PERFORMANCE ANALYSIS OF DIFFERENTIAL PHASE MODULATION FOR HF COMMUNICATION

NORHASHIMAH BINTI MOHD SAAD

A project report submitted in fulfillment of requirements for the award of the degree of Master of Engineering (Electrical-Electronics & Telecommunications)

Faculty of Electrical Engineering Universiti Teknologi Malaysia

OCTOBER 2004

iii

Dedicated to my beloved Mak and Abah: Hj. Mohd Saad b. Hj. Kasim and Hjh. Siti Jeliha bt. Hj. Zakaria

iv

ACKNOWLEDGEMENTS

I wish to express my deepest gratitude and appreciation to my supervisor, Associate Professor Dr. Ahmad Zuri Shaameri, for his guidance, suggestions and encouragements throughout this study.

I would like to thank the DSP Lab Technician, Mr. Jefri Ismail for the cooperation, help and constant support throughout this study.

Very special appreciation and gratitude to Abdul Rahim Abdullah, and all my colleagues in DSP Lab: Ahmad Sazali Senawi, Nurulfadzilah Hasan, Abdul Rahim Mat Sidek and Fitri Dewi Jaswar, for all valuable suggestions, encouragements and unconditional supports to complete this study.

Last but not least, a special thanks to my parents, Hj. Mohd Saad b. Hj. Kasim and Hjh Siti Jeliha bt. Hj. Zakaria who always pray for my success, and all my colleagues in UTM for sharing ideas and knowledge to complete the Master study in UTM. Without them, this research would not have been possible.

ABSTRACT

Data transmission using HF spectrum (3-30 MHz) is widely used due to its ability in providing long distance communications at low cost. Due to multipath fading problems in HF channels, the maximum symbol rate of data transmission is limited to 100 baud per second. Differential multiple phase modulation techniques can be used to increase the transmission rate without changing the baud rate. Advanced digital modulation techniques based on PSK is used due to its reliability in providing lower error rate compared to other modulation techniques, such as modulation based on FSK. Unlike coherent detection, phase synchronization is not critical for the differential detection, and implementation can be made simpler in differential multiple phase modulations. For this study, the BER and PER performance of DPSK, DQPSK, D8PSK and D16PSK modulation techniques are presented. The performance evaluation for each modulation are investigated in additive white Gaussian noise environment and random phase delay is included that is based on uniform distribution. In general, the BER and PER performance for differential multiple phase detection decrease for every doubling of phases, but the main advantage is the reliability in data transmission in achieving higher transmission rate.

vi

ABSTRAK

Penghantaran data melalui spektrum HF (3-30 MHz) digunakan secara meluas kerana keberkesanannya di dalam sistem komunikasi jarak jauh dengan kos yang rendah. Walaubagaimanapun, saluran HF terdedah kepada masalah pemudaran multipath, yang menghadkan kadar maksimum penghantaran data kepada 100 baud per saat. Bagi mengatasi masalah ini, pemodulatan pembezaan berbilang fasa boleh digunakan, di mana kadar penghantaran data ditingkatkan tanpa mengubah kadar simbol data. Pemodulatan digital berdasarkan fasa, PSK dipilih kerana kebolehannya memberikan BER yang lebih rendah berbanding kaedah lain, seperti pemodulatan frekuensi, FSK. Tidak seperti pengesanan secara koheren, pengesanan secara perbezaan tidak dipengaruhi oleh lengah fasa, dan perlaksanaannya menjadi lebih ringkas. Di dalam kajian ini, prestasi BER dan PER bagi DPSK, DQPSK, D8PSK dan D16PSK di analisis di dalam persekitaran hingar putih Gaussian, manakala lengah fasa secara rawak dikenakan pada isyarat mengikut taburan normal. Secara umumnya, nilai BER dan PER merosot bagi setiap peningkatan gandaan fasa dalam pemodulatan, tetapi kelebihannya adalah keberkesanannya meningkatkan kadar penghantaran data.

