e-Education, e-Business, e-Management and e-Learning, pp. 368-372. Peter C. Verhoef 2003, Understanding the Effect of Customer Relationship Management Efforts on Customer Retention and Customer Share Development, Journal of Marketing, Vol. 67 (October 2003), pp. 30-45. Scullin, S., Fjermestad, J. & Romano, N. C. 2004, E-relationship marketing: changes In traditional marketing as an outcome of electronic customer relationship Management, The Journal of Enterprise Information Management, Vol. 17, No. 6, pp. 410-415. C.Y. Lim &Vincent Khoo 2010, Customer Relationship Management: Lifecycle of Predicting Customer Lifetime Value, Second International Conference on Computer Research and Development, pp. 88-92. Colleen Cunningham & Il-Yeong Song 2007, A Taxonomy of Customer Relationship Management Analyses for Data Warehousing, Tutorials, Posters, Panels and Industrial Contributions at ER 2007. Galbreath, J & Rogers, T 1999, Customer relationship leadership: a leadership and motivation model for the twenty-first century business, The TQM Magazine, Vol. 11, No. 3, pp. 161-71. Huang Y. 2004, The new management things [M], Beijing: Democracy and construction Press, 2004. (In Chinese). G.C. Moore and I. Benbasat, Development of an instrument to measure the perceptions of adopting and information technology innovation. Information Systems Research, 1991. 2(3), pp. 192-222. H. Rosenbaum and B.Y. Huang, A framework for web-based ecommerce customer relationship management. In Proceedingsof the 8th AMCIS Conference. 2002. V. McKinney, K. Yoon, and F. Zahedi, The measurement of web-customer satisfaction: an expectation and disconfirmation approach. Information Systems Research, 2002. 13(3), pp. 296- 315. P.B. Seddon, S. Staples, R. Patnayakuni, and M. Bowtell, Dimensions of information systems success. Communications ofAIS, 1999. 2(Article 20). Kabiraj, S., Electronic Customer RelationshipManagement: Origin and Opportunities, Engineering Management Conference, IEEE xplore, Nov. 2004, pp. 484-488. Khalifa, M.; Ning Shen, Effects of Electronic Customer Relationship Management on Customer Satisfaction: A Temporal Model, 38th Annual Hawaii International Conference on Systems Sciences, Jan. 2005, pp. 171a - 171a. T. Chang, L. Liao, W. Hsiao, An Empirical Study on the e-CRM Performance Influence Model for Service Sectors in Taiwan, Proceedings of the 2005 IEEE International Conference on e Technology, e-Commerce and e-Service, IEEE Xplore, 29 March-1 April, 2005, pp. 240 245. R. Feinberg and R.Kadam, E-CRM Web Service attributes as determinants of customer satisfaction with retail Web sites, International Journal of Service Industry Management, 2002. A. Dutta, Managing Customer Turnover In Online Services: A Systems Thinking Approach, International Conference on Information Systems, 2000, pp. 681 687.
[4] [5] [6] [7] [8] [9] [10]
[11] [12]
[13]
[14]
[15]
[16] [17] [18] [19] [20] [21] [22] [23] [24]
[24]
[25] [26]
503
Vol. 3, Issue 1, pp. 500-504
[27] Z. Song, Y. Wang, B. Yu, Z. Zhang, How Customer Relationship Creates Value, International Conference on Services Systems and Services Management, IEEE Xplore, June 2005, Vol.1, pp.144150. Greve, G. and Albers S. (2006). Determinants of Performance in Customer Relationship Management Assessing the Technology Usage Performance Link. Paper presented at the Proceedings of the 39th Hawaii International Conference on System Sciences - 2006.Article in a journal: Wang, Y., Lo, H. P., Chi, R., & Yang, Y. (2004)., An integrated framework for customer value and customer-relationship management performance: a customer-based perspective from China. Managing Service Quality, 14(2/3), 169-182 D. Bradshaw and C. Brash, "Managing customer relationships in the e-business world: how to personalize computer relationships for increased profitability," International Journal of Retail &Distribution Management, vol. 29, pp. 520- 530, 2001. Anderson, James C., Business marketing: understand what customers value, Harvard Business Review, November-December, 1998, pg. 54. Cunningham S.M, The Major Dimensions of Perceived Risk InF.C.Donald (Ed.) Risk Taking and Information Handling InConsumer behaviors,[M],Boston Harvard University Press, 1967:82108. Bennett, the American marketing Research Foundation, 1995, pg.59.
[28]
[29]
[30]
[31] [32]
[33]
Authors
Usman Musa Zakari Usman is a M.Sc. research student at the Faculty of Computer Science and Information Systems of University Technology Malaysia (UTM), Skudai, Johor Malaysia. He holds his B.Sc. in business Information Technology from Limkokwing University of Creative Technology Cyberjaya Malaysia (LUCT). His research interests are innovative solutions for "knowledge-based" information systems that span several areas applying ontology and knowledge management for interoperating information systems, software engineering and enterprise system.
Abdullah Nabeel Jalal is a M.Sc. research student at the Faculty of Computer Science and Information Systems of University Technology Malaysia (UTM), Skudai, Johor Malaysia. He holds his B.Sc. in ManagementInformation System from University of Masul Iraq.
Mahdi Alhaji Musa is currently a Master Student in IT-Management at Faculty of Computer science and information system University Technology Malaysia, and working in the Department of Computer Science at the Yobe State University Damaturu-Nigeria.Mahdi holds a B.Eng. from Bayero University Kano Nigeria in Electrical Engineering. . His main research interests are innovative solutions for "knowledge-based" information systems that span several areas applying ontology and knowledge management for interoperating information systems, e-learning and M-learning.
504
Vol. 3, Issue 1, pp. 500-504
PERFORMANCE & EMISSION STUDIES ON A SINGLE CYLINDER DI DIESEL ENGINE FUELED WITH DIESEL & RICE BRAN OIL METHYL ESTER BLENDS
Bhabani Prasanna Pattanaik1, Basanta Kumar Nanda2 and Probir Kumar Bose3
1
Assistant Professor, Department of Mechanical Engg., Gandhi Institute for Technological Advancement, Bhubaneswar - 752054, Orissa, India, 2 Associate Professor, Department of Mechanical Engineering, Maharaja Institute of Technology, Bhubaneswar, Orissa, India, 3 Director, National Institute of Technology, Agartala, Jirania, Tripura 799055, India,
ABSTRACT
The Present research work studies about the production of Rice bran oil biodiesel (RBOBD), study of its characteristics and potential as a substitute for Diesel fuel in CI engines. In the present investigation biodiesel is produced by base catalyzed transesterification of rice bran oil is known as Rice Bran Oil Methyl Ester (RBOME). Various proportions of RBOME and Diesel are prepared on volume basis and used as fuels in a four stroke single cylinder direct injection Diesel engine to study the performance and emission characteristics of these fuels. The results obtained show a 49% reduction in smoke, 35% reduction in HC and 37% reduction in CO emissions for the blends whereas the brake power and BTE are reduced by 2.4% and 3.2% respectively with 4.3% increase in the SFC . Therefore it is concluded from the present experimental study that the blends of RBOME and Diesel fuel can successfully be used in Diesel engines as an alternative fuel without any modification in the engine and it is also environment friendly by the emission standards.
KEYWORDS: RBOBD, Transesterification, Blend, RBOME, Performance & Emission.
I.
INTRODUCTION
The use of vegetable oils in Diesel engines replacing petroleum diesel is being studied over the last century. Many scientists and researchers over the years have studied various types of vegetable oils and their use in Diesel engines. However some physico-chemical properties of vegetable oils like high density and viscosity, low volatility and formation of carbon deposits tend to limit their use as fuel in diesel engines. It was experimentally proven and worldwide accepted that the transesterification process is an effective method for biodiesel production and reduce in viscosity and density of vegetable oils. The transesterification process is a reversible reaction between the triglycerides of the vegetable oil and alcohol in the presence of an acid or base as catalyst. As a result of transesterification the monoalkyl esters of the vegetable oil are formed and glycerin is produced as a byproduct in the process. The monoalkyl or methyl esters of the vegetable oil produced during transesterification are popularly known as biodiesel. In India efforts are being made for using non-edible and under exploited oils for production of methyl esters or biodiesel [7]. Blending petroleum Diesel fuel with methyl esters of vegetable oils is the most common practice of using biodiesel in diesel engines in present time. There have been reports that significant reduction in the exhaust gas are achieved with the use of blends in Diesel engines [19]. Several studies have shown that diesel and biodiesel blends reduce smoke opacity, particulate matters, un-burnt hydrocarbons, carbon dioxide and carbon monoxide emissions, but the NOx emissions have slightly increased [32]. It was reported from several previous studies that
505
Vol. 3, Issue 1, pp. 505-513
the transesterification of the crude vegetable oil with alcohol in the presence of catalyst is the easiest method for production of biodiesel [14]. The present experimental work investigates about the production of biodiesel from rice bran oil by transesterification with methanol, preparation of test fuels for the engine experiments in the form of three blends of rice bran oil biodiesel (RBOBD) and Diesel as B20, B30 and B50 and measurement of various engine performance parameters and exhaust emissions. The structure of the present research paper is as follows. Section 2 represents the detailed experimental procedure for biodiesel production, characterization of the test fuels and the description of the experimental setup of the present work. Section 3 represents the results of the series of experiments conducted during the present research work followed by discussions on the behavior of the test fuels used in the engine. Section 4 discusses the conclusions and findings of the ongoing experimental investigation along with their implications.
II.
EXPERIMENTAL PROCEDURE
2.1 Biodiesel Production by base-catalyzed transesterification method

Rice bran oil and methanol were mixed in a molar ratio of 3:1 and the mixture was poured into the test reactor. Then base catalyst (KOH) in 1% w/w was added into the already present mixture in the reactor. The mixture inside the reactor was heated to a temperature of 65oC and stirred continuously. The mixture in the reactor was allowed to remain at the same temperature for a period of 3 hrs and then it was allowed to settle under gravity. After settling two layers were formed, the upper layer was found to be Rice bran oil methyl esters (RBOME) and the lower layer being glycerol. The glycerol was separated out and the RBOME was mixed with 10% (by vol) hot water and shaken properly and allowed to settle again for 24 hrs. After settling was over the lower layer was separated which contained water and KOH. The part was biodiesel and moisture. After complete removal of moisture, pure biodiesel or RBOME was obtained.
2.2 Preparation of biodiesel blends

After production the RBOME was blended with neat diesel fuel in various volume concentrations to prepare biodiesel blends. These blends were subsequently used in the engine tests. The level of blending for convenience is referred as BXX. Where XX indicates the percentage of biodiesel present in the blend. For example a B20 blend is prepared with 20% biodiesel and 80% diesel oil by volume. During the present engine experiments the blends prepared and used were B20, B30 and B50.
2.3 Characterization of Test Fuel

The test fuels used in the engine during the experiments were B20, B30, B50 and Diesel oil. Before application on the engine, various physico-chemical properties of all the above test fuels were determined and compared to each other.
Table 1. Properties of Diesel and RBOME Properties Density at 20oC Kinematic Viscosity at 40oC Heating Value (MJ/Kg) Flash Point (oC) Cloud Point (oC) Cetane Index Diesel 0.82 2.7 42.50 67 -6 50 RBOME 0.87 4.81 38.81 166 -1 47
2.4 Experimental Setup

A schematic diagram of the experimental setup and test apparatus are given below.
506
Vol. 3, Issue 1, pp. 505-513
Figure 1. The Experimental Setup
1. Test engine, 2. Dynamometer, 3. Diesel tank, 4. Fuel blend tank, 5. Diesel burette, 6. Fuel blend burette, 7. Air tank, 8. Air flow meter, 9. Air intake manifold, 10. Exhaust, 11. Smoke meter, 12. Exhaust Gas analyzer, 13. Stop watch, 14. RPM indicator, 15. Exhaust temp. indicator, 16. Coolant temp. indicator, 17. Lub. Oil temp. indicator, 18. Rotameter, 19. Pressure sensor, 20. Charge amplifier, 21. Computer.
Table 2. Test Engine Specification Sl. No. 1 2 3 4 5 6 7 8 9 10 11 12 Item Make Engine type No. of cylinders Cooling Method Bore Stroke Compression ratio Injection Pressure Engine speed Rated output Injection nozzle opening Dynamometer Exhaust gas analyzer Description Kirloskar oil engines Four stroke CI One Water cooled 80 mm 110 mm 16.7:1 170 bar 1500 rpm 5.1 kW 23o bTDC Eddy current type AVL
2.5 Description of Experimental Setup

The present set of experiments were conducted on a four stroke single cylinder direct injection water cooled diesel engine equipped with eddy current dynamometer. Two separate fuel tanks with fuel flow control valves were used for the operation of the engine on diesel and biodiesel. One of the fuel tanks contained diesel and the other tank was filled with individual fuel blends of B20, B30 and B50. The engine was operated with full load and constant speed and the performance parameters like brake power, torque, specific fuel consumption and brake thermal efficiency were measured for diesel and all the test fuels. The CO and HC emissions were also measured for diesel and all the test fuels by using the data obtained from the exhaust gas analyzer.
507
Vol. 3, Issue 1, pp. 505-513
III.
RESULTS & DISCUSSION
3.1 Engine Performance Analysis

3.1.1 Brake Power
Brake Power vs Load 4 3.5 Brake Power (kW) 3 2.5 2 1.5 1 B50 0.5 0 0 20 40 60 80 100 Diesel B20 B30
Load (%) Figure 2. Variation of Brake Power with Engine Load
The power developed by the engine at varying load is higher for Diesel and slightly less for the blends of RBOME. However with B20 blend the brake power developed is very close to that with diesel. A 2.4% reduction in brake power is obtained with the B50 blend. The lower value of power output with biodiesel blends is due to the higher viscosity and lower heating value of the blends as compared to petro diesel [10]. 3.1.2 Brake Thermal Efficiency
BTE vs BMEP 30 BTE (%) 20
Diesel B20 10 B30 B50 0 0
BMEP (bar) Figure 3. Variation of BTE with BMEP
The brake thermal efficiency (BTE) increases with increase in brake power for all types of fuels. The BTE is observed higher in case of Diesel than all the three blends of RBOME. As the percentage of biodiesel increases in the blend it results in a slight decrease in BTE. A 3.2% decrease in BTE is observed for the B50 blend during the experiments. This is due to the fact that with higher blends of biodiesel, the fuel is more viscous hence lower is the heating value [6].
508
Vol. 3, Issue 1, pp. 505-513
3.1.3 Brake Specific Fuel Consumption The brake specific fuel consumption (BSFC) was found to be lowest for diesel and tend to increase a little with the RBOME blends. The BSFC is more with higher blends of biodiesel. A 4.3% increase in BSFC is observed in case of the B50 blend. This is because of lower heating value and higher viscosity of the blends as compared to petro diesel [27].
35 30 BSFC (kg/kWhr) 25 20 15 10 B20 5 0 0 1 2 3 4 5 6 7 Diesel B50 B30 BSFC vs BMEP
BMEP (bar) Figure 4. Variation of BSFC with BMEP
3.2 Engine Exhaust Emission Analysis

3.2.1 CO Emissions
%CO vs Load 0.35 0.3 0.25 CO (%) 0.2 0.15 0.1 0.05 0 0 40 60 80 100 Load (%) Figure 5. Variation in CO emissions with engine load 20 Diesel B20 B30 B50
The variation in CO emission at different loads with all the test fuels is shown in fig 5. At low and medium loads, CO emissions of the blends were not much different from those of diesel fuel [28]. However at full load conditions the CO emissions of the blends decrease significantly when compared to those of diesel [28]. This type of behavior can be attributed towards the complete combustion occurring in case of blends due to the presence of oxygen in the methyl esters of rice bran oil. 3.2.2. HC Emissions The HC emission from the engine at different loads is shown in the above result. At lower loads the HC emissions are usually less and at higher loads they are more. However at full load operation the
509
Vol. 3, Issue 1, pp. 505-513
HC emission is maximum for diesel [30]. With higher blends of biodiesel the HC emission gradually reduces [29].
%HC vs Load 120 100 80 HC (ppm) 60 40 B30 20 0 0 20 40 60 Load (%) Figure 6. Variation in HC emissions with engine load 80 100 B50 Diesel B20
3.2.3 Smoke Opacity

Smoke Opacity vs Load
18 16 14 12 10 8 6 4 2 0 0 20 40 60 80 100 Load (%)

Figure 7. Variation in Smoke emissions with engine load
Smoke Opacity (%)
Diesel B20 B30 B50
Fig 7. Shows the variation in smoke emissions at different loads for all the test fuels used in the experiments. The smoke is formed due to incomplete combustion of fuel in the combustion chamber [27]. It is seen from the above set of results that the smoke emissions are less with blends of RBOME in comparison to that of diesel fuel. This is because of better combustion of blends due to the availability of more oxygen in biodiesel [24].
IV.
CONCLUSIONS
The objective of this study was production and characterization of biodiesel from Rice bran oil and preparation of B20, B30, and B50 blends for use in a single cylinder DI diesel engine. Based on the experimental results found the following conclusions are drawn:
510
Vol. 3, Issue 1, pp. 505-513
1. The physico-chemical properties of biodiesel obtained from rice bran oil are little different from those of diesel oil. The viscosity of biodiesel is higher than that of diesel especially at low temperatures. 2. The brake power of the engine using all the blends of RBOME is very close to the value obtained with diesel. A nominal 2.4% reduction in brake power is obtained with the B50 blend. 3. The BTE of the test engine for the three blends was found little lower than the value obtained with diesel. The engine with B50 blend shows 3.2% reduction in BTE. 4. The BSFC of the blends are little higher than that of Diesel. The higher fuel consumption with the blends reflect to the lower heating value and higher viscosity of the biodiesel. The BSFC increases linearly with the increase in biodiesel percentage in the blend. A 4.3% increase in BSFC is observed with the B50 blend. 5. As per the exhaust emissions with the blends, it was found that the CO, HC and smoke emissions were reduced significantly when compared to those of diesel. The results obtained show a 49% reduction in smoke, 35% reduction in HC and 37% reduction in CO emissions at full load. Form the above conclusions drawn, it is found that the performance of the test engine when operating with RBOME blends were very satisfactory and close to that of Diesel oil and significant improvement was noticed in the exhaust emissions of CO, HC and Smoke when the engine was operating with the blends. Therefore it can be concluded that the blends of RBOME can be successfully used as alternative fuel in diesel engines without any engine modifications.
Abbreviations
CI DI RBOBD RBOME KOH B20 B30 B50 BP BTE BSFC CO HC Compression Ignition Direct Injection Rice Bran Oil Biodiesel Rice Bran Oil Methyl Ester Potassium Hydroxide 20% biodiesel + 80% diesel 30% biodiesel + 70% diesel 50% biodiesel + 50% diesel Brake Power Brake Thermal Efficiency Brake Specific Fuel Consumption Carbon Monoxide Hydro Carbons
ACKNOWLEDGEMENT
The authors thank the Department of Mechanical Engg., Jadavpur University, Kolkata for providing laboratory facilities for the conduct of experiments and Prof. (Dr) Probir Kumar Bose, Director, NIT Agartala for his valuable guidance and help during the course of the present research work.
REFERENCES
[1]. Chandan Kumar, Gajendra Babu MK, Das LM. Experimental investigations on a Karanja oil ester fueled DI diesel engine. SAE Paper No. 2006-01-0238; 2006. p.11724. [2]. Ramadhas AS, Muraleedharan C, Jayaraj S. Performance and emission evaluation of a diesel engine fueled with methyl esters of rubber seed oil. Renewable Energy 2005;30:1789800. [3]. Puhan Sukumar, Vedaraman N, Sankaranarayanan G, Bharat Ram Boppana V. Performance and emission study of Mahua oil (Madhuca indica oil) ethyl ester in a 4-stroke natural aspirated direct injection diesel engine. Renewable Energy 2005;30:126978.
511
Vol. 3, Issue 1, pp. 505-513
[4]. Chang DYZ, Van Gerpen JH, Lee I, Johnson LA, Hammond EG, Marley SJ. Fuel properties and emissions of soybean oil esters as diesel fuel. Journal of the American Oil Chemists Society 1996;73:154955. [5]. Gerhard V. Performance of vegetable oils and their monoesters as fuels for diesel engines. SAE 1983:831358. [6]. Rao PS, Gopalkrishnan KV. Vegetable oils and their methyl esters as fuel in diesel engines. Indian J Technol 1991;29:2927. [7]. Bhattacharya SK, Reddy CS. Vegetable oils as fuels for internal combustion engines: a review. J Agr Eng, Indian Soc Agr Eng (ISAE) 1994;57(2):15766. [8]. Senthil Kumar M, Ramesh A, Nagalingam B. Complete vegetable oil fuelled compression ignition engine. SAE paper No. 2001-28-0067, 2001. [9]. Kalam MA, Masjuki HH. Biodiesel from palmoil an analysis of its properties and potential. Biomass and Bioenergy 2002;23:4719. [10]. Canakci Mustafa, Erdil Ahmet, Arcaklioglu Erol. Performance and exhaust emissions of a biodiesel engine. Applied Energy 2006;83:594605. [11]. Nwafor OMI. Emission characteristics of diesel engine operating on rapeseed methyl ester. Renew Energy 2004;29(1):11929. [12]. Lapuerta M, Armas O, Rodriguez FJ. Effect of bio-diesel fuels on diesel engine emissions. Prog Energy Combust Sci 2008:198223. [13]. Agarwal AK. Biofuels (alcohols and bio-diesel) applications as fuels for internal combustion engines. 2007;33(3):23371. [14]. Agarwal AK, Das LM. Bio-diesel development and characterization for use as a fuel in compression ignition engines. J Eng Gas Turbines Power 2001;123(2):4407. [15]. Tsolakis A, Megaritis A, Wyszynski ML, Theinnoi K. Engine performance and emissions of a diesel engine operating on dieselRME (rapeseed methyl ester) blends with EGR (exhaust gas recirculation). Energy 2007;32:207280. [16]. Barnwal BK, Sharma MP. Prospects of bio diesel production from vegetable oils in India. Renew Sust Energy Rev 2005;9:36378. [17]. Sahoo PK, Das LM, Babu MKG, Naik SN. Biodiesel development from high acid value polanga seed oil and performance evaluation in CI engine. Fuel 2007;86:44854. [18]. Fukuda Hideki, Kondo Akihiko, Noda Hideo. Biodiesel fuel production by transesterification of oils. Journal of Bioscience and Bioengineering 2001;92(5): 40516. [19]. DK Bora, M Polly, V Sanduja, LM Das. Performance evaluation and emission characteristics of a diesel engine using Mahua oil methyl ester. SAE; 2004-28-0034. 2004. [20]. Meher LC, Naik SN, Das LM. Methanolysis of Pongamia pinnata oil for production of biodiesel. J Sci Ind Res India 2004;63:9138. [21]. Ramadhas AS, Jayaraj S, Muraleedharan C. Biodiesel production from high FFA rubber seed oil. Fuel 2005;84:33540. [22]. Heywood JB. Internal combustion engine fundamentals. New York: McGraw-Hill Book Co; 1988. [23]. Senatore A, Cardone M, Rocco V, Prati MV. A comparative analysis of combustion process in DI diesel engine freed with biodiesel and diesel fuel. SAE Paper 2001:3947. 2001-01-0691. [24]. Ski Lee Chang, Wook Park Sung, Kwon II Sang. An experimental study on the atomization and combustion characteristics of biodiesel-blended fuels. Energy & Fuels 2005;19:22018. [25]. ShahidEM,JamalY.Areviewofbiodieselasvehicularfuel.Renewable and Sustainable EnergyReviews2008;12:2484e94. [26]. Meher, L.C., Vidya Sagar, D., Naik, S.N. 2006. Technical aspects of biodiesel production by transesterificationa review. Renew Sustain Energy Rev 10:248268. [27]. Graboski, M.S., McCormick, R.L. 1998. Combustion of fat and vegetable oil derived fuels in diesel engines. Prog Energy Combust Sci 24:125164. [28]. Alam, M.; Song, J.; Acharya, R.; Boehman, A. Combustion and emissions performance of low sulfur, ultra low sulfur and biodiesel blends in a DI diesel engine. SAE paper 2004-01-3024. [29]. Nabi, M. N.; Shahadat, M. M. Z.; Rahman, M. S.; Beg, M. R. A. Behavior of diesel combustion and exhaust emission with neat diesel fuel and diesel-biodiesel blends. SAE paper 2004-01-3034. [30]. Hribernik, A. The influence of biodiesel fuel on injection, combustion, emissions, and performance of direct-injected diesel engine. J. Mech.Eng. 2006, 52, 3-14. [31]. Sanatore, A.; Cardone, M.; Rocco, V.; Prati, M. V. A comparative analysis of combustion process in DI diesel engine fueled with biodiesel and diesel fuel. SAE paper 2000-01-0691. [32]. Y. Zhang, A.L. Boehman, Impact of biodiesel on NOx emissions in s common rail direct injection diesel engine, Energy and Fuels 21 (2007) 20032012.
512
Vol. 3, Issue 1, pp. 505-513
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 Authors Biographies
Bhabani Prasanna Pattanaik was born on 14th June 1977. He obtained his Bachelors Degree in Mechanical Engg. from Utkal University, Vanibihar, Bhubaneswar in the year 2000 and M.Tech in Heat Power Engg. from Jadavpur University, Kolkata in the year 2002. He is presently pursuing Ph.D. in the Department of Mechanical Engg., Jadavpur University, Kolkata. His research areas are Non-Conventional and Renewable Energy Sources, Internal Combustion Engines, Engine Combustion, Emission, Alternative Fuels and Engine Tribology. Presently he is working as Assistant Professor in the Department of Mechanical Engg., Gandhi Institute for Technological Advancement, Madanpur, Bhubaneswar, Odisha, India.
Basanta Kumar Nanda was born on 23rd June 1967. He obtained his Bachelors Degree in Mechanical Engg. from REC, Silchar in the year 1992 and M.Tech in Production Engg. from Bengal Engineering College (D.U.), Shibpur in the year 2001. He is presently pursuing Ph.D. in the Department of Mechanical Engg., Jadavpur University, Kolkata. His research areas are Alternative Fuels for I.C.Engines, Heat Transfer and Thermodynamics. At present he is working as Associate Professor in the Department of Mechanical Engg., Maharaja Institute of Technology, Khurda, Odisha, India.
Probir Kumar Bose was born on 24th December 1955. He obtained his B.E. in Mechanical Engg. from REC, Durgapur and M.M.E. in Heat Power Engg. from Jadavpur University, Kolkata. He obtained his Ph.D. in the field of Thermal Engg. from IIT Powai, Mumbai. He has supervised 39 Master Degree and 29 Ph.D. thesis. He has 87 publications in International and National Journals. He was formerly professor in the Department of Mechanical Engg. and formerly Director, School of Automotive Engg., Jadavpur University, Kolkata. Presently he is the Director, National Institute of Technology, Agartala, Tripura, India.
513
Vol. 3, Issue 1, pp. 505-513
POWER FLOW ANALYSIS OF THREE PHASE UNBALANCED RADIAL DISTRIBUTION SYSTEM

Puthireddy Umapathi Reddy1, Sirigiri Sivanagaraju2, Prabandhamkam Sangameswararaju3
Department of Electrical and Electronics Engineering, Sree Vidyanikethan Engineering College, Tirupati-517102, India. 2 Department of Electrical and Electronics Engineering, Jawaharlal Nehru Technological University College of Engineering Kakinada, Kakinada-533 003, India. 3 Department of Electrical and Electronics Engineering, Sri Venkateswara University College of Engineering, Tirupati-517502, India.
1
ABSTRACT
This paper provides a new approach for power flow and modeling analysis of three phase unbalanced radial distribution systems (URDS) using the simple forward/backward sweep-based algorithm. A three phase load flow solution is proposed considering voltage regulator and transformer with detailed load modeling, for the transformer modeling symmetrical components theory is used and zero sequence-voltage and-current updating for the sweep-based methods is shown. The validity and effectiveness of the proposed method is demonstrated by a simple 19-bus unbalanced system for grounded wye-delta and delta grounded wye transformer connections. Results are in agreements with the literature and show that the proposed model is valid and reliable.
KEYWORDS:
Distribution System, Forward-Backward sweep-based methods, Three-phase load model analysis, Power flow analysis.
I.
INTRODUCTION
Power distribution systems have different characteristics from transmission systems [1],[2].They are characterized as Radial/weakly meshed structures, Unbalanced networks/loads: single, double and three phase loads, High resistance/reactance(R/X) ratio of the lines, Extremely large number of branches/nodes, Shunt capacitor banks and distribution transformers, Low voltage levels compared with those of transmission systems and distributed generators[3].[4]. Because of the inherent unbalanced nature of the power distribution system, each bus may be having loads that can be threephase grounded wye or ungrounded delta connected, two-phase grounded or single-phase grounded [5]. The unbalanced nature of power distribution systems requires special three phase component and system models [6]. The operation and planning studies of distribution system requires a steady state conditions of system can be obtained from the load flow solution[7],[8].The efficiency of the entire process depends heavily on the efficiency and capability of the load flow program used for this purpose[9]. Most of the researchers presented techniques, especially to obtain the load flow solution of distribution networks [10],[11] have proposed a load flow solution method by writing an algebraic equation for bus voltage magnitude. However this method is suitable for single-phase analysis [12]. A few researchers have proposed load flow solution techniques[17] to analyze unbalanced distribution networks [13],[14] have formulated load flow problem as a set of non linear power mismatch equations as a function of the bus voltages. These equations have been solved by Newtons method [15],[16] have proposed three phase power flow algorithm based on the forward and back word walk along the network. The method considers some aspects of three phase modelling of branches and detailed load modelling [18],[19]. P. Aravindhababu, proposed a method [20], A new fast decoupled
514
Vol. 3, Issue 1, pp. 514-524
power flow method for distribution systems. Improvements in the representation of PV buses on three-phase distribution power flow [21] are proposed. A new approach have given [22], [23] for three-Phase Fast Decoupled Load Flow for Unbalanced Distribution Systems. T.H. Chen, N.C. Yang, proposed three-phase power-flow by direct ZBR method for unbalanced radial distribution systems [24]. A Simple and Direct Approach for Unbalanced Radial Distribution System three phase Load Flow Solution [25] have been explained. The significance of this power flow analysis is to apply load flow data for capacitor placement, network reconfiguration, voltage regulator placement etc. in URDS. This paper presents an algorithm for solving load model and power flow analysis of three-phase unbalanced radial distribution systems. The algorithm is capable of solving for systems with many feeders emanating from grid substation with large number of nodes and branches. This paper considers all types of load modelling i.e distribution system line model, line shunt admittance model distributed load model, capacitor model, transformer modeling, Forward-Backward Sweep (FBS) load flow, algorithm for load flow, results and discussion, conclusions and references. Based on the proposed algorithm, a computer program has been developed using MATLAB and results are presented for typical network of 19-node URDS.
II.
MODELING OF UNBALANCED RADIAL DISTRIBUTION SYSTEM
Radial distribution system can be modeled as a network of buses connected by distribution lines, switches or transformers. Each bus may also have a corresponding load, shunt capacitor and/or cogenerator connected to it. This model can be represented by a radial interconnection of copies of the basic building block shown in Figure 1. Since a given branch may be single-phase, two-phase, or three-phase, each of the labeled quantities is respectively a complex scalar, a 2 1, or a 3 1 complex vector. The model consist of distribution line with are without voltage regulator or Switch or Transformer.
Figure 1. Basic building block of unbalanced radial distribution system inclusion of all models
2.1 Distribution system line model For the analysis of power transmission line, two fundamental assumptions are made, namely: Threephase currents are balanced and Transposition of the conductors to achieve balanced line parameters. A general representation of a distribution system with N conductors can be formulated by resorting to the Carsons equations, leading to a N N primitive impedance matrix. The standard method used to form this matrix is the Kron reduction, based on the Kirchhoffs laws. For instance a four-wire grounded star connected overhead distribution line shown in figure 2 results in a 4 4 impedance matrix. The corresponding equations are
515
Vol. 3, Issue 1, pp. 514-524
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 Bus p a b c n
a Vp b Vp
Ia j z Ib j I
c Vp c j
Bus q
aa j
z bb j z cc j z
nn j
In j
}z }z }z
a
ab j bc j cn j
z ac j z bn j
an j
Vqa Vqc Vqb
b c n
Figure 2. Model of the three-phase four wire distribution line
V pa V qa z aa z ab z ac z an I a j j j j j b b ba b bb bc bn z j z j z j I j V p V q z j c = c + ca (1) z cb z cc z cn I c V p V q z j j j j j n n na nb nc nn n z j z j z j I j V p V q z j It can be represented in matrix form as abc z n abc V abc V abc Z j j I j p q + (2) = n n nT nn I n Vp Vq z z j j j n n If the neutral is grounded, the voltage V p and V q can be considered to be equal. From the lst row of
eqn. (2) it is possible to obtain
1 T I n = z nn z n I abc j j j j
abc = V abc + Ze abc I abc Vp q j j
Where
ze aa ze ab ze ac j j j 1 T Ze abc = Z abc z n z nn z n = ze ba ze bb ze bc j j j j j j j j ca cb cc ze j ze j ze j
(3)
and substituting eqn.(3) into eqn. (2), the final form corresponding to the Krons reduction becomes (4)
(5)
I abc is the Current vector through line between nodes p and q can be equal to the sum of the load j
currents of all the nodes beyond line between node p and q plus the sum of the charging currents of all the buses beyond line between node p and q, of each phase. Therefore the bus q voltage can be computed when we know the bus p voltage, mathematically, by rewriting eqn. (4)
516
Vol. 3, Issue 1, pp. 514-524
aa ab ac a a V q V p ze j ze j ze j b b ba V q = V p ze ze bb ze bc j j j c c V q V p ze ca ze cb ze cc j j j I a j b I j I c j
(6)
2.2 Line shunt admittance model

