Sei sulla pagina 1di 7

World Applied Programming, Vol (2), Issue (5), May 2012.

342-348
ISSN: 2222-2510
2011 WAP journal. www.waprogramming.com

Simulated Study of QoS Multicast Routing


Using Genetic Algorithm
Mala C*

A. Anurag Mahesh

R. Aravind

Department of Computer
Science and Engineering,
National Institute of
Technology, Tiruchirappalli,
Tamil Nadu, India-620015
mala@nitt.edu

Department of Computer
Science and Engineering,
National Institute of
Technology, Tiruchirappalli,
Tamil Nadu, India-620015
CS10501@nitt.edu

Department of Computer
Science and Engineering,
National Institute of
Technology, Tiruchirappalli,
Tamil Nadu, India-620015
CS10542@nitt.edu

R. Rajgopal

Narendran Rajagopalan

B. Nithya

Department of Computer
Science and Engineering,
National Institute of
Technology, Tiruchirappalli,
Tamil Nadu, India-620015
CS10543@nitt.edu

Department of Computer
Science and Engineering,
National Institute of
Technology, Tiruchirappalli,
Tamil Nadu, India-620015
narenraj1@gmail.com

Department of Computer
Science and Engineering,
National Institute of
Technology, Tiruchirappalli,
Tamil Nadu, India-620015
nithya@nitt.edu

Abstract: The evolution of networks has been rapid with increase in the number of network devices.
Multicasting was introduced to increase the bandwidth efficiency of the networks. Multicast routing is the
process of determining an efficient path from the sending node to a group of receiving nodes. In this work,
Genetic Algorithm is used to determine the optimal multicast routing satisfying the quality of service
requirements such as total delay, total delay jitter, packet loss and total bandwidth. New parameters Total
Buffer Size and Queuing Delay are added along with the other parameters and the simulation results through
MATLAB suggest that convergence is faster with better results.
Key word: Search algorithms linear search selection search binary search
I.

INTRODUCTION

Due to the increase in the number of devices and users in the Network, the number of requests for data transfer has
increased exponentially. Basically, network communication can be classified into unicast communication[1], multicast
communication and broadcast communication. In case of unicast communication, there is a one to one correspondence
between the requester and transmitter. Each request being treated as a unicast request is expensive in terms of network
resources and time. In case of broadcasting, the source sends the requested data to all. The applications of broadcasting
are limited within local area networks. In multicasting, if a group of users request for the same data, only one copy of
the data is transferred until a branching is encountered towards the destinations. During branching, data is replicated
and transferred to different destinations. This results in improved efficiency and better resource utilization. In this
paper optimization of multicast routing is performed using Genetic Algorithm taking into consideration some of the
Quality of Service[2] parameters such as total delay, total delay jitter, total packet loss and total bandwidth. Then two
more Quality of Service parameters Buffer Space and Queuing delay are added and the performance of it using
Genetic Algorithm is analyzed.
Genetic Algorithm[3,7,8,9,10] was introduced by John Holland in 1960. It is a meta-heuristic adaptive search
algorithm based on natural biological evolution theory. It can be applied to a wide range of engineering problems to
obtain faster results. Genetic Algorithm involves the simple operations of representing the search space as strings and
performs operations such as,
Selection : It is a process of selecting some of the strings from the search space. There are different methods of
selection that can be applied in Genetic Algorithm, such as stochastic Uniform Selection, Remainder Selection,
Uniform Selection, Roulette Selection, Tournament Selection, Custom Selection.

342

Mala C et al., World Applied Programming, Vol (2), No (5), May 2012.

Reproduction : It is a process of selecting strings for the next generation based upon their fitness values so that the
performance can be improved. Reproduction involves operations like crossover and mutation.
Crossover : It is a process of mixing and matching two strings to evolve with a better one. Crossover can be of
different types like single point crossover, scattered crossover, two point crossover, intermediate crossover, heuristic
crossover, arithmetic crossover.
Mutation : It is a process of changing a small part of the string synonymous to genetic mutation in biological theory.
The probability of mutation is very low. The different types of mutation functions are Gaussian, uniform, adaptive
feasible. Mutation enables the search to be extensive rather than getting constrained within a local optima.
After performing the above mentioned operations, the newly evolved string is evaluated with the fitness function and
accepted or subjected to further evolution.
II.