vii

CONTENTS

CHAPTER

ITEM

PAGE

TITLE PAGE TESTIMONY DEDICATION ACKNOWLEDGEMENT ABSTRACT (ENGLISH) ABSTRACT (MALAY) CONTENTS LIST OF TABLES LIST OF FIGURES LIST OF TERMS LIST OF APPENDIX

i ii iii iv v vi vii xi xii xiv xv

CHAPTER 1

INTRODUCTION

1.1 1.2 1.3 1.4 1.5 1.6 1.7

Introduction Purposes of Study Scope of Work Definition of Terms Problem Statements Research Methodology Organization of Thesis

1 2 3 3 4 4 5

viii

CHAPTER II

REVIEW OF LITERATURE

2.1 2.2 2.3 2.4 2.5 2.6

Introduction HF Propagation Characteristics Effects of Multipath Fading HF Digital Protocols Error Detection, Correction and Control Recent Developments in HF Communications

6 6 8 9 10 11

CHAPTER III

THEORY IN HF DIGITAL COMMUNICATION

3.1 3.2 3.3 3.4 3.5

Introduction Probability of Error Match Filter Poisson Distribution Function Coherent Detection 3.5.1 Phase Shift Keying Coherent Detection

14 15 18 21 22 22

3.6

Phase Synchronization Error in Coherent Detection

24

3.7 3.8

Differential Phase Shift Keying Robustness To Phase Synchronization Error in DPSK Detection

25 29

ix

CHAPTER IV

DESIGN OF DIFFERENTIAL MULTIPLE PSK MODULATION

4.1 4.2 4.3 4.4

Introduction Differential Multiple Phase Shift Keying Union Bound On Probability of Error Differential Quadrature Phase Shift Keying 4.4.1 DQPSK BER Performance

30 31 33 35 41 43 46 49 57 58

4.5

Differential 8 Phase Shift Keying 4.5.1 D8PSK BER Performance

4.6

Differential 16 Phase Shift Keying 4.6.1 D16PSK BER Performance

4.7

Robustness to Phase Synchronization Error in Differential Multiple PSK Detection 4.7.1 4.7.2 Case for DQPSK Detection Case for Differential Multiple Phase Detection

59 60

4.8

Signal Representation in Time and Frequency Domain

63

CHAPTER V

RESULTS AND DISCUSSIONS

5.1 5.2 5.3

Introduction Analysis of BER Performance Analysis of PER Performance

64 66 69

CHAPTER VI

CONCLUSIONS 6.1 Conclusions 73

6.2

Suggestions

74

REFERENCES

76

APPENDICES

80

xi

LIST OF TABLES

TABLES

TITLE

PAGE

3.1 3.2 4.1 4.2

Encoded sequence for DPSK transmitter DPSK detected sequence Combination for DQPSK Transmitted Signal Encoded sequence for Inphase channel of DQPSK transmitter

27 27 36 37

4.3

Encoded sequence for Quadrature channel of DQPSK transmitter

37

4.4 4.5 4.6 4.7 5.1 5.2

DQPSK detected sequence for Inphase channel DQPSK detected sequence for Quadrature channel Lookup table for D8PSK detection Lookup table for D16PSK detection Theoretical BER Performance for a fixed BER of 10 BER performance of simulation result for a fixed BER of 10-4 Theoretical PER Performance for a fixed PER of 10-2 Simulation PER Performance for a fixed PER of 10-2
-4

37 37 46 56 67 68

5.3 5.4

70 71

xii

LIST OF FIGURES

FIGURE

TITLE

PAGE

2.1 3.1 3.2 3.3 3.4 3.5 4.1 4.2 4.3 4.4

Types of HF propagation Gaussian distribution for symbol x0 and x1 Block diagram of a system Block diagram for PSK coherent detection Block diagram of DPSK transmitter Block diagram of DPSK receiver Constellations diagram for differential multiple PSK Basic differential multiple PSK modulator Basic differential multiple PSK demodulator Union bound signal space diagram for differential M-ary PSK

7 16 19 23 26 26 32 32 33 34

4.5 4.6 4.7 4.8 4.9

DQPSK constellations diagram DQPSK receiver structure Union bound for DQPSK BER performance D8PSK constellations diagram Demodulation structure for D8PSK detection