These current injections for representing line charging, which should be added to the respective compensation current injections at nodes p and q, are given by
a Ishq y aa + y ab + y ac j j j b 1 y ba Ishq = j 2 c ca yj Ishq
) (
y ab j y ba + y bb + y bc j j j y cb j
y ac j
) (
y bc j y ca + y cb + y cc j j j
Vqa b Vq c Vq
(7)
2.3 Distribution System Load Model

Constant Power: Real and reactive power injections at the node are kept constant. This load corresponds to the traditional PQ approximation in single-phase analysis. Constant Impedance: These types of loads are useful to model large industrial loads. The impedance of the load is calculated by the specified real and reactive power at nominal voltage and is kept constant. Constant Current: The magnitude of the load current is calculated by the specified real and reactive power at nominal voltage and is kept constant.
2.3.1 Distributed load model

Sometimes the primary feeder supplies loads through distribution transformers tapped at various locations along line section. If every load point is modeled as a node then there are a large number of nodes in the system. So these loads are represented as lumped loads. At one fourth length of line from sending node, where two thirds of the load is connected. For this a dummy node is created. One third loads is connected at the receiving node. In the unbalanced distribution system, loads can be uniformly distributed along a line. When the loads are uniformly distributed it is not necessary to model each and every load in order to determine the voltage drop from the source end to the last loads.
2.4 Capacitor model

Shunt capacitor banks are commonly used in distribution systems to help in voltage regulation and to provide reactive power support. The capacitor banks are modeled as constant susceptances connected in either star or delta. Similar to the load model, all capacitor banks are modeled as three-phase banks with the currents of the missing phases set to zero for single-phase and two-phase banks.
2.5 Transformer modeling

Three-phase transformer is represented by two blocks shown in Figure 3. One block represents the per unit leakage admittance matrix Y abc and the other block models the core loss as a function of
voltage on the secondary side.
517
Vol. 3, Issue 1, pp. 514-524
Figure 3. General Three-phase Transformer Model
Now that Y abc is not singular, the non zero sequence components of the voltages on the primary
SP
side can be determined by
abc I Y abc V abc SS S s Similar results can be obtained for forward sweep calculation abc = Y abc VS SS

abc = Y abc VP SP
-1
(8)
-1
abc I Y abc V abc SP P s
(9)
S SS SS htat the last row is replaced with [1 1 1 ], I abc and Y abc are obtained by setting the elements in s SP the last row of I abc and Y abc to 0, respectively. Once the nonzero-sequence components of S SP abc or V abc are calculated, zero-sequence components are added to them to form the line-toVP S
neutral voltages so that the forward/backward sweep procedure can continued.
Where V abc is the nonzero sequence component of V abc , Y abc is same as Y abc , except
III.
FORWARD - BACKWARD SWEEP (FBS) LOAD FLOW METHOD
3.1 Backward Sweep:

The purpose of the backward sweep is to update branch currents in each section, by considering the previous iteration voltages at each node. During backward propagation voltage values are held constant at the values obtained in the forward path and updated branch currents are transmitted backward along the feeder using backward path. Backward sweep starts from extreme end branch and proceeds along the forward path.
Ia j
zeaa j Vqa
Ia j
ILa q
a Sq
a ICq
Figure 4.Single phase line section with load connected at node q between phase a& neutral n.
Figure 4 shows phase a of a three-phase system where lines between nodes p and q feed the node q and all the other lines connecting node q draw current from line between node p and q.
518
Vol. 3, Issue 1, pp. 514-524
During this propagation different load currents and capacitor currents (if exist) are calculated using mathematical models of loads and capacitors presented in section 2.1. The line charging currents of all the branches are added to the load current. Figure 5 shown a branch j of the distribution network, connected between two nodes p and q and M sub-laterals are connected to it. The parent branch current feeds the load at the node and the sub-laterals connected to the parent branch. This current can be calculated using Eqn. (10).
th
Figure 5. Branch j of distribution network connected to M sub-laterals
k k I abc = IL abc + q j
Where
abc Im m M
m M
abc k Y sh m
abc k 1 V q m
(10)
M represents the set of line sections connected to jth branch If capacitor bank is placed at the receiving end of the branch then capacitor current should also be included. Table 1. shows a mathematical Models of different loads(star & delta connected) which gives constant power, constant impedance and constant current. Another advantage of the proposed method is all the data is stored in vector form, thus saving an amount of computer memory. The proposed method finds extensive use in network reconfiguration, capacitor placement and voltage regulator placement studies.
Table 1. Mathematical Models of different loads
k is the half line shunt admittance of the branch in kth iteration. Y abc sh m k I abc is the branch current vector in line section j in kth iteration. j abc k is the current vector in branch m before updating in kth iteration. Im abc k 1 is the voltage vector of the branch m in (k-1)th iteration. Vq m
519
Vol. 3, Issue 1, pp. 514-524
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 3.2 Forward Sweep
The purpose of the forward sweep is to calculate the voltages at each node starting from the source node. The source node voltage is set as 1.0 per unit and other node voltages are calculated as abc k = V abc k + Ze abc Y abc V abc k I abc k Vq (11) sh p p j j Where
abc Vq
abc k are the voltage vectors of phases for pth and qth nodes respectively in kth ,Vp
ze ab j bb ze j ze cb j ze ac j bc ze j ze cc j
iteration.
ze aa j abc = ze ba Ze j j ze ca j
I abc j
is the current vector in jth branch in kth iteration.
These calculations will be carried out till the voltage at each bus is within the specified limits. Therefore the real and reactive power losses in the line between nodes p and q may be written as:
abc V abc )(I abc )* S abc = (V p q j j

Where
(12)
S abc is a vector of power loss with three, two or single phase j

abc and V abc are voltage vector of three phases at nodes p and q Vp q
I abc is the branch current vector of three phases for the section connected in between pth and qth j
node
3.3 Forward Backward sweep method algorithm

Step 1: Step 2: Step 3: Step 4: Step 5: Step 6: Step 7: Step 8: Step 9: Step10: Step11: Step12: Step13: Read input data regarding the unbalanced radial distribution system. Determine forward Backward propagation paths. Initialize the voltage magnitude at all nodes as 1 p.u and voltage angles to be 00, -1200, and 1200 for phase A, phase B and phase C respectively. Determine forward Backward propagation paths. Initialize the voltage magnitude at all nodes as 1 p.u and voltage angles to be 00, -1200, and 1200 for phase A, phase B and phase C respectively. Set iteration count k=1 and = 0.0001 Calculate load currents and capacitor currents(if exist) at all nodes. Calculate the branch currents using eqn. (10) in the backward sweep. Calculate node voltages using eqn. (11) in the forward sweep Check for the convergence, if the difference between the voltage magnitudes in two consecutive iterations is less than then go to step 9 else set k=k+1 and go to step 5. Calculate real and reactive power loss in each branch. Print voltages and power losses at each node. Stop.
520
Vol. 3, Issue 1, pp. 514-524
IV.
This method computes the power flow solution for its given radial network with its loadings and illustrated with 19 node test system of unbalanced radial distribution system. The outcome
of this paper is to apply load flow data for capacitor placement, network reconfiguration, voltage regulator placement etc. in URDS are very useful. 4.1 Example 1 A 19 node unbalanced radial distribution system is shown in Figure 6. The line and load data are given in [23]. For the load flow the base voltage and base MVA are chosen as 11 kV and 1000 kVA respectively.
Table2: Voltage and Phase angles of 19 node URDS
Existing method [23] Node No. Phase A Phase B Phase C Phase A Proposed method Phase B Phase C
Va
(p.u) 1.0000 0.9875 0.9854 0.9824 0.9820 0.9793 0.9786 0.9728 0.9659 0.9560 0.9550 0.9548 0.9544 0.9545 0.9526 0.9535 0.9536 0.9537 0.9516
V a
deg 0.00 0.01 0.00 0.03 0.03 0.04 0.04 0.06 0.08 0.10 0.10 0.11 0.10 0.10 0.11 0.13 0.10 0.10 0.13
Vb
(p.u) 1.0000 0.9891 0.9887 0.9839 0.9837 0.9808 0.9803 0.9738 0.9660 0.9555 0.9543 0.9538 0.9534 0.9539 0.9510 0.9514 0.9533 0.9531 0.9498
Vb
deg -120.06 -120.04 -120.04 -120.02 -120.03 -120.02 -120.02 -120.00 -119.97 -119.93 -119.92 -119.92 -119.90 -119.91 -119.91 -119.91 -119.91 -119.92 -119.91
Vc
(p.u) 1.0000 0.9880 0.9863 0.9830 0.9828 0.9801 0.9796 0.9735 0.9657 0.9550 0.9533 0.9536 0.9521 0.9528 0.9512 0.9522 0.9522 0.9522 0.9505
Vc
deg 120.06 120.11 120.14 120.12 120.12 120.13 120.13 120.14 120.14 120.16 120.17 120.16 120.17 120.17 120.15 120.15 120.16 120.16 120.16
Va
(p.u) 1.0000 0.9874 0.9854 0.9823 0.9820 0.9791 0.9786 0.9727 0.9657 0.9562 0.9548 0.9547 0.9544 0.9543 0.9526 0.9533 0.9534 0.9536 0.9515
V a
deg 0.00 0.01 0.00 0.03 0.03 0.04 0.04 0.06 0.08 0.09 0.10 0.11 0.10 0.10 0.11 0.13 0.10 0.10 0.13
Vb
(p.u) 1.0000 0.9890 0.9885 0.9838 0.9836 0.9805 0.9801 0.9737 0.9658 0.9552 0.9543 0.9536 0.9535 0.9537 0.9510 0.9514 0.9531 0.9530 0.9496
Vb
deg -120.00 -119.98 -119.98 -119.97 -119.97 -119.96 -119.96 -119.94 -119.91 -119.86 -119.86 -119.87 -119.85 -119.86 -119.83 -119.86 -119.90 -119.82 -119.86
Vc
(p.u) 1.0000 0.9878 0.9862 0.9829 0.9826 0.9799 0.9794 0.9733 0.9656 0.9548 0.9533 0.9535 0.9521 0.9528 0.9511 0.9521 0.9519 0.9520 0.9503
Vc
deg 120.00 120.05 120.06 120.06 120.07 120.07 120.08 120.08 120.09 120.09 120.10 120.10 120.11 120.11 120.12 120.10 120.11 120.10 120.10
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Figure 6. Single line diagram of 19 node URDS
521
Vol. 3, Issue 1, pp. 514-524
Table3: Active and Reactive Power flows of 19 node URDS
Table4: Summary of test results of 19 node URDS Description Phase A Phase B Phase C Minimum Voltage 0.9515 0.9496 0.9503 Max. Voltage regulation (%) 4.82 5.01 4.93 Total Active Power Loss (kW) 4.34 4.42 4.54 Total Reactive Power Loss (kVAr) 1.95 1.90 1.94 Total Active Power Demand (kW) 126.32 116.14 123.17 Total Reactive Power Demand (kVAr) 61.12 56.13 59.65 Total Feeder Capacity (kVA) 140.23 129.21 136.67
Voltage profile with comparison of the proposed method with existing method and active and reactive Power flows of 19 node URDS are given in table 2 and 3. Voltage variation is given in table 2, which gives better magnitudes are obtained in proposed method. The active power flow gives higher power flow capacity with proposed method shown in Table 3. The Table 4 gives summary of test results for 19 node unbalanced radial distribution systems. From table 4 it has been observed that the minimum voltage in phases A, B, C is 0.9515, 0.9496 and 0.9502 at node 19. The maximum percentage voltage regulation in phases A, B and C are 4.82%, 5.01% and 4.93%. The total active power loss in phases of A, B and C are 4.34, 4.42 and 4.54 kW and the total reactive power loss in phases of A, B and C are 1.95,1.90 and 1.94 kVA respectively. The real power losses in phases A, B and C are 3.64%, 3.96% and 3.78% and the reactive power losses are 3.32%, 3.34% and 3.33% of their total loads. The solution is converged in 4 iterations and time taken is 0.00645 seconds for 19 node URDS. The proposed method is capable of solving for systems with many feeders emanating from grid substation with large number of nodes compared with the existing method [23] and results are found satisfactory.
V.
CONCLUSIONS
In this paper, a simple algorithm has been presented to solve power flow and load modeling i.e distribution system line model, line shunt admittance model, Distributed load model, capacitor model and transformer modeling of unbalanced radial distribution networks. The proposed method has good
522
Vol. 3, Issue 1, pp. 514-524
convergence property for practical distribution networks with practical R/X ratio. Computationally, this method is extremely efficient; as it solves simple algebraic recursive equations for voltage phasers and another advantage is all the data is stored in vector form, thus saving computer memory. The Forward-Backward Sweep (FBS) algorithm is capable of solving for systems with many feeders emanating from grid substation with large number of nodes and branches. A computer program has been developed using MATLAB and results are presented for typical network of 19-node URDS.
REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] W.H.Kersting, A method to the design and operation of a distribution system, EEE Trans. on Power Apparatus and Systems, Vol.PAS-103 1984, pp.1945-1952. Chen. T. H, Chen M. S, Hwang. K. J, Kotas. P, Chebli. E. A., "Distribution system power flow analysis-a rigid approach," IEEE Trans. on Power Delivery, vol. 6, no. 3, pp. 1146-1153, Jul. 1991. Chen. T. H, Chen M. S, Hwang. K. J, Kotas. P, Chebli. E. A, Three phase generator and transformer models for distribution analysis IEEE Trans. power Del., vol. 6, no. 4, pp- 1671-1681, Oct. 1991. Bishop. M.T, Foster J.D, Down D. A, "The application of single phase voltage regulators on threephase distribution systems," Rural Electric Power Conference, pp. C2/1- C2/7, April 1994. Das.D, Kothari.P, Kalam.A, "Simple and efficient method for load flow solution of radial distribution systems," Electrical Power and Energy Systems, vol. 17, no. 5, pp. 35-346, Oct. 1995. Shirmohammadi. D, Carol S. Cheng. A, A Three phase power flow method for real time distribution system analysis, IEEE Trans. on Power System, Vol. 10, No. 2, May 1995. Zimmerman. R. D, Chiang. H. D, "Fast decoupled power flow for unbalanced radial distribution systems," IEEE-PES Winter Meeting, paper no. 95, New York, 1995. Ghosh. S, Das.D, Method for load flow solution of radial distribution networks, IEE Proc. Generation, Transmission and Distribution, Vol. 146, No. 6, pp. 641-648, 1999. Goswamy. S. K, Basu, "Direct solution of distribution systems," IEE Proc., pt. C, vol. 188, no. 1, pp. 78-88, 1999. Thukarm. D, Wijekoon Banda. H.M, Jerome.J, "A Robust three phase power flow algorithm for radial distribution systems," Electric Power System Research, vol. 50, no. 3, pp. 227-236, Jun. 1999. Lin. W.M, Su. Y.S, Chin .H.C, Teng. J.H , "Three-Phase unbalanced distribution power flow solutions with minimum data preparation," IEEE Trans. on Power Systems, vol. 14, no. 3, pp. 1178-1183, Aug. 1999. Lin.W. M, Teng. J. H., "Three-Phase distribution networks fast decoupled power flow solutions," Electric Power and Energy Systems, vol. 22, no. 5, pp. 375-380, Jun. 2000. Garika. P. A. N, Pereia. J. L. R, Camerio. S, Dacosta . V. M, Martin.S N., "Three-Phase power flow calculations using the current injection method," IEEE Trans. on Power Systems ,vol. 15, no. 2, pp. 508514, May 2000. Mui K. N, Chiang H. D, Existence, uniqueness and monotonic properties of the feasible power flow solution for radial three phase distribution networks, IEEE Trans. Circuit and Syst., vol.47, no.10, pp.1502-1514, Oct. 2000. Teng. J. H, A Network-Topology-based Three-Phase Power Flow for Distribution Systems, Proceedings of National Science Council ROC (A), vol.24, no.4, pp.259-264, 2000. Kersting W . H, Distribution System Modeling and Analysis, CRC Press 2000. Garika. P. A. N, Pereia. J. L. R, Camerio., Voltage control devices models for distribution power flow analysis, IEEE Trans. Power Syst., vol. 16, no. 4, pp. 586594, Nov. 2001. Teng. J. H, "A Modified gauss- seidel algorithm of threephase power flow analysis in distribution network," Electrical Power and Energy Systems, vol. 24, no. 2, pp. 97-102, Feb. 2002. S. Herraiz, L. Sainz, and J. Clua, Review of harmonic load flow formulations, IEEE Trans.Power Del., vol. 18, no. 3, pp. 10791087, Jul.2003. P. Aravindhababu, A new fast decoupled power flow method for distribution systems, Electric power Components and Systems, vol.31, no. 9, pp. 869878, 2003. Garika. P. A. N, Pereia. J. L. R, Camerio, Improvements in the representation of PV buses on three-phase distribution power flow, IEEE Trans. Power Del., vol. 19, no. 2, pp. 894896, Apr. 2004. Jen-Hao Teng, Chuo-Yean Chang,Backward/Forward Sweep-Based Harmonic Analysis method for Distribution Systems IEEE Trans. Power Del.,vol. 22, no. 3, pp. 16651672, July.2007. A. R. Hatami and M. Parsa Moghaddam, Three-Phase Fast Decoupled Load Flow for Unbalanced Distribution Systems, Iranian Journal of Electrical and Computer Engineering, vol. 6,no.1,winter-spring 2007.
[12] [13]
[14]
[15] [16] [17] [18] [19] [20] [21]
[22] [23]
523
Vol. 3, Issue 1, pp. 514-524
[24] T.H. Chen,N.C. Yang, Three-phase power-flow by direct ZBR method for unbalanced radial distribution systems, IET Generation Transmission Distribution, Vol. 3, Issue 10, pp. 903 910, 2008. J.B.V. Subrahmanyam, C. Radhakrishna and K. Pandukumar, A Simple and Direct Approach or Unbalanced Radial Distribution System three phase Load Flow Solution Research Journal of applied Sciences, Engineering and Technology, vol. 2, no.5, pp. 452459, Augest,2010.
[25]
Authors P.UMAPATHI REDDY: He Received B.E from Andra University and M.Tech.,(Electrical Power Systems) from Jawaharlal Nehru Technological University, Anantapur, India in 1998 and 2004 respectively, Now he is pursuing Ph.D. degree. Currently he is with Department of Electrical and Electronics Engineering, Sree Vidyanikethan Engineering College, Tirupati, India. His research interest includes Power distribution Systems and Power System operation and control. He is Life Member of Indian Society for Technical Education.
S.Sivanaga Raju: He received B.E from Andra University and M.Tech.degree in 2000 from IIT, Kharagpur and did his Ph.D from Jawaharlal Nehru Technological University, Anantapur, India in 2004. He is presently working as Associate professor in J.N.T.U.College of Engineering Kakinada,(Autonomous) Kakinada, Andrapradesh, India. He received two national awards (Pandit Madan Mohan Malaviya memorial Prize and best paper prize award from the Institute of Engineers (India) for the year 2003-04. He is referee for IEEE journals. He has around 75 National and International journals in his credit. His research interest includes Power distribution Automation and Power System operation and control. P. Sangameswara Raju: He is presently working as professor in S.V.U. College Engineering, Tirupati. Obtained his diploma and B.Tech in electrical Engineering, M.Tech in power system operation and control and Ph.d in S. V. University, Tirupati. is areas of interest are power system operation, planning and application of fuzzy logic to power system, application of power system like non-linear controllers.
524
Vol. 3, Issue 1, pp. 514-524
HYPERSPECTRAL IMAGE CLASSIFICATION USING M-BAND WAVELET TRANSFORM

K. Kavitha, S. Arivazhagan, R. Dhivya Priya
Electronics and Communication Engineering Department, Mepco Schlenk Engineering College, Sivakasi, India
ABSTRACT
This work presents a new technique that classifies various land cover classes present in the hyperspectral images by making use of the features from transformed version of the input image. A hyperspectral image is characterized by a large dimensionality data, recorded at very fine spatial resolution in hundreds of narrow frequency bands. These bands provide a wealth of spatial and spectral information of the scene, imaged by the sensors. Being inspired by the properties of M-band wavelets, a transform called M-Band Wavelet Transform (MBWT) is used. All the bands of the hyperspectral dataset are decomposed into sub-bands by using MBWT. Then, the features such as mean, standard deviation, variance, skewness and kurtosis are extracted from all the sub-bands of M-Band Wavelet transformed image. The extracted features are used for classification. Support Vector Machines with Binary Hierarchical Tree (BHT) is developed to classify the data by One Against All (OAA) methodology. Then, the Performance of MBWT is evaluated. The experiment is conducted on the AVIRIS hyperspectral dataset taken over the North-western Indianas Indian Pine Site.
KEYWORDS:
M-Band Wavelet Transform (MBWT), Hyperspectral Image Classification, Multi-class Classifier, Support Vector Machine (SVM), OAA, BHT
I.
INTRODUCTION
Over the past decade hyperspectral image analysis has matured into one of the most powerful and fastest growing technologies in the field of remote sensing. The hyper in hyperspectral means over as in too many and refers to the large number of measured wavelength bands. Hyperspectral images are spectrally over determined, which means that they provide ample spectral information to identify and distinguish spectrally unique materials and each pixel in a Hyperspectral image has its own characteristic spectrum. It provides the potential of wealthy information extraction which was not possible with any other type of remotely sensed data. David Landgrebe proposed that high dimensional data can have the substantially increased capability for deriving more detailed and more accurate information [1]. A generalized procedure for hyperspectral data analysis is suggested and it involves identification and labeling of training samples followed by feature extraction and classification [1]. Classification of heterogeneous classes present in the Hyperspectral image is one of the recent research issues in the field of remote sensing. Classifying the pixels in the Hyperspectral Images and identifying their relevant class belongings depends on the feature extraction and classifier selection processes. Feature extraction is an agnate process while classifying images. In Remote Sensing, the number of training samples available is often limited and in order to avoid the problems caused by the limited training samples, several feature extraction methods based on the wavelet transform have been proposed for hyperspectral images [2]. DWT gives better accuracy when used for texture classification [3]. P.S.Hiremath discussed the concept of extracting the features from wavelet sub-bands [4]. A wavelet transform-based texture classification algorithm has several important characteristics like the ability of decorrelating the data, providing orientation sensitive information and reduced computational complexity [4]. One of the drawbacks of standard wavelets is that they are not suitable for the analysis of high- frequency signals with relatively narrow bandwidth [5]. To overcome this problem, M-band orthonormal wavelet were created as a direct generalization
525
Vol. 3, Issue 1, pp. 525-533
by Daubechies [6]. It has been reported that M-band orthonormal wavelets are able to zoom onto narrow band high frequency components of a signal as they give a better energy compaction than 2band wavelets [7]. In general, the M-Band wavelet transforms are more efficient in signal decomposition than traditional wavelet transforms [8]. Chitre and Dhawan [9] have used the M-band wavelet for texture classification. Classification of medical images using M-band wavelet transforms is a promising technique for getting higher accuracy [10]. Always, Hyperspectral datasets are non-linear in nature. The Kernel methods used in machine learning models converts these non-linear datasets into a linear one, thereby making it useful for applications like classification, regression and clustering [11-12]. Kernel methods are suitable for the classification of high dimensional data while the availability of the training samples is limited. Many types of kernels like linear, polynomial, radial Basis Function (RBF), Sigmoid etc., are available. Selection of proper kernel gives proper results. Support Vector Machine (SVM) with Radial Basis Function (RBF) is a preferred combination which balances the complexity and accuracy [13]. The usage of SVM classifier for Hyperspectral images is shown by J.Gualtieri [14]. Multiclass classifier for Hyperspectral images is explained by T.Joachims [11]. The support vector machine (SVM) with kernel trick has been successfully used in hyperspectral image classification [15]. Yuliya uses probabilistic Support Vector Machine (SVM) for pixel wise classification of Hyperspectral images followed by the Markov Random Field (MRF) regularization method for refining the classification results based on spatial contextual information [16]. From the literature, it is obvious that, for classification, the textural information is important and it can be extracted with the help of wavelets. So, in this proposed methodology, transform domain is used for classification. The transform domain enhances the classification rate by decorrelating the pixels in an image into different sub-band images. First the hyperspectral image is transformed using M-Band Wavelet Transform. Then the features are extracted from the transformed images. The extracted features are used for efficient classification of hyperspectral image. Rest of the paper is organized in the following manner. Section-II deals with the Proposed Work followed by the background of SVM as Section-III. Section-IV describes the Experiment Design and Section-V is dedicated to the Results and Discussions. SectionVI gives the conclusion about the work.
II.
PROPOSED WORK
2.1. M-Band Wavelet Transform

Standard M-band wavelet is a new part of wavelet analysis theories and it can offer wider range of wavelet base selection and wavelet function with better properties than two band wavelet. Present applications of M-band wavelet include image compression, edge detection, image fusion, image classification, etc. One of the drawbacks of standard wavelets is that they are not suitable for the analysis of high- frequency signals with relatively narrow bandwidth. Unlike the standard wavelet decomposition, which gives a logarithmic frequency resolution, the M-band decomposition gives a mixture of a logarithmic and linear frequency resolution. Further as an additional advantage, M-band wavelet decomposition yields a large number of sub-bands, which is required for improving the classification accuracy.
Figure 1. M-Band Wavelet decomposition of an image (M=3)
M-band wavelet transform is used to decompose the image into MXM sub-bands [18] i.e., an M-level decomposition yields M2 sub-bands as shown in figure 1. Because M2 channel will be generated through M-band decomposition at the first level, further decomposition of M-band is not suggested.
526
Vol. 3, Issue 1, pp. 525-533
Compared with traditional two-band wavelet, the M-Band wavelets have three advantages, when applying to feature extraction. M-band wavelet overcomes the limitation of two-band wavelet in wavelet base selection. Besides logarithmic frequency resolution, M-band wavelet can offer linear frequency resolution. More sub-bands are generated and more detailed information is included at each level, making it suitable for enhancing the accuracy of classification. M-channel filters used for decomposing the time-scale space into MM sub-bands play a vital role in increasing the classification accuracy and are essentially selected as frequency and direction oriented band pass filters. The different combinations of M-Band Wavelet filters decompose an image at different scales and orientations in frequency space. Figure 2 shows the M-channel filter bank for M=3. In general, the wavelet transform of an image is defined as the 2D signal decomposition onto a set of basis functions called wavelets. A full wavelet expansion of a 2D signal can be represented by a set of basis functions called scaling and wavelet functions ( and ), respectively and are associated with the analyzing (or synthesizing) filters h and g. M-band wavelets are a set of M-1 basis functions. Scaled and translated versions of basis functions forms a tight frame for the set of square integrable functions defined over the set of real numbers L2(R) [17]. For the M-1 wavelets, l(x), l=1,2,...M-1, given any function f(x) L2(R), it has been shown that, = < , l,m,n > l,m,n (1) where l(x) and l,m,n(x) are scaled and translated wavelet functions and Z represents the set of integers. Multiresolution analysis is also defined with the scaling function and the M-1 wavelet functions [17]. A multiresolution analysis is a sequence of approximation spaces for L2(R). If the spaces spanned by the scaling function and wavelet function at different resolution for fixed m and n Z is denoted by Vm and Wm, then for M=3, it can be shown that [17], V1=V0 W10 W20 (2) V2=V1 W11 W21 (3) A typical M-channel filter bank structure for M=3 is shown in figure 2.
Figure 2. M-channel filter bank (M=3)
2.2. Feature Extraction

Transforming the input data into the set of features is called feature extraction. The extracted feature vectors are given as input to a classifier. By applying transforms, the pixels present in the image can be decorrelated and this in turn increases the classification rate. All the bands of the hyperspectral dataset are decomposed into sub-bands by using MBWT. Then, the features such as mean, standard deviation, variance, skewness and kurtosis are extracted from all the sub-bands of M-Band Wavelet transformed image.
527
Vol. 3, Issue 1, pp. 525-533
International Journal of Advances in Engineering & Technology, March 2012. Engineering IJAET ISSN: 2231-1963
HYPERSPECTRAL IMAGE
M-BAND WAVELET TRANSFORM
FEATURE EXTRACTION
CLASSIFICATION Figure 3. Proposed Methodology
The feature named Mean is used to average out the image thus eliminating the noise. The Standard deviation feature is used to find how far the pixels are deviated or dispersed from the mean. The Variance feature of a data set is calculated by taking the arithmetic mean of the squared differences between each value and the mean value. The feature that is used to find whether the pixel distribution is skewed to the left or right of the mean is termed as Skewness. The Kurtosis feature is the measure of whether the distribution is tall and skinny or short and squat, compared to the normal distribution of the same variance. The features are calculated using the formulas given in Table 1.
Table 1. Feature Set Feature Mean Standard Deviation Variance Skewness Kurtosis Formula ij = (
ij= 2 ij ij
)/MN
2
( =(
ij - ij)
) 0.5/MN
2
ij - ij)
) /MN
3 ij 4 ij
Skew(xij )= Kurt(xij )= (
ij
- ij )3/ - ij )4/
ij
)-3
III.
SUPPORT VECTOR MACHINE
SVM is a good candidate for remote sensing data classification for a number of reasons. SVM outperforms other classifiers as they work well with a small training data. They can support more number of features during the classification task. SVM performs nonlinear classification using kernel nonlinear trick. Kernel-based methods are based on mapping data from the original input feature space to a based kernel feature space of higher dimensionality and solving a linear problem in that space. The aim is to find a linear separating hyper plane that separates classes of interest. The hyper plane is a plane in a arating multidimensional space and is also called a decision surface or an optimal separating hyper plane. Radial Basis Function Kernel is used here. This kernel nonlinearly maps samples into a higher dimensional space, unlike the linear kernel, and so it can handle the case when the relation between class labels and attributes is nonlinear as shown in figure 4.
528
Vol. 3, Issue 1, pp. 525-533
Figure 4. Separable classification using RBF kernel in a) Original Space b) Feature Space
Here, One-against-all (OAA) Binary Hierarchical Tree strategy is used by SVMs while classifying images. It separates the classes hierarchically by considering how one class is separated from others.
Figure 5. OAA Binary Hierarchical Tree (Wi represents Classi)
Likewise so many SVMs can be run to find out the interested classes. While training, care is shown towards the over fitting of the samples.
IV.
EXPERIMENTAL DESIGN
The experiment is conducted on the AVIRIS hyperspectral dataset taken over the Northwestern Indianas Indian Pine Set. The dataset consists of 220 bands and each band consists of 145x145 pixels. The original dataset contains 16 classes and ground truth is available for that.
Figure 6. Ground Truth Classes
V.
The Feature set derived from the Feature Extraction method is used for classification. Randomly chosen pixels from each class and their corresponding feature vectors are used for training. The classifier produces output based on, whether the particular pixel under test belongs to the interested trained class or not. Thus, the pixels under the same class are separated from whole dataset.
529
Vol. 3, Issue 1, pp. 525-533
International Journal of Advances in Engineering & Technology, March 2012. Engineering IJAET ISSN: 2231-1963
Similarly other classes also trained and by this method, the classes are separated hierarchically. After that the pixels of interested class are assigned white gray level while others are assigned black. By comparing each class of pixels with the ground truth, the number of misclassified pixels can be found and from that the average accuracy can be cal calculated.
Figure 7. Classified output
From figure 8, it is inferred that, for nearly two fourth of the classes, the average accuracy obtained two-fourth by the proposed feature set reaches more than 50%. In specific, classes like Alfalfa, Grass Pasture Mowed, Oats, Soy clean, Wheat and Steel, gives an overall accuracy of more than 50% whereas the classes like Hay-Widrowed, Corn Widrowed, Corn-min, Corn, Grass Trees and Building-Grass Tree Drives attains Grass average accuracy of more than 45%. By comparing the classification result with the ground truth available, the overall accuracy of classification is obtained and is found to be 68.4%. ,
530
Vol. 3, Issue 1, pp. 525-533
Figure 8. Average Accuracy
Figure 9. Classified output with OA= 68.4%
VI.
CONCLUSIONS
The proposed work developed a classification scheme that uses statistical features alone while classifying an image. It is predicted that those features alone gave an overall classification accuracy of 68.4%. When this application is extended to sub-pixel levels, it is possible to reduce the misclassifications and hence there comes the chance of improvement in overall classification accuracy. For this case of analysis, the knowledge of how much a pixel belongs to a particular class is important. Therefore, proper soft classification algorithms can be identified and can be used to classify the land cover classes, as they are expected to yield good accuracies.
ACKNOWLEDGEMENTS
The authors are grateful to the Management and Principal of our college for their constant support and encouragement.
REFERENCES
[1]. David Landgrebe, G.E. Cohn, J.C. Owicki, (1999) Some Fundamentals and Methods for Hyperspectral Image Data Analysis, SPIE Photonics West, San Jose CA.
531
Vol. 3, Issue 1, pp. 525-533
[2]. Hsu, P.H. , (2003) Spectral Feature Extraction of Hyperspectral Images Using Wavelet Transform, Ph.D. Thesis, Department of Surveying Engineering, National Cheng Kung University, Tainan, Taiwan, R.O.C. Kamarul Hawari Ghazali, Mohd. Marzuki Mustafa, Aini Hussain, (2009 ) Image Classification using Two Dimensional Discrete Wavelet Transform, International Conference on Instrumentation, Control & Automation, pp71-74. P. S. Hiremath, S. Shivashankar, (2006) Wavelet Based Features for Texture Classification, GVIP journal, Vol.6, Issue-3. P. Steffen et. al., (1993) Theory of regular M-band wavelets bases, IEEE Trans. on Signal Processing, Vol. 41, No. 12, pp3497-3510. I. Daubechies, (1988) Orthonormal bases of compactly supported wavelets, Communication Pure Appl. Math. XLI, pp909-996. H. Zou, A.H. Tewfik, (1992) Discrete orthogonal M-band wavelet decompositions, Proc. Int. Conf. on Acoustic Speech and Signal Processing, Vol.4, ppIV-605-IV-608. Ganesh D. Bhokare, Raghuram J. Karthik, Vikram M. Gadre, (2005) Modified EZW Coding for the M band Wavelet Transform and its application to Image Compression, IEEE Workshop on Statistical Signal Processing Proceedings, pp883-888. Y. Chitre and A.P. Dhawan, (1999) M-band wavelet discrimination of natural textures, Pattern Recognition, Vol. 32, pp773-789. Wen-Li Lee, Yung-Chang Chen, Kai-Sheng Hsieh, (2003) Ultrasonic liver tissues classification by fractal feature vector based on M-band wavelet transform, IEEE Transactions on Medical Imaging, Vol. 22, Issue-3. B. Scholkopf, A. Smola, (2002) Learning with KernelsSupport Vector Machines, Regularization, Optimization and Beyond, MIT Press Series. G. Camps-Valls, J. L. Rojo-Alvarez, and M. Martinez-Ramon, (2007) Kernel Methods in Bioengineering, Signal and Image Processing, Idea Group Publishing, Hershey, PA. G. Camps-Valls, L. Bruzzone, (2005) Kernel-based methods for hyperspectral image classification, IEEE Trans. Geosci. Remote Sens. Vol. 43, No. 6, pp1351-1362. J.Gualtieri and R. Cromp, (1998) Support vector machines for hyperspectral remote sensing classification, Proc. 27th AIPR Workshop Advances in Computer Assisted Recognition, Washington DC, pp121-132. G. Camps-Valls, L. Gomez-Chova, J. Munoz-Mar, J. Vila-Frances, J. Calpe-Maravilla, (2006) Composite kernels for hyperspectral image classification, IEEE Geosci. Remote Sens. Lett. , Vol. 3, No. 1, pp93-97. Y. Tarabalka, M. Fauvel, J. Chanussot, J. A. Benediktsson, (2010) "SVM and MRF-based method for accurate classification of hyperspectral images, IEEE Geoscience and Remote Sensing Letters, Vol. 7, Issue-4, pp736-740. Mausumi Acharya, Malay. K. Kundu, (2000) Two Texture Segmentation using M-Band Wavelet Transform, 15th International Conference on Pattern Recognition (ICPR00), Vol. 3, pp.3405 Turgut Aydm, Yucel Yemez, Emin Ananm, Bulent Sankur, (1996) Multidirectional and Multiscale Edge Detection via M-Band Wavelet Transform, IEEE Transactions on Image Processing, Vol. 5, No. 9, pp1370-1377.
[3].
[4]. [5]. [6]. [7]. [8].
[9]. [10].
[11]. [12]. [13]. [14].
[15].
[16].
[17]. [18].
Authors K. Kavitha is Assistant Professor, Department of ECE, Mepco Schlenk Engineering College, Sivakasi. She received her B.E degree in Electronics and Communication Engineering from Mepco Schlenk Engineering College in 1998 and M.E. degree in Microwave and Optical Engineering from Alagappa Chettiar College of Engineering and Technology, Karaikudi in 2000. She is currently working toward the Ph.D. degree with the Anna University, Tirunelveli. Her research interests include Image Classification and Change Detection. S.Arivazhagan is Professor & Head, Department of ECE, Mepco Schlenk Engineering College, Sivakasi. He received his B.E degree in Electronics and Communication Engineering from Alagappa Chettiar College of Engineering and Technology, Karaikudi in 1986 and M.E. degree in Applied Electronics from College of Engineering, Guindy, Anna University, Chennai in 1992. He has been awarded with Ph.D. degree by Manonmaniam Sundaranar University, Tirunelveli in the year 2005. He has twenty six years of teaching and research experience and currently working as Professor & Head, Department of Electronics and Communication
532
Vol. 3, Issue 1, pp. 525-533
Engineering, Mepco Schlenk Engineering College, Sivakasi.. He has been awarded with Young Scientist Fellowship by TNSCST, Chennai in the year 1999. He has published around 130 Technical papers in International / National Journals and Conferences. He has completed 9 Research and Development Projects, funded by ISRO, Trivandrum, DRDL, Hyderabad, ADE, Bangalore, DRDO, New Delhi, DST, New Delhi and AICTE, New Delhi. R. Dhivya Priya is Post Graduate Student in Communication Systems, Mepco Schlenk Engineering College, Sivakasi. She received her B.E degree in Electronics and Communication Engineering from Arulmigu Kalasalingam College of Engineering, Krishnankoil in 2010. Her area of research is Hyperspectral Image Classification and Analysis.
533
Vol. 3, Issue 1, pp. 525-533
MODELING AND CONTROLLER DESIGN OF AN INDUSTRIAL OIL-FIRED BOILER PLANT