RELATED WORKS

In [5] Ziqiang Wang et al proposed a PSO Based Multicast Routing Algorithm with Bandwidth and Delay as the
Quality of Service Constraints for optimization. In [6] Liu Jing et al have addressed the Quality of Service Multicast
routing problem using Particle Swarm Optimization with the Quality of Service Parameters under consideration were
Cost, Delay, Delay Jitter and Packet Loss. In [4] Mala et al discuss solving the Multicast Routing problem with
Quality of Service parameters Buffer Space and Queuing Delay in addition to the other basic Quality of Service
Parameters using Particle Swarm Optimization. The next section discusses the proposed work and the algorithm for
solving the problem under study. The simulation and results are discussed in the Simulation and performance analysis
section, followed by conclusion.
III.

PROPOSED WORK

Multicast routing is used in many Internet Applications and Live video streaming. Delay or delay jitter in data transfer
is intolerable in networks. Packet loss can reduce the efficiency of the network. So quality of service constraints such
as delay, delay jitter, packet loss and bandwidth are mentioned. In this work, Genetic Algorithm is applied to choose
the optimal multicast route confirming to quality of service requirements. Network is represented as directed acyclic
graph G= (V, E), where V represents the set of vertices and E represents the set of edges. Each node is associated with
delay, delay jitter and packet loss and each edge with delay, delay jitter and bandwidth.
Delay(G) = Delay(V)+Delay(E)
Delay Jitter(G) = Delay Jitter(V) + Delay Jitter(E)
Packet Loss(G) = Packet Loss(V)
Bandwidth(G) = Bandwidth(E)
Queuing Delay(G) = Queuing Delay(V)
Buffer Size(G) = Buffer Size(V)
The Quality of Service constraints on the tree are summarized as follows,
Delay Constraint : Delay(G) < D
Delay Jitter Constraint : Delay Jitter(G) < DJ
Packet Loss Constraint : Packet Loss(G) < P
Bandwidth Constraint : Bandwidth(G) > B
Queuing Delay Constraint : Queueing Delay(G) < Q
Buffer Size Constraint : Buffer Size(V) > BS
IV.

ALGORITHM

Step 1 : The network under consideration is represented as a graph and using Depth First Search algorithm, different
possible trees are recorded as the solution set.
Step 2 : The different parameters such as Delay, Delay Jitter, Packet Loss, Bandwidth, Queuing Delay and Buffer Size
are calculated for each of the tree in the solution set.
Step 3 : The solution set is represented as string set and subjected to
Step 3.1 : Selection : Twenty random strings are selected from the string set. Stochastic Uniform
Selection is used in this process.
Step 3.2 : Reproduction : It involves the process of creating new strings from the existing ones
after Crossover and Mutation. The crossover probability is equal to 80 percent.

343

Mala C et al., World Applied Programming, Vol (2), No (5), May 2012.

Step 3.3 : Crossover : It is a process of cutting and pasting parts of two strings into new evolved
off springs. Scattered crossover is used in this case.
Step 3.4 : Mutation : It is the process of changing a part of the string to emulate the mutation
behavior of genetic evolution. The probability of Mutation is very low, about 0.001.
Step 3.5 : The fitness value of the string is evaluated with the fitness function F and if the
number of iterations is less than N, steps 3.1 to 3.4 is repeated.
Step 4 : The tree corresponding to the string with best fitness value is taken as the multicast tree.
V.