36 38 40 42 43

xiii

4.10 4.11 4.12 4.13 5.1 5.2 5.3 5.4

D8PSK union bound signal space plane D16PSK constellations diagram D16PSK receiver structure Signals representation in time and frequency domain Theoretical BER performance Simulation result of BER performance Theoretical PER performance Simulation result of PER performance

47 49 50 64 66 68 69 71

xiv

LIST OF TERMS

BER DPSK DQPSK D8PSK D16PSK FSK HF PER PSK SNR

Bit Error Rate Differential Phase Shift Keying Differential Quadrature Phase Shift Keying Differential 8-Phase Shift Keying Differential 16-Phase Shift Keying Frequency Shift Keying High Frequency Packet Error Rate Phase Shift Keying Signal to Noise Ratio

xv

LIST OF APPENDICES

APPENDIX

TITLE

PAGE

BER Performance

81

PER Performance

82

Detection based on FSK

83

CHAPTER I

INTRODUCTION

1.1 Introduction

Ionospheric propagation is responsible for the ability to do broadcasting and communications. The long distance transmission is carried out on the HF spectrum (330 MHz) using skywave propagation, while for the short distance transmission, the groundwave propagation will be used [Goodman, 1992]. Nowadays, the HF communication system is widely used, not only for the tactical and strategic military purposes, but also by the commercial world, amateur radios, maritime and aeronautical operators.

The advantages of this type of communication arise from its relative simplicity, its ability to provide communication over thousand of miles and its moderate cost per circuit mile. HF communication involves minimum infrastructure and inexpensive maintenance compared to other technology such as satellite communication [AbdullahHusni et al, 2003].

2 Due to variability of ionosphere, the HF signal is subjected to multipath fading phenomenon, which limits the data transmission rate to 100 baud per second [Goodman, 1992]. In order to overcome this problem, the advanced modulation techniques can be used to ensure the reliability in data transmission. Thus, the focus of this study is to design a HF communication system that can improve the reliability in data transmission using differential multiple phase modulation techniques.

1.2 Purposes Of The Study

The purpose of this study was to design and simulate a HF communication system that can increase data transfer rate that is limited by using HF channel using advanced modulation techniques specifically in differential multiple phase modulations. The performances of the techniques are analyzed in term of the bit error rate and packet error rate of the modulation. Differential detection is used to overcome phase synchronization error in coherent detection.

1.3 Scope Of Work

This study was focused on differential multiple phase digital modulation, which is important to design a system that can increase data transfer rate that is limited by using the HF transmission channel. The modulation techniques used are DPSK, DQPSK, D8PSK and D16PSK.

3 System was designed to process within the voice band frequency and not on radio band frequency. Sampling frequency used is 8000 Hz and the carrier frequency is 1000 Hz. The bandwidth of the signal is 4000 Hz.

Data format used is PACTOR, containing 8 characters or 64 bits of data and 16 bits for error control in a packet for 100 baud data transmission rate. The system was designed to test in a present of additive white Gaussian noise and random phase delay in received signals.

1.4 Definitions of Terms

For the purpose of this study, the following operational definitions are used:

BER

Bit error rate number of error present within the period of data transmission

DPSK DQPSK D8PSK D16PSK FSK HF PER PSK SNR

Differential phase shift keying Differential Quadrature phase shift keying Differential 8-phase shift keying Differential 16-phase shift keying Frequency shift keying High frequency band channel Packet error rate number of packet with at least an error presents Phase shift keying Ratio of signal power to noise power

4 1.5 Problem Statements

In HF communication system, the variability of ionosphere results multipath fading phenomenon. This phenomenon gives several affects in the communication, which are frequency selective fading and time selective fading [Goodman, 1992].

Frequency selective fading problems will cause for inter symbol interference (ISI). Due to this problem, the maximum data transmission rate is limited to 100 baud per second [Goodman, 1992][Willink et al, 1996]. By limiting the data transmission rate to 100 baud per second, inter symbol interference (ISI) problem can be avoided. As a solution, to increase the data transmission rate without changing or increasing the baud rate, the differential multiple phase modulation can be used.