Orosun Rapheal.1 and Adamu Sunusi Sani2
2
Savannah Sugar Company Numan, Nigeria Department of Electrical Engineering, Bayero University Kano, Nigeria
ABSTRACT
In this study, a physical boiler system is modeled as a multivariable plant with two inputs (feed water rate and oil-fired flow rate) and two outputs (steam temperature and pressure). The plant parameters are modeled by identification based on experimental data collected directly from the plant. The routines of System Identification Toolbox with structure selection for Autoregressive Moving Average together with Recursive Least Square (ARX) were used to identify the model. The identified ARX model was validated using Akaike's Final Prediction Error (FPE) criterion. The identified model was further subjected to test, using the validation input data; simulated model outputs for both temperature and pressure agree closely with the actual plant outputs with error of 8% and 9% respectively. Furthermore, Proportional Integral Derivative (PID) Controller was developed to control the identified model. Simulation studies were carried out; the results obtained indicate the effectiveness of the technique. The controller was able to track the temperature and pressure set points steadily and rapidly.
KEYWORDS: Boiler Modeling, Identification, ARX model, Akaikes Criterion, PID controller.
I.
INTRODUCTION
The function of a boiler is to deliver steam of a given quality (i.e. temperature and pressure) to a steam turbine and other process equipments [17; 20]. For instance, in Savannah Sugar Company Limited Numan, Nigeria, power generation is by thermal power plant whereby electric power is generated from combustion of black oil. Investigation of the dynamics of power plant boiler requires detailed models with representation of plant components [8; 15; 16]. Large-scale models are generally based on first-principle equations (i.e. mass, momentum and energy balances), with phenomenological correlations (e.g. heat transfer correlations), and may be considered as knowledge models [19; 22]. However, to justify the basic structure of oil-fired boilers control systems, a different kind of model may be very helpful where only first-cut dynamics is captured in order to reveal the essential input-output interactions [6]. Such models may be called interpretation or identification models. Successful identification of nonlinear plant models is more problematic since, in addition to stimulating the dynamics, the plant must now be exercised across the operating range over which modeling is needed. One strategy which gave acceptable results in earlier work was to step the control inputs to drive the plant through its range of operating points while superimposing pseudo random binary signals at the same time [6]. This operation is difficult to effect while the boiler is in operation. On the other hand, to handle such a complex system with several inputs and outputs is complicated [11]. Therefore, the input and output data collected from the SSCLN power plant will be used for system modeling. The oil-fired boiler system (OFBS) is very complex for which well organized control functions may result in real benefit both in terms of efficiency and plant availability [21]. Hence, sophisticated distributed control systems (DCS) have been applied to OFBS, incorporating traditional PID controllers, model-based feed forward compensators, parameter scheduling, Boolean functions and so on [7; 24].
534
Vol. 3, Issue 1, pp. 534-541
The Industrial Oil Fired Boiler System (OFBS) continuously supplies 50tonns/hr of steam at a temperature of 420oC and pressure of 3200KPa to a 3.2MWatts turbine for power generation and production process use. At the time of carrying out this study, the boiler plant at SSCLN was operated mostly on open-loop with human operators manually adjusting the plants variable so as to achieve some form of control. This result in frequent boiler failures with prolonged production stoppages, not to mention the health and environmental impact of excess heavy black smoke (containing CO2 and CO which is a green house gas) released into the atmosphere due to incomplete combustion of black oil.
II.
DATA COLLECTION AND PRE-PROCESSING
During operation a sampling period of 20 seconds was selected as a good compromise between the need of a low sampling rate according to the control design and wishes of the boiler engineers to have new control actions computed sufficiently rapidly after important perturbation. The input-output data for model construction and validation had been collected during different days. Therefore, the involved input sequence was detected by production constraints. The data was taken from where inputs suffered significant variations. However, it was checked a posteriori that the considered sequence was persistently excited. The input raw data was converted to equivalent input flow rate data as follows: The raw data was taken using pressure and temperature gauges, but these quantities are controlled by flow control valves hence the need to convert the input variables into flow rates. The control valve characteristics are related to the pressure by [4];
P = 20 10 3 Qi
(1)
Where P , is pressure of the fluid, i is the flow rate of the fluid. From equation (1), the flow rate is obtained as:
Qi =
P 20 10 3
( 2)
III.
BOILER SYSTEM IDENTIFICATION
System identification allows one to build a mathematical model of a dynamic system based on measured data [26]. This is achieved by adjusting parameters within a given model until its output coincides as close as possible with measured output.
Figure 1. Components of a steam boiler plant
535
Vol. 3, Issue 1, pp. 534-541
To obtain a model that is satisfactory under a given environmental condition, measured input output data from the system were used to provide the equivalent estimated input and output parameters which are used for building the parametric model. Parametric model are linear and the recursive nature of self-tuning algorithms suggests that discrete time modeling is appropriate [7; 12]. The model identification approach is based on the fundamental assumption that the plant dynamics can be well approximated by a finite order slowly varying linear system [25].
3.1
Model choice
Linear models are simple and easy to implement on wide range of controllers. A linear model is often sufficient to accurately describe system dynamics and, in most cases of system identification, fitting linear models are tried first [25; 27]. The initial trial of fitting linear models gave acceptable preliminary results. Hence, an ARX model structure (Autoregressive Moving Average together with Recursive Least Square) has been chosen as;
y (t ) + a1 y (t 1) + ... + a na y (t n a ) = b1u (t 1) + ... + bnb u (t n k nb + 1) + e(t )

The parameters na and nb are the orders of the ARX model, and nk is the delay. A more compact way to write the difference equation is:
(3)
y1 u1 e1 A(q) = B(q) + y2 u 2 e2
where q is the delay operator.
( 4)
[ y1 , y 2 ] is the output vector, with y1 as the steam temperature and y2 as steam pressure. [u1 ,u 2 ] is input vector, with u1 as feed water flow rate and u2 as the oil flow rate. [e1 , e2 ] is the model error vector which is white noise.
A(q) = 1 + a1 q 1 + ... + a na q na B(q) = b1 + b2 q + ... + bnb q
1 nb +1
(5) ( 6)
A(q) and B(q) parameters was estimated using MATLAB Identification Toolbox. The ARX model structure is sufficient to represent random drift disturbances, which influence the system performance [25].
3.2
Design of PID Controller
A PID controller is a simple three-term controller. The transfer function of the most basic form of PID controller was used [14; 23]. The schematic diagram of the PID control of the boiler plant system is shown in Figure 2. The two non-interactive loops presented in Figure 2 are each controlled by independent PID. The PID controllers are chosen because of their good performance and robustness features in many industrial applications [14; 18; 23].
Pl ant
PID PID Con trol l er 1
In 1
Ou t1
Scope PID PID Control l er 2 In 2 Ou t2
M od el
Figure 2. Schematic diagram of the PID control boiler plant
536
Vol. 3, Issue 1, pp. 534-541
IV.
IDENTIFICATION AND VALIDATION RESULTS
The dynamic model of the boiler process is obtained in which the main variables of the boiler process are identified using ARX model structure around operating points. The routines of identification toolbox with structure selection for ARX model were used [19; 27].
Figure 3. Output y1 due to input u1
Figure 4. Output y2 due to input u2
The data of Figure 3 and Figure 4 are divided into two equal half with the first part as working data and the other half as validation data. Sampling time is twenty seconds and the excitation signal is a discrete white noise. Figure 5 represents the estimated data obtained for validation set using ARX model for the output temperature. A good prediction behavior is observed as indicated in Figure 6, which shows the error plot of measured and simulated model output temperature.
Figure 5. Measured and simulated model output temperature
537
Vol. 3, Issue 1, pp. 534-541
Figure 6. Error plot of measured and simulated model output temperature
Figure 7. Measured and simulated model output pressure
Figure 8. Error plot of measured and simulated model output pressure
Similarly, Figure 7 represents the estimated data obtained for validation set using ARX model for the output pressure. A good prediction behavior is observed as indicated in Figure 8, which shows the error plot of measured and simulated model output pressure. Discrete-time ARX model:
A( q ) y (t ) = B ( q )u (t ) + e(t )
(7 )
A(q ) = 1 1.016q B(q ) = 13.33q

1
+ 0.03743q
2
(8) (9)
+ 5.668q
Equation (8) and (9) are the estimated A( q ) and B ( q ) parameters for temperature sub-model using MATLAB System Identification Toolbox.
A(q ) = 1 1.046q 1 + 0.04763q 2 B(q ) = 1717 q 1697 q

1 2
(10) (11)
Equation (10) and (11) are the estimated A( q ) and B ( q ) parameters for pressure sub-model using MATLAB System Identification Toolbox.
538
Vol. 3, Issue 1, pp. 534-541
The ARX model was validated using Akaike's Final Prediction Error (FPE) criterion. According to Akaike's theory, the most accurate model has the smallest FPE [9; 25]. Akaike's Final Prediction Error (FPE) is defined by [9; 25]:
1+ FPE = V 1
d N d N
(12)
Where V is the loss function, d is the number of estimated parameters, and N is the number of values in the estimation data set. The toolbox assumes that the final prediction error is asymptotic for d<<N and uses the following approximation to compute FPE:
FPE = V (1 +
2d N
(13)
The loss function V which is the normalized sum of squared prediction errorsis defined by [25]:
1 N T V = det . N (t , N )( (t , N )) 1
Where,
N
(14)
represents the estimated parameters.
V = 1.21 and FPE = 1.22 for the temperature sub-model. V = 6.578 and FPE = 6.622 for the pressure sub-model.
4.1 PID Simulation Result and Discussion
A simulation study was made to determine the degree of improvement that could be gained in SSCLN boiler plant by the application of model proportional integral derivative control. Figure 9 and Figure 10 show the tracking response for the steam temperature and pressure using PID controller.
Figure 9. PID output of boiler plant (Temperature)
Figure 10. PID output of boiler plant (Pressure)
539
Vol. 3, Issue 1, pp. 534-541
PID was able to overcome the startup instability experienced by some industrial controllers [5]. The temperature and pressure track the set point steadily. Any fluctuations and disturbance was handled without affecting the output from the controller. It is evident that the performance of the PID controller is satisfactory.
Table 1. PID controller parameters PID Tuning Parameters Proportional Gain, KP Integral Gain, KI Derivative Gain, KD Output Limit Upper Output Limit Lower Output Initial Value ARX Model Temperature 350 220 10 0.475 0 0.1
Pressure 1390 1230 5 0.31 0 0.1
The PID controller was tuned using Ziegler-Nichols tuning method [14; 23].
V.
CONCLUSION
The estimated data obtained for the validation set using ARX model for the output temperature and pressure shows a good prediction behavior. The simulated model output lies mostly within the confidence line indicating that the validated model is good enough for measuring the fitness of the plant control system. It can be seen that the relationship between the measured and the simulated validation model is excellent. The error plot shows the error between the two as 9%. The study also presented the application of proportional integral derivative control PID techniques for an industrial oil-fired boiler system. Simulation results obtained with the PID are very attractive in terms of rapidity, tracking error, stability and smoothness of control signals. The proposed structure is very simple and easy to implement.
REFERENCES
[1]. Bitmead, R. R., Gevers, M and Wertz, V., (1996), Adaptive Optimal Control, The Thinking Mans GPC, Prentice Hall, International Systems and Control Engineering. [2]. Chai T., Mao K. and Qin X., (1994). Decoupling Design of Multivariable Generalized Predictive Control, IEEE Proc-Control Theory Application Vol. 141 (3), pp197-201. [3]. Clark D. W., Montadi C. and Tuffs P. S., (1987), Generalized Predictive Control Part 1 & 2, Pergamon Journal Ltd, pp. 137-148 & 149-160. [4]. Gumpy J. M. and Jiya J. D., (2009), Identification of an Industrial Oil-Fired Power Plant Boiler System, Journal of Engineering Technology, vol. 4, No. 1, pp 56. [5]. Gumpy J. M. and Jiya J. D., (2008), Design of a Decentralized Generalized Predictive Controller for an Industrial Oil-Fired Boiler System, Journal of Engineering Technology, vol. 3, No. 2, pp 1. [6]. Irwin G., Brown, Hoggand B., and Swidenbank E., (1995), Coal combustor Neural Network Modeling of a 200MW Boiler System, IEEE proc-control theory and applications, vol. 142, pp. 529-536. [7]. Maffezzoni C. (1997) Boiler-Turbine Dynamics in Power-Plant Control, Pergamon, Elsevier Science Ltd, Great Britain, pp. 301-311. [8]. Maffezzoni, C., Bolis and Ferrarini I., (1993), Synthesis of the overall Boiler-Turbine control system by single loop Auto-Tuning Technique, pp. 663-666. [9]. MathWorks, (2009), MATLAB Documentation Set Giude for Windows, pp. 1. [10]. Morari M. and Lee J. H. (1999), Model Predictive Control: Past, Present and Future, Journal of Computer & Chemical Engineering, pp. 667-682. [11]. Oluwande G. and Boucher A. R., (1999), Implementation of a Multivariable Model Based Predictive Controller for Super Heater Steam Temperature and Pressure Control on a Large Coal Fired Power Plant, National Power Plc, UK Swindon Wiltshire, pp. 1-4. [12]. Voda A. and Ray D., (2000), Issues in Identification, Validation in Closed-loop and Control Design, Laboratoire de Automatique de Grenoble ENSIEG, France, pp. 1-3.
540
Vol. 3, Issue 1, pp. 534-541
[13]. Yugeng X. (1998), Generalization of Predictive Control Principles in Uncertain Dynamic Environments, Institute of Automation, Shanghai Jiao University, 954 Hua Shan Road, Shangai 2000, 30, P. R. China, pp. 1-2. [14]. Zhong J., (2006), PID Controller Tuning; A Short Tutorial, Mechanical Engineering, Purdue University. [15]. Kim S. (2003), Modeling and Simulating Fire Tube Boiler Performance, Aalborg University, Institute of Energy Technology, Pontoppidanstrde 101, Aalborg. [16]. Gordon P. and Joseph B. (1996), Nonlinear Control Oriented Boiler Modeling-A Benchmark Problem for Controller Design, IEEE Transactions on Control Systems Technology, Vol. 4, No. 1. [17]. Goodall P. M. (1979), The Efficient Use of Steam, IPC Science Technology Press Chideock. [18]. Astrom K. J. and Hagglund T. H. (1995), New Turning Methods for PID controller, Proceedings of the 3rd European Control Conference. [19]. Li M., Chen C. and Liu W. (2009), Identification Based on MATLAB, Proceedings of the International Workshop on Information Security and Application (IWISA), Qingdao, China. [20]. Boiler (2011), Encyclopdia Britannica, Ultimate Reference Suite, Chicago, Encyclopdia Britannica. [21]. Daz S. P. (2001), Modelling and Simulation of an Industrial Steam Boiler with Ecosimpro, 1st Meeting of EcosimPro Users, UNED, Madrid. [22]. Li W., Xiaoguang, and Chang A. (1997), Neuro-Fuzzy Controller for a Stoke-Fired Boiler Based on Behavior Modeling, Control Engineering Practice, Vol. 7, (4). [23]. Xue D. et al (2007), Linear Feedback Control, Society for Industrial and Applied Mathematics. [24]. Yugeng X. (1998), Generalization of Predictive Control Principles in Uncertain Dynamic Environments, Institute of Automation, Shanghai Jiao University, 954 Hua Shan Road, Shangai 2000, 30, P. R. China, pp. 1-2. [25]. Ljung L. (2011), System Identification Toolbox Users Guide R2011b, The MathWorks, Inc. [26]. Cirka L., Fikar M., and Petru P. (2006), Idtool 4.0A Dynamical System Identification Toolbox for Matlab/Simulink, Faculty of Chemical and Food Technology, Slovak University of Technology in Bratislava. [27]. McLellan J. (2004), Using Matlab and the System Identification Toolbox to Estimate Time Series Models, Department of Chemical Engineering, Queens University, Canada.
Authors
Rapheal Orosun receives the B.Eng Electrical Electronics from Federal University of Technology Yola, Nigeria in 2008; Currently studying for a MEng Electrical Engineering at Bayero University Kano, Nigeria. He is presently a Shift Electrical Engineer (TE) at Savannah Sugar Co. Ltd Numan, Nigeria. He is also an active member of IEEE Nigeria Section. His research interest lies in the field of Industrial Automation and Renewable Energy.
Sunusi Sani Adamu receives the B.Eng Electrical from Bayero University Kano, Nigeria in 1985; the MSc degree in electrical power and machines from Ahmadu Bello University, Zaria, Nigeria in 1996; and the PhD in Electrical Engineering, from Bayero University, Kano, Nigeria in 2008. He is currently a senior lecturer in the Department of Electrical Engineering, Bayero University, Kano. His main research area includes power systems simulation and control, and development of microcontroller based industrial retrofits. Dr Sunusi is a member of the Nigerian Society of Engineers and a registered professional engineer in Nigeria.
541
Vol. 3, Issue 1, pp. 534-541
INTERNET BASED REMOTE MONITORING AND CONTROL SYSTEM

Monita N. Jadhav and G. R. Gidveer
Department of Electronics Engineering, J.N.E.C., BAMU, Aurangabad, India
ABSTRACT
With the growing popularity of internet and development of embedded technology, Web technology has been extended to the development and application of embedded system. It is the end of the network era which takes PC equipment as the basic network node. Besides PC equipment, the network nodes include various types of embedded devices. How to remotely monitor, control, diagnosis, manage and maintain operations of embedded devices from different subnets and physical areas, is the problem needed to be solved. Web based remote monitoring and controlling system should directly connect the equipment to the network as node using Atmega16 processor and network module. The clients do not need to install special software and may monitor and control the current condition of equipment through web browsers. This structure has the following advantages- it truly realizes seamless connection between equipment and their management. It greatly decreases the system building cost. With the development of industrial Ethernet technology, the real time performance of system is improved further.
KEYWORDS:
ASP.net, Atmega16, Internet based systems, Wiznet
I.
INTRODUCTION
Integrating web and embedded technology, the embedded equipment monitoring and controlling system based on web management can be done. Managers can remote access, monitor and maintain the on-site equipment through the network and using a web browser without limit of region and time It can realize the inter-access between the heterogeneous equipment. The function of Web-based equipment monitoring system is to collect real-time data of the on-site equipment, publish it through a Web form, and remotely send the data in the form of the user-defined data transmission style [1]. It will provide flexible remote monitoring and diagnosis function combining the configuration software based on standard browser. The data will be published through web page form by the web server in various user defined formats. If the parameter value is different from the original set value by the user, a corrected signal will be send to the control unit automatically by the processor. The limit value can be set by the user through the web browser at any remote location. Thus the defined parameter can be controlled through the processor via internet. In a similar fashion various other parameters for control and monitoring can be added to customize the hardware to suite the user requirements. With the growing popularity of Internet and the development of embedded technology, Web technology has been extended to the development and application of embedded system. It is the end of the network era which takes PC equipment as the basic network node. Besides PC equipment, the network nodes include various types of embedded devices. How to use Embedded and Web technology to perform remote monitoring, diagnosis, management, and controlling and maintenance operation of embedded devices from different subnets and physical areas, is a problem that needs to be solved. Embedded Web based equipment condition monitoring and controlling system directly connects the equipment to network as a node [2]. The clients do not need to install special software and may monitor and control the current condition of equipment through browsers. This structure has the following advantages:
542
Vol. 3, Issue 1, pp. 542-548
a) It truly realizes seamless connection between equipment and their management. b) It greatly decreases the system building cost. c) With the development of industrial Ethernet technology, the real-time performance of system is improved further. The section two will elaborate the minute details of the whole system. The hardware contents of the system like microcontroller and sensors and the software design, languages and database systems of the remote monitoring and control system is also covered.
II.
INSIDE THE SYSTEM
2.1. The Function Designing of System

The idea was to take a system with some parameters and connect the system directly to internet/server and now monitor or control these various parameters through remote location via embedded web or internet. The function of Web-based equipment monitoring system is to collect real-time data information of the onsite equipment, publish it through a Web form, and remotely send the data in the form of the user-defined data transmission style. It will provide flexible rich remote monitoring and diagnosis function combining the configuration software based on standard browser. The data will be published through web page form by the web server. The remote Computer will collect the data and running status through the network and will provide the comparison on the historical data. If the parameter value is different from the original set value a signal will be given to the fan or heater accordingly and will turn it on. Thus the processor sends the control signal to the control unit. Thus the defined parameters can be controlled through the processor via internet [3]. In order to connect the embedded devices to the Internet and enable users to monitor the embedded devices using a standard Web browser, a website will be developed and ported onto the remote computer (ISP server). The Web server provides the Web-based graphical interfaces to users of the Internet to carry out unified supervision and management of various devices linked to network. In this project, the analog parameter planned for monitoring and control is temperature. The heart of the hardware is the microcontroller ATmega 16[4].The signal is picked by the sensor (LM35). The sensed signal is fed to the analog to digital (A/D) converter in built in the microcontroller. The A/D converter converts the analog input signal into digital signal and supplies to the processor. The ATmega16 based chip set controls the A/D collections and data transmission. The LM35 series are precision integrated-circuit temperature sensors, whose output voltage is linearly proportional to the Celsius (Centigrade) temperature. The LM35 thus has an advantage over linear temperature sensors calibrated in Kelvin, as the user is not required to subtract a large constant voltage from its output to obtain convenient Centigrade scaling. The LM35 does not require any external calibration or trimming to provide typical accuracies of 14C at room temperature and 34C over a full 55 to +150C temperature range [5].
2.2. Hardware Definition

For the purpose of definition of hardware architecture for sensor data acquisition, digitization and network establishment for internet connectivity, various books on microcontrollers / microprocessors and chip sets were studied to identify the components best suited for the intended application. The heart of the hardware architecture is the processor (ATmega16) around which the chip set is developed. The ATmega16 exchanges the sensor data with Data port module which consists of Monitoring module (A/D sub-module) and Control module (Relay driver circuit). The A/D convertor converts the sensor data into digital format and supplies to ATmega16 for data processing. The control module converts the digital command generated by ATmega16 into power signals for regulating the target outputs. The figure 1 shows the detailed structure of hardware of the project. The hardware part of system consists of Atmega16 core board, A/D convertor, signal conditioning, sensors, and communications interface. The Atmega16 chip consists of FLASH, 8-channel 10-bit ADC, one/two 16/8-bit timers. It is an 8-bit RISC microprocessor with the characteristics of high costperformance, low power, small size and high integration. The ATmega16 provides the following features:
543
Vol. 3, Issue 1, pp. 542-548
Figure 1 The Hardware architecture block diagram
16K bytes of In-System Programmable Flash Program memory with Read-While-Write capabilities, 512 bytes EEPROM, 1K byte SRAM, 32 general purpose I/O lines, 32 general purpose working registers, a JTAG interface for Boundary scan, On-chip Debugging support and programming, three flexible Timer/Counters with compare modes, Internal and External Interrupts, a serial programmable USART, a byte oriented two-wire Serial Interface, an 8-channel, 10-bit ADC with optional differential input stage with programmable gain (TQFP package only), a programmable Watchdog Timer with Internal Oscillator, an SPI serial port, and six software selectable power saving modes. The Idle mode stops the CPU while allowing the USART, Two-wire interface, A/D Converter, SRAM, Timer/Counters, SPI port, and interrupt system to continue functioning. The Power-down mode saves the register contents but freezes the Oscillator, disabling all other chip functions until the next External Interrupt or Hardware Reset. In Power-save mode, the Asynchronous Timer continues to run, allowing the user to maintain a timer base while the rest of the device is sleeping. The ADC Noise Reduction mode stops the CPU and all I/O modules except Asynchronous Timer and ADC, to minimize switching noise during ADC conversions. The On-chip ISP Flash allows the program memory to be reprogrammed in-system through an SPI serial interface, by a conventional non-volatile memory programmer, or by an On-chip Boot program running on the AVR core. The boot program can use any interface to download the application program in the Application Flash memory. Software in the Boot Flash section will continue to run while the Application Flash section is updated, providing true Read-While-Write operation. By combining an 8-bit RISC CPU with InSystem Self-Programmable Flash on a monolithic chip, the Atmel ATmega16 is a powerful microcontroller that provides a highly-flexible and cost-effective solution to many embedded control applications. The various signal of equipment are picked up by sensors and converted to digital signals by the A/D converter chip after Signal conditioning. The programmable control chip controls the A/D collection and the data transmission with Atmega16. For the purpose of streaming equipment data along with equipment operating condition information to the remote server, the Ethernet module converts the digital data into Ethernet format and transmits it to the web server through RJ45 connector. This part will be embedded in the devices as an independent hardware module. A relay is an electrically operated switch. Many relays use an electromagnet to operate a switching mechanism mechanically, but other operating principles are also used. Relays are used where it is necessary to control a circuit by a low-power signal (with complete electrical isolation between control and controlled circuits), or where several circuits must be controlled by one signal. The first relays were used in long distance telegraph circuits, repeating the signal coming in from one circuit and retransmitting it to another. Relay selected is cube type (JM-2P) which has the features like Smaller size compared to RW series (sugar cube relays), but with 25A Inrush current and has application for automotive electrical systems [6].
544
Vol. 3, Issue 1, pp. 542-548
The hardware details can be actually seen in the figure 2.
Figure 2 Hardware components of the system
2.3. Software Definition

To achieve integration of the embedded hardware with World Wide Web, various papers / books were referred to identify the right operating system and other software platforms for development of data processing and Graphical user interface (GUI) definition. The sensor data processing/control, GUI and the Databases are developed by integrating the standardized technologies, such as embedded C, ASP.NET, MySQL, and the dynamic linking library (DLL) files [7]. Remote monitoring station has includes several functions, they are initialization of serial port, sending and accepting data, database interface and so on. There can be various possibilities scripts, Microsoft has introduced ASP. Asp.net enables to access information from data sources, such as backend database and text files that are stored on a web server or a computer that is accessible to a web server. Asp.net enable to use a set of programming code called templates to create HTML documents. Asp.net also enables to separate HTML design from data retrieved mechanisms. Therefore, changing the HTML design does not affect the program that retrieves data from the databases. Asp.net is based on the .net frame work. Web form enables to include user interface, such as text box, list box controls and application logic of Web applications, and configuration files enable to store the configuration settings of an asp.net application. MySQL is the free database software available on net so decided to use it. It is also easy to use software and user friendly.
2.4. Ethernet Driver Module

Ethernet drivers are software programs that provide hardware-software interaction between the operating system of a personal computer (PC) and its network port.WIZ812MJ is the network module that includes W5100 (TCP/IP hardwired chip, include PHY), MAG-JACK (RJ45 with Transformer) with other glue logics. It can be used as a component and no effort is required to interface W5100 and Transformer. The WIZ812MJ is an ideal option for users who want to develop their Internet enabling systems rapidly. The W5100 is a full-featured, single-chip Internet enabled 10/100 Ethernet controller designed for embedded applications where ease of integration, stability, performance, area and system cost control are required. The W5100 has been designed to facilitate easy implementation of Internet connectivity without OS. The W5100 is IEEE802.3 10BASE-T and 802.3u 100BASETX compliant.
545
Vol. 3, Issue 1, pp. 542-548
The W5100 includes fully hardwired, market-proven TCP/IP stack and integrated Ethernet MAC & PHY. Hardwired TCP/IP stack supports TCP, UDP, IPv4, ICMP, ARP, IGMP and PPPoE which has been proven in various applications for several years. 16Kbytes internal buffer is included for data transmission. No need of consideration for handling Ethernet Controller, but simple socket programming is required [8]. For easy integration, three different interfaces like memory access way, called direct, indirect bus and SPI, are supported on the MCU side. The structure of web based remote monitoring system is given in figure 3.
Figure 3 The structure of web based remote monitoring system
2.5. Result and Discussion