SIMULATION AND PERFORMANCE ANALYSIS

Genetic Algorithm is applied for Quality of Service Optimization with parameters Total Bandwidth, Total Delay, Total
Delay Jitter and Total Packet Loss. The fitness function is given as
F = C1/TBW * ( C2 * TD + C3 * TDJ + C4 * TPL )
(1)
Where F is the fitness function
C1, C2, C3 and C4 are constants.
TBW is Total Bandwidth.
TD is Total Delay.
TDJ is Total Delay Jitter.
TPL is Total Packet Loss.
In equation (1), the function is directly proportional to Total Delay, Total Delay Jitter and Total Packet Loss and
inversely proportional to Total Bandwidth.
F TD, TDJ, TPL
F 1/TBW
For optimal performance of a network, Total Delay, Total Delay Jitter and Total Packet Loss must be minimized and
Total Bandwidth must be increased. Hence the fitness function F is a minimization function.
The capacity of intermediate nodes to store and forward data increases the performance of the network by decreasing
packet loss and also reducing delay jitter. Hence a new parameter Total Buffer Space is added with the parameters in
equation (1) and the fitness function is given as
F = C1/TBW * ( C2*TD + C3*TDJ + C4*TPL ) * C5/TBS
Where C5 is a constant.
TBS is Total Buffer Space.

(2)

In the Equation (2), F is inversely proportional to Total Buffer Space.


F 1/TBS
But the performance of the network improves with increase in Buffer Space, hence the fitness function is a
minimization function.
Queuing Delay is the average time the packets spend in the queues in the intermediate nodes between the source and
the destination. Queuing delay increases the overall delay and delay jitter of the network. So another new Parameter
Total Queuing Delay is added with the parameters in equation (1) and the fitness function is given as
F = C1/TBW * ( C2*TD + C3*TDJ + C4*TPL + C5*TQD )
Where C5 is a constant.
TQD is Total Queuing Delay.
In Equation (3), F is directly proportional to Total Queuing Delay.
F TQD

344

(3)

Mala C et al., World Applied Programming, Vol (2), No (5), May 2012.

The performance of the network improves with decrease in Queuing Delay, hence the fitness function is a
minimization function.
Considering both the parameters Total Buffer Space and Total Queuing Delay, the fitness function is given as,
F = C1/TBW * ( C2*TD + C3*TDJ + C4*TPL ) * C5/TBS + C6 * TQD
(4)
The fitness function in equation (4) is a minimization function as derived from equations (2) and (3).
Genetic Algorithm is applied to the set of parameters Total Bandwidth, Total Delay, Total Delay Jitter and Total
Packet Loss with fitness function as per equation (1) and tabulated in Tables. These parameters in equation (1) are
called as Reference Parameters. Similarly Genetic Algorithm is applied to optimize Quality of Service with an
additional Parameter Total Buffer Size with fitness function as per equation (2) and tabulated in Tables as With
Reference Parameters and Buffer Space. With Total Queuing Delay as another parameter, genetic algorithm is
applied with fitness function given in equation (3) and named as With Reference Parameters and Queuing Delay.
The results obtained with all parameters with fitness function (4) are tabulated under With Reference Parameters,
Buffer Space and Queuing Delay in the Tables.
The Objective Function Values obtained using Genetic Algorithm with different Parameter set are tabulated in Table
1. Figure 1 depicts the graphical comparison of Objective function values for different parameter sets.
Table 1. Table with Objective Function values tabulated for Different Parameter Sets

Sl. No

Number of
Nodes

With
Reference
Parameters(
RP)

With Reference
Parameters and
Buffer
Space(RPBS)

With Reference
Parameters and
Queuing
Delay(RPQD)

With Reference
Parameters, Buffer
Space and Queuing
Delay(RPBQ)

20

1.3655

0.082376

0.84102

0.0047813

30

1.7118

0.17059

1.7059

0.0071779

40

3.2563

0.3584

3.0588

0.012315

50

4.8041

0.56439

4.8039

0.021235

60

9.0864

0.8133

6.9412

0.027229

70

9.4729

1.173

9.4706

0.037140

Figure 1. Graph depicting Objective Function Values for Different Set of Parameters

345

Mala C et al., World Applied Programming, Vol (2), No (5), May 2012.