1.6 Research Methodology

There are several approaches taken in order to achieve the objective of this study, which are:

1. Literature of review on HF communication system for understanding the concept and problem that occur in this particular type of communication. 2. Understanding the basic theory on digital signal processing and digital communication system to find ways on solving research problems. 3. Designing differential multiple PSK system which are DPSK, DQPSK, D8PSK and D16PSK. 4. Programming in MATLAB for performance analysis purposes.

5 5. Data analysis and simulation of the detection using MATLAB to analyze the performance of modulation techniques in term of BER and PER. 6. Calculation and performance comparison between theory and simulation. 7. Thesis and report writing.

1.7 Organization Of Thesis

This thesis is divided into six chapters. The first chapter contains an overview of this project. Some explanations about the literature and recent development in HF were covered in chapter 2. Chapter 3 describes the theory in HF digital communication. The design of differential multiple phase modulations were described in chapter 4. Chapter 5 presents the analysis of results. This thesis ends with the conclusion and suggestions for further research.

CHAPTER II

REVIEW OF LITERATURE

2.1 Introduction

This chapter describes the HF propagation characteristics, effects of multipath fading, HF digital protocols, error detection, correction and control, and recent developments in HF digital modulation techniques.

2.2 HF Propagation Characteristics

HF propagation is divided into two basic modes, which are [Goodman, 1992]:

a) Ground wave propagation b) Sky wave propagation

Figure 2.1

Types of HF propagation

Ground wave propagation is the dominant mode of propagation for short distance communication. Here the electromagnetic tends to follow the contour of the earth. The ground wave travels in direct contact with the earths surface, and it suffers a severe frequency-dependent attenuation because of absorption by the ground. Sky wave propagation is the dominant mode of propagation in the 3-30 MHz frequency range. Here long distance communication is obtained by reflecting the wave at the ionosphere regions.

There may be four regions in ionosphere present called D, E, F1 and F2 regions [McNamara, 1991]. Among these regions, only E, F1, sporadic E, and F2 refract HF waves. Sky wave propagation is caused primarily by reflection from F layer [Couch, 1997]. Because of this layer, international broadcast stations in HF band can be heard from the other side of the world at almost any time during day and night.

8 2.3 Effects of Multipath Fading

Multipath fading may occur when the transmitted signals are diffracted by different layers of ionosphere [Goodman, 1992]. When the path length of transmission differs, the carrier frequency of the signal will also change. There might be loss data on the detection part because of these changes. Multipath fading will lead to frequency selective fading and time selective fading.

Frequency selective fading occurs when the received signals has a time delay in millisecond unit due to diffraction in transmitted signals. In time domain, the delay will introduce time delay spread problem that will cause for inter symbol interference (ISI) and limit the data transmission rate to 100 baud per second [Willink et al, 1996]. In frequency domain, this problem makes system attenuate certain frequency, which can affect the sub-carrier frequency of the signal, and the signal might be loss.

Time selective fading or flat fading is caused by delay due to the diffraction signals arrived at the receiver in different phases [McNamara, 1991]. This delay in phase will result in cancellation of waves or attenuation in wave amplitude. The worst case happens if the same signal from different paths arrive at the receiver at 1800 phase

different which resulting the wave practically cancels each other.

The Doppler shift is caused by the motion of the electrons in the ionosphere layer which introduce changes in radio frequency [Goodman, 1992]. The frequency shift will cause error when the receiver is unable to recognize the signal, as the frequency that has been shifted is too large. If the receiver is using a band-pass filter to capture the signal, the frequency shift will reject some of the signal.

9
2.4 HF Digital Protocols

Several types of data format exist in HF communication to produce a system which performs robustly under wide range of typical HF channel conditions. Some of the most frequently used data formats are listed as follows:

a) RTTY b) AMTOR c) PACTOR d) CLOVER e) G-TOR

Radio Teletypewriter (RTTY) has 5 bits code became available in the market in the years following World War 2 [Kasser, 1991]. Radio amateurs experimented with using that equipment for communications. There is no error detection technique in RTTY. RTTY is a half duplex communication mode. RTTY is also a character transmission mode that each character is transmitted as soon as it is typed. AMTOR is a half duplex communication mode and a specialized form of RTTY [Kasser, 1991]. It has 5 bits in a packet and cannot transfer ASCII. AMTOR provides good performance in bit error rate particularly error correction in real time, but does not effectively compete with the speed and error correction of more modern modes [Henry, 1992].