The application is taken of temperature of a room, where it is needed to have constant temperature. So the required temperature is set point. If the temperature is below set point heater will be on and if it is above set point then cooler will be on. The figure 4 shows admin panel of the web on which the current temperature is displayed which is less than the set maximum temperature. So heater is on and cooler if off. We can upgrade the set point if needed with the administrative rights. The Issue of Communication Security: The main design considerations were security, performance, ease of use, availability, and scale. Security is paramount when we consider exporting private and confidential information outside of the firewall. We must assure that only valid users can access the web, while active attackers on the Internet cannot. We describe our techniques for authenticating clients and maintaining the privacy of the information from illegitimate outsiders. The issue of communication security at the present scenario is handled by having the user id and password to access the website controls. The technology is advancing day by day. The security issues involved in the system will be completely removed out by applying latest concepts of internet. Then the system like this can be easily implemented by small to large scale industries for remote monitoring and controlling the plants.
546
Vol. 3, Issue 1, pp. 542-548
Figure 4 Monitoring and control panel
III.
CONCLUSIONS
In this application, a low-cost, Internet-based control system has been designed and implemented. The application possibilities are virtually unlimited by attaching modules with appropriate interfaces, although the usage of the system is demonstrated with only a few sample devices. Compared with other applications, this system has advantages in terms of allowing direct bidirectional communication and reducing overhead, which can be vitally important for some real-time applications. The operational costs can be reduced by relinquishing the storage of large data to an FTP server on the Internet.
ACKNOWLEDGEMENTS
The authors would like to thank to the college who made us available all the needful. Monita is thankful to her husband; Nitin for his support all the time.
REFERENCES
[1] Yong-tao ZHOU, Xiao-hu CHEN, Xu-ping WANG, Chun jiang YAO, 2008: Design of Equipment remote monitoring System Based on Embedded Web, IEEE, pp 73-78. [2] Ali Ziya Alkar, Mehmat Atif Karaca, 2009, An Internet Based Interactive Embedded Data- Acquisition System for Real-Time Applications, IEEE, Vol .58.No.3, pp 522- 529. [3] Mo Guan, Wei, Ying Bao, 2008 A Monitoring System Based on Embedded Internet Technology for Embedded Devices, IEEE ,Vol. pp.5-8. [4] www.alldatasheet.com/datasheet-pdf/pdf/78532/ATMEGA16.html [5] www.national.com/ds/LM/LM35.pdf [6] www.goodsky.com.tw/contents/36-gm.pdf [7] M. Babita Jain, Amit Jain, M.B. Srinivas, 2008A Web based expert systems shell for fault diagnosis and control of power systems equipment , IEEE, pp1310-1313. [8] www.wiznet.co.kr
547
Vol. 3, Issue 1, pp. 542-548
Authors
Monita N. Jadhav. She has received the B.E. degree from the Electronics and Telecommunication Engineering Department, Government College of Engineering, Jalgaon, North Maharashtra University, Maharashtra, India in 2003. She is currently pursuing M. E. from the Electronics Engineering Department, Dr. Babasaheb Ambedkar Marathwada University of Aurangabad, Maharashtra, India. Her main areas of interests are microprocessor architecture design, access control systems, and embedded systems. G. R. Gidveer. He is currently working as Assistant Professor Department of Electronics & Telecommunication at Jawaharlal Nehru Engineering College. He had received degree B.E. in first Div and PG in Distinction. His total teaching experience in years is 34 Years. He has guided 10 Masters level project. He is also acting as consultant for Dept. of Police and Municipal Corporation Aurangabad.
548
Vol. 3, Issue 1, pp. 542-548
ADVANCED CONTROL TECHNIQUES IN VARIABLE SPEED STAND ALONE WIND TURBINE SYSTEM
V. Sharmila Deve and S. Karthiga
Department of Electrical and Electronics Engineering Kumaraguru College of Technology, Coimbatore, Tamil Nadu, India
ABSTRACT
This project presents an advanced control strategy for the operation of a direct-drive synchronous generatorbased stand-alone variable-speed wind turbine. The control strategy for the generator-side converter with maximum power extraction is presented. The stand-alone control featured is constant output voltage and frequency that is capable of delivering to variable load. The potential excess of power is dissipated. The PI controller in switch mode rectifier can be replaced with fuzzy control technique to improve the output voltage level. The simulation results show this control strategy gives better regulating voltage and frequency under sudden varying load conditions. Dynamic representation of dc bus and small signal analysis are presented. The dynamic controller shows very good performance.
KEYWORDS: PMSG, boost converter, inverter, driver circuit and PIC/DSP
I.
INTRODUCTION
In this paper to design advance control techniques in variable speed to give continuous Supply to load. Variable-speed wind turbines have many advantages over fixed-speed generation such as increased energy capture, operation at maximum power point, improved efficiency, and power quality [1]. The reliability of the variable-speed wind turbine can be improved significantly by using a direct-drive synchronous generator. PMSG has received much attention in wind-energy application because of their property of self-excitation, which allows an operation at a high power factor and high efficiency [2]. Optimum power/torque tracking is a popular control strategy, as it helps to achieve optimum windenergy utilization [3][7]. The switch-mode rectifier has also been investigated for small-scale variable-speed wind turbine[10], [11].It is very difficult to obtain the maximum voltage level by using PI controller. In order to obtain the maximum output, Fuzzy logic control can be used.
II.
BLOCK DIAGRAM
A. Block diagram description

Generator converts the variable speed mechanical power produced by the wind turbine into electrical power. The power produced in the generator having variable frequency and voltage AC power. This Ac power converted into DC power with the help of uncontrolled rectifier. The dc power will be have variable voltage. This variable voltage is boosted to rated level with the help of boosted converter. Boosted dc power is converted into fixed frequency AC power and it is delivered to load. Between load and inverter as storage system with converter and inverter is used to store the energy. This storage system will store the energy at the time of load lesser than maximum level. Also this storage system is used to deliver power to load when the boost converter unable to boost up the voltage.
549
Vol. 3, Issue 1, pp. 549-557
Fig 1.Block diagram of the project
B. Wind-turbine characteristics
The amount of power captured by the wind turbine (power delivered by the rotor) is given by,
where is the air density (kilograms per cubic meter), is the wind speed in meters per second, is the blades swept area, and p is the turbine-rotor-power coefficient, which is a function of the tipspeed ratio and pitch angle . m rotational speed of turbine rotor in mechanical radians per second, and radius of the turbine. If the wind speed varies, the rotor speed should be adjusted to follow the change. The target optimum torque can be given by
The mechanical rotor power generated by the turbine as a function of the rotor speed for different wind speed is shown in Fig. 2.
Fig 2.Mechanical power generated by the turbine as a function of the rotor speed for different wind speeds.
The optimum power is also shown in this figure. The optimum power curve (Popt) shows how maximum energy can be captured from the fluctuating wind. The function of the controller is to keep the turbine operating on this curve, as the wind velocity varies. It is observed from this figure that there is always a matching rotor speed which produces optimum power for any wind speed. If the controller can properly follow the optimum curve, the wind turbine will produce maximum power at any speed within the allowable range.
III.
SYSTEM OVERVIEW
Fig. 3 shows the control structure of a PMSG-based stand alone variable-speed wind turbine which include a wind turbine, PMSG, single-switch three-phase switch-mode rectifier, and a vector-
550
Vol. 3, Issue 1, pp. 549-557
)(
) (
controlled PWM voltage-source inverter. A constant dc voltage is required for direct use, storage, or conversion to ac via an inverter.
Fig 3.Control structure of a PMSG-based stand-alone variable-speed wind turbine.
In this paper, a single-switch three-phase switch-mode rectifier is used to convert the ac output voltage of the generator to a constant dc voltage before conversion to ac voltage via an inverter. The single-switch three-phase switch-mode rectifier consists of a three-phase diode bridge rectifier and a dc to dc converter. The output of the switch-mode rectifier can be controlled by controlling the duty cycle of an active switch (such as IGBT) at any wind speed to extract maximum power from the wind turbine and to supply the loads. A vector-controlled IGBT inverter is used to regulate the output voltage and frequency during load or wind variations. Voltage drop due to sudden fall in wind speed can be compensated by the energy-storage system. During wind gust, the dump-load controller will be activated to regulate the dc-link voltage to maintain the output load voltage at the desired value.
IV.
CONTROL OF SWITCH-MODE RECTIFIER WITH MAXIMUM POWER EXTRACTION
The structure of the proposed control strategy of the switch mode rectifier is shown in Fig. 4. The control objective is to control the duty cycle of the switch S in Fig. 2 to extract maximum power from the variable-speed wind turbine and transfer the power to the load. The control algorithm includes the following steps.
Fig 4. Control strategy of the switch-mode rectifier.
1) Measure generator speed g. 2) Determine the reference torque (Fig. 4) using the following equation:
551
Vol. 3, Issue 1, pp. 549-557
3) This torque reference is then used to calculate the dc current reference by measuring the rectifier output voltage Vd as given by 4) The error between the reference dc current and measured dc current is used to vary the duty cycle of the switch to regulate the output of the switch-mode rectifier and the generator torque through a proportionalintegral (PI) controller. The generator torque is controlled in the optimum torque curve as shown in Fig.5 according to the generator speed.
Fig. 5. Generator torque reference versus speed.
If the PMSG is operating at point a and the wind speed increases from vw1 to vw2 (point b), the additional power and, hence, torque causes the PMSG to accelerate. Finally, the generator will reach the point c where the accelerating torque is zero. A similar situation occurs when the wind velocity decreases. In the proposed method, the wind speed is not required to be monitored, and, therefore, it is a simple output-maximization control method without wind-speed sensor (anemometer).
V.
SIMULATION MODEL USING PI CONTROLLER
Fig 6. Matlab model of the PMSG based variable-speed wind-turbine system using PI Controller
552
Vol. 3, Issue 1, pp. 549-557
VI.
OUTPUT WAVEFORMS
Fig 7.a output voltage of PMSG
Fig 7.b Output voltage of uncontrolled rectifier
Fig 7.c boosted output voltage.
553
Vol. 3, Issue 1, pp. 549-557
Fig 7.d load voltage
Fig 7.e load current
VII.
FUZZY LOGIC CONTROLLER
A control strategy for a rectifier with variable speed direct driven permanent magnet synchronous generator. The fuzzy logic controller is used to track generator speed with varying wind speed to optimize turbine aerodynamic efficiency in the outer speed loop. The voltage space vector PWM in field-oriented control is adopted in the control of the generator side converter. By means of the fieldoriented control, the highest efficiency of wind turbine can be reached. The Fuzzy controller of inner current loop is used instead of the traditional PI controller to improve the performances of current loop. Both simulation and experiments have been conducted to validate the performance.
TABLE- FUZZY-RULE-BASED MATRIX
554
Vol. 3, Issue 1, pp. 549-557
VIII.
SIMULATION DIAGRAM USING FUZZY CONTROLLER
Fig 8.Matlab model of PMSG based variable-speed wind-turbine system using Fuzzy logic controller.
IX.
OUTPUT WAVEFORMS
Fig 9.a output voltage of PMSG
Fig 9.b Output voltage of uncontrolled rectifier
555
Vol. 3, Issue 1, pp. 549-557
Fig 9.c Boosted Output voltage
Fig 9.d load voltage
Fig 9.e load current
X.
CONCLUSION
A control strategy for a direct-drive stand-alone variable speed wind turbine with a synchronous generator has been presented in this project. The controller is capable of maximizing output of the variable-speed wind turbine under fluctuating wind. The generating system with the proposed control strategy is suitable for a small-scale stand alone variable-speed wind-turbine installation for remote-
556
Vol. 3, Issue 1, pp. 549-557
area power supply .The simulation results has proves that Regulating the o/p voltage & frequency under sudden load variations and typical wind movement.
ACKNOWLEDGMENT
The authors wish to thank the family members who have provided full support.
REFERENCES
[1] S. Mller, M. Deicke, and R. W. De Doncker, Doubly fed induction generator system for wind turbines, IEEE Ind. Appl. Mag., vol. 8, no. 3,pp. 2633, May 2002. [2] T. F. Chan and L. L. Lai, Permanent-magnet machines for distributed generation: A review, in Proc. IEEE Power Eng. Annu. Meeting, 2007. [3] M. De Broe, S. Drouilhet, and V. Gevorgian, A peak power tracker for small wind turbines in battery charging applications, IEEE Trans. EnergyConvers., vol. 14, no. 4, Dec. 1999. [4] R. Datta and V. T. Ranganathan, A method of tracking the peak power points for a variable speed wind energy conversion system, IEEE Trans. Energy Convers., vol. 18, no. 1, Mar. 1999. [5] K. Tan and S. Islam, Optimal control strategies in energy conversion of PMSG wind turbine system without mechanical sensors, IEEE Trans. Energy Convers., vol. 19, no. 2, Jun. 2004. [6] S. Morimoto, H. Nakayama, M. Sanada, and Y. Takeda, Sensorless output maximization control for variable-speed wind generation system using PMSG, IEEE Trans. Ind. Appl., vol. 41, no. 1, pp. 6067, Jan. 2005. [7] M. Chinchilla, S. Arnaltes, and J. C. Burgos, Control of permanentmagnet generators applied to variablespeed wind-energy systems connected to the grid, IEEE Trans. Energy Convers., vol. 21, no. 1, pp. 130135, Mar. 2006. [8] D. M. Whaley, W. L. Soong, and N. Ertugrul, Investigation of switchedmode rectifier for control of smallscale wind turbines, in Conf. Rec. IEEE IAS Annu. Meeting, 2005, pp. 28492856. [9] E. Muljadi, S. Drouilhet, R. Holz, and V.Gevorgian, Analysis of permanent magnet generator for wind power battery charging, in Conf. Rec. IEEE IAS Annu. Meeting, 1996, pp. 541548. [10] W. Lixin , Fuzzy System and Fuzzy Control, Beijing: Tsinghua Universtiy,2003. [11] Y Liyong,L.Yinghong,C Yaai,and L.Zhengxi, "A Novel Fuzzy Logic Controller for Indirect Vector Control Induction Motor Drive ,Proceedings of the 7th World Congress on Intell igent Control and Automation, Chongqing, China, pp.24-28, June 2008 .
Authors Biography
V. SHARMILADEVE received his B.E. degree in Electrical and Electronics Engineering from Bharathiyar University, India in 1999 and ME degree in Power systems and from Annamalai university, India in 2002. He is currently working as Senior grade Assistant professor in Kumaraguru college of Technology, Coimbatore, India. Her main research is in Economic dispatch of power system and her areas of interest is in applications of smart grid, Distributed Generation, FACTS devices and their control
S. KARTHIGA was born in Coimbatore India, in 1989. She received the B.E (Electrical and Electronics) degree in Park College of Technology, Coimbatore, India in 2010. Now she is doing M.E (Power Electronics and Drives) in Kumaraguru College of Technology, Coimbatore, India. Her areas of interest are Power Electronics and Drives, Power Quality, and Renewable Energy
557
Vol. 3, Issue 1, pp. 549-557
ANALYSIS OF UNIFIED POWER QUALITY CONDITIONER DURING POWER QUALITY IMPROVEMENT

B. Sridhar1 , Rajkumar Jhapte2
1
Asstt. Prof., Electrical Engg., Christian college of Engg. & Technology, Bhilai, India 2 Sr. Asstt. Prof., EEE., Shri Shankaracharya Technical Campus, Bhilai, India
A BSTRACT
The non linear loads in industry cause an increasing deterioration of the power system voltage and current waveforms. As a result, harmonics are generated from power converters or nonlinear loads. This causes the power system to operate at low power factor, low efficiency, increased losses in transmission and distribution lines, failure of electrical equipments, and interference problem with communication system. So, there is a great need to mitigate these harmonic and reactive current components. Active Power filters are a viable solution to these problems. In this manuscript a three-phase unified power quality conditioner (UPQC), with a combination of shunt active power filter and series active power filter is used to eliminate supply current harmonics, compensate reactive power, voltage sag and voltage swell compensation on distribution network. The performance of the active power filter mainly depends on control strategy used to generate reference current for shunt active power filter (APF) and generate reference voltage for series active power filter. Simulation results of these two active power filters are carried out separately and as a combination also.
INDEX T ERMS: Unified power quality conditioner (UPQC), shunt active power filter (APF), series active
power filter (APF), point of common coupling (PCC), voltage source converters (VSC)
I.
INTRODUCTION
The non linear equipments during switching ON/OFF of high rated load connected at the point of common coupling (PCC) highly distorted may result into voltage sags or swells. There are several sensitive loads, such as computer or microprocessor based AC/DC drive controller, with good voltage profile requirement; can function improperly can lose valuable data or in certain cases get damaged due to these voltage sag and swell conditions. One of the effective approaches is to use a unified power quality conditioner (UPQC) at point of common coupling (PCC) to protect the sensitive loads. A unified power quality conditioner is a combination of shunt and series active power filters, sharing a common dc link shown in Fig.1. [1]. It is a versatile device that can compensate almost all power quality problems such as voltage harmonics, voltage unbalance, voltage flickers, voltage sags & swells, current harmonics, current unbalance, reactive current, etc. The continuous usage of nonlinear loads injects current and voltage harmonic components into the power system and increases reactive power demands and power system voltage fluctuations. This analysis will be extended as well to the optimum UPQC with minimum VA requirement. [15] Harmonic current components create several problems like, Increase in power system losses. Overheating and insulator failures in transformers, rotating machinery, Conductor and cables. Reactive power burden, Low system efficiency, Poor power factor, System unbalances and causes excessive neutral currents.
558
Vol. 3, Issue 1, pp. 558-565
Malfunctioning of the protective relays and untimely tripping. The solution is to install line conditioning systems that suppress the power system disturbances.[3]
Fig.1. System configuration of UPQC
II.
OPERATION PRINCIPLE
The principle of operation of active filters is based on the injection of the harmonics required by the load. Thus the basic principle of Active Power filter is that it generates a current equal and opposite in polarity to the harmonic current drawn by the load and injects it to the point of coupling thereby forcing the source current to be pure sinusoidal.
Fig 2.Simulink Model for UPQC
559
Vol. 3, Issue 1, pp. 558-565
Active power filter consists of two blocks, first one reference harmonic component calculation block, which calculates harmonic content on the load side and the second one is inverter block .Calculated harmonic currents are given to inverter block. Active power filter requires inverter to inject the harmonics in opposite polarity at point of coupling .The characteristics of the harmonic compensation are strongly dependent on the filtering algorithm employed for the calculation of load current harmonics. Many filtering algorithms are developed to extract the harmonic components. Control method also plays an important role in active power filter control. Reduced Switching losses, compensation of higher harmonics, less current ripple etc needs a control method to meet all these requirements.[2]
III.
CONTROL ALGORITHM
Unified power quality conditioner is a combination of shunt APF and series APF. Control algorithm is been used to generate gating pulses to shunt APF and series APF. Generating gating pulses to shunt APF for that reference current generator and hysteresis current controller is required. Similarly, for generating gating pulses to series APF, reference voltage generator and hysteresis voltage controller is required and complete control strategy.[5] 3.1 Reference Current Generator The control strategy is based on the extraction of Unit Vector Templates from the distorted input supply. These templates will be then equivalent to pure sinusoidal signal with unity (p.u) amplitude. [8] In order to have distortion less load current, the load current must be equal to these reference signals. The measured load currents are compared with reference load current signals. The error generated is then taken to a hysteresis current controller to generate the required gate signals for shunt APF. The unit vector template can be applied for shunt APF to compensate the harmonic current generated by nonlinear load. The shunt APF is used to compensate for current harmonics as well as to maintain the dc link voltage at constant level. To achieve this task the dc link voltage is sensed and compared with the reference dc link voltage. The error is then processed by a PI controller. The output signal from PI controller is multiplied with unit vector templates. The source current must be equal to this reference signal. In order to follow this reference current signal, the 3phase source currents are sensed and compared with reference current signals. The error generated is then processed by a hysteresis current controller with suitable band, generating gating signals for shunt APF. 3.2 Hysteresis Current Controller APF eliminates system harmonics through injecting a current to the system that is equal to the load harmonic current; therefore the system side will almost have no harmonic current remaining. Since the load harmonics to be compensated may be very complex and changing rapidly and randomly, APF has to respond quickly and work with high control accuracy in current tracking. 3.3 Reference Voltage Generator The control strategy is based on the extraction of Unit Vector Templates from the distorted input supply. These templates will be then equivalent to pure sinusoidal signal with unity (p.u) amplitude. The 3 phase distorted input source voltage at PCC contains fundamental component and distorted component. To get unit input voltage vectors, the input voltage is sensed and multiplied by gain equal to I/Vm where Vm is equal to peak amplitude of fundamental input voltage. 3.4 Hysteresis Voltage Controller APF eliminates system voltage sag and voltage swell through injecting a voltage to the system that is equal to the decrease load voltage magnitude; therefore the system side voltage will almost maintains constant. Since the load voltage to be compensated may be very complex and changing rapidly and randomly, APF has to respond quickly and work with high control accuracy in voltage tracking.
560
Vol. 3, Issue 1, pp. 558-565
IV.
POWER INVERTERS & CONTROL TECHNIQUES
Inverters are used to inject compensation current in to the system to cancel the harmonics in source current. The current waveform for canceling harmonics is achieved with the voltage source inverter (VSI) in the current controlled mode and an interfacing filter. The desired current waveform is obtained by accurately controlling the switching of the insulated gate bipolar transistors (IGBTs) in the inverter. Control of the current wave shape is limited by the switching frequency of the inverter and by the available driving voltage across the interfacing inductance.[9] In voltage source converters (VSC), variables to be controlled are the amplitude and frequency of the converter output voltage. PWM techniques enable the modulation of the controllable switches of the converter. By proper modulation of the controllable switches, a high frequency converter output voltage can be generated, whose low frequency or localized average during each switching cycle is same as that of the desired output voltage of the converter at that cycle.
V.
SYSTEM DATA
415 V 50 Hz 750 V 10 mH 5000 uF 2460 watts 1000 Vars 10 kVA
Supply Voltage, Vs Supply frequency DC Bus voltage, Vdc, Injection Inductor, L DC side capacitor, C Sensitive load active power Sensitive load reactive power Series filter transformer
VI.
RESULTS & DISCUSSIONS
This result shows Shunt APF controller effectively compensates harmonics, reactive power compensation and maintains D.C link voltage constant. The current waveform for canceling harmonics is achieved with the voltage source inverter in the current controlled mode and an interfacing filter. The filter provides smoothing and isolation for high frequency components. The desired current waveform is obtained by accurately controlling the switching of the insulated gate bipolar transistors (IGBTs) in the inverter. Control of the current wave shape is limited by the switching frequency of the inverter and by the available driving voltage across the interfacing inductance.
Fig.3. Load current from 0.04 sec to 0.2 sec
Fig.4. Unit vector Templates Generation 0.4 sec to 0.5sec
561
Vol. 3, Issue 1, pp. 558-565
Fig.5. Reference current from 0.4 sec to 0.5 sec
Fig.6. Shunt APF compensating currents from 0.4 sec to 0.5 sec
Fig.7. DC link capacitor voltage from 0.3 sec to 1 sec
Fig.8. Source current from 0.4 sec to 0.6 sec
Fig.9. Source voltage & current from 0.4 sec to 0.6 sec
6.1 Results For Series APF during voltage sag This result shows Series APF controller effectively compensates voltage sag and maintains load voltage constant.
Fig.10. Load Voltage from 0.04sec to 0.2sec during Sag
562
Vol. 3, Issue 1, pp. 558-565
Fig.11. Load Voltage from 0.04sec to 0.2sec during Sag
6.2 Results For Series APF during voltage swell This result shows Series APF controller effectively compensates voltage swell and maintains load voltage constant.
Fig.12. Load Voltage from 0.04sec to 0.2sec during Swell
Fig.13. Load Voltages from 0.04sec to 0.2sec during Swell
6.3 Results during voltage sag and voltage swell The powers due to harmonics quantities are negligible as compared to the power at fundamental component, therefore, the harmonic power is neglected and the steady state operating analysis is done on the basis of fundamental frequency component only. The unified power quality conditioner is controlled in such a way that the voltage at load bus is always sinusoidal and at desired magnitude. Therefore the voltage injected by series active power filter must be equal to the difference between the supply voltage and the ideal load voltage. Thus the series active power filter acts as controlled voltage source. The function of shunt active power filter is to maintain the dc link voltage at constant level. In addition to this the shunt active power filter provides the VAR required by the load, such that the input power factor will be unity and only fundamental active power will be supplied by the source[1].This result shows Series APF controller effectively compensates both voltage sag and swells and maintains load voltage constant.
Fig.14. Load from 0.04sec to 0.24sec during Sag and Swell
563
Vol. 3, Issue 1, pp. 558-565
Fig.15. Load Voltages from 0.04sec to 0.24sec during Sag & Swell
VII.
CONCLUSION
From the MATLAB / Simulink based simulation responses, it is evident that the Shunt APF, reference current generator, hysteresis current controller and also for Series APF, reference voltage generator, hysteresis voltage controller are performing satisfactorily. The source current waveform is in phase with the utility voltage and free from harmonic components. The load voltage waveform is maintaining constant during voltage sag and swell conditions. The three phase terminal voltages, three phase load voltages, three phase source currents and the dc link voltage are sensed and used to generate the switching patterns for shunt and series APFs. The shunt active power filter helps series active power filter during voltage sag and swell condition by maintaining the dc link voltage at set constant level, such that series active power filter could effectively supply or absorb the active power. In addition to this shunt active power filter also provides the required load VARs and thus making the input power factor close to unity.
REFERENCES
[1] Khadkikar.V, Chandra.A, Barry.A.O and Nguyen.T.D: Analysis of power flow in UPQC during voltage sag and swell conditions for selection device ratings Industrial Electronics. 2006 IEEE International Symposium on Volume 2, 9-13 May 2006. [2] H. Akagi: New trends in active filters for improving power quality. Proceedings of the 1996 International Conference, Vol .1, Jan 1996, pp. 417 425. [3] B. Singh, K, Al-Haddad and A. Chandra: "A Review of Active Power Filters for Power Quality improvement. IEEE Trans on Industrial Electronics, Vol.45, No.5, Oct1999, pp. 960-071. [4] Khadkikar.V, Agarwal.P, Chandra.A, Barry.A and Nguyen.T.D: A simple new control technique for unified power quality conditioner (UPQC). Harmonics and Quality of Power, 2004. 11th International Conference on 12-15 Sept. 2004. [5] A.Chandra, B. Singh, B.N.Singh and Al-Haddad.K: "An Improved Control Algorithm of Shunt Active Filter for Voltage Regulation, Harmonic Elimination, Power Factor Correction, and Balancing of Nonlinear Loads". IEEE Trans on Power Electronics, Vol. 15, No. 3,May 2000,495-507. [6] M T Tsai: Analysis and design of a cost-effective series connected AC voltage regulator, IEE Proc. Electr. Power Appl., Vol. 151, No. 1, Jan 2004, pp. 107-114. [7] M.S.Kim: Steady-state operating characteristics of unified active power filters. J.M.S, Applied Power Electronics Conference and Exposition, 1997. APEC '97Conference Proceedings 1997, Twelfth Annual, Volume: 1, 23-27 Feb 1997, [8] Li.R, A.T.Johns, and M.M.Elkateb: Control concept of Unified Power Line Conditioner. Power Engineering Society Winter Meeting, 2000. IEEE, Volume: 4, 23-27 Jan 2000, pp.2594 -2599. [9] Elnady A and Salama M.M.A: New functionalities of the unified power quality conditioner. Transmission and Distribution Conference and Exposition, 2001 IEEE/PES, Volume: 1, 28 Oct.-2 Nov 2001, pp. 415-420. [10] Haque M.T., Ise T., Hosseini S.H., A novel control strategy for unified power quality conditioner (UPQC), PESC 02, vol. 1, june 2002, pp. 94-98. [11] Basu.M, S.P.Das and Dubey G.K., Performance study of UPQC-Q for load Performance study of UPQCQ for load. IECON 02, IEEE, Volume: 1 5-8 Nov 2002, pp. 698 -703. [12] M.M. Morcos, J.C. Gomez, Electric power quality-the strong connection with power electronics, Power & Energy Magzine, IEEE, volume 1, Issue: 5, Sep-Oct 2003, pp. 18-25. [13] Ghosh.A, Jindal.A.K and Joshi.A: A unified power quality conditioner for voltage regulation of critical load bus. Power Engineering Society General Meeting, 2004. IEEE 6-10June 2004, pp. 471 476. [14] B. Singh, V. Verma, A. Chandra and K. Al Haddad, Hybrid filters for power quality improvement, Proc. IEE on Generation, Transmission and Distribution, vol. 152, pp. 365-378, May 2005.
564
Vol. 3, Issue 1, pp. 558-565
[15] Jayanti N.G., Basu Malabika, Conlon Micheal F., et al, Rating Requirements of a Unified Power Quality Conditioner for Voltage Ride Through Capability Enhancement, The 3rd IET International Conference on Power Electronics, Machines and Drives, pp.632-636, 2006. [16] M. Basu, S. P. Das, G. K. Dubey, Investigation on the performance of UPQC-Q for voltage sag mitigation and power quality improvement at a crucial load point, IET on Generation, Transmission and Distribution, vol. 2, no. 3, pp. 414-423, May 2008.
Authors Biographies
B. Sridhar was born in Bilaspur, India on November 27, 1979. He graduated from RCET, Bhilai with First division degree in B.E. Electrical Engineering in 2006 and pursuing Masters of Engineering (Power System), SSCET, Bhilai. He is currently Assistant Professor at CCE&T, Bhilai, India. His research interests include Power Quality.
Rajkumar Jhapte was born in Bhilai, India on February 02, 1982. He graduated from MPCCET, Bhilai with First division degree in B.E. Electrical Engineering in 2004 and Masters of Engineering (Power System), SSCET, Bhilai in 2009. He is currently Assistant Professor at SSCET, Bhilai, India. His research interests include Power Quality, Power Electronics, Power system protection etc.
565
Vol. 3, Issue 1, pp. 558-565
DESIGN OF MULTI BIT LFSR PNRG AND PERFORMANCE COMPARISON ON FPGA USING VHDL
Amit Kumar Panda, Praveena Rajput, Bhawna Shukla
Department of Electronics & Communication Engineering, Guru Ghasidas University Bilaspur, Chattishgarh
ABSTRACT
The main purpose of this paper is to study the FPGA implementation and performance analysis of 8, 16, and32 bit LFSR pseudo random number generator system. We have used FPGA to explain how FPGAs ease the hardware implementation part of communication systems. The analysis is conceded out to find number of gates, memory and speed requirement in FPGA as the number of bits is increased. The comparative study of 8, 16 and 32 bit LFSR on FPGA is shown here to understand the on chip verification. Recently the field programmable gate arrays have enjoyed wide spread use due to several advantages related to relatively high gate density, short design cycle and low cost. The greatest advantage of FPGAs are flexibility that we reconfigured the design many times and check the results and verify it on-chip for comparing with others PN sequence generators. The logic of PN Sequence Generator presented here can be changed any time, if we want a PN generator of more length all we need to do is change the number of shift register and adjust the taps. In this paper we have used one XOR operation for taping. Also we can use XNOR operation for taping. By increasing the number of tapping we can generate more randomness in the sequence.
KEYWORDS: LFSR, FPGA, PRNSG, VHDL
I.
INTRODUCTION
Random numbers are useful for a variety of purposes such as generating data encryption keys, simulating and modelling complex phenomena and for selecting random samples from larger data sets which are more explain in [1]-[4]. They have also been used aesthetically, for example in literature and music, and are of course ever popular for games and gambling. When discussing single numbers, a random number is one that is drawn from a set of possible values, each of which is equally probable, i.e. a uniform distribution. When discussing a sequence of random numbers, each number drawn must be statistically independent of the other. Random number generator is a computational device to generate a sequence of numbers or that lack any pattern. Random number generator may be software based or Hardware-based systems [5], [6], [8]. Hardware-based systems for random number generation are widely used, but often fall short of this target, though they may meet some of the statistical tests for randomness to confirm that they do not have any easily decipherable patterns. Methods for generating random results have existed since ancient times, including dice, coin flipping, the shuffling of playing cards, the use of yarrow stalks, and many other techniques. Various methods to compute pseudo-random numbers are known [2], [7]. Most of them are based, on linear congruential equations [9], [10] and require a number of time consuming arithmetic operations. In contrast, the use of feedback shift registers permits very fast generation of binary sequences. Shiftregister sequences of maximum length (m-sequences) are well suited to simulate truly random binary sequences [11]. However, they cannot be used in applications which require a decimal random sequence. This paper presents a fast random generator which is based on shift-register sequences and which generate decimal representation of binary digits in a pseudo-random mode.
566
Vol. 3, Issue 1, pp. 566-571
A field-programmable gate array (FPGA) is an integrated circuit created to be configured by the customer after manufacturing hence "field-programmable". The FPGA configuration is generally defined using a hardware description language (HDL), similar to that used for an application-specific integrated circuit (ASIC) (circuit diagrams were previously used to specify the configuration, as they were for ASICs, but this is increasingly rare). FPGAs can be used to implement any logical function that an ASIC can perform. The ability to update the functionality after shipping, partial reconfiguration of the portion of the design and the low non-recurring engineering costs relative to an ASIC design offer advantages for many applications. The most common HDLs are VHDL and Verilog. Though these two languages are similar but we prefer VHDL for programming because of its widely in use ([15]-[18]). Here we implemented 8, 16 and 32 bit LFSR using double taping as in the feedback path on Spartan 3S1000 FPGA of Xilinx using Xilinx ISE 10.1 [19]. As Xilinx provides automation tools for designing and implementing any logical as well as hardware on a single chip to get faster prototype, so it is widely used to implement any digital logic on FPGA. The structure of this paper is as follows. Section II contains a brief description of the Linear Feedback Shift Register and Rules for selecting feedback polynomial. Section III shows the Synthesis and timing simulation result and also performance comparison between 8, 16 and 32 bit LFSR using double tapping is presented.
II.
LINEAR FEEDBACK SHIFT REGISTER
A linear feedback shift register (LFSR) is a shift register whose input bit is a linear function of its previous state. The most commonly used linear function of single bits is XOR. Thus, an LFSR is most often a shift register whose input bit is driven by the exclusive-or (XOR) of some bits of the overall shift register value. Xor shift is a category of pseudorandom number generators designed by George Marsaglia. It repeatedly uses the transform of exclusive or on a number with a bit shifted version of it [11], [12]. The initial value of the LFSR is called the seed, and because the operation of the register is deterministic, the stream of values produced by the register is completely determined by its current (or previous) state. Likewise, because the register has a finite number of possible states, it must eventually enter a repeating cycle. However, an LFSR with a well-chosen feedback function can produce a sequence of bits which appears random and which has a very long cycle. Applications of LFSRs include generating pseudo-random numbers, pseudo-noise sequences, fast digital counters, and whitening sequences. Both hardware and software implementations of LFSRs are common.
2.1. Implementation of PRNSG