Table 2. Objective Function values at Different Generations of Genetic Algorithm with Different Parameter sets
Sl.
No

Number
of
Generations

1
2
3
4
5
6

10
20
30
40
50
60

Objective
Function Values
for
Reference
Parameters
0.7451
0.7451
0.7451
0.7451
0.7451
0.7451

Objective
Function
Values for Reference
Parameters
with
Buffer Space
0.0745
0.0745
0.0745
0.0745
0.0745
0.0745

Objective
Function
Values for Reference
Parameters
with
Queuing Delay
0.9313
0.825
0.7451
0.7451
0.7451
0.7451

Objective Function Values


for Reference Parameters
with Buffer Space and
Queuing Delay
0.005078
0.003972
0.003628
0.003325
0.003128
0.002925

Figure 2. Objective Function Values at Different Generations for Different Parameter Sets

Figure 3. Graph shows the convergence of Genetic Algorithm with RPBQ Parameters

346

Mala C et al., World Applied Programming, Vol (2), No (5), May 2012.

The above discussed multicast routing with quality of service constraints is also solved using Particle Swarm
Optimization, another multi objective optimization algorithm and the time taken for Convergence is listed in Table 3
and figure 4 depicts it graphically.

Table 3. Time taken for Convergence Using Particle Swarm Optimization and Genetic Algorithm

Sl No

Number of Nodes

Time in seconds
Using Particle Swarm
Optimization

Time in Seconds
Using Genetic
Algorithm

1
2
3
4
5
6

20
30
40
50
60
70

52.3136
46.5248
35.376
25.5136
17.384
14.1504

56.1728
52.528
39.2352
25.9424
16.7232
13.5072

Figure 4. Graph depicting the performance of Genetic Algorithm and Particle Swarm Optimization

VI.

CONCLUSION

In this work, an attempt is made to choose the optimal multicast route satisfying the quality of service requirements
such as Total Delay, Total Delay Jitter, Packet loss, Bandwidth, Queuing Delay. The Simulation results depicting the
Best fitness Value with respect to generations and Mean fitness values with respect to generations clearly indicates that
the addition of Buffer Size and Queuing delay increases the efficiency of the Genetic Algorithm to a considerable
level. Hence, with the addition of these parameters, Genetic Algorithm can be used to select the optimal multicast
route from a list of routes satisfying the quality of service constraints much faster.

347

Mala C et al., World Applied Programming, Vol (2), No (5), May 2012.

REFERENCES
[1]

Dragos Ilie, Optimization Algorithms with Applications to Unicast QoS Routing in Overlay Networks, Research Report No.2007:09, ISSN
1103-1581, 2007.
[2] Junwei Wang, Xingwei Wang, Min Huang, An intelligent QOS Multicast Routing Algorithm under Inaccurate Information, IEEE 2006, pp
1073-1077, 2006.
[3] D.E. Goldberg, Algorithms in Search, Optimization and Machine Learning, Pearson Education, India, 2004.
[4] Mala C, Narendran Rajagopalan, Simulated Study of QoS Multicast Routing Using Particle Swarm Optimization, World Applied
Programming, Vol (1), No (3), pp.176-182, August 2011.
[5] ZiqiangWang, Xia Sun, Dexian Zhang, A PSO Based Multicast Routing Algorithm, IEEE Third International Conference on Natural
Computation(ICNC), pp. 664-667, 2007.
[6] Liu Jing, Jung Sun,Wenbo Xu, QOS Multicast Routing based on Particle Swarm Optimization, E.Corchado et al.(Eds.): IDEAL LNCS
4224, pp. 936-943, 2006.
[7] D. Bhandari, C. A. Murthy, and S. K. Pal, Genetic algorithm with elitist model and its convergence, Int. J. Pattern Recognit. Artif. Intell.,
vol. 10, pp. 731747, 1996.
[8] K .F.Man, KS. Tang and S. Kwong, Genetic Algorithms, Springer, 1999.
[9] K. F. Man, K. S. Tang, and S. Kwong, Genetic algorithms: Concepts and designs, IEEE Trans. Ind. Electron., vol. 43, no. 5, pp. 519
534,Oct. 1996.
[10] Zbigniew Michalewicz, Genetic Algorithms + Data Structure = Evolution Programs, Springer-Verlag, 1995.

348

Potrebbero piacerti anche