CLOVER is a PSK mode, which uses a full duplex simulation. It is well suited for operation under good conditions. Clover uses Reed-Solomon error correction codes to correct a moderate number of errors but it can also switch to ARQ whenever the conditions are very bad and number of error exceeds the capacity of Reed-Solomon error corrector.

10 PACTOR is an improvement over AMTOR and packet radio. PACTOR is cheap and reliable. In this context, this means fast, robust and error free data transfer over HF. PACTOR employs structured packet which contains 8 characters of data for 100 baud per second or 20 characters for 200 baud per second of transmission rate. PACTOR uses Automatic Repeat Request (ARQ) method with Cyclic Redundancy Checking (CRC-16) which consists of 16-bit frame check sequence (FCS) for error correction method [Stalings, 2000]. PACTOR-1 is a powerful FSK mode especially in a very poor propagation condition. PACTOR-11 is a robust and powerful PSK mode which operates effectively under both very good and poor conditions, and is as much as 8 times faster then PACTOR-1.

Golay-Transmission over Radio (G-TOR) is a FSK mode that offers a high transfer rate under good conditions. It combines the error correcting properties of Automatic Link Establishment (ALE), including Forward Error Correction (FEC) coding and the ARQ cycle of packet and a new application of the inevitability of the Golay code, to produce a faster new mode. Military is a major user in HF communication system.

2.5 Error Detection, Correction and Control

In digital communication, signals that are affected with noise, tends to produce error. Error occurs when signal detected at the receiver are not the same as the signal sent from transmitter [Haykin, 1988]. Error control refers to mechanism to detect and correct errors that occur in packets of data.

11 The most prevalent methods for error detection include the application of redundancy, exact count encoding schemes, parity checking in various forms, and cyclic redundancy checking (CRC) [Goodman, 1992]. CRC is the most reliable method of error detection. This method, although not simple to implement, will detect about 99.95% of all transmission errors. The most generally accepted code of this type is the CRC-16, which employs 16 bits in the frame check sequence [Tomasi, 1987] [Stalings, 2000].

Error correcting methods include Automatic Repeat Request (ARQ) and Forward Error Correction (FEC) schemes. ARQ is generally viewed to be the most reliable method for insuring the integrity of transmitted messages under most circumstances, while FEC detects and corrects errors without the necessity to call for retransmission [Goodman, 1992].

2.6 Recent Developments in HF Communications

[Raos et al, 2003] has presented the performance of a modem designed for HF communications. The aim was to attain the real time communication for voice transmission. In this paper, frequency selective HF channel was converted into a set of frequency flat channels with multi-carrier modulation orthogonal frequency division multiplexing (OFDM). Spread spectrum technique was applied to improve modems performance. The performance of the modem with 16 sub-carriers and QPSK modulation was analyzed via simulation in Rayleigh fading channel in terms of BER. The data rates are approximately 2400 and 3600 bps and digital voice can be transmitted interactively.

12 [Jaswar-Shaameri, 2003] has explored the FPGA (field programmable gate array) implementation for CPFSK (continuous phase FSK) modulation techniques for HF communication. The paper describes the implementation of an FSK modem based on CPFSK modulation techniques on the FLEX10K board EPF10K70RC240. Both transmitter and receiver modules are adopted with the objective to minimized the system size. Two types of modems were developed with different detection scheme: conventional noncoherent and square wave detection. The square wave detection has 60% less components because of the parallel structure and no multiplier is used in the algorithm. Thus, the system size was minimized.

[Charles-Tri, 2003] has discussed the performance of a Direct Sequence DPSK (DS-DPSK) spread spectrum system over a Rician frequency non-selective, slowly fading channel in the presence of pulsed noise interference and additive white Gaussian noise. They indicate that the receiver is effective in mitigating the effects of pulse noise jamming for all fading conditions that they considered.

The study on error probability for coherent PSK and noncoherent DPSK over Rician fading channels was done by [Yao-Teng et al, 2002]. They have discussed the performance of communication systems using both coherent PSK and noncoherent DPSK modulation, in correlated with Rician fading channels with diversity reception. The method with both coherent and non-coherent detections resulted better performance in Rician fading channels compared to Rayleigh fading channels.