Pseudo random number sequence generator is generated in VHDL according to the following circuit based on the concept of shift register.
Figure 1. Single Taping LFSR
The bits in the LFSR state which influence the input are called taps A maximum-length LFSR produces an m-sequence (i.e. it cycles through all possible 2n 1 states within the shift register except the state where all bits are zero), unless it contains all zeros, in which case it will never change. The sequence of numbers generated by this method is random. The period of the sequence is (2n - 1), where n is the number of shift registers used in the design. For 32 bit design the period is 4294967295.This is large enough for most of the practical application. The arrangement of taps for feedback in an LFSR can be expressed in finite field arithmetic as a polynomial mod 2. This means that the coefficients of the polynomial must be 1's or 0's. This is called
567
Vol. 3, Issue 1, pp. 566-571
the feedback polynomial or characteristic polynomial. For example, if the taps are at the 16th, 14th, 13th and 11th bits (as shown), the feedback polynomial is X16 + X14 + X13 + X11 + 1
2.2. The rules for selecting Feedback Polynomial

The rules for selecting feedback polynomial [11] are as follows
The 'one' in the polynomial does not correspond to a tap it corresponds to the input to the first bit (i.e. x0 which is equivalent to 1). The powers of the terms represent the tapped bits, counting from the left. The first and last bits are always connected as an input and output tap respectively. The LFSR will only be maximum-length if the number of taps is even; just 2 or 4 taps can suffice even for extremely long sequences. The set of taps taken all together, not pairwise (i.e. as pairs of elements) must be relatively prime. In other words, there must be no common divisor to all taps. Once one maximum-length tap sequence has been found, another automatically follows. If the tap sequence, in an n-bit LFSR, is [n, A, B, C, 0], where the 0 corresponds to the x0 = 1 term, then the corresponding 'mirror' sequence is [n, n C, n B, n A, 0]. So the tap sequence [32, 7, 3, 2, 0] has as its counterpart [32, 30, 29, 25, 0]. Both give a maximum-length sequence.
III.
SYNTHESIS AND SIMULATION
In this design, we describe the RTL-level of the LFSR pseudo-random number generator for 8-bit, 16bit and 32 bit using VHDL language, and use the Xilinxs chip XC3S 1000 Sparta3 as the target chip. Then we synthesize, place and route on the Xilinx ISE platform. Finally we use ISE Simulator to do a timing simulation [19].
3.1. Timing Simulation

Input the netlist file generated from synthesizing, placing and routing on the Xilinx ISE 10.1 software and the standard delay file, the simulation waveform is shown in Fig. 2, Fig. 3, Fig. 4 under the simulation clock is 371.747MHz. Although there are a certain delay and glitch phenomena in the waveform, but they dont affect the normal operation of the whole module. So the test results are correct on the function and timing. Here the simulation clock period is taken as 20 ns. Simulation waveform for 8-bit, 16-bit and 32-bit are shown in Fig. 2, Fig. 3, Fig. 4. 3.1.1 Simulation Result of 8 bit LFSR The actual random output for 8-bit LFSR should come 28 = 256. But from the waveform simulation analysis it is found 63 random states is coming in 8-bit LFSR using single taping which is shown in Fig. 2. The timing simulation is shown from 20 ns to 1280 ns. After this time clock the random output is repeating again.
Figure 2. Timing Simulation waveform of LFSR-8 Bit (20 ns to 1280 ns): Total 63 periods
568
Vol. 3, Issue 1, pp. 566-571
3.1.2. Simulation Result of LFSR 16-Bit The actual random output for 16-bit LFSR should come 216-1 = 65535. But from the waveform simulation analysis it is found 255 random states is generating in 16-bit LFSR using single taping which is shown in Fig. 2. The timing simulation is shown from 20 ns to 5120 ns. After this time clock the random output is repeating again.
3.1.3. Simulation Result of LFSR 32-Bit The actual random output for 32-bit LFSR should come 232-1= 4,29,49,67,295. But from the waveform simulation analysis it is found 2046 random states is generating in 32-bit LFSR using single taping which is shown in Fig. 4. The timing simulation is shown from 20 ns to 40940 ns. After this time clock the random output is repeating again.
3.2. Performance Analysis

The performance based on simulation and synthesis report for 8 bit, 16 bit and 32 bit LFSR using single tapping are given in Table 1.
569
Vol. 3, Issue 1, pp. 566-571
Table 1. Simulation and Synthesis Result. Performance Time to complete the total states Total no. of Random States generating Clock Shift Register Xor gate Number of Slices No. of Slice Flip Flops No. of 4 i/p LUT GCLK (Gate + Net) Delay Total pin 8 Bit 20 ns to 1280 ns = 1260 ns 63 20 ns 08 01 04 08 01 01 2.690 ns 10 16 Bit 20 ns to 5120 ns = 5100 ns 255 20 ns 16 01 09 16 01 01 2.690 ns 18 32 Bit 20 ns to 40940 ns = 40920 ns 2046 20 ns 32 01 18 32 01 01 2.690 ns 34
IV.
CONCLUSIONS
It is clearly found from the synthesis and simulation result that 8 bit 16 bit and 32 bit LFSR with single tapping can only generate 63, 255 and 2046 random output rather than 255, 65535 and 4,29,49,67,295. Also the performance is better in 32 bit LFSR as compared to other. In future enhancement by increasing the number of tapping the degree of randomness as well as the run length can be increased and again we can compared for maximum length sequence between different length LFSR by implementing on FPGA.
REFERENCES
[1]. [2].
.] 3[
[4]. [5]. [6]. [7]. [8]. [9].
[10]. [11]. [12].
[14].
[15].
M. Luby, Pseudorandomness and Cryptographic Applications, Princeton University Press, 1996. Random Number Generator (2011). Wikipedia website [Online]. Available:http://en.wikipedia.org/wiki/Hardware_random_number_generator. Jiang Hao, Li Zheying, On the Production of Pseudo-random Numbers in Cryptography in Journal Of Changzhou Teachers College of Technology, Vo1. 7 , No. 4, Dec. 2001. D. E. Knuth, The Art of Computer Programming, Vol. 2: Seminumerical Algorithms. Reading, MA: Addison-Wesley, 1969. F. James, A Review of Pseudo-random Number Generators, Computer Physics Communications 60, 1990. P. L'Ecuyer, Random Numbers for Simulation, Comm. ACM, 33:10, 1990. Pseudo-Random-Generator. Wikipedia website [Online] Available: http://en.wikipedia.org/wiki/Pseudorandom_number_generator Katti, R.S. Srinivasan, S.K., Efficient hardware implementation of a new pseudo-random bit sequence generator IEEE International Symposium on Circuits and Systems, 2009. ISCAS 2009. C. Li and B. Sun, Using linear congruential generators for cryptographic purposes, In Proceedings of the ISCA 20th International Conference on Computers and Their Applications, pp. 13-18, March 2005. L'Ecuyer, Pierre, Tables of Linear Congruential Generators of Different Sizes and Good Lattice Structure, Mathematics of Computation, Vol. 68, No. 225, 1999, Pages 249-260. Linear Feedback Shift Register (2011), Wikipedia website. [Online]. Available: http://en.wikipedia.org/wiki/Linear_feedback_shift_register. Goresky, M. and Klapper, A.M. Fibonacci and Galois representations of feedback-with-carry shift registers, IEEE Transactions on Information Theory, Nov 2002, Volume: 48, On page(s): 2826 2836 Ding Jun, Li Na, Guo Yixiong, A high-performance pseudo-random number generator based on FPGA 2009 International Conference on Wireless Networks and Information Systems. K.H. Tsoi, K.H. Leung and P.H.W. Leong, Compact FPGA-based True and Pseudo Random Number Generators, 11th Annual IEEE Symposium on Field-Programmable Custom Computing Machines (FCCM03), 2003. Jiang Hao, Li Zheying, FPGA design flow based on a variety of EDA tools in Micro-computer information, 2007(23)11-2:201-203.
.] 3 1[
570
Vol. 3, Issue 1, pp. 566-571
[16]. Chan K., Samiha Mourad Digital Design using Field Programmable Gate Arrays Prentice Hall, Inc 1994 [17]. Field programmable gate array (2011). Wikipedia website. [Online]. Available:http://en.wikipedia.org/wiki/Field-programmable_gate_array. [18]. Bhasker J, A VHDL Primer, P T R Prentice Hall, Pages 1-2, 4-13, 28-30 [19]. Xilinx, Inc. Xilinx Libraries Guide, 2011.
Authors Amit Kumar Panda was born on 08 Jul 1982. He has done M. Tech (ELDT) in 2009. His teaching Experience is 2.5 year in Guru Ghasidas Vishwavidyalaya (A Central University), Dept. of ECE, Bilaspur with the designation of Asst. Professor.
Praveena Rajput is working as an Assistant Professor in the Department of Electronics & Communication Engg., Guru Ghasidas Vishwavidyalaya, Bilaspur (C.G). She has done her B.E from I.E.T.E. She is also life member of IETE Institute. She has done his M.E from SSCET Bhilai with Hons.
Bhawna Shukla is working as an Asst. Professor & HOD in the Department of ECE, IT, GGV, Bilaspur (CG). She has done M. Tech from SSCET Bhilai. Her teaching experience is 13 year.
571
Vol. 3, Issue 1, pp. 566-571
NUMERICAL ANALYSIS OF THE VISCOELASTIC FLUID IN PLANE POISEUILLE FLOW

N. Khorasani and B. Mirzalou
1
Department of Mechanical and Aerospace engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
ABSTRACT
The state of flow parameters and bifurcation pictures is evaluated for the flow of a nonlinear viscoelastic fluid in a closed channel. The fluids constitutive equation is assumed to be a derivative of Oldroyd-B and the flow is in between two parallel planes driven by a controlled pressure gradient. In simulation, a set of coupled ordinary differential equations with high degree of non-linearity is resulted. The shear stress/ shear rate curve of the fluid has a maximum and a minimum in the steady state condition. The velocity and stress are represented by orthogonal functions in the direction of the flow and transverse to it. The flow field is obtained from the conservation and constitutive equations truncated by Chebychev functions. Normal stress effects are also taken into account. The steady state and transient solutions are obtained for different Weissenberg numbers. It is observed that for a certain Weissenberg number, the system exchanges its stability with a deviation flow. At a second critical number however, the deviation flow diminishes. Finally, it is found that the flow responds oscillatorily when the Reynolds number is small, and monotically at large Reynolds number.
KEYWORDS: Viscoelastic, Poiseuille flow, Johnson-Segalman, Galerkin, Chandrasekhar
I.
INTRODUCTION
While the problem of PPF has been extensively investigated for Newtonian fluids, respectively little attention has been devoted to the flow of viscoelastic fluids especially Johnson-Segalman model. Laminar poiseuille flows have gained interest in the past two decades due to their applications in many processes involving liquid flows in relatively long tubes, such as waxy crude oil transport [1,2] and polymer extrusion [3,4]. Flow through a channel with fixed boundary conditions and the constant applied negative pressure gradient is known as the Poiseuille flow that contains viscoelastic fluid. Several constitutive equations, such as Doi-Edwards model with a rouse relaxation mode [5,6], the Gieskus model [7], The Maxwell and Oldroyd predict shear stress/ shear rate diagram behaviour, although the Johnson-Segalman model has non-monotonic curves. Theoretical studies on Couette and Poiseuille flows, using the more popular constitutive models, such as Maxwell fluids, found that the base flow is linearly stable at any level of elasticity, in the absence of inertia [8,9]. Recent experiments have shown that non-monotonic constitutive models are suitable to describe plane flows behaviour [10]. The Johnson-Segalman model is used for modelling a high molecular weight fluid composed of a Newtonian solvent and a polymer solute. It is predicted that the bifurcation diagram is influenced dramatically by the viscosity value. The present paper explores and analyses stress parameters and important factors of Johnson-Segalman fluid in a Poiseuille flow. Fluid elasticity and normal stresses give rise to additional nonlinearities and coupling among the flow variables, making an already complex problem even more difficult to solve. Because of this we should solve the equations with a more efficient numerical solution method that will be explained later. The present study focuses on the steady state and pre-critical range of transient behaviour of the PPF of a JS fluid. Fluid elasticity and normal stresses give rise to additional nonlinearities and coupling among the flow variables, making an already complex problem even more difficult to solve. The interplay between inertia and elasticity was examined using a system of six degrees of freedom. A judicious mode selection was later carried out to include the dominant velocity and stress terms. The
572
Vol. 3, Issue 1, pp. 572-580
results show good agreement with the behaviour based on the work of Ashrafi [11].Jeorgiou and Vlassopoulos [12] solved the time-dependant plane shear flow of a Johnson-Segalman fluid using mixed finite elements in space and a fully- implicit scheme in time. Their numerical results agree with findings of Kolkka et al. [13].The present work completes the work of Ashrafi and Khayat [14].
II.
PROBLEM FORMULATION
The plane-Poiseuille flow of a Johnson-Segalman fluid is formulated in this section. Consider a flow of an incompressible viscoelastic fluid of density , relaxation time and viscosity in a channel of width d. the x axis is parallel to the walls and is located exactly between the plates while the y axes is perpendicular to it and directs upward. Two boundary plates are constant and the applied pressure gradient drives the fluid toward in the closed channel. In this study, only fluids that can be reasonably represented by a single relaxation time and constant viscosity are considered. The fluid is a polymer solution composed of a Newtonian solvent and a polymer solute of viscosities and , respectively. Therefore = + . In general, the flow is governed by the continuity and conservation of momentum equations for an incompressible fluid with the applied constitutive equation of JohnsonSegalman model [15,16]. The JS model in dimensional form is given by: D (1) + = [ + ( ) ] Dt D (2) = + . 1 (( ) . + . ) + (U. T + T. (U) ) Dt t 2 2 U is the velocity vector, P is the pressure, T is the polymeric contribution of the stress tensor, t is the time, is the gradient operator and [0,2] is a dimensionless material (slip) parameter. Equation (2) includes both lower and upper-convective terms. The value of is a measure of the contribution of non-affine motion to the stress tensor. For = 0 the motion is affine and the Oldroyd-B model is recovered, whereas for = 2, the motion is completely non-affine and the model is reduced to the Oldroyd-Jaumann model [17]. When = 0 and = 0, the upper-convected Maxwell model is recovered. Thus equation (1) belongs to the class of Oldroyd constitutive equations for incompressible viscoelastic fluids and can also be cast in terms of the Jaumann derivative [15]. After implementing the continuity and momentum equations with respect to the two above equations, a set of ODEs will be achieved: Re dP = We + + t dx y y + (2 ) + y y (3) (4) (5) u + (1 ) 2 y (6)
= t = t
= t
u u y 2 y
Here, the Reynolds number, Re, the Weissenberg number, We, and the solvent-to-solute viscosity ratio, , are the three similarity groups of the problem, which are given respectively by Re = , We = , = (7)
U is the maximum velocity of the basic poiseuille flow at the axis of the channel. Note that in the case of = 1, the first term of equation (3) turns out to be simply equivalent to the We number. The boundary condition is defined as: (8)
573
Vol. 3, Issue 1, pp. 572-580
1 y = ,t = 0 2 Equations (3) to (6) are further simplified if they are cast in terms of the normal stress difference, N, and a combination of normal stresses, Z. the last is found to decay exponentially with time and exhibits uninteresting transient behavior and will be set equal to zero. N= ,Z = + (1 ) (9)
It is convenient to seek the departure in velocity u and stress from the base flow, which are introduced as follows: u = u + u , = + In this case the relevant equations become Re u S S u = + + t y y y (11) (12) (13) (10)
N u u = N + 2(AyS + S + S) t y y
S u u u = S + + (AyN + N + N) t y y y
Note that the Prime is droped in above Equations. Below parameters are carried out from the base flow solution. In this paper, the index b refers to the base flow. u = Ay y N = 2( ) ) (14)
1 + (2 )(
(15)
S =
1 + (2 )(
(16)
The following abbreviations, used in the equations, are introduced: = ( 1) 2 We dp A= dx (17) (18)
2.1. Galerkin projection method and the dynamical system

The solution of Equations (11) to (13) is carried out using the Galerkin projection method. Similarly to any flow in the transition regime, the PPF of viscoelastic fluids involves a continuous range of excited spatio-temporal scales. In order to assess the effect of the motion of the arbitrarily many smaller length and time scales, one would have to resolve in detail the motion of the small scales. This issue remains unresolved, since despite the great advances in storage and speed of modern computers, it will not be possible to resolve all of the continuous ranges of scales in the transition regime. Because of this, nowadays, low order dynamical systems can be a viable alternative to conventional numerical methods to probe the nonlinear range of flow behaviour [18,19]. Low order dynamical systems are carried out using Galerkin projection method. The velocity and stress components assume
574
Vol. 3, Issue 1, pp. 572-580
truncated Fourier or other orthogonal representations in space, depending on the boundary conditions. Expansion coefficients are functions of time alone, thus leading to a nonlinear system upon p projection of equations onto the various modes. In this method the variables u(y,t), N(y,t) and S(y,t) are represented by series of Chandrasekhar functions that satisfy the homogenous boundary conditions. The suitable level of truncation that is imposed here, is M=2 that leads to a six he e, six-dimensional model. The general series representation for the velocity and normal stress difference are given by Equations (19) to (21). ( , )= ( , )= ( , )= ( ) ( ) ( ) ( ) ( ) ( ) (19)
(20)
(21)
[20], Where ( ) are the even and odd Chandrasekhar functions, for i even and odd, respectively [20 and ( ) are defined: (y) = , where are constants. M is the number of modes. Finally after applying the Galerkin projection method on the equations, a set of six ODEs are generated and solved. on s
III.
RESULT & DISCUSSION
3.1. Steady state solution and bifurcation pictures

In this section, the steady-state solution is obtained for any parameter range and some limit cases will state be considered in detail. The steady state solution of equations (11) to (13) are derived by setting the idered steady-state time derivatives equal to zero. Equation (16) is the most revealing result of the poiseuille flow of the JS model. It reflects the possibility of non non-monotonic behaviour for the stress/shear rate relation. r Figure 1 shows the behaviour of the shear stress as a function of for = 0 r 0.2and [0,1]. The figure indicates that the stress curves tend to have two extrema, which merge as increases. This situation is reminiscent of the load/deformation behavio r in elasticity. The case of an Oldroyd behaviour Oldroyd-B fluid ( = 0) is similar to the case of Neo Hookean solid and the curve for a Newtonian fluid ( = 1) ) Neo-Hookean is similar to a curve of a Hookean solid solid.
Figure 1. Steady-state shear stress/shear rate diagram for state
= 0.2and
[0,1].
Obtaining the bifurcation pictures for different values of We and is very difficult and the resulted ordinary differential equations has resolved using a powerful software, spending a lot of time. Recall ns powerful that the Re number, affect only the transient solution and does not take important role in the steadyimportant state solution. Figure 3 and 4 show the steady state normal and shear stresses value as a function of steady-state
575
Vol. 3, Issue 1, pp. 572-580 ,
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 [0,7]for [0.2,1]. It has been shown that for = 0, the curves are linear and the system behaviour tends to the Newtonian fluid. With respect to the initial part of Figure 2, the system loses its stability much fasterat a certain We, as increases and in fact the separation occurs earlier. All the curves start at the origin point. It is apparent from the figures that, in the phase spacethe base flow solution separates in to three parts: pre-critical region ( < ), Critical region ( < < ) and post-critical region ( > ). The range of first critical We value is [0.4 , 0.6] and the second range is [4.8 , 5]. For the critical range, an additional fixed branch emerges, which coincides with the loss of stability of the base flow and the emergence of a stable nonlinear velocity profile. These figures are symmetric with respect to the horizontal axis. By applying departures in velocity and stress, one sudden extremum at the beginning part ofthe curves in the Figure 3 occurs. This figure is comparable to those in Ref. [21]. Shear stress at the start of the motion is very high due to sudden applied pressure gradient and decreases after. It is obvious that there is a nonlinear relation between increasing and the value of normal or shear stresses. While the grow steps of are equal to 0.2, with increasing this number, stress value decreases much slower. 3.2. Pre-critical transient behaviour
Numerical solution of Equations (11) to (13) with the mentioned method, leads to a nonlinear transient solution of the system. Chandrasekhar series representation is used to introduce the flow field and stress along with the Galerkin projection, to drive dynamical system that governs the expansion coefficients. All results of this section, belongs to the pre-critical We number. In this region the base flow is unconditionally stable [18]. The Reynolds number is fixed at Re=0.1 and We is assumed to equal to 0.2. Transient flow behaviour of the system has been shown in the Figure 4. The departure of system to the base flow is shown with dimensionless time scales. In this region, the velocity profile is similar to the Newtonian Poiseuille flow but the curves slightly deviate from the central symmetry axis. Considering the parameters = 1.1and = 0.1, it has been observed from the figure that until the time t=1.9, the velocity profile is rapidly increasing but after that, the growth of the velocity profile has stopped and the curves tend to steady-state condition. From this time to t=3.9, velocity magnitude has slightly reduced and finally at the mentioned time, the system is completely steady. Black dashed curve shows the final steady-state condition. Figure 5 indicates that, the flow does not converge monotonically towards the poiseuille base flow. There is in fact an oscillatory behaviour about the origin (base flow). According to the figure, the flow tends to grow in amplitude at the early stages, exceeding the initial amplitude. After that, the curves are damped after some oscillations according to the implemented pressure gradient value and the steady state velocity is attained approximately at t=5.5. it can be mentioned that while the Re number is constant, the applied pressure gradient do not affect the system to reach steady-state condition. In this region, with increasing the pressure gradient, the maximum velocity increases rapidly. However, due to the high complexity and the direct and indirect influence of this parameter in terms of obtained equations, this is not always true. For example, as is observed in the critical range analysis, transverse behaviour happens. The pressure that is imposed to the system, force the flow to move towards the uniform flow. Figure 6 shows the 3-dimensional transient behaviour of flow in the pre-critical range. The oscillating behaviour of the profile is clearly shown in the figure. The oscillations near axis of the channel are much more than near the walls. This figure gives an appropriate general side to review the flow parameters and velocity distribution in the channel.
576
Vol. 3, Issue 1, pp. 572-580
Figure 2. Bifurcation picture for normal Stress N as function of We for
= 0.04,
= 0.5, = 1
Figure 3. Bifurcation picture for shear Stress S as function of We for
= 0.04,
= 0.5, = 1
Figure 7 shows the velocity profile against the time for = 0.04, = 0.2 and We=0.2 at three different Reynolds number, respectively Re= 0.01, Re= 0.1and Re= 0.5. In the pre-critical Region, increasing pressure gradient of the system, increases the velocity and consequently stress value and the system attains the steady-state condition after some oscillations. There is a slight overshoot initially but the monotonous decay toward the steady state is obvious. Less oscillation occurs at a constant Reynolds number to reach the steady state condition, when the applied pressure gradient decreases and the fluid uniforms with less oscillation but at the same time as previous. The time to attain steady-state condition is independent of the pressure gradient. Number of oscillations is directly in related with the Reynolds number. The curve, depicted at Re=0.5 has less oscillations in comparison with the figure that is depicted in Re=0.1 but needs more time to uniform. For small Re, the convergence to the steady state is monotonic and quick as a result of the dominant viscous effects. Typically at small Re, there is a strong overshoot near t=0, which particularly forces the motion to be monotonic soon afterwards.
577
Vol. 3, Issue 1, pp. 572-580
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.4 0.2 0.0 y 0.2 0.4 t t t t t t t t 0 0.05 0.1 0.5 1.9 2.5 3.9 inf
Figure 4. Transient behaviour of flow in the pre-critical range ( < = 1.1, = 0.04and = 0.2
U y,t
) for
= 0.1,
= 0.2
Figure 5. Influence of implemented pressure gradient on the velocity profile against the time in the pre-critical range ( < ) for = 0.1, = 0.2, = 0.04and = 0.2
Figure 6. 3-dimensional transient behaviour of flow in the pre-critical range (We < We ) for = 0.2, = 0.04 and = 0.2
= 0.1,
578
Vol. 3, Issue 1, pp. 572-580
Figure 7. Transient flow curves at different Reynolds number for = 10, = 0.04 and
= 0.2
IV.
CONCLUSIONS
The steady state and transient nonlinear responses are examined for the plane-poiseuille flow of a viscoelastic fluid that obeys the Johnson-Segalman constitutive equation with added Newtonian viscosity. Given the complexity of the constitutive equation and the potentially intricate interplay between fluid inertia and elasticity, only one-dimensional disturbances are considered. Four dimensionless groups emerge in the formulation, namely, the Reynolds number, Re, the Weissenberg number, We, the solvent-to-solute viscosity ratio, , and the slip parameter, , in the constitutive model. A dynamical system approach based on the Galerkin projection method is adopted to describe the flow as Re and We are varied. The nonlinear dynamical system is derived by expanding the flow field (velocity and stress) into odd and even Chandrasekhar functions along the direction perpendicular to the flow. The equations governing the time-dependent expansion coefficients are derived by projecting the various modes onto the original equations. A suitable truncation level is adopted to close the hierarchy of the resulting set of ordinary differential equations. The stress curves tend to have two extrema, which merge as increases. This situation is reminiscent of the load/deformation behavior in elasticity. For given and , two critical values of the Weissenberg number are found, at which an exchange of slopes and system behaviour takes place between the base flow and a non-trivial steady state branch (nonlinear velocity profile). The two critical points coincide with the extrema of the shear stress/shear rate curve. There are thus three ranges of Weissenberg numbers of interest that are investigated, namely, the pre-critical range, the critical range and the post critical range. Here, we consider only the pre-critical range in the transition regime that is always stable. Although the Reynolds number does not influence the value of the critical We numbers and the bifurcation picture, it has a significant influence on the transient behaviour. The time that the system attains steady state condition is independent of the applied pressure gradient. Number of oscillations is in related with the Reynolds number. For small Re, the convergence to the steady state is monotonic and quick as a result of the dominant viscous effects. Typically there is a strong overshoot near t=0, which particularly forces the motion to be monotonic soon afterwards. As the pressure gradient of the system increases, the magnitude of the stress will be increased and the system attains the steady state condition after some oscillations. There is a slight overshoot initially but the monotonous decay toward the steady state is obvious. Less oscillation occur at a constant Reynolds number to attain the steady state condition, when the applied pressure gradient decreases and the fluid uniforms with less oscillation but at the same time as previous. Various scenarios are explored to understand transient and nonlinear behaviour by varying Re and We. The results are depicted in Figures 4 to 7, where the flow in the precritical range of We is investigated. By increasing the Reynolds number, the flow takes more time to reach the steady-state condition. Thus the more viscous fluids, for a given level of elasticity, tend to relax faster.
579
Vol. 3, Issue 1, pp. 572-580
REFERENCES
[1]. G. Viney, A. Wachs, J. F. Agassant, (2006),Numerical simulation of weakly compressible flows: the restart of pipeline flows of waxy crude oils J. Non-Newtonian, Fluid Mech. 136 (2006)93 oils, Newtonian, (2006)93-105. [2]. M. R. Davidson, Q. D. Nguyen, H. P. Ronningsen, (2007), Restart model for a multi Restart multi-plug gelled waxed oil pipeline, J. petrol. Sci Eng.59 (2007) 1-16. , Sci. [3]. G. C. Georgiou, M. J. Crochet, (1994), Compressible viscous flow in slits with slip at the wall J. Compressible wall, Rheology, 38 (1994) 639-654. 654. [4]. G. C. Georgiou, (2003), The time dependent compressible poiseuille and extrudate The extrudate-swell flows of a carreau fluid with slip at the wall J. Non-Newtonian Fluid Mech.109 (2003)93eau wall, -114. [5]. T. C. B Mcleish, R. C. Ball, (1986), J. polym. Sci. B 24(1986) 1375-1745 [6]. T. C. B. Mcleish, (1987), J. Poly. Sci. B 25, 2253-2264 [7]. D. S. Malkus, J. A. Nohel, B. J. Plohr, (1993), Proc. IUTAMsymp, Numerical simulation of non Numerical nonisothermal flow of viscoelastic fluids vol. 28,pp.57-71 fluids, [8]. M. M. Denn, (1990), Issuesin viscoelastic instability Annu. Rev. Fluid mech. 22,13 Issuesin instability, [9]. R. G. Larson, (1992), Instabilities in viscoelastic flows Rheol. Acta 31, 213 Instabilities flows, [10] N. A. Spenley, X. F. Yuan, M. E. Cates, (1996), J. Phys. . France 6, 551-571 [11] N.Ashrafi, (1999), A low order dynamical systems approach to non linear stability of non A non-linear non-Newtonian fluids, Doctorate Thesis, London s, [12] G.C.Georgiou, D. Vlassopoulos, (1998), J. Non ou, Non-Newtonian Fluid Mech. 75, 77-97 [13] R.W.Kolkka, D.S.Malkus, M.G.Hansen, G.R.Ierley,R.A.Worthing, (1998), J. Non Non-Newtonian Fluid Mech, 29, 303-335 [14] Roger E. khayat, NarimanAshrafi, (2002), A low Dimensional approach to nonlinear plane approach plane-poiseuille flow of viscoelastic fluids [15] R. B. Bird, R. C. Armstrong, O. hassager, (1987), Dynamics of polymeric fluids 2nd edition, Wiley, ] Dynamics fluids, NewYork, Vol. 1 [16] M. W. Johnson , D. Segalman, (1977), A model for viscoelastic fluid behavior which allows non ] d nonaffine deformation, J. Non-Newtonian fluid mech.2, 278 ewtonian [17] M. W. Johnson , D. Segalman, (1977), A model for viscoelastic fluid behavior which allows non ] A nonaffine deformation, J. Non-Newtonian fluid mech.2, 278 ewtonian [18] G.R.Sell, C.Foias and R.T Temam, (1993), Turbulance in Fluid flows: A dynami system approach, Turbulance dynamical Springer-Verlag,New York [19]P.Holmes, J.L.Lumley, G.Berkooz, (1996), Turbulance Coherent structures Dynamical systems and Berkooz, symmetry, Cambridge [20] A. H. Nayfeh and B. Balachadran, (1995), Applied Nonlinear Dynamics, John Wiley & Sons, New , York [21] Roger E.khayat, N. Ashrafi, (2002), A low-dimensional approach to non-linear plane ] linear plane-Poiseuille flow of viscoelastic fluids,J. Physcs of fluids, Vol 14, 1757 ,J. 1757-1766
Authors Nariman Khorasani is an assistant professor in Islamic Azad University Science and University, Research branch, Tehran, Iran since 2005. He teaches graduate and undergraduate courses. His research area is Non-N Newtonian flow.
Babak Mirzalou received the B.Sc. degree in Mechanical engineering from Bahonar University, Kerman, Iran, in 2009, The M.Sc. degree in Mechanical engineering from Science and Research branch of Islamic Azad University, Tehran, Iran, in 2012. His research interests include Non-Newtonian fluids and Rheology. Newtonian
580
Vol. 3, Issue 1, pp. 572-580 ,
LASER DIODE MODELLING FOR WIRELESS POWER TRANSMISSION

Emanuele Scivittaro and Anna Gina Perri
Electrical and Electronic Department, Electronic Devices Laboratory, Polytechnic of Bari, via E. Orabona 4, Bari - Italy
ABSTRACT
Wireless power transmission using laser beaming can be realized with arrays of high power semiconductor laser diodes. In this paper we present a complete model for a single power emitter diode characterized by a tapered amplifying section and a quantum well active layer. The band mixing technique has been used for gain calculation and the Transmission Line Laser Modelling (TLLM) for evaluating the electromagnetic propagation inside a tapered waveguide. Moreover we present a complete time domain analysis, which allows to include non linear effects. The model has been completed with a proper set of rate equation for carrier distribution and sources to take into account the Amplified Spontaneous Emission (ASE) noise. A real device has been simulated and the results are reported and discussed. In particular we have demonstrated that mode conversion becomes overestimated when less than 600 sections are used in the proposed model, or when the step size is greater than 1.5 m.
KEYWORDS: Wireless Power Transmission, Laser Diode Modelling, Tapered Structures, Laser Beaming,
Amplified Spontaneous Emission Noise.
I.
INTRODUCTION
The wireless transmission of large amount of power, as can be 100 KW for example, is not a new concept. Since the beginning of 1900s, Tesla made the first attempt to transfer power using just the earth surface as conductor [1]. In 1968 it has been explained for the first time by Glaser [2] the well known SPS (Solar Power Satellite) concept, which consists in producing energy on a power station in earth orbit, where there is not the day-night cycle, and transmit it wirelessly where necessary on earth. This concept was supported by the great improvement, in the previous years, of microwave technology that made feasible the wireless power transmission. Nevertheless today it is still not possible to make the wireless power transfer in an efficient and easy way as demonstrated by several attempts from different authors [3] [4]. However, in the last years, the gained reliability and the improvements in beam quality emission of power semiconductor lasers opens a new scenario: laser beaming, respect to microwaves systems, allow directionality, big distances, small and light devices for power sources and receivers. Furthermore laser beaming is more safe than microwave power transmission [5]. In this paper we present a complete model for a single power emitter diode characterized by a tapered amplifying section and a quantum well active layer. The band mixing technique has been used for gain calculation and the Transmission Line Laser Modelling (TLLM) for evaluating the electromagnetic propagation inside a tapered waveguide. Moreover we present a complete time domain analysis, which allows to include also non linear effects. The proposed model has been completed with a proper set of rate equation for carrier distribution and sources to take into account the Amplified Spontaneous Emission (ASE) noise.
581
Vol. 3, Issue 1, pp. 581-591
In Section 2 we have described the main feature of a single emitting laser diode and the tapered diode modelling, while in Section 3 the obtained results are analyzed and discussed. The conclusions and future developments are illustrated in Section 4.
II.
LASER DIODE FOR POWER BEAMING
The main feature that a single emitting laser diode has to fulfil is to keep good laser beam quality with the increasing amount of power emitted. It allows either to maximize the conversion efficiency of photovoltaic system either to minimize the receiving surface and consequently the dimensions of photovoltaic receiver. Several works are available in literature [6][7], which describe laser diodes with a quasi diffraction limited laser beam and a power from 7 to 9 W emitted. It has been possible to obtain these results by using a tapered laser, which consists of an index guided straight section joined to a gain guided tapered section, as shown in Fig. 1.
Figure 1. Scheme of a tapered diode structure with typical geometric dimensions.
The tapered section is the amplifying section and allows to distribute energy in a broad area delaying the occurring of the detrimental non linear effects as Catastrophical Optical Damage (COD), Beam Filamentation, etc. The straight section is a mono-modal waveguide which cut off higher modes. Generally speaking, when the taper spreads, the number of allowed higher modes increases and can occur a power transfer from fundamental mode to higher order modes. When the modes are reflected at the wider facet and propagate backwards, they are progressively no more supported and are radiated away. Hence only the fundamental mode remain when the narrower end is reached. It implies that only the fundamental mode experiments the laser amplification.
2.1. Tapered Diode Modelling

In the proposed model, the waveguide is divided into a large number of small rectangular sections, and then, by using the effective index method, the local modes can be obtained. We assume that the fundamental mode is the only guided mode in the straight guide and the power of this mode is normalized to unity. Let An and An+1 be the complex electric field amplitude of the nth and the (n+1)th sections normalised such that |A| = 1. In Fig. 2 we show two incident (Ei,n, Ej,n) , two reflected (ERi,n, ERj,n) and three transmitted (Ei,n+1, Ej,n+1, Ek,n+1) local guided modes in the nth and (n+1)th sections. In addition to these modes reflected radiation modes and transmitted radiation modes also exist. By applying the boundary conditions to the interface between adjacent sections n and n+1 we have:
A j ,n +1 = cij Ai , n
i =1
Mn
(1)
where Mn is the maximum higher-order local guided mode number in the nth section. cij is the coupling coefficient which relates the complex amplitude coefficient of mode j of section (n+1) to mode i of section n.
582
Vol. 3, Issue 1, pp. 581-591
Figure 2. An approximation of tapered waveguides using a series of steps.
The cij coefficient can be evaluated for a TE polarized mode by matching of the transverse field components at the interface between the adjacent sections. That is [8]:
2 i , n j ,n +1 ( j ,n + i ,n +1 ) I i , n; j ,n +1 ( j , n + j ,n +1 ) ( i ,n + i ,n +1 ) I i , n;i ,n I j ,n +1; j ,n+1 cij ( TE ) = 2 i ,n j , n+1 I i ,n; j ,n+1 ( i ,n + i,n+1 ) Ii,n;i ,n I j ,n+1; j ,n+1
where
for j M n
(2)
for j > M n
I m , a ; n ,b =
m,a
En ,b dx .
(3)
In Eq. (2) p,q denotes the propagation constant of the pth mode of the qth section and Ep,q is the associated lateral electric field solution. This power transmission between allowed modes is known as mode conversion. If the mode conversion is negligible the power is conserved mainly in the fundamental mode and the propagation will be adiabatic. For little taper angles, as in Fig. 1, it is possible to have a quasi-adiabatic propagation. It is well known that quantum well structures allows lower threshold currents, narrower gain curve, higher peak gain and less temperature dependence. Of course all these features makes quantum wells very attractive gain medium for our purpose. Moreover the introduction of strain, by using crystal junctions with unmatched lattice, allows band engineering that enhance the possibilities of the designer to define the wonted parameters, as the emission wavelength or prefer one between TE and TM fundamental node. The implemented model gives the possibility to predict the gain spectra in presence of strain for different materials and keep in account, by using the perturbative method suggested by Kane [9], the existing degeneration between heavy hole and light hole at the edge band. Implementing this gain model as purposed in [10], it is possible to obtain the band distribution for a certain quantum well as depicted in Fig. 3, in which we have reported the first five calculated bands of a QW structure composed by a 80 GaAs layer between two thicker layers of Al0.2Ga 0.8As in both crystal direction <110> and <100>. The continuous lines are referred to Heavy Hole bands, the dotted lines are referred to Light Holes.
583
Vol. 3, Issue 1, pp. 581-591
Figure 3. Hole band profiles for a 80 GaAs/Al0.2Ga 0.8As QW in <110> and<100> directions.
Moreover it is possible to see that band mixing occurs and the bands are no more parabolic. Knowing band profiles, it is possible to calculate the density of states, the transitions strengths and the gain. In Fig. 4 an example of gain calculation for the structure of Fig. 3 is reported.
Figure 4. TE gain spectra, for a 80 GaAs/Al0.2Ga 0.8As QW, calculated at various carrier density.
For dynamic considerations and carrier distribution in a QW structure, it has been implemented a complete set of rate equations [11]. In a power device, with broad area sections, the electromagnetic propagation is due to a combination of several modes, as it happens in wider sections of a taper, so we need to take it into account in our model. We can write one rate equation related at the carrier density in each main section of the structure, the well and the two barriers. Of course for each section there are different mechanism of capture/escape characterized by a typical time. Briefly, we can say that the carrier injection in barrier region is due to bias current and carriers from the well; the carrier escape is associated at carriers captured by well and non radiative recombination. In the well injection from barriers increase the density and radiative recombination together with nonradiative phenomena like thermionic emission and tunnel effect deplete the area. In formulas it is possible to write:
584
Vol. 3, Issue 1, pp. 581-591
1 1 d 1 N SCH N N I N V = 1 i 1 SCH SCH 1SCH + 1 1 1 well dt s n e VSCH VSCHq N SCH
2 N + 2 SCH s
2V SCH 1 V SCH
1 dN 1 N = (1 ) 1 SCH dt s
2 1V SCH N SCH 1VSCH N 1 N 1 N1 + (1 ) 2 1 2 V V s well e n (N 1 ) e well w w w w TE v g g TE ( N 1 ) 1 TE TE S TE S TE TM v g g TM ( N 1 ) 1 TM TM S TM S TM
2 2 d 2 N SCH N SCH N N V = 2 SCH 2SCH + 2 1 2 well dt N SCH s n e V SCH dS TE S TE w w = TE v g g TE (N 1 ) 1 TE TE S TE S TE TE dt p
1 N + 1 SCH s
1VSCH 2 V SCH
(4)
+ TE BN 12
dS TM S TM w w = TM v g g TM ( N 1 ) 1 TM TM S TM S TM TM + TM BN 12 dt p
where 1 is the injected current, q is the electronic charge, VSCH is the volume of one side of the SCH layer, s is the carrier transport time in the SCH region, e is the carrier thermionic emission/escape time, is the leakage factor, w is the optical confinement factor of the single quantum well, g(N) is the carrier-density dependent gain, S is the photon density normalized with respect to the volume of the quantum well, Vwell, p is the photon lifetime, is the gain compression factor, is the spontaneous emission factor, i is the internal quantum efficiency, vg is the group velocity and both nsch and n are the carrier recombination lifetimes in the SCH region and in the quantum well, respectively. The carrier-density dependent carrier recombination lifetimes are given by [A + BN + CN2]-1 where A,B, and C are the usual recombination coefficients. Using the same symbolism, the equation (4) can be easily generalized to a MQW (Multi Quantum Well) structure and for a multi modal waveguide as:
585
Vol. 3, Issue 1, pp. 581-591
d ( 1 N SCH dt
)=
i I
1
V SCHq
N SCH
1
2 2V N SCH N N 1V + 1 1 1 well + 2 SCH 1 SCH s V SCH nSCH ( 1 N SCH ) e V SCH 1
1 N SCH 1V SCH dN 1 = (1 ) 1 dt s 1V well
2V N 1 N 2 1 well N1 N V well 1 1 1 2 c n ( N1 ) e
V w w 1 TE v g g TE ( N 1 ) (1 TE TE S TE ) S TE active V well V w w 1 TM v g g TM ( N 1 ) (1 TM TM S TM ) S TM active V well dN i = dt i 1V N i 1 i well V well

i 1 i
Ni
i +1V N i N i +1 i well V well i i +1
Ni n ( Ni )
V w w i TE v g g TE ( N i ) (1 TE TE S TE ) S TE i active V well V w w i TM v g g TM ( N i ) (1 TM TM S TM ) S TM i active V well

2 2V N dN M = (1 ) 2 SCH M SCH s V well dt w M TE v g g TE ( N M
(5)
2V N M 1 1 well N M N V well NM 2M M 1 M c n (NM ) e Vactive M V well Vactive M V well
w ) (1 TE TE S TE ) S TE
w M TM v g g TM ( N M
w ) (1 TM TM S TM ) S TM
d ( 2 N SCH dt
)=
M
N SCH
2
1 N SCH N 1 M V well N SCH 1V SCH SCH 2 + + 1 s 2V SCH n ( N SCH ) 2 e 2V SCH 2
dS TE = dt dS TM = dt
i =1
w TE v g g TE ( N i ) (1 TE TE S TE ) S TE
S TE
TE p
M 1V + TE B N i2 well i =1 Vactive M 1V + TM B N i2 well i =1 Vactive
i =1
w TM v g g TM ( N i ) (1 TE TM S TM ) S TM
S TM
TM p
The propagation analysis has been performed by the TLLM (Transmission Line Laser Modelling) [12] [13]. The basic principle of the TLLM is that the laser cavity length is longitudinally divided into a number of equal length sections s with a length L, as shown in Fig. 5. In each section, a scattering matrix S is used to represent the optical process, including stimulation emission, spontaneous emission, and attenuation. The matrices of these sections are then connected by transmission-lines, which account for propagation delays of the optical wave.
586
Vol. 3, Issue 1, pp. 581-591
Scattering matrices depending on the position Forward wave
backward wave
Figure 5. The Transmission Line Laser Model of the tapered structure.
Each section includes a propagation delay, denoted by T. This is related to the cavity length of laser and the group velocity vg inside the cavity, given by:
L = vg T
(6)
The connecting processes can be represented by a connecting matrix C. From the iterations of scattering and connecting processes, the output optical field in the time domain can be obtained. Then, by applying fast Fourier transform (FFT), the laser output spectrum can be easily acquired. The independent carrier rate equation is used in each section to govern the carrierphoton interaction. In the following, we introduce the scattering and connecting matrices as well as the carrier density rate equation, separately. The calculated scattering matrix is:
VM ( n ) 1 VC ( n ) = VL ( n ) Y (1 + Z S ) VP ( n ) k KC t ( g + Y )( Z s 1) g ( Z S + 1) g ( Z S + 1) 2 K C t ( g + Y ) Z S VM ( n ) VC ( n ) VL ( n ) VP ( n ) k
i
2 KC tYC ( Z S 1)
2 KC tYC ( Z S 1)
2YL ( Z S + 1) ( 2YC Y )( Z S + 1) 2YC ( Z S + 1) ( 2YL Y )( Z S + 1) 4 ( K C tYC Z S ) 4 ( KC tYL Z S )
0 0 Y (1 Z S )
2Y
(7)
587
Vol. 3, Issue 1, pp. 581-591
The proposed algorithm is described by flowchart of Fig.6.
Figure 6. The flow chart of the tapered structure transmission line model.
The scattering matrix includes the frequency-dependent gain, attenuation, mode conversion, beamspreading, and dynamic chirp. The continuous distance, z, is discretized, and replaced by its corresponding section, n. The complete scattering matrix may be derived using the microwave circuit concepts.
III.
ANALYSIS OF RESULTS
With the proposed method, by simulating the real device described in [14], at first we have made a search for the minimum number of section needed for having a reliable solution. The normalized TE0 mode amplitude along the 900 m-long tapered waveguide, for different values of the total number S of sections, is shown in Fig. 7.
588
Vol. 3, Issue 1, pp. 581-591
Figure 7. Normalized TE0 mode amplitude along the 900m-long tapered waveguide for different values of the total number S of sections.
It is easy to see that mode conversion becomes overestimated when less than 600 sections are used in our model, or when the step size is greater than 1.5 m. In order to maintain low the number of section used and to achieve favorable computing speed, it is possible to use a compensating factor, as indicated in [15]. In fact, as shown in Fig. 8, there is a good agreement between the results for the 24-section and 300section model by using an appropriate gain compensating factor.
Figure 8. Gain versus output power for two S values (24 and 300).
Finally, the effects of higher order lateral modes on the light-current characteristics of the linear TSTLLM amplifier model are shown in Fig. 9.
589
Vol. 3, Issue 1, pp. 581-591
Figure 9. Output power versus total current.
From Fig. 9 we can deduced that inclusion of higher order lateral modes in the ASE model causes a smoother transition from the non-lasing to lasing operation.
IV.
CONCLUSIONS
We have presented a complete model for a single power emitter diode characterized by a tapered amplifying section and a quantum well active layer. The band mixing technique has been used for gain calculation and the Transmission Line Laser Modelling (TLLM) for evaluating the electromagnetic propagation inside a tapered waveguide. Moreover the time domain analysis allowed us to include in the analysis also the non linear effects. The proposed model has been completed with a proper set of rate equation for carrier distribution and sources to take into account the Amplified Spontaneous Emission (ASE) noise. The method has been applied to simulate a real device, searching at first the optimal section number to have a reliable solution. We have demonstrated that mode conversion becomes overestimated when less than 600 sections are used in our model, or when the step size is greater than 1.5 m. In the proposed method it is possible to consider also thermal effects [16], which requires however a complex mathematical procedure, currently under development.
REFERENCES
[1] [2] [3] [4] [5] [6] Tesla N., (1905) The Transmission of Electrical Energy Without Wires as a Means for Furthering Peace, Electrical World and Engineer., n. 7, pp. 21. Matsumoto H., Hascimoto K., (2006) Supporting Document for the URSI White Paper on Solar Power Satellite Systems, URSI. Steinsiek F., Weber K. H., Foth W. P., Foth H. J., Schafer C., (2005) Wireless Power Transmission Technology Development and Demonstrations,, Proceedings of 2nd International Conference on Recent Advanced in Space Technologies, pp.140 -149. Summerer L., Purvell O. (2009) Concepts for wireless energy transmission via laser, ACT RPR NRG 2009 SPS ICSOS. Inter-commission Working Group on SPS, URSI White Paper on a Solar Power Satellite (SPS), (2005). Dittmar F., Klehr A., Sumpf B., Knauer A., Fricke J., Erbert G., Trnkle G., (2007) 9-W Output Power From an 808 nm Tapered Diode Laser in Pulse Mode Operation With Nearly DiffractionLimited Beam Quality ; IEEE Journal of Selected Topic in Quantum Electronics, vol .13, n. 5, pp. 316-327.
590
Vol. 3, Issue 1, pp. 581-591
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] Paschke K., (2005) Nearly Diffraction Limited 980-nm Tapered Diode Lasers With an Output Power of 7.7 W, IEEE Journal Of Selected Topics In Quantum Electronics, vol. 11, n. 5, pp.2-16. Ghafouri-Shiraz H., Tan P. W., (1996) Analysis of a semi-linear taperedwaveguide laser-diode amplifier, Microwave and Optical Technology Letters, vol. 12, pp. 53-56. Yablonovitch, E., Kane E.O., (1998) Band structure engineering of semiconductor lasers for optical communications, Journal of Lightwave Technology, vol. 6, n. 8, pp. 1292-1299. Zory P. S., (1993) Quantum Well Lasers; Academic, Boston, MA. Nguyen L. V. T., Lowery A. J., Gurney P. C. R., Novak D., (1995) A Time-Domain model for Highspeed Quantum-Well Lasers Including Carrier Transport Effects; IEEE Journal Of Selected Topics In Quantum Electronics, vol. 1, n. 2, pp. 319-327. Christopoulos C., (2006) The Transmission-Line Modeling (TLM) Method in Electromagnetics, Editor Morgan & Claypool. Nguyen L. V. T., Lowery A. J., Gurney P. C. R., Novak D., (1999) A Tima Domain Model for High Speed Quantum Well Lasers Including Carrier Transport Effects, IEEE Journal of Selected Topics in Quantum Electronics, vol. 1, n. 2, pp. 494-504. Bendelli G., Komori K., Arai S., (1992) Gain Saturation and Propagation Characteristics of IndexGuided Tapered-Waveguide Traveling-Wave Semiconductor Laser Amplifiers (TTW-SLAs), IEEE J. Quant. Electron., vol. 28, n.. 2, pp. 447-457. Lowery A. J., (1988) New inline wideband dynamic semiconductor laser amplifier model, IEE Proceedings Pt. J., vol. 135, n. 3, pp. 242-250. Borruel L., Sujecki S., Moreno P., Wykes J., Krakowski M., Sumpf B., Sewell P., Auzanneau S. C., Wenzel H., Rodriguez D., Benson T. M., Larkins E. C., Esquivias I., (2004) "Quasi-3-D simulation of high-brightness tapered lasers", IEEE Journal of Quantum Electronics, , vol.40, n. 5, pp. 463-472.
Authors
Emanuele Scivittaro received the Master of Science degree in Electronic Engeneering from Polytechnic of Bari in 2006 with a thesis regarding the design, realization and characterization of a PBG-based accelerator for cancer Hadrontherapy, developed in the Electronic Device Laboratory of Bari Polytechnic. During his Ph.D. period, he studied Laser systems for wireless power transmission.
Anna Gina Perri received the Laurea degree cum laude in Electrical Engineering from the University of Bari in 1977. In the same year she joined the Electrical and Electronic Department, Polytechnic of Bari, where she is Professor of Electronics from 2002. Her current research activities are in the area of numerical modelling and performance simulation techniques of electronic devices for the design of GaAs Integrated Circuits and in the characterization and design of optoelectronic devices on PBG. Moreover she works in the design, realization and testing of nanometrical electronic systems, quantum devices and FET on carbon nanotube. Prof. Perri is the Director of Electron Devices Laboratory of the Electronic Engineering Faculty of Bari Polytechnic. She is author of over 250 book chapters, journal articles and conference papers and serves as referee for many international journals. Prof. Perri is a Member of Advisory Editorial Board of International Journal of Advances in Engineering & Technology.
591
Vol. 3, Issue 1, pp. 581-591
EFFECT OF PROCESS PARAMETERS ON THE PHYSICAL PROPERTIES OF WIRES PRODUCED BY FRICTION EXTRUSION METHOD
A. Hosseini, E.Azarsa, B.Davoodi, Y.Ardahani
Department of Mechanical Engineering, University of Tabriz, Tabriz City, Iran
ABSTRACT
The aim of this experiment is the study of the process parameters on wires produced by a novel friction stir extrusion approach. Friction stir extrusion relies on the intrinsic frictional heating and extensive plastic deformation generated by process to stir, consolidate, and extrude the powders, chips directly into useable product forms in onestep. In this work it was attempted to find effects of process parameters such as traverse speed rate and die rotational speed on physical and micro structural properties. It has been observed that hot crack and cold crack defects were existed on wires produced using too high and too low die rotational speed, respectively. Furthermore, the dimension and amount of defects such as voids in the wires generally increases with the increase of die transverse speed.
KEYWORDS: Aluminum Chips, Friction Stir Extrusion, Microstructure, Process Parameters
I. INTRODUCTION
There have been immense advances in most aspects of material recycling specially in the field of Metallic chips that produced through various manufacturing processes; therefore in the last decade many studies have been carried out in developing an effective way to recycle metallic chips. During the manufacture of aluminum components, much of the original material is discarded as machine waste, e.g. swarf, metal chips, etc. Techniques that can compact waste products thereby allow more material to be stored. In addition, recycling and re-processing techniques that help recover some of the material costs are being encouraged. For aluminum and its alloys, the energy required to extract the metal from the primary ore is substantial. The recycling and re-processing of aluminum waste products is, therefore, becoming commercially more attractive. Processes utilized to recycle aluminum chips are divided into two main groups: Conventional method and the direct conversion method. The conventional method requires melting of the material to be recycled, casting of a billet, and then hot extrusion of the billet to form a consolidated product in wire or rod form. The direct conversion method utilizes conventional hot extrusion which may or may not be preceded by a static pressing stage. Among these methods direct conversion is much under attention because it may result in great reduction in consumed material, energy and labor costs [1-3]. In recent works for recycling aluminum chips, it was stated that the difficulty of recycling the selected aluminum scrap depends on scrap type, scrap size distribution, contaminant, and the ratio of surface area to body volume and also influence of chips preparation in the aluminum alloy recovery rate and dross production had been investigated [4-6]. In some other works additional tasks utilized to make improvements in recycling processes. These works include cold press pressure prior to melting process, additional sintering process before the final extrusion of aluminum chips, addition of aluminum powder as a soft matrix in the 7075 aluminum chips in direct conversion process [6-8]. Friction stir welding (FSW) and friction stir processing (FSP) are new welding and processing methods which were developed by TWI (1991) and Mishra et al (2003), respectively. These are solid state processes for joining and microstructure modification, respectively. Both methods rely on plastic
592
Vol. 3, Issue 1, pp. 592-597
dissipation for heat generation and may be considered severe plastic deformation processes. In both technologies, welding and processing, a rotating, non-consumable tool, is used to produce a plasticized layer which undergoes severe plastic deformation and heating. Another advantage of application of FSP to metals is that this process results fine grains in the processed zone due to substantial deformation at hot working temperatures. A novel process that is developed from FSP and FSW is friction extrusion (also named friction stir extrusion). Friction extrusion was developed at TWI in the early 1990s, and employs frictional heat to generate the necessary heat required for extrusion. It has been demonstrated successfully with several aluminum alloys. The required temperature is generated and maintained by frictional contact between the material to be extruded and the extrusion die. A high shear strain rate is developed at the boundary between the die and the extruded material, and within the plastic zones of the extruded material. However, there are some attractive aspects of the friction extrusion process and it may yet prove to be an industrially useful technique for the achievement of several purposes, e.g. recycling of machining waste, consolidation of powder product and, potentially, as a method of mechanical alloying [3]. The friction extrusion process may be considered a variant of the direct conversion method of recycling since a consolidated product is produced without melting and casting. In present study, it was attempted to produce AA2025 aluminum wires by friction stir extrusion of machining chips. Also, effects of process parameters such as traverse speed of rotating die and die rotational speed on physical and micro structural properties had been investigated.
II.
EXPERIMENTAL WORKS
The aluminium chips used as extruded material are AA2025-T8 aluminium which was fabricated through milling of AA2025-T8 shaft. Relatively uniform and clean chips were milled from aluminium shaft without any lubricant or cutting fluid. Then, these chips were dried and cleaned. Schematic illustration of designed setup can be seen in Fig.1. Furthermore, the experimental setup of process is demonstrated in Fig.2. Production of wire from machined aluminium chips includes four main steps: As it can be seen in Fig.1a, stationary billet chamber was located in stationary side of conventional turning machine. It should be mentioned that a conventional turning machine had been utilized in this study to supply required movements for friction stir extrusion die. In second step aluminium chips filled in stationary side of extrusion die and compressed into it (Fig.1b). The third stage is directing rotating die with a load into the stationary billet chamber. At last produced aluminium wire was pushed out through rotating die. Both stationary billet chamber and rotating die were fabricated from H13 hot worked steel. Inner diameter of stationary billet chamber and outer diameter of rotating die are same and 30 mm. The rotating die has a scrolled face and a central through hole. The scroll is similar to patterns machined into friction stir welding tool shoulders designed for operation at a 0 tilt angle. The central hole has a diameter of 2.5mm which defines the resulting wire diameter.
Figure 1. Schematic of friction stir extrusion process to produce aluminum wires from chips
593
Vol. 3, Issue 1, pp. 592-597
A post-extrusion heat treatment (PEHT) of 15 h at 160C was applied to the extruded wires. The PEHT is similar to the artificial aging step used to produce T8 tempers in the alloy plates. However, in normal industrial practice the aging step is typically preceded by a solution heat treatment, water quenching, and stretching to several percent plastic deformations.
Figure 2. Experimental setup of friction stir extrusion process
In order to study the effects of process parameters to properties of fabricated wires various die rotational and traverse speeds had been selected which can be seen in table 1.
Table 1. Samples produced under various process parameters
Sample Number 1 2 3 4 5 Die Rotational Speed (rpm) 200 400 600 400 400 Traverse Speed (mm/min) 20 20 20 10 30
III.
Macroscopic photograph of produced wires in different process parameters can be seen in Fig.3. There are three common imperfection encountered in friction stir extrusion: twist, tear (or cold tear) and cracks [3]. The presence of cold cracks easily can be seen in Fig.3a. This can be attributed to low rotational speed of extrusion die. If the rotational speed is selected with lower amounts, generated heat will not be sufficient to reach the softening temperature of aluminum. Therefore cold cracks will be present on the surface of fabricated wires (see Fig.3a). As it can be seen in Fig.3b with increasing rotational speed from 200 to 400 rpm, this imperfection has been eliminated from surface of wire. It is clear that in samples produced with this rotating speed enough heat generated and conducted to chips to reach them to softening temperature of AA2025-T8 aluminum. In the other words, the required temperature is generated and maintained by frictional contact between the material to be extruded (or chips) and the extrusion die. In high rotational speeds, a high shear strain rate is developed at the boundary between the die and the extruded material (or chips), and within the plastic zones of the extruded material. The ensuing temperature rise can lead to deformation at localized regions. Since the material softens locally, further deformation is concentrated within the region and under an axial
594
Vol. 3, Issue 1, pp. 592-597
load it is continuously extruded through the open ended die without this imperfection. From Fig.3b, it is obvious that the entire surface is quite smooth but a slight twist can be seen. This imperfection can be eliminated by increasing rotational speed to 600 rpm (see Fig3.c), but in this one produced wires surface suffers from big cracks which was named as hot crack. In general, the twist is not obvious on any of the wires made at die rotation rates greater than 400 rpm but becomes more apparent with decreasing rpm below 400 rpm. The wires extruded 600rpm exhibited closely spaced, circumferential, surface, cracks as shown in Fig. 3c. It is worth to mention that in extruded wires with 600 rpm, the cracks appeared after 130mm of extrusion.
Figure 3. Macroscopic photograph of produced wires with different process condition
Fig.4 shows microscopic observation of samples taken by scanning electron microscopy. As it can be seen in this figure, in samples produced with traverse speed of 20 mm/min there are large voids that may reduce mechanical properties. With increasing traverse speed of die to 30 mm/min, these voids change into channel and can exist through all length of wire (see Fig4.c). However in wires fabricated with low transverse speed there is not any imperfection such as voids formation or channeling effect (see Fig4.a). So the optimum traverse speed for rotating die is 10 mm/min or lower.
595
Vol. 3, Issue 1, pp. 592-597
Figure 4. 4 microscopic observation of samples taken by scanning electron microscopy
IV.
CONCLUSIONS
Taking all above-mentioned discussion into consideration the following conclusion can be drawn about issue: 1) By friction stir extrusion processing method defect free aluminum wires can be fabricated from aluminum chips that had been produced through other manufacturing processes. 2) Limits on the process appear to be related to the extrusion temperature which, if too low, results in cold tearing and, if too high, causes what appears to be a form of hot cracking. 3) From macroscopic and microscopic evaluations it can be inferred that the optimum condition for
596
Vol. 3, Issue 1, pp. 592-597
processing is 400 rpm rotational speed and 10 mm/min traverse speed for rotating die.
REFERENCES
[1]. Gronostajski, J., Matuszak, A., 1999. The recycling of metals by plastic deformation: an example of recycling of aluminum and its alloy chips. Journal of Materials Processing Technology 9293, 3541. [2]. Tekkaya, A.E, Schikorra, M., Becker, D., Biermann, D., Hammer, N., Pantke, K., 2009. Hot profile extrusion of AA-6060 aluminum chips. Journal of Materials Processing Technology 209, 33433350. [3]. W.Tang, A.P.Reynolds., Production of wire via friction extrusion of aluminum alloy machining chips, Journal of Material Processing Technology 0924-0136, 2010 [4]. Xiao, Y., Reuter, M.A., 2002. Recycling of distributed aluminum turning scrap. Minerals Engineering 15, 963970. [5]. Puga, H., Barbosa, J., Soares, D., Silva, F., Ribeiro, S., 2009. Recycling of aluminium swarf by direct incorporation in aluminium melts. Journal of Materials Processing Technology 209, 51955203. [6]. Mashhadi, H., Amini, Moloodi, A., Golestanipour, M., Karimi, E.Z.V., 2009. Recycling of aluminium alloy turning scrap via cold pressing and melting with salt flux. Journal of Materials Processing Technology, 31383142. [7]. Samuel, M., 2003. A new technique for recycling aluminium scrap. Journal of Materials Processing Technology 135, 117124. [8]. Sherafat, Z., Paydar, M.H., Ebrahimi, R., 2009. Fabrication of Al7075/Al, two phase material, by recycling Al7075 alloy chips using powder metallurgy route. Journal of Alloys and Compounds 487, 395399.
Authors
Ahmadreza Hosseini was born in Tabriz, Iran in 1986. He received the B.E. degree in mechanical engineer from Islamic Azad University, Tabriz, Iran in 2009. He is studying in M.S. degree in manufacturing engineer at University of Tabriz, Tabriz, Iran. His research interests are computational fluids dynamic, solids mechanic, metal forming and machining and friction stir welding.
Behnam Davoodi was born in Tehran, Iran in 1963. He is assistant professor at Mechanical engineering faculty of University of Tabriz. He received Bachelor of Science degree in manufacturing engineering at university of Tabriz, Master of Science at Tarbiat Modarres University and PHD degree at Institut National des Sciences Appliques (INSA). His fields of research are advanced machining methods, metal forming technologies and advanced processing techniques. Ehsan Azarsa was born in Tabriz, Iran at 1986. He was received his Bachelor of Science degree in Mechanical engineering in 2009 and Master of Science degree in 2011 from University of Tabriz, Iran. His research interests are friction stir welding and processing of metallic and polymeric materials.
Yashar Ardahani was born in Tabriz, Iran at 1986. He was received his Bachelor of Science degree in Mechanical engineering in 2009 and Master of Science degree in 2012 from University of Tabriz, Iran. His research interests are friction stir welding and processing of metallic and polymeric materials, metal forming, machining and non-destructive testing.
597
Vol. 3, Issue 1, pp. 592-597
FABRICATION OF PP/AL2O3 SURFACE NANOCOMPOSITE VIA NOVEL FRICTION STIR PROCESSING APPROACH
1
Shahram Alyali1, Amir Mostafapour2, Ehsan Azarsa3