[John-Mohamed et al, 2002] discussed the data recovery in Differential Encoded QPSK (DEQPSK). The research presented a new algorithm for recovering step phase changes for a DEQPSK modulated signal by computing the phase angle from the arctangent of the ratio Quardrature/Inphase, and comparing it with previous phase.

13 STANAG 5066 was developed in the course of modernizing the communication services within NATO. A subnetwork profile for HF radio data communications (STANAG 5066) was introduced through NATO to provide improved HF communications for broadcasting [NATO][Reynolds-Gillespie, 1997]. [Trinder-Brown, 1999] was proposed the STANAG incorporates a data rate change (DRC) mechanism to change the data rate of the link between the range of 75 to 2400 bps and above. This was to optimize the modem data rate to changes in HF channel conditions, to maintain the maximum throughput. The DRC algorithm was described as a requirement for an effective data rate changes.

[Nilsson-Giles, 1997] explored the potential of multi-carrier or orthogonal frequency division multiplexing (OFDM) for military HF communication. The proposed system uses a differential binary PSK (DBPSK) modulation method on each carrier. Number of carriers used in this system is 1024 over a bandwidth of 125 kHz and symbol duration is 8.2ms. The modem proposed by Nilsson is capable of operating until 1.22Mbps.

CHAPTER III

DIGITAL COMMUNICATION THEORY

3.1 Introduction

This chapter describes the theoretical foundation in HF digital communication. The probability of error, match filter, Poisson distribution function for packet error rate, PSK coherent detection and differential PSK detection, phase synchronization error in coherent detection and robustness to phase synchronization error in differential PSK detection were discussed in detail in this chapter.

15 3.2 Probability of Error

In digital communication system design, the main objective is to receive data as similar as the data sent from the transmitter. It is important to analyze the system in term of probability of error to view the systems performance. Each modulation technique has different performance while dealing with signals, which normally are affected with noise. General explanation for probability of error is explained in this section.

General equation for the output signal that is affected with the additive white Gaussian noise can be shown in Equation (3.1) [Shanmugan, 1988][Proakis, 1995].

If x(t ) is the true signal, then y (t ) is the signal corrupted by noise, n(t )

y (t ) = x(t ) + n(t )

(3.1)

The performance of each modulation is measured by calculating its probability of error with assumption that systems are operating with this additive white Gaussian noise [Shanmugan, 1988][Proakis, 1995].

The probability density function (pdf) for noise can be represented as a Gaussian distribution [Proakis, 1995]
n0 2 1 exp 2N 0 2N0

PN (n0 ) =

(3.2)

16 where n0 = n(t 0 ) is noise at time t 0 and N 0 = E[ x(t )]2 is noise power at time t 0 .

Consider x(t ) to be two possibilities, x0 (t ) and x1 (t ) , at time t = t0 , x0 = x0 (t 0 ) and x1 = x1 (t 0 ) . Conditional probability for symbol x0 and x1 can be represented as

Py | x0 ( y ) = Py| x1 ( y ) =

( y x0 ) 2 1 exp 2N 2N 0 0 ( y x1 ) 2 exp 2N 2N 0 0 1

(3.3)

(3.4)

The probability of occurrence for both symbols, also known as priori probability is shown as follows

P ( x0 ) = P ( x1 ) =

1 2

(3.5)

Probability of error Reference point x + x1 T= 0 2 x0 Figure 3.1

x1

Gaussian distribution for symbol x0 and x1

17 Figure 3.1 shows the Gaussian distribution for probability of getting error for x0 and x1. Probability of error is area under the graph can be calculated as

Pe =

P
T

y | x0

( y ) P( x0 )dy + Py| x1 ( y ) P( x1 )dy

T

(3.6)

= Py| x0 ( y ) P( x0 )dy + Py| x1 ( y ) P ( x1 )dy

( y x0 ) 2 1 1 exp 2N 2N 2T 0 0

dy +

T ( y x1 ) 2 1 1 exp 2N 2 2 N 0 0

dy

Since the chances of occurrence for both symbol is the same, Equation (3.6) can be expressed as
( y x0 ) 2 1 1 Pe = 2. exp dy 2N 2 T 2N 0 0