Department of Mechanical Engineering, Islamic Azad University, Science and Research Branch, Tehran, Iran 2,3 Faculty of Mechanical Engineering, University of Tabriz, Tabriz, Iran
ABSTRACT
In the present study, a novel friction stir processing method was utilized to produce surface PP/Al2O3 nanocomposite. This approach is a new variant of friction stir processing technology. A new tooling system has been devised which include rotating pin, stationary shoulder and a close-loop heating section. The volume percentage of nano Al2O3 particles ranged from 5% to 15% in polypropylene matrix. From microscopic observations, it was clear that the distribution of reinforcing particles was uniform in polypropylene matrix. Moreover, tensile and micro-hardness tests have been utilized to investigate the mechanical properties of prepared samples in different volume percentages. It has been observed that the samples with high percentage of nano Al2O3 powder show higher micro-hardness number as well as higher ultimate tensile strength. Furthermore, it was found that distribution of reinforcement particles has magnificent effect on mechanical properties of fabricated surface composites.
KEYWORDS: Nanocomposite, PP/Al2O3, Friction stir processing, Mechanical Properties
I.
INTRODUCTION
Polymers have extensive applications in aeronautics, automobiles, constructions, oil and gas industries, and so on. Polypropylene (PP) is one of these materials that widely used in automotive industry and electronic applications due to its good performance, high strength to weight ratio and excellent processing properties as well as low cast. However, its application is somewhat limited due to its weak abrasive properties, relatively poor impact resistance at room or low temperatures and low hardness [1-5]. The use of inorganic fillers has been a common practice in the plastics industry to improve the mechanical properties of thermoplastics, such as heat distortion temperature, hardness, toughness, stiffness, and mould shrinkage. Therefore, so much attention has been paid to improve the mechanical properties of PP thermoplastic in the last few decades [5]. The effects of filler on the mechanical and physical properties of the composites strongly depend on its shape, particle size, aggregate size, surface characteristics and degree of dispersion. In general, the mechanical properties of composites filled with micron-sized filler particles are inferior to those filled with nano particles of same filler [6,7]. Nano particles are entities with diameters in the range of 1100 nm. When nano particles are embedded in polymer, the resulted composite material is known as polymer nanocomposite. Recently, the methods utilized for preparation of nanocomposites and properties of manufactured nanocomposite are much under attention [8]. Among reinforcement utilized to modify mechanical properties, alumina (Al2O3) has received large interest because of its excellent dielectric properties, good thermal conductivity, high strength, and resistance to strong acid and bases even at elevated temperatures [9]. In last few years, much attention has been paid to a new surface modification technique named friction stir processing (FSP) [10-13]. FSP is a solid-state processing technique, which was developed based on friction stir welding (FSW) [13]. The main aim of this process is to obtain a finegrained microstructure and it has been advanced as a grain refinement technique. Moreover, it is a very
598
Vol. 3, Issue 1, pp. 598-605
attractive process for fabricating surface composites [14]. As mentioned above friction stir processing ss of metallic materials is a solid solid-state technique and temperature of material does not reach to its melting point. However, friction stir processing of polymeric materials is not a solid solid-state process because polymers consist of molecules of different lengths and the materials do not have single melting point, but melting ranges. Thus, as FSP applied to polymeric materials, some shorter chains reach their melting point while longer chains do not. Therefore, bits of solid material are suspended in oint enough molten material to render the mixture easy to move and form. Another problem that occured during friction stir processing is to promote uniform cooling rate throughout the processing zone. If throughout outer layer of a plastic sheet cools much quicker than inner, a hard shell is formed. As the inner layers then cool, the material contracts and pulls away from the shell. So, large voids will be formed which remarkably reduce the mechanical properties of fabricated composite. For this reason friction stir processing of these materials requires additional cares. In this study, an attempt has been made to investigate the mechanical properties of nano nano-alumina filled PP via a new variant of friction stir processing technique and three volume percentages of nano ant nanopowder in polymeric matrix has been compared in mentioned properties. The method used in the present work utilizes friction stir processing concept in which a heating system is added to process for promote a uniform cooling rate which significantly affects the mechanical and microstructural properties of these materials. The designed tool is consists of a shoe, a rotating pin and a heater, which located at the back of pin. It provides the mixing and joining of plastic parts together in the presence provides of heat. The shoulder is stationary relative to pin, whereas in FSW and FSP of metals th shoulder the rotates with the pin. The tooling system is shown in f figure 1. Tensile and microhardne tests were microhardness carried out to evaluate mechanical properties of fabricated nanocomposites. Furthermore, in order to make a judgment about the dispersion state of alumina particles in the system produced samples has system, been studied by light and electron micros microscopy.
Figure 1. Picture of designed tool for FSW of polymers: a) Schematic illustration b) Photograph of tool igned
II.
EXPERIMENTAL WORKS
Polypropylene sheets utilized were commercial grade and supplied from Germany Company. The Al2O3 powder used in this work has particle size <90 nanometer, which packaged by Nano Pars Lima Chemical Company in Iran. In order to produce PP/Al2O3 nanocomposite, alumina particles were composite, contrived in a groove with various dimensions in the middle of samples. Then the Al2O3 particles
599
Vol. 3, Issue 1, pp. 598-605
were compressed into the groove. It is important to note that various groove dimensions result in different volume fraction of alumina powder in nanocomposites. Volume fraction of particles was estimated by dividing the amount of particles in the slot and area over which it was distributed in the matrix [15]. The final stage is plunging of tool with pin inside the plate for producing nanocomposite. In contrast to conventional friction stir processing technique, in this novel approach there isn't any requirement to close material upper surface with a FSP-shaped tool (this tool is similar to FSP tool but there isn't any pin on it) which prevents outpouring of particles, because the tooling system has a shoe shape shoulder that provide this purpose. In the other words, with utilizing this tooling system one of the friction stir processing steps could be eliminated. This mechanism in turn saves the production time and lessens the costs. The photograph of tooling system is shown in figure 1. The tool used is consists of a shoulder which is look like a shoe, a rotating pin and a heater which located at the back of pin. Additionally, a specially designed fixture was utilized to assure that the tool works in its best performance and keep shoulder stationary relative to pin (Figure 2). Furthermore, the shoulder surface was coated with PTFE. The main role of pin is to produce frictional heat for softening the workpiece and stirring material within the joint. The tool's shoulder is similar to a shoe, which is utilized to contain the displaced material and hold it on the sheet, while it is cooled. A heater, equipped with a closed-loop thermo-controller, is primarily responsible for providing a uniform cooling rate.
Figure 2. producing of Al2O3/PP nanocomposite via novel friction stir processing technique
Samples' thickness, width and length were 10mm, 100mm and 300mm, respectively. The tool pin was made of H13 hot worked steel but shoulder's material was made of 7075 aluminum alloy due to its high thermal conductivity and stability in elevated temperatures. Pin diameter and length were 9 and 6 mm respectively and threaded with standard pitch. The shoulder's width and length were 28 and 250 mm, respectively. A thrust bearing separated the pin from the shoulder and its main purpose was to hold the shoulder stationary relative to pin. Transverse speed of tool, pin rotational speed and shoulder temperature were 1000 rpm, 50 mm/min and 190c, respectively. To ensure that shoulder applies enough pressure on work pieces, a tool-offset depth is required during plunge step of the process. The optimum value of 0.5 mm was achieved for plunge depth through experimental tests. It is worth noting that a dwell time is necessary to start the process. Contact of the pin with the workpiece creates frictional heating and softens the workpiece material; contacting the shoe shaped shoulder to the
600
Vol. 3, Issue 1, pp. 598-605
workpiece increases the workpiece heating, expands the zone of softened material. In general, this step is necessary to heat the material in order to create a pool of semi molten polymer. semi-molten In present study, Al2O3/PP nanocomposites were produced in three different Al2O3 percentages, PP5% Al2O3, PP-10% Al2O3 and PP PP-15% Al2O3 by making 3 grooves with 1*2.8, 2*2.8 and 3*2.8mm dimensions, respectively. Tensile tests were performed by Zwick/Roll (model TIFR010THA50) nsions, device with autograph capability and samples were extracted from each welded part in accordance with ASTM D 638 standard. The three specimens were obtained for evaluating properties of each sample. These tests were conducted at a crosshead speed of 50 mm/min. In addition, the micro microhardness properties of the produced composites were measured on the cross section of processed zone and perpendicular to the processing direction with Struers Duramin device. For this purpose, an Struers-Duramin indenter with a 100 g load for 15 s on a standard Vickers micro hardness tester had been utilized. The micro-hardness standard Vickers micro-hardness (H) was d hardness determined by the equation [16]: H=KP/d2 (1) Where P is the applied load, k is a geometric factor equal to 1.85 and d is the mean diagonal length of the imprint after removing the indenter. At least five imprints were made under each load. For making judgment about particles distribution in polypropylene plates, an OLYMPUS, SZ61 Stereo microscope was utilized. For microscopic evaluations, samples were cut through a LEICA RM2135 evaluations, microtome device.
Figure 3. Diagrams of tensile tests of samples with different alumina contents Table 1. Data obtained from tensile test of fabricated nanocomposites in different volume percentage Sample Youngs modulus (MPa) Ultimate Strength (MPa) Deflection %(Ft) PP 1491 34.7 14.7 PP+5%Al2O3 1378 35.8 13.3 PP+10%Al2O3 1430 36.2 12.5 PP+15%Al2O3 1493 36.9 12.4
601
Vol. 3, Issue 1, pp. 598-605
III.
Tensile test results of samples are shown in figure 3. Also table 1 illustrated the date which has been obtained through tensile test. Young modulus, ultimate tensile strength and percentage of deflection Youngs , are shown in this table. In all processing conditions, the plunge depth was same and 0.5 mm. In other words, designed tool is operated with about 0.5 mm of the tool on contact with the workpiece; any additional workpiece contact will produce significant amount of flash around the shoe. As the penetration depth of tool shoulder increases, more pressure applied to the material exists at the surface of sheets. Since these surface materials are completely melted, increasing pressure will results in outpouring of them from weld nugget, consequently leads to thickness reduction. Therefore, tensile strength decreases due to stress concentration in this area. If the penetration depth is selected lower than 0.5 mm, the shoulder will ride on a cushion of material that will smear across the jo line, joint which in turn lead to low quality joints. From Figure.3 and Table 1, it is clear that increasing volume , fraction of Al2O3 nanopowder leads to an enhancement in ultimate tensile strength of samples. Comparing tensile propeties of samples with 15% alumina with pure polypropylene, it can be inferred ropylene, that ultimate tensile strength of fabricated nanocomposite shows about 10% enhancement. This can be explained as follows: As Al2O3 powder incorporated to polymer matrix, the polymer chains had not enough space and time to orientate which results in a material with shorter molecular chains and consequently reduction of mechanical strength is occurred. On the other hand, addition of a chanical reinforcement material such as alumina powder to the matrix of a polymer c increase tensile can properties due to nucleating effect of this particles As polypropylene cools, in crystallization step particles. existense of nucleating particles like alumina will cause to a reduction in sperulite size and therfore an enhancement in tensile strength. From above mentined points it is clear that in fabricated composites ength. via this method the second mechanism has dominant effect on mechanical properties. However, this second phase generally reduce the alongation because of poor interface bonding bet between polymeric matrix and reinforcement. So, there is a general trend of decreasing elongation as volume fraction of alumina powder increases. The obtained micro-hardness values of produced PP/Al2O3 nanocomposites are reported in figure 4 in hardness terms of Al2O3 contents. According to the results, a 54% increase was observed in the case of micro microhardness values of the nanocomposites with 10% nano alumina powder. The high values of micro microhardness obtained by this method could be attributed to the good dispersion and higher level of nano alumina particles, which is known as hard materials in polymeric matrix. Figure 5 that shows microstructure of PP+10%Al2O3 nanocomposite that was taken through Stereo microscope are a proof to above-mentioned statement. Furthermore, it is obvious that in samples with alumina percentage mentioned Furthermore, above 10% agglomeration of nano powder is responsible for a little reduction in micro micro-hardness value
Figure 4. Micro-hardness of fabricated composites in different nano alumina contents hardness nano-alumina
602
Vol. 3, Issue 1, pp. 598-605
Figure 5. Microscopic observation of fabricated nanocomposites
Figure 6 is demonstrated the SEM images of fabricated nanocomposites in different volume percentage. The images have been taken in two different magnifications (300 and 1 m scales). Good distribution of particles in polymeric matrix is one the main criteria during fabricating of nanocomposites by various technique. Figure 6. A. is shown the nanocomposite with 5% of Al2O3 volume percentage. From this figure it is obvious that particles are distributed uniformly in PP matrix. Although, in some regions there are some bonding between alumina and polypropylene, this in turn demonstrated nanocomposite formation in its best form. In some processing techniques that utilized to onstrated produce polymeric nanocomposite, formation of small voids around reinforcement particles is a around drawback to achieve high mechanical properties. However, according to SEM images, there is not any according void inside processing zone, which is a major reason to improved tensile and micro micro-hardness properties. It is worth to mention that in nanocomposite with high volume percentage of alumina powder agglomeration phenomenon was occurred that resulted in a reduction in mechanical properties. This can be clearly seen in figure 6. F.
603
Vol. 3, Issue 1, pp. 598-605
A B
Polymeric Mass
Polypropylene layers
Alumina Particles
Agglomeration Phenomenon
Figure 6. Scanning electron microscopy of fabricated nanocomposites a) PP- 5% Al2O3 with 1m scale b) PP-5%Al2O3 with 300 m scale c) PP- 10% Al2O3 with 1m scale d) PP-10%Al2O3 with 300 m scale e) PP- 15% Al2O3 with 1m scale f) PP-15% Al2O3 with 300 m scale
IV.
CONCLUSIONS
Taking all above-mentioned discussion into consideration the following conclusion can be drawn about issue: In this study a novel friction stir based method is developed for producing polymeric nanocomposites. Furthermore, it is a very effective method for surface modification of polymers and makes an improvement in mechanical properties of these materials such as tensile strength and hardness value. Comparing with conventional friction stir processing method, designed tooling system leads to a great reduction in manufacturing time because this system eliminate second step of FSP (movement of FSP like tool to close upper surface of samples) and shoe shaped shoulder prevent outpouring of powder during process. A 54% increase in micro-hardness value and 10% enhancement in ultimate tensile strength were observed for the sample with 10% Nano Al2O3 content. Good distribution of reinforcement alumina particles is responsible for this enhancement. In fabricated nanocomposites with higher volume percentage agglomeration phenomenon is occurred with leads to a reduction in mechanical properties.
604
Vol. 3, Issue 1, pp. 598-605
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 References
[1]. Zhen. J & Ding. YZ & Du. JQ, (2001) Plast. Proc. Appl, vol. 23, no. 1, pp. 5-8. [2]. Zheng. Yp & Wang. B, (2002) Studies of TiO2/epox nano composites. Compos. Mater.vol. 9, no. 4, pp. 3-11. [3]. Liang. JZ, (2002) Toughening and reinforcing in rigid inorganic particle filled polypropylene: a review. Appl. Polym. Sci.vol. 83, pp. 1547-1555. [4]. J.Z. Liang, (2007) Evaluation of dispersion of nano-CaCO3 particles in polypropylene matrix based on fractal method, Compsites. Part A, vol. 38, no. 6, pp. 1502-1506 [5]. Morel. R, (1985) Handbook of properties of technical and engineering ceramics, part 1: an introduction for the engineer and designer. England, Her Majestys Stationary Office [6]. Sumita. M & Shizuma. T & Miyasaka. K & Ishikawa. K, (1983) Macromol. Sci. Phys. ;B22:601. [7]. Sumita. M & Tuskurmo. T & Miyasaka. K & Ishakawa. K, (1983) Mater. Sci. ;18:1758. [8]. Chi. MC & Jingshen. W & Jian. XL & Ying. KC, (2002) Polypropylene/calcium carbonate nanocomposites, Polymer, vol. 43, no.10, pp. 2981-2992. [9]. H.J. Liu & H. Fuji & K. Nogi, (2004) Mater. Sci. Technol. 20 pp. 399-402. [10]. K. Ohishi & T.R. Mcnelley, (2005) Metal. Trans. A 35 A pp. 2951-2961. [11]. J.Q. Su & T.W. Nelson & C.J. Sterling, (2005) Scripta Mater. 52 pp. 135-140. [12]. D.C. Hofmann & K.S. Vecchio, (2005) Mater. Sci. Eng. A 402 pp. 234-241. [13]. R.S. Mishra & Z.Y. Ma (2003) Mater. Sci. Eng. A341 307-310. [14]. A. Shafiei-Zarghani & S.F. Kashani-Bozorg & A. Zarei-Hanzaki (2009) Mater. Sci. Eng. A 500 pp. 8491 [15]. Uma. Ramadorai & Joseph. W Newkirk & Rajiv S. Mishra (2005) ''Modification of aluminium alloys to create In-Situ surface composites'', Proceedings of 19th international conference on surface modification technologies, August 1-3, , St. Paul, Minnesota, USA [16]. H. Dong & T. Bell (1999) Surf. Coat. Technol. 111pp. 2940.
Authors
Shahram Alyali was born in Naghadeh, Iran in 1985. He received the B.E. degree in manufacturing engineer from Islamic Azad University, Tabriz, Iran in 2009. He is studying in M.S. degree in manufacturing engineer in Islamic Azad University, Science and Research Branch, Tehran, Iran. His research interests are FSW and FSP onto polymeric materials.
Amir Mostafapour was born in Khoy, Iran in 1971. He is assistant professor at Mechanical engineering faculty of University of Tabriz. He received Bachelor of Science degree in manufacturing engineering at university of Tabriz and Master of Science and PHD degree at Amir Kabir University of Technology. His fields of research are advanced welding and processing methods, metal forming technologies and nondestructive testing methods.
Ehsan Azarsa was born in Tabriz, Iran at 1986. He was received his Bachelor of Science degree in Mechanical engineering in 2009 and Master of Science degree in 2011 from University of Tabriz, Iran. His research interests are friction stir welding and processing of metallic and polymeric materials.
605
Vol. 3, Issue 1, pp. 598-605
FAILED STEAM TRAPS: FIRST STEPS TO REPLACEMENT

Kant E Kanyarusoke1 and Ian Noble-Jack2
Mechanical Engineering Department, Cape Peninsula University of Technology, Cape Town, South Africa. Town,
ABSTRACT
Current Energy, Environmental and Economic concerns are exerting pressure on steam plant managers to re examine their plant maintenance practices. South Africa, big components of the steam line such as boilers practices.In have always received legitimate attention But, the smaller ones especially Steam Traps have tended to be attention. taken for granted. A Literature search reveal lack of adequate Trap knowledge in industry - with the result revealed that up to 55% of steam in some plants is lost through them! In this paper, traps are reviewed.Also, described,is Industrial experimentation carried out at two Cape Town factories on them. Based on th research and the experimentation, current practice of replacing traps with identical ones was challenged withone of the urrent s challengedfactories- advised to try a mix of old and newer Technology traps in its refurbishment.
K EY W ORDS:
Steam Trap, Steam Trap replacement,Energy efficiency,Thermodynamic trap, Venturi Thermodynamic Trap, Condensate removal
I.
INTRODUCTION
Regular Steam line maintenance in industry revolves around keeping heat transfer surfaces clean, preventing line steam leakages and preparing boilers, pressure vessels and heat exchangers for legal inspections. Steam traps tend to receive little attention partly because they are tiny items - often installed in quite obscure areas of the plant. But also, Sandeep [1] has observed that lack of steam trap knowledge is the weakest link in steam line maintenance. Moreover, when properly selected and installed often with expert involvement they give reliable service during their design life. This tends to shield attention from them. And if installed to discharge to the atmosphere as was the situation in this case study flash steam discharge is part of the normal operation. So, an ordinary their observer may not easily detect trap progress to total failure. Figure 1 shows flash and live steam coming out of a trap. It takes a skilled eye to detect the gradual transformation from a to b. mation
Figure 1: Normal operation a to fully failed b condition of a steam trap discharge to atmosphere
Many steam traps are designed for a 3 to 5 year life [2]. This fact is not always appreciated in industry. For example, at two leading East African Industries one of us worked in, the majority had not been replaced for over twenty years. Even in the South African factory under consideration in this
606
Vol. 3, Issue 1, pp. 606-617
paper, they have been on line since the pre 1994 days. Yet according Gardner [3] and other experts like in some of the US Federal Government departments[2& 4],by the 3rd to 5th year, 30 to50% of them may have failed.In an inspection of a Belfast oil refinery steam plant, McKay and Holland [5] found 25% of the traps leaking live steam. They estimated that up to 121 500 could be saved on annual basis by prevention of the steam waste.The implication is that at some stage, traps have to be replaced. This paper illuminates the first steps on their replacement. It summarises the theory behind the functioning and failure of commonly available traps and then focuses on essential industry based (as opposed to laboratory) testing and experimentation before trap selection for replacement. First, a brief review of steam traps and their uses is presented. A discussion of three of commonly available ones follows. It is these types that are later referred to in the experimentation section of the paper. Bimetallic strip and Thermostatic traps are also quite common but are not discussed because the factories in point did not use them. Methods of investigating trap performance are reviewed before a description and discussion of experiments done during the research. Finally - after specific recommendations, several issues arising out of the results and needing further investigation are pointed out.
II.
STEAM TRAPS
Steam trapsare devices installed in a steam-condensate circuit primarily to prevent steam from leaking through the drains provided for condensate - i.e. they trap and block steam in a condensate sub circuit. By doing this, they are able to minimise energy wastage. And by draining condensate, they serve several purposes: Theyhelp keep the steam as dry as possible so that heat transfer rates at surfaces of steam use are maximised They help lower steam sub circuit heat losses to the atmosphere by thinning the liquid film in the pipeline They prevent steam pipeline annular flow, thereby availing maximum flow area and minimising line steam pressure losses They prevent water hammer downstream the steam line thereby protecting fittings and the entire pipeline from possible catastrophic failure. In a series of articles and papers, Kirsner[6 - 8] explains how condensation induced water hammer can be up to 100 times as destructive as ordinary steam induced water hammer in slug flow which in itself is tens of times as destructive as ordinary liquid water hammer!
A secondary trap function is to purge and vent air and non condensable gases from the steam line. This helps keep heat transfer surfaces at point of use, air/gas- free so that overall thermal resistance is minimised. Non condensable gases include carbon dioxide which forms corrosive carbonic acid with water. Its expulsion along with that of oxygen in the air helps prevent corrosion in the entire steam condensate circuit. In effecting its discharge/expulsionduty, some steam inadvertently gets lost even from a brand new trap as evidenced by trap manufacturers catalogues [9 -10] and Swagelok Energy Advisors [11]. In steam systems energy audits, Bhatt for example emphasises correct choice of steam traps as one of ways to improve overallenergy circuit efficiency[12]. Einstein et al. [13] also advise adoption of more modern traps to reduce energy waste and improve system reliability.The loss though normally small - is different for different trap types and makes.In trap selection therefore, there is a fundamental need for balance between effectiveness (i.e. duty) and efficiency (cost of duty in terms of steam leakage).
2.1 Brief Summary of available Trap types

Traditionally, there have been three major types of traps grouped according to their principle of operation [9, 10, and 12]. Mechanical employing the buoyancy principle to shut out steam-laden lightcondensate; Thermodynamic employing the Bernoulli steady flow energy (SFE) principle and Thermostatic using temperature difference between condensate and steam as the driver. The Fixed Orifice Plate trap being simple and cheap- used to be popular decades ago. But its propensity to leakage and blockageshas since led to its replacement by more effective and efficient types.
607
Vol. 3, Issue 1, pp. 606-617
According to Trytten [16] and Hollscher [17], a afourth type, the Venturi orifice trap trapwas first developed for the US navy in early 1960s. It has now been adapted for industry since the mid . mid-1990saccording to the latter.Like the original plate orifice trap, i ithas no moving parts. Itis a form of c convergent-divergent nozzleemploying the first and second laws of Thermodynamics under meta stable flow conditions to metaminimise steam leaks.It is the focus of this paper. The fifth and most recent entrant is a patented It Electronic Condensate Controller [ lectronic [17].It is claimed to be able to monitor and respond to condensation and ensure zero steam discharge into the condensate line. For purposes of this paper, Table 1 givesa summary commentary on three of the categories b based on some manufacturers literature [9, 10 and on our own experiences in industry. 9, 10],
Table 1: Three of the commonly available Traps
III.
TRAP PERFORMANCE INVESTIGATION
Steam trap performance is receiving much attention these days mainly because of current world energy crisis and carbon foot print concerns. According to the May 2010 International Energy ccording Outlook release, [18] about 86% world energy production is from fossil fuels. Most of this energy is 86%of . produced with help ofsteam plants steam plants.Yet according to a new computation by Shafiee and Topal [19], it fiee is estimated that current knownoil gas and coal reserves will have been fully depleted by 2040, 2042 knownoil, and 2112respectively. At the same time the same plants are responsible for the present annual 30 time, Billion Tonnes of carbon dioxide (CO2)load into the atmosphere. The concerns on proper steam plant . management to minimise not only their need but also the overall energy cost component in industrial e he products and services are therefore well founded. Locally - in South Africa - other reasons for proper energy management were highlighted by Fawkes, [ ]. They include: Profitability, Competitiveness [20]. in an increasingly open world economy and capital cost constraints necessitating delays in investments. While the latter may need further scrutiny in light of increasing demand by non hile
608
Vol. 3, Issue 1, pp. 606-617
industrial sectors and of capital equipment cost inflation, the fact remains: every positive effort on steam trap maintenance will reward the organization and society handsomely and in a short period. Estimates by SpiraxSarco[10]indicate that in a given steam plant, if no trap preventive maintenance activities are done, up to 30% of the traps may fail within 3 to 5 years and that at any one time, up to 50% may be blowing live steam! Experts further agree that suitable Preventive Maintenance can reduce resulting steam losses to between 1 and 3% [1, 2 and 21].In a report on adoption of Trap Maintenance in 10 Textile industries in India for example, Sandeep [1] reports a 2008 annual steam saving of 2000 Tonnes from a combined 64 Tonne/hr capacity working 6000 hours a year. In 1999, the US Department of Energy issued a Federal Technology Alert on trap testing. In that alert, a case study on three plants in which a total of 103 traps - out of 511 tested - had failed open wasdescribed. Total annual steam losses were estimated at 20 885 000 lb (Appr. 9500 Tonnes) and subsequent savings on adoption of simple maintenance were estimated at US$ 92 323. Clearly, there is money to be saved (and made too!) by taking a closer lookat maintenance of steam traps. There are also many Tonnes of CO2, Nitrogen and Sulphur oxides (NOx and Sox) not to be offloaded into the atmosphere. The issue we discuss in this paper however is: after identifying a failed trap, do we simply replace it with a new one of the same specifications?If not, how do we go about comparing performances of different traps which may - or may not be of the same type? Best Practice procedures on steam trap maintenance and testing have been described by many authorities e.g. [1, 2, 9-11].Four approaches are now available to trap-testing: Visual using site glasses and/or discharges to atmosphere (the latter - for traps discharging to atmosphere or at return to a hot well); Temperature checks for steam and condensate pressures before and after the trap respectively; Ultrasonic listening devices for flow through the traps; and a most recent one fluid conductivity testing inside the trap.Usually, a combination of some of these approaches with varying sophistication is used depending on the numbers of traps, the skills available and the overall economics of the maintenance programme. First rate organisations and manufacturers like, Armstrong, SpiraxSarco and NASA [9-10 and22] recommend suitable actions to take depending on the results of the inspections and tests. Most of these actions involve repairs and internal cleaning. However, in many sub Sahara African steam plants such as those in East Africa referred to in I above, and the particular case in Cape Town - previous maintenance may have been inadequate. In such cases, simple repairs and cleaning will not be the option because of severe pipeline corrosion and/or previous water and steam hammer effects on the traps.New traps (and probably piping and insulation) become imperative. Normal practice would be to replace failed traps by identical ones (especially if some are held in stock). But this can lead to fresh failures if not well thought out. Also, it cuts out new technology traps which might be able to perform better. It is therefore necessary to take a fresh look at the replacement options available. In this endeavour, theASME code PTC 39.1 2005 is helpful. It describes a method for determining condensate flow rate and steam leakage rate through the trap. These two are the most important (but not the only ones) parameters in comparison of trap performance. In Europe and South Africa however, International Standard ISO 7841:1988 guidelines are the more used procedures. Experimentation in this paper was influenced by this standard.
3.1 Examples of Previous setups on Trap performance comparison

Several researchers and practitioners e.g. [21, 23 -26] have in the past reported comparative performances of different traps. Tsyrulnikov [24] and Shada Abu-Halimeh [25] carried out their tests under laboratory conditions Tucker [21] and Joy on the other hand did their experiments on line but tended to compare industrial performances after replacement of faulty traps with those immediately before. On one hand, therefore, traps were being tested in their prime giving no indication of how comparative performance would change after several months/years of running. Yes - it is accepted that a traps steam leakage rate will increase with time of service. But internal wear or ageing characteristics of different traps can be different. Hence laboratory conditions on their own cannot give conclusive data on trap overall performance. This is why code PTC 39.1-2005 lists three other attributes to be factored into overall performance. They are: Air discharge rate; Susceptibility to failure and Effects of back pressure.
609
Vol. 3, Issue 1, pp. 606-617
On the other hand, the commonestcomparisons are those of lines with some failed traps before and after refurbishment. These are exemplified by practitioners such as Sandeep[1], Webb [23], Dickerson [27] and Hanekom [28]. It is to be expected in such cases that performance would improve. The more realistic approach is to install two new correctly sized traps on a line and monitor their performance over a period of time. Examples of such attempts in the literature includeWardell [29] and Currie [30].It is significant to note that these particular practitioners were motivated by a Technology change in the traps. The same change in South Africa is the reason behind this paper.
IV.
EXPERIMENTS: VENTURI TRAP VS. THERMODYNAMIC TRAP
4.1Background
Company A Ltd. in Cape Town (details of which - including processes - we are not allowed to disclose for confidentiality reasons) was considering replacing some of the steam traps along its extensive steam distribution network. It runs two boilers with a combined capacity of 18 Tonne per hour at a name plate pressure of 1034 kPa (gauge). Normally it fires them to 800 kPa (gauge) or 903 kPa absolute. Saturated steam is then transported to various usage points over a large area. Some of the lines are longer than 1 km.They were old and in many places, insulation was broken down. A big fraction of condensate was not returned and for the particular trap stations in this experiment, discharge was to the atmosphere. Prior to their inviting us, plant management had done comparative tests between thermodynamic (TD) and venturi (VT) type traps at two trap stations on different lines leaving the boiler house. They had also set up the stations for our observation.After observing the alternative traps on line in action, we were presented with results of their previous experiments for interpretation (Tables2 and 3). Using an analysis we illustrate below in 4.4, it seemed within limitations of the data - that the Venturi traps were leaving condensate in the lines and therefore it was not possible to determine their energy efficiencies at that stage. The Thermodynamic traps appeared to perform with respective energy efficiencies of 82% and 90% on lines 1 and 2. It was therefore suggested and agreed that the experiments be repeated at a later date with our involvement.
Table 2: LINE 1: 1 1/2" (40 mm line: D3 VT Vs. 1/2" TD Trap:
Reported steam pressure: Psat = 4 bar (gauge) = 501.3 kPa (abs); Temperature: Tsat = 151.9oC; g-sat = 2.681 kg/m3
Initial bucket mass m1 (kg) Initial bucket Temperature T1 (oC) Final bucket mass m2 (kg) Final bucket Temperature T2 (oC) Condensate Temp. before trap Tcond-trap (oC) Condensate collection time t (s) THERMODYNAMIC (TD) 4.42 5.13 4.93 19.2 25.3 17.4 7.53 10.16 10.53 96.3 97 95.2 135 138 135 600 189 149 VENTURI (VT) 5.526 5.34 18.2 18.2 10.453 10.15 76.5 74.8 139 138 600 600
Table 3: LINE 2: 8" (200 mm line: D2 VT Vs. 1/2" TD Trap:
Reported steam pressure: Psat = 5 bar (gauge) = 601.3 kPa (abs); Temperature: Tsat = 158.9oC; g-sat = 3.178 kg/m3
Initial bucket mass m1 (kg) Initial bucket Temperature T1 (oC) Final bucket mass m2 (kg) Final bucket Temperature T2 (oC) Condensate Temp. before trap Tcond-trap (oC) Condensate collection time t (s) THERMODYNAMIC (TD) 5.12 4.1 17.2 17.4 9.57 8.26 85.9 92.3 148 149 300 300 VENTURI (VT) 5.01 4.41 18.7 17.6 9.01 8.31 70.4 72.5 150 146 600 600
4.2 Theory
Figure 2 gives a sketch of a general arrangement of drip traps along a steam main. If dry saturated steam enters the pipeline at Pressure Psat-in, it soon loses heat ql per metre length of run - and
610
Vol. 3, Issue 1, pp. 606-617
condensate begins to form at a rate ql/hfg per metre length. Earlier on in the mains, a big percentage of this condensate will bypass the traps along the mains because the conveying velocity is in the range of conveying 20 40 m/s [9].Further downstream however, sufficient condensate will have formed to get Further appreciable quantities and fractions into the drip legs and hence into the traps. The issue then is to discharge as much of this condensate as possible while passing minimal steam.
Figure 2: Condensation and Drip Traps along a steam main
Thermodynamic and Venturi traps achieve this in totally different ways. Referring to sketches with in Table 1, condensate pressure in a TD lifts the disc off its seat and flows out through the resulting opening. This continues until condensate has reduced so much that wet steam begins to pass through . the opening. But then, the big gain in velocity through the opening lowers the static pressure below the disc and consequently, the latter shuts off until another sufficient load of condensate a ufficient accumulates. The VT on the other hand passes condensate as a normal venturi device handling an incompressible fluid until saturation (point a in Figure 3). After saturation, meta-stable or flash flow results as pressure continues to drop. Then the flow rate drops rapidly owing to partial conversion of both liquid kinetic energy and static enthalpy to vapour stagnation enthalpy. Either of w or dry steam wet passesthrough as a compressible fluid. In that case, the expansion at the VT throat to a critical at pressure Pcrlimits the throughput to negligible values. The processes are illustrated on a T diagram T-s as in Figure 3. Bailey (1951) is reported by Tucker [ ] to have determined the mass flow rate . [18] variation with inlet dryness fraction. His results showed that the flowrate dropped threefold on fraction. reaching liquid saturation (point bin), halved further at 10% dryness inlet. By dry saturation (point din), , it had reduced to less than 4% of the initial sub cooled rate.
Figure 3: Effect of Venturi inlet state on the thermodynamic Temperature (T) Entropy (s) process diagram
The above brief explanations show that both TD and VT operate by opening for condensate but shut out steam differently. One closes while the other remains open utilising the back pressure created by fluid acceleration to retard oncoming steam. The concern at Co. A Ltd was whether, it was true the
611
Vol. 3, Issue 1, pp. 606-617
VT was wasting less steam and if so, whether it was not leaving condensate in the line. On our second visit, therefore experiments to answer these concerns were set up.
4.3 Experimental set up

We used the bucket condensate collection method to compare the two traps on both lines. The equipment used was: probe: Omega Infra red thermometer with thermocouple probe Make:Omega Professional; Resolution to 0.1oC; Accuracy: Ambient Temp: 25C (77F) 1C (1.8F); 33 to 1000C to2% of reading : or 2C (4F) whichever is greater Electronic weigh scale: Make LBK 12; Capacity;12 kg; Readability to 2 gm; Repeatability; Make: SD 2 gm; Linearity; 4 gm; Stabilisation time; 2 s 2 Plastic buckets: Capacity 15 litres; Empty weight 0.51 kg and 0.50 kg : Digital Stop watch: Make - Casio; Precision to 0.01 s 2 New Delta Venturi traps (D and D4 with D4 as the larger capacity trap 1 new (D3 trap); Spirax SarcoThermodynamic trap; 1 old but functional Spirax Sarco inverted bucket (IB) trap modynamic installed downstream Test station Number 2. est
The set up at a trapping station is shown in Figure 4. The mass of the empty bucket with its cover w . was taken and recorded as mb in kg. Some ordinary tap water was poured into the bucket and a mass m1 obtained.The initial water temperature was taken with help of the probe to give T1. Then the water The bucket was placed to collect discharge from one of the traps and covered. The pipe inlet hole on the cover was a running clearance fit with the pipe and so discharge in the bucket was at atmospheric pressure. The isolating valve to the alternat trap was confirmed closed and then that to the test trap alternative opened. Simultaneously, the stop watch was started and run until closure of the valve. This gave a discharge time t s. The water bucket was then removed and the final water temperature T2, recorded immediately.Finally, the water bucket was weighed to give a mass m2. During the course of the Finally, experiment at least 3 readings of the pipe surface temperature at the bottom of the drip leg were taken using the infra red option on the thermometer - and the average of these approximated to the condensate temperature, Tcondtrap. The average bucket surface temperature Ts, was determined in like trap manner. Also, the site glass was constantly watched for trap cycling.
Figure 4: Setup at a trapping station
The above procedure was repeated for the second trap but with the second bucket as the first one had to be given time to cool. This was repeated on each of two steam lines three times. Some of the precautions taken to get consistent results included: Starting off with as nearly equal amounts of cold water as possible so that the initial bucket thermal masses and wetted area were nearly the same areas
612
Vol. 3, Issue 1, pp. 606-617
Running the experiments for the same amount of time (except where it was evident water quantities and flashing were being quite excessive) - so that bucket heat leakages to the atmosphere were as nearly equal as we could try to make
On the second line, we also checked operation of the inverted bucket trap downstreamwhile the test traps were running. This was to verify whether any of the two alternative traps performance was having an effect downstream. The results of these investigations, together with computations for energy efficiency are given in the sample spreadsheet for line 2 in Table 4.
Table 4: Results of Tests on Line 2: (200 mm main steamline)
COMPANY 'A Ltd' STEAM TRAPS TESTS OF 15th OCT 2009