(3.7)

If x0 is shifted to 0, the probability of error can be expressed as

Pe =
T

y2 exp 2N 2N 0 0 1

dy

(3.8)

where T =

x1 x0 y2 into Equation (3.8) and probability of error . Substitute z 2 = 2 N0

can also be shown as

Pe = Q = 2

z2 exp 2 dz 2 1

(3.9)

18

2 = x1 x 0 2 N0

where

. Q( ) is referred as the Q function and is given in table form. To

maximize the probability of correct detection, value of 2 has to be increased

2 [ x1 x0 ] =

N0

(3.10)

Factor of time is inserted since x(t ) changes with time

2 [ x1 (t ) x0 (t )] =

N0

(3.11)

The probability of correct detection is obtained from Q(/2). To maximize probability of correct detection, the match filter structure is adopted.

3.3 Match Filter

Match filter is a method to detect signal by maximizing signal power in the presence of noise [Proakis, 1995][Rodger, 2000]. The idea behind a match filter is correlation using convolution. The output of a match filter does not necessarily look like the signal being detected, but the amplitude of each point in the output signal is a measure of how well the filter kernel matches the corresponding section of the input signal. Theory for match filter is shown as follows [Rodger, 2000][Haykin, 1988].

19

x(t) + n(t)

y( t ) h(t)

Figure 3.2 Block diagram of a system

Figure 3.2 shows the relationship between input and output of a system. The relationship between the output y (t ) and input x(t ) in the presence of noise n(t) that is a convolution process is y (t ) = h(t ) * [ x(t ) + n(t )]

(3.12)

where x(t ) is true signal and deterministic, n(t ) is additive white Gaussian noise, h(t ) is system impulse response and y (t ) is output. If noise is not considered, the relationship between input and output as follows y (t ) = h(t ) * x(t )

(3.13)

The objective is to find h(t) that maximizes the probability of correct detection of x(t). Referring to Equation (3.11), this is done by maximizing the quantity of 2 where y (t0 ) is value of y(t) at time t0 and E [n(t0 )] is the noise power at time t0 .
2

=
2

E [n(t0 )]

y (t0 )

2 2

(3.14)

20
2 Applying Schwartz Inequality function, MAX can be expressed as follows

2 MAX

H ( f ) df

X(f )
2

df

S nn ( f )

H( f )

df

1 2 X ( f ) df N0

(3.15)

where N 0 = S nn ( f ) and N 0 represents noise power. Based on [Couch L.W. 1997] the
2 is obtained when h(t) is chosen such that equality is attained. maximum value of MAX

This occurs when system impulse response or filter kernel h(t) h(t ) = x(t0 t )

(3.16)

where filter kernel h(t) is in a non-causal form. Output y (t ) is a convolution of system impulse response and input signal. y (t ) = h(t ) * x(t )

(3.17)

x( )h(t )d

The end result is correlation of x(t ) . Thus, the matched filter can be realized using a correlater.

y (t ) = x(t ).x(t )dt

0

(3.18)

21
3.4 Poisson Distribution Function

Poisson distribution is the discrete probability density function used in communication to determine the packet error rate within the transmitted data especially for large packet size and low bit error rate. Two important parameters used to evaluate the Poisson distribution are bit error rate and the packet size used in transmitted data.

Discrete probability density function can be expressed as

x ( x) = e b

bk (x k ) k = 0 k!

(3.19)

The distribution of error within a packet of binary data assuming that b<1.
b = T

(3.20)

is the bit error rate of the respective data and T is the packet size of the transmitted data. Packet error rate is Pe = 1 x (0)

(3.21)

22
3.5 Coherent Detection

Coherent detection involves the use of matched filter to detect the transmitted signals [Haykin, 1988]. The detection method will be explained here and the bit error rate (BER) will be derived. Coherent detection is optimum in terms of the BER performance but requires the exact form of the reference signal at the receiver. The BER performance increases if exist any phase synchronization error between the received and reference signal at the receiver. This problem is resolved at increase complexity by introducing a carrier recovery circuit [Shaameri-Jaswar, 2003].

3.5.1

Phase Shift Keying Coherent Detection

Phase shift keying, or PSK for coherent detection is a basic method for transmitting and receiving digital signals, which the phase of transmitted signal is varied to convey information [Roden, 1988][Rodger, 2000]. There are several schemes that can be used to accomplish PSK. The simplest method uses two signal phases: 0 degrees and 180 degrees. The state of each bit is determined according to the state of the preceding bit. If the phase of the wave does not change, then the signal state stays the same (low or high). If the phase of the wave phase reverses, then the signal state changes.

Figure 3.3 is derived from Equation (3.22) [Proakis, 1995]

y (t ) = x(t )[x1 (t ) x0 (t )]dt

0

Tb

(3.22)

23 where x1 (t ) and x0 (t ) are represented as

x1 (t ) = A cos 2f1t

0 t Tb

"1"

x0 (t ) = A cos 2f1t

0 t Tb

"0"

x(t )

x1 (t ) = A cos 2f 1t

Tb

x0 (t ) = A cos 2f 1t Figure 3.3

Tb

Block diagram for PSK coherent detection

From Equation (3.11), to maximize the probability of correct detection

2 MAX =

1 N0

Tb

(x (t ) x
1 0

(t ) ) dt
2

(3.23)

Value of x1 (t ) and x0 (t ) are substituted into Equation (3.23) and result of the calculation is

2 MAX =

2 A2 dt N0 0

Tb

2 A2 cos 2 2 f1t dt N0 0

Tb

(3.24)

24 From Equation (3.24), the following equation can be derived

2 MAX =

2 A 2Tb N0

(3.25)

BER for PSK can be shown as follows BER = Q MAX 2 A2Tb = Q 2N0 (3.26)

where Q denotes for Q function. Signal-to-noise ratio can be calculated as follows

A2Tb SNR (dB ) = 10 log 2N 0

(3.27)

3.6 Phase Synchronization Error in Coherent Detection

Synchronization is critical to ensure that the BER performance is optimal for coherent detection [Shaameri-Jaswar, 2003]. Given that a received signal within bit duration is
y (t ) = A cos(2f1t + )

0 t Tb

(3.28)

25 where is the phase present in the signal. If the reference signal is

x1 (t ) = A cos(2f1t )

0 t Tb

(3.29)

The output signal is

z (t ) = y (t ) x1 (t )dt
0

Tb

= A cos(2f 1t + ) y (t ) A cos(2f 1t )dt

0

Tb

= =

A2 2

Tb

A cos(4f t + ) + cos( )dt

1 0

A2Tb cos( ) 2

(3.30)

The result shows how critical the phase shift affects the output signal. Thus, it is desired to have the detection scheme a compromise between robustness to phase synchronization error and BER performance.

3.7 Differential Phase Shift Keying

Differential phase shift keying (DPSK) is a type of phase modulation noncoherent detection, where it uses the received phase of the previous signal as a reference signal. DPSK is developed as a solution to the phase synchronization error in coherent PSK detection [Goodman, 1992] [Roden, 1988]. The problem phase synchronization can be resolved by introducing differential encoding in the modulation

26 signal. The structure for the DPSK transmitter and receiver are shown in Figure 3.4 and Figure 3.5.

Figure 3.4

Block diagram of DPSK transmitter

Figure 3.5

Block diagram of DPSK receiver

27 The DPSK is developed to simplify the PSK by not requiring carrier recovery circuit. The input sequence is encoded as
ak = ak 1 sk a k 1 s k

(3.31)

Table 3.1 shows the example of encoded sequence for DPSK transmitted signals. The process for detecting the transmitted sequence is shown in Table 3.2.

Table 3.1:

Encoded sequence for DPSK transmitter

k sk ak x(t)

1 1

2 1 1
cos 2f1t

3 0 0 - cos 2f1t

4 1 0 - cos 2f1t

5 0 1
cos 2f1t

1
cos 2f1t

1
cos 2f1t

Table 3.2:

DPSK detected sequence

k
x(t)

0
cos 2f1t

1
cos 2f1t

2
cos 2f1t

3 - cos 