15TH OCTOBER 2009; 13 30 TO 16 00; MODERATE WINDY, 20.9oC
TD Trap
(VT) D3
(VT) D4
INV. BKT NEXT TO:
Empty Bucket mass: 0.5 kg; Conv. & Radiation Surf. area: 0.2992 m2
Initial bucket mass m1 (kg) 3.2 22.6
94.8 5.2
93.6
TD
3.29 22.9
96.1 5.39
95.3
D3
3.2 22
93.1
D4
3.37 23
96.5 6.19
89.4
2.95 23.5
98.6 4.98
97.1
3.13 18.9
79.3 6.95
95.5
3.0 3 21. 9
91.9 7.22 98.1 411 165
3.0 1 22. 3
93.5 6.75 97.5 409 166 57.7 699 276 3 600 6.23
2.9 8 23. 1
96.9
3.0 9 24. 1
101 7.1 3
95.6
3.0 4 23. 3
97. 7 5.3 5
86
Initial bucket Temperature T1 (oC) Initial specific enthalpy h1 (kJ/kg)

Final bucket mass m2 (kg)
6.7 2
93.5
4.46
69.6
Final bucket Temperature T2 (oC) Final specific enthalpy h2 (kJ/kg) Condensate Temp. before trap Tcond-trap (oC) Average bucket surface Temperature Ts(oC) Cond. spec. enthalpy at Trap hf (kJ/kg) Live steam spec. enthalpy at Trap hg (kJ/kg) Condensate collection time t (s) Average trap discharge rate: m=(m2-m1)/t (10-3kg/s) Appr. Bucket heat loss rate to atm.: Qs = 0.022*0.2992*(Ts-20.9) (kW) Energy rejection rate: Qrej = Qs+[(m2-0.5)h2-(m1-0.5)h1]/t (kW) Energy rejected per kg discharge: qrej (kJ/kg) Trap Energy efficiency: hf/qrej (%) Hourly live steam rejected: mg=3600(Qrej-mhf)/(hg-hf) (kg/hr.)
392 157
53
399 154
54.1
407 157
57
400 166
58.2
392 156
57 662
275 3
401 160
55.3
360 154
50. 2
291 NA
NA
375 160
51.2
56
697 276 3 600 6.98
671
2754
650
2751
672
2754
702
2764
676
275 8
650
275 1
NA
NA
676
2758
300
6.67
300
7.00
300
6.77
600
6.37
600
6.23
600
6.73
300
7.7 0
300
4.20
300
9.40
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2 NA
0.2
5.5
824
81.4
5.8
832
78.1
5.5
813
82.7
4.2 658
107
4.4
636 110
4.1
657 106
3.9
624 106
4.2
625 108
5.2 NA
673 96. 6
6.4 678 99.7
NA
NA
1.8
2.2
1.6
-0.5
-0.7
-0.5
-0.4
-0.6
0.3 NA
0.0
613
Vol. 3, Issue 1, pp. 606-617
International Journal of Advances in Engineering & Technology, March 2012. IJAET ISSN: 2231-1963 4.4 Results Discussion and Follow up Tests elsewhere:
From Table 4, the key findings were: Both Venturi Traps D3 and D4 were undersized for the duty on line 2 as evidenced by the unexpected 100+ % efficiency (and consequent meaningless negative steam discharge rate) highlighted in the table. Trap D3 was actually observed to be continuously discharging and that is the reason D4 was tried. Though D4 was observed to cycle, it still left condensate in the line as evidenced by the increased load onto the inverted bucket trap downstream (see last column of table). The thermodynamic trap was effective in its duty of removing condensate. However, it passed steam at a rate up to eight times the manufacturersloaded condition estimation of 0.25 kg/hr! (No load condition = 0.75kg/hr) [9].Whether this was due to wrong sizing or otherwise is not within of the scope of this report. The inverted bucket trap though old and not a primary target of this investigation did its duty very well and efficiently too! This opens a window for marginally undersized Venturi traps to be used in conjunction with an efficient, resilient trap. Such a combination could combine the positive attributes of both to optimise the total line performance. However, some consideration would have to be given to the effects on part of the line between the traps.
After the above findings on an old and long steam line several hundreds of metres from the boiler house, a second set of experiments was done in another factory (Co. B Ltd.) just a few metres after the steam stop valve in the boiler house itself. The results of this investigation are given in Table 5.
Table 5: Results of Tests at Company B Ltd just after the steam header in the boiler house:
Company 'B Ltd' Steam Trap Tests of 22nd Nov 2009 22nd Nov 2009; 14 30 TO 16 30; BOILER THERMODYNAMIC HOUSE STEAM HEADER; 22.4oC (TD) Empty Bucket mass: 0.5 kg Initial bucket mass m1 (kg) Initial bucket Temperature T1 ( C) Initial specific enthalpy h1 (kJ/kg) Final bucket mass m2 (kg) Final bucket Temperature T2 ( C) Final specific enthalpy h2 (kJ/kg) Condensate Temp. before trap Tcond-trap ( C) Condensate spec. enthalpy at Trap hf (kJ/kg) Live steam spec. enthalpy at Trap hg (kJ/kg) Condensate collection time t (s) Average trap discharge rate: m=(m2-m1)/t (kg/s) Energy rejection rate: Qrej = (m2-0.5)h2-(m10.5)h1)/t (kW) Energy rejected per kg discharge: qrej (kJ/kg) Trap Energy efficiency: hf/qrej (%) Live steam rejected: mg= 3600*(Qrej-mhf)/(hg-hf) (kg/hr.)
o o o
VENTURI (VT) D3
3.614 3.485 3.682
3.697 21.6 90.6 5.334 92.3 386.7 174.8 740.1 2772 300 0.0055 5.266 965.0 76.7 2.16
3.418 22.1 92.7 4.648 89 372.8 177 749.8 2774 300

0.0041
3.704 19.5 81.8 4.652 73.8 309.0 172.2 728.7 2767 300
0.0032
21 88.1
4.339
19.6 82.2
4.215
19.5 81.8
4.599
55.6
232.8 176.7 726.5
56.1
234.9 175 741.0
57.1
239.0 173 732.4
2774
2772
2769
300
0.0024
300
0.0024
300
0.0031
4.253
1037.4
3.403
1076.8
2.064 854.2
2.090 858.9
2.398 784.6
72.3 2.09
67.7 1.94
85.1 0.54
86.3 0.50
93.3 0.29
The main finding here was that the Venturi trap D3 had the necessary capacity for this particular trap location and that it was more energy efficient than the TD it was being compared with. The steam
614
Vol. 3, Issue 1, pp. 606-617
losses through D3 were well within what manufacturers of other traps specify for their properly selected traps. However again, whether the TD was oversized for the location or not is outside the scope of this report.
V.
CONCLUSION AND RECOMMENDATIONS
From the results above, it can be said that a Venturi trap can be an energy efficient trap if properly sized for specific locations. It would also appear that the VT would most likely work best under constant load conditions much upstream the steam line. The theoretical and other practitioners and researchers work cited in this paper however cannot be ignored. There is sufficient evidence to justify installation of VTs in properly maintained plants with well treated boiler feed water. Even downstream a steam main - provided the pipe line insulation is sound - they could be installed. In that case however, we would recommend that they be alternated with more resilient traps like the inverted buckets to maximise total line effectiveness and energy efficiency. To Company A Ltd., we therefore recommended as follows: Begin by looking at the steam line ahead of the stations where you would wish to install new traps. Renew the piping and/or insulation in that part. This is anyway necessary irrespective of which trap type you will select. Aim to use a trap mix (including a suitable VT upstream) along the line. We know this can be problematic in terms of having more than one competing suppliers on site and also in terms of stocking spares etc. But this is your plant and the potential benefits of using the best offers from each can be significant. In any case, work out a total life cycle cost-benefit analysis for various proposals before final selection.
VI.
FURTHER WORK FOR RESEARCH
In this paper, we have reviewed some of the literature on steam traps and given a bit of our own experiences with various traps. An industry based experiment to compare performance of two different traps was described and some of the key results analysed with a view to making specific recommendations to one factory. This work was by no means exhaustive. There are issues it raised but are not answered here-in. For example; in Table5: why did the Venturi condensate temperature appear consistently higher than that for the Thermodynamic trap yet results suggested the formerwas leaving condensate in the line? And if it was leaving condensate in the line, what mechanism was at play to cause it to visibly cycle? Other areas for investigation are more economic than scientific. The use of sub optimal Traps with oversized resilient neighbours to maximise total line utility could be an interesting area to look into. That notwithstanding, we hope we have shed some light on initial steps towards failed trap replacement. The key point is: Do not simply replace the failed trap with its replica. Apply a bit of thought and experimentation where necessary. The results could be more rewarding than you might expect. ACKNOWLEDGEMENT We would like to thank management teams of Companies A and B for having let us do the experiments at their factories. We have in line with the confidentiality agreement with company A kept their identities anonymous. Help by our Thermodynamics Lab Technician, Mark Jenkins is also appreciated.
REFERENCES:
[1] Sandeep, B.N. (2008) Energy & fuel savings by adapting steam trap maintenance in textile processing industries - Experiences from user group, Tirupur User group conference presentation, Tirupur, India, 2008: Available on line at: http://www.emt-india.net/Presentations2010/05Buildings_20Feb2010/Conquest.pdf US Federal Energy Management Programme (1999) Steam Trap Performance Assessment: Advanced technologies for evaluating the performance of steam traps Internet: http://www.eren.doe.gov/femp Tim Gardner, (2007) Wasting energy through steam loss Gardner Energy Management: Available on
[2] [3]
615
Vol. 3, Issue 1, pp. 606-617
[4] [5] [6] [7] [8] line at:www.gemtrap.co.uk US Department of Energy: Steam Tip Number 1 Available on line at: http://www.p2pays.org/ref/08/07612.pdf Mc Kay, G. & Holland, C.R.(1980) Energy savings from Steam losses on an oil refinery, Engineering Costs and Production Economics, Vol.5, pp193203. Kirsner, W. ( 1999): Condensation-InducedWater hammerHPAC Heating/Piping/AirConditioning January 1999 pp 112 122 Kirsner, W. (2005a) Condensation induced Water hammer Canadian Institute of Power Engineers, April, 2005 Kirsner, W. (2005b) Condensation induced water hammer in District steam systems Proceedings of 2005 ASME Pressure vessels and Piping Division Conference; July 17-21, 2005, Denver, Colorado USA Armstrong Steam and Condensate Group: Steam Trapping and Steam Tracing Equipment Catalogue pp 64; Available on line at: www.armstrong-intl.com Spirax Sarco, (2007) Energy Losses in Steam Traps Module 11.15 in The Steam and Condensate Loop Book, 2007 Swagelok Company (2009) Steam System Best Practices Steam - Trap Leak Rates Swagelok Energy Advisors, Inc. Document No. 3, 2009: Available on line at: WWW.SWAGELOKENERGY.COM Bhatt, M. Siddahartha, (2000) Energy audit case studies I steam systems, Applied Thermal Engineering, Vol. 20, pp285-296. Einstein Dan, Worrel Ernst &Khrushch Marta, (2001) Steam systems in industry: Energy use efficiency improvement potentials Lawrence Berkley National Laboratory. Available on line at: http://escholarship.org/uc/item/3m1781f1 Iranian Ministry of Petroleum: Iranian Engineering Standard for Process Design of Steam Traps: IPSE-PR- 845 - Original edition Oct. 1996 Revised 2003. Available on line at:http://217.174.18.60/ips/PR/e-pr-845.pdf Elite, A. LLC (1996) History of Venturi Steam Traps Available on line at: http://www.anchorelite.com/SteamSpecialistHistory.htm Trytten, D. (2008) Venturi Orifice Steam Traps The Engineer's Place for News and Discussion Available on line at: http://cr4.globalspec.com/thread/19163/Venturi-Orifice-Steam-Traps Donald G Holscher, G. D. (2008) Venturi Orifice Steam Traps The Engineer's Place for News and Discussion Available on line at: http://cr4.globalspec.com/thread/19163/Venturi-Orifice-Steam-Traps International Energy Agency , (2010) Report #:DOE/EIA-0484(2010) International Energy Outlook 2010 - Highlights Shafiee, S&Topal, E. (2009) When will fossil fuel reserves be diminished?Journal of Energy Policy Vol. 37 (2009), pp 181189 Fawkes, H. (2005) Energy efficiency in South African industry Journal of Energy in Southern Africa Vol 16 No 4 November 2005 Tucker, J. (2000) Energy and Cost Savings through the installation of low maintenance traps New Practice Case study of the UK Energy Efficiency Best Practice Programme . Aug, 2000 NASA, 2001: Steam Traps Facilities standardized Maintenance task guide. NASA 2001 edition pp26. Webb, T. (1977) Steam trapping International Journal of Energy Research Volume 1, No. 2, pp179185. Tsyrul'nikov, I.M. (1998) Methods of and means for testing steam traps Chemical and Petroleum Engineering, Vol. 34, Nos. 7-8, 1998 Shada Abu-Halimeh (2004) SummaryOfThePerformance AnalysisOfVenturi Orifice Steam Traps MSc Thesis, Queens University Belfast Faculty of Engineering Chemical Engineering Department Joy, D.A. (1987)"Evaluating steam trapperformance data in light of the ANSJ/ASME Test Code," Chemical Processing. (USA), 50, No. 13, pp 92-94, 96 (1987). Dickerson, D. (2003) Clariant Experience of GEM Steam Traps Letter from Clariant UK Ltd to June 2003: Available on line at Gardner Energy Management Ltd, 25th http://www.gemtrap.com/industry_references.php Hanekom, Hans (2009) Un titled: -Letter from SwartlandBoudienste, to Delta Industries, 31st March 2009: Available on line athttp://www.delta-industries.com/Swartland%20letter%20for%20website.pdf Wardell, L. Lee (2002) GEM Steam Trap Efficiency Test Letter from DSM Composite Resins UK Ltd to Gardner Energy Management Ltd, 21st June 2002: Available on line at http://www.gemtrap.com/industry_references.php Currie, R., (1998) GEM Trap Reliability Letter from BAA Glasgow to Gardner Energy Management Ltd, 7th Aug 1998: Available on line at http://www.gemtrap.com/industry_references.php
[9] [10] [11] [12] [13]
[14]
[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
[28] [29]
[30]
616
Vol. 3, Issue 1, pp. 606-617
Authors
Kant E Kanyarusoke is a Ugandan Industrial Mechanical Engineer. He received his BSc in Engineering (1st class Honours) degree from Makerere University, Kampala in 1982 and his MSc in Mechanical Engineering (Design and Production Eng) from the University of Lagos, Nigeria in 1985. He then worked in several Foods, Beverages and Chemical factories in East Africa. Since 2006, he has done fulltime lecturing in Mechanical Engineering Design and Thermo Fluids courses within the Southern African region. He is currently doing doctoral research on solar energy utility maximization for sub Sahara Africa.
Ian Noble-Jack is a South African Professional Engineer with wide experience in industrial processes and machinery. In 1978, he joined the South African Department of Labour as Inspector of Machinery. In 1990, he joined - as lecturer - the Cape Technikon, (the predecessor of current Cape Peninsula University of Technology). He has since been maintaining close contact with industry on consultative basis. He consults mainly on boilers and steam plants.
617
Vol. 3, Issue 1, pp. 606-617
International Journal of Advances in Engineering & Technology. IJAET
ISSN: 2231-1963
MEMBERS OF IJAET FRATERNITY

Editorial Board Members from Academia
Dr. P. Singh, Ethiopia. Dr. A. K. Gupta, India. Dr. R. Saxena, India. Dr. Natarajan Meghanathan, Jackson State University, Jackson. Dr. Syed M. Askari, University of Texas, Dellas. Prof. (Dr.) Mohd. Husain, A.I.E.T, Lucknow, India. Dr. Vikas Tukaram Humbe, S.R.T.M University, Latur, India. Dr. Mallikarjun Hangarge, Bidar, Karnataka, India. Dr. B. H. Shekar, Mangalore University, Karnataka, India. Dr. A. Louise Perkins, University of Southern Mississippi, MS. Dr. Tang Aihong, Wuhan University of Technology, P.R.China. Dr. Rafiqul Zaman Khan, Aligarh Muslim University, Aligarh, India. Dr. Abhay Bansal, Amity University, Noida, India. Dr. Sudhanshu Joshi, School of Management, Doon University, Dehradun, India. Dr. Su-Seng Pang, Louisiana State University, Baton Rouge, LA,U.S.A. Dr. Avanish Bhadauria, CEERI, Pilani,India. Dr. Dharma P. Agrawal University of Cincinnati, Cincinnati. Dr. Rajeev Singh University of Delhi, New Delhi, India. Dr. Smriti Agrawal JB Institute of Engineering and Technology, Hyderabad, India Prof. (Dr.) Anand K. Tripathi College of Science and Engg.,Jhansi, UP, India. Prof. N. Paramesh
pg. A
ISSN: 2231-1963
University of New South Wales, Sydney, Australia. Dr. Suresh Kumar Manav Rachna International University, Faridabad, India. Dr. Akram Gasmelseed Universiti Teknologi Malaysia (UTM), Johor, Malaysia. Dr. Umesh Kumar Singh Vikram University, Ujjain, India. Dr. A. Arul Lawrence Selvakumar Adhiparasakthi Engineering College,Melmaravathur, TN, India. Dr. Sukumar Senthilkumar Universiti Sains Malaysia,Pulau Pinang,Malaysia. Dr. Saurabh Pal VBS Purvanchal University, Jaunpur, India. Dr. Jesus Vigo Aguiar University Salamanca, Spain. Dr. Muhammad Sarfraz Kuwait University,Safat, Kuwait. Dr. Xianbo Qui Xiamen University, P.R.China. Dr. C. Y. Fong University of California, Davis. Prof. Stefanos Gritzalis University of the Aegean, Karlovassi, Samos, Greece. Dr. Hong Hu Hampton University, Hampton, VA, USA. Dr. Donald H. Kraft Louisiana State University, Baton Rouge, LA. Dr. Veeresh G. Kasabegoudar COEA,Maharashtra, India. Dr. Nouby M. Ghazaly Anna University, Chennai, India. Dr. Paresh V. Virparia Sardar Patel University, V V Nagar, India. Dr.Vuda Srinivasarao St. Marys College of Engg. & Tech., Hyderabad, India. Dr. Pouya Derakhshan-Barjoei Islamic Azad University, Naein Branch, Iran. Dr. Sanjay B. Warkad Priyadarshini College of Engg., Nagpur, Maharashtra, India. Dr. Pratyoosh Shukla Birla Institute of Technology, Mesra, Ranchi,Jharkhand, India. Dr. Mohamed Hassan Abdel-Wahab El-Newehy King Saud University, Riyadh, Kingdom of Saudi Arabia. Dr. K. Ramani K.S.Rangasamy College of Tech.,Tiruchengode, T.N., India. Dr. J. M. Mallikarjuna
pg. B
ISSN: 2231-1963
Indian Institute of Technology Madras, Chennai, India. Dr. Chandrasekhar Dr.Paul Raj Engg. College, Bhadrachalam, Andhra Pradesh, India. Dr. V. Balamurugan Einstein College of Engineering, Tirunelveli, Tamil Nadu, India. Dr. Anitha Chennamaneni Texas A&M University, Central Texas, U.S. Dr. Sudhir Paraskar S.S.G.M.C.E. Shegaon, Buldhana, M.S., India. Dr. Hari Mohan Pandey Middle East College of Information Technology, Muscat, Oman. Dr. Youssef Said Tunisie Telecom / Sys'Com Lab, ENIT, Tunisia. Dr. Mohd Nazri Ismail University of Kuala Lumpur (UniKL), Malaysia. Dr. Gabriel Chavira Jurez Autonomous University of Tamaulipas,Tamaulipas, Mexico. Dr.Saurabh Mukherjee Banasthali University, Banasthali,Rajasthan,India. Prof. Smita Chaudhry Kurukshetra University, Kurukshetra, Harayana, India. Dr. Raj Kumar Arya Jaypee University of Engg.& Tech., Guna, M. P., India. Dr. Prashant M. Dolia Bhavnagar University, Bhavnagar, Gujarat, India. Dr. Dewan Muhammad Nuruzzaman Dhaka University of Engg. and Tech., Gazipur, Bangladesh. Dr. Hadj. Hamma Tadjine IAV GmbH, Germany. Dr. D. Sharmila Bannari Amman Institute of Technology, Sathyamangalam, India Dr. Jifeng Wang University of Illinois, Illinois, USA. Dr. G. V. Madhuri GITAM University, Hyderabad, India. Dr. T. S. Desmukh MANIT, Bhopal, M.P., India. Dr. Shaikh Abdul Hannan Vivekanand College, Aurangabad, Maharashtra, India. Dr. Zeeshan Ahmed University of Wuerzburg, Germany. Dr. Nitin S. Choubey M.P.S.T.M.E.,N.M.I.M.S. (Shirpur Campus), Dhule, M.S., India.
pg. C
ISSN: 2231-1963
Dr. S. Vijayaragavan Christ College of Engg. and Technology, Pondicherry, India.
Editorial Board Members from Industry/Research Labs.

Tushar Pandey, STEricsson Pvt Ltd, India. Ashish Mohan, R&D Lab, DRDO, India. Amit Sinha, Honeywell, India. Tushar Johri, Infosys Technologies Ltd, India. Dr. Om Prakash Singh , Manager, R&D, TVS Motor Company, India. Dr. B.K. Sharma Northern India Textile Reserch Assoc., Ghaziabad, U.P., India. Mr. Adis Medic Infosys ltd, Bosnia.
Advisory Board Members from Academia & Industry/Research Labs.

Prof. Andres Iglesias, University of Cantabria, Santander, Spain. Dr. Arun Sharma, K.I.E.T, Ghaziabad, India. Prof. Ching-Hsien (Robert) Hsu, Chung Hua University, Taiwan, R.o.C. Dr. Himanshu Aggarwal, Punjabi University, Patiala, India. Prof. Munesh Chandra Trivedi, CSEDIT School of Engg.,Gr. Noida,India. Dr. P. Balasubramanie, K.E.C.,Perundurai, Tamilnadu, India. Dr. Seema Verma, Banasthali University, Rajasthan, India. Dr. V. Sundarapandian, Dr. RR & Dr. SR Technical University,Chennai, India. Mayank Malik, Keane Inc., US. Prof. Fikret S. Gurgen, Bogazici University Istanbul, Turkey. Dr. Jiman Hong Soongsil University, Seoul, Korea.
pg. D
ISSN: 2231-1963
Prof. Sanjay Misra, Federal University of Technology, Minna, Nigeria. Prof. Xing Zuo Cheng, National University of Defence Technology, P.R.China. Dr. Ashutosh Kumar Singh Indian Institute of Information Technology Allahabad, India. Dr. S. H. Femmam University of Haute-Alsace, France. Dr. Sumit Gandhi Jaypee University of Engg.& Tech., Guna, M. P., India. Dr. Hradyesh Kumar Mishra JUET, Guna , M.P., India. Dr. Vijay Harishchandra Mankar Govt. Polytechnic, Nagpur, India. Prof. Surendra Rahamatkar Nagpur Institute of Technology, Nagpur, India. Dr. B. Narasimhan Sankara College of Science And Commerce, Coimbatore, India. Dr. Abbas Karimi Islamic Azad University,Arak Branch, Arak,Iran. Dr. M. Munir Ahamed Rabbani Qassim University, Saudi Arabia. Dr. Prasanta K Sinha Durgapur Inst. of Adva. Tech. & Manag., Durgapur, W. B., India. Dr. Tole H. Sutikno Ahmad Dahlan University(UAD),Yogyakarta, Indonesia. Dr. Anna Gina Perri Politecnico di Bari, BARI - Italy. Prof. Surendra Rahamatkar RTM Nagpur University, India. Dr. Sagar E. Shirsath Vivekanand College, Aurangabad, MS, India. Dr. Manoj K. Shukla Harcourt Butler Technological Institute, Kanpur, India. Dr. Fazal Noorbasha KL University, Guntur, A.P., India. Dr. Manjunath T.C. HKBK College of Engg., Bangalore, Karnataka, India. Dr. M. V. Raghavendra Swathi Institute of Technology & Sciences, Ranga Reddy , A.P. , India. Dr. Muhammad Farooq University of Peshawar, 25120, Khyber Pakhtunkhwa, Pakistan. Prof. H. N. Panchal L C Institute of Technology, Mehsana, Gujarat, India.
pg. E
ISSN: 2231-1963
Dr. Jagdish Shivhare ITM University, Gurgaon, India.
Research Volunteers from Academia

Mr. Ashish Seth, Ideal Institute of Technology, Ghaziabad, India. Mr. Brajesh Kumar Singh, RBS College,Agra,India. Prof. Anilkumar Suthar, Kadi Sarva Viswavidhaylay, Gujarat, India. Mr. Nikhil Raj, National Institute of Technology, Kurukshetra, Haryana, India. Mr. Shahnawaz Husain, Graphic Era University, Dehradun, India. Mr. Maniya Kalpesh Dudabhai C.K.Pithawalla College of Engg.& Tech.,Surat, India. Dr. M. Shahid Zeb Universiti Teknologi Malaysia(UTM), Malaysia. Mr. Brijesh Kumar Research Scholar, Indian Institute of Technology, Roorkee, India. Mr. Nitish Gupta Guru Gobind Singh Indraprastha University,India. Mr. Bindeshwar Singh Kamla Nehru Institute of Technology, Sultanpur, U. P., India. Mr. Vikrant Bhateja SRMGPC, Lucknow, India. Mr. Ramchandra S. Mangrulkar Bapurao Deshmukh College of Engineering, Sevagram,Wardha, India. Mr. Nalin Galhaut Vira College of Engineering, Bijnor, India. Mr. Rahul Dev Gupta M. M. University, Mullana, Ambala, India. Mr. Navdeep Singh Arora Dr B R Ambedkar National Institute of Technology, Jalandhar, Punjab, India. Mr. Gagandeep Singh Global Institute of Management and Emerging Tech.,Amritsar, Punjab, India. Ms. G. Loshma Sri Vasavi Engg. College, Pedatadepalli,West Godavari, Andhra Pradesh, India. Mr. Mohd Helmy Abd Wahab Universiti Tun Hussein ONN Malaysia, Malaysia.
pg. F
ISSN: 2231-1963
Mr. Md. Rajibul Islam University Technology Malaysia, Johor, Malaysia. Mr. Dinesh Sathyamoorthy Science & Technology Research Institute for Defence (STRIDE), Malaysia. Ms. B. Neelima NMAM Institute of Technology, Nitte, Karnataka, India. Mr. Mamilla Ravi Sankar IIT Kanpur, Kanpur, U.P., India. Dr. Sunusi Sani Adamu Bayero University, Kano, Nigeria. Dr. Ahmed Abu-Siada Curtin University, Australia. Ms. Shumos Taha Hammadi Al-Anbar University, Iraq. Mr. Ankit R Patel L C Institute of Technology, Mahesana, India. Mr.Athar Ravish Khan Muzaffar Khan Jawaharlal Darda Institute of Engineering & Technology Yavatmal, M.S., India. Prof. Anand Nayyar KCL Institute of Management and Technology, Jalandhar, Punjab, India. Mr. Arshed Oudah UTM University, Malaysia. Mr. Piyush Mohan Swami Vivekanand Subharti University, Meerut, U.P., India. Mr. Mogaraju Jagadish Kumar Rajampeta, India. Mr. Deepak Sharma Swami Vivekanand Subharti University, Meerut, U.P., India. Mr. B. T. P. Madhav K L University, Vaddeswaram, Guntur DT, AP, India. Mr. Nirankar Sharma Subharti Institute of Technology & Engineering, Meerut, U.P., India. Mr. Prasenjit Chatterjee MCKV Institute of Engineering, Howrah, WB, India. Mr. Mohammad Yazdani-Asrami Babol University of Technology, Babol, Iran. Mr. Sailesh Samanta PNG University of Technology, Papua New Guinea. Mr. Rupsa Chatterjee University College of Science and Technology, WB, India. Er. Kirtesh Jailia Independent Researcher, India. Mr. Abhijeet Kumar MMEC, MMU, Mullana, India.
pg. G
ISSN: 2231-1963
Dr. Ehab Aziz Khalil Awad Faculty of Electronic Engineering, Menouf, Egypt. Ms. Sadia Riaz NUST College of E&ME, Rawalpindi, Pakistan. Mr. Sreenivasa Rao Basavala Yodlee Infotech, Bangalore, India. Mr. Dinesh V. Rojatkar Govt. College of Engineering, Chandrapur, Maharashtra State, India. Mr. Vivek Bhambri Desh Bhagat Inst. of Management & Comp. Sciences, Mandi Gobindgarh, India. Er. Zakir Ali I.E.T. Bundelkhand University, Jhansi, U.P., India.
pg. H
Volume-3,Issue-1
URL : http://www.ijaet.org

International Journal of Advances in Engineering &amp; Technology (IJAET) Volume 3 Issue 1

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

International Journal of Advances in Engineering &amp; Technology (IJAET) Volume 3 Issue 1

Caricato da

Copyright:

Formati disponibili

International Journal of Advances in Engineering & Technology (IJAET)

Waveform Analysis of Pulse Wave Detected in the Fingertip with PPG

Subhash Bharati & Girmallappa Gidveer

Vol. 3, Issue 1, pp. i-vi

Vol. 3, Issue 1, pp. i-vi

Vol. 3, Issue 1, pp. i-vi

Vol. 3, Issue 1, pp. i-vi

Vol. 3, Issue 1, pp. i-vi

Vol. 3, Issue 1, pp. 1-13

Figure 1. Quadrotor configuration

Vol. 3, Issue 1, pp. 1-13

Vol. 3, Issue 1, pp. 1-13

From (7) it easy to show that :

d = arcsin (u x sin ( d ) u y cos ( d ) ) (u x cos ( d ) + u y sin ( d ) ) d = arcsin cos (d )

k k C kr k , 1 = e m , 2 = s and = m Jr JrR Jr JrR

Vol. 3, Issue 1, pp. 1-13

CONTROL STRATEGIES OF QUADROTOR

Vol. 3, Issue 1, pp. 1-13

3.2. Robust fault tolerant control with sensor faults of quadrotor

Vol. 3, Issue 1, pp. 1-13

& & V&1 = e1 ( x 1d ( y 2 f s 1 ) ) + d%1 ( 1 )

(28) (29) (30) (31)

x id x i i [3,5,7,9,11] ei = x (i 1)d + k (i 1)e(i 1) + (i 1)(i 1) y i i [ 4,6,8,10,12] &

Vol. 3, Issue 1, pp. 1-13

( si ) represent the unknowns parts related to velocity sensor faults.

6 Angular velocity of roll [rad/s]

(a) evolution of angular velocity of roll

(b) evolution of angular velocity of pitch

(c) evolution of angular velocity of yaw

-0.1 x dp y8-Backstepping y8-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50

-0.1 y dp y 10-Backstepping y 10-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50

(d) evolution of linear velocity along X axis

(e) evolution of linear velocity along Y axis

(f) evolution of linear velocity along Z axis

Vol. 3, Issue 1, pp. 1-13

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 -0.05

Angular velocity of roll [rad/s]

(a) evolution of angular velocity of roll

(b) evolution of angular velocity of pitch

(c) evolution of angular velocity of yaw

-0.1 x dp -0.15 y8-Backstepping y8-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50

-0.1 y dp -0.15 y 10-Backstepping y 10-RFTCS 0 5 10 15 20 25 30 temps[sec] 35 40 45 50

(d) evolution of linear velocity along X axis

(e) evolution of linear velocity along Y axis

(f) evolution of linear velocity along Z axis

(a) evolution of roll angle

(b) evolution of pitch angle

(c) evolution of yaw angle

0.1 X position [m] Y position [m]

0.1 Z attitude [m] Yd Y-Backstepping Y-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50

-0.2 Xd X-Backstepping X-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50

(d) evolution of x position

(e) evolution of y position

(f) evolution of z position

Vol. 3, Issue 1, pp. 1-13

0.8 Yaw angle[rad]

(a) evolution of roll angle

(b) evolution of pitch angle

(c) evolution of yaw angle

0.1 X position [m] Y position [m]

0.1 Z attitude [m] Yd Y-Backstepping Y-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50

-0.2 Xd X-Backstepping X-RFTCS 0 5 10 15 20 25 30 Time[sec] 35 40 45 50

(d) evolution of x position

(e) evolution of y position

International Journal of Advances in Engineering & Technology (IJAET) Volume 3 Issue 1

International Journal of Advances in Engineering & Technology (IJAET) Volume 3 Issue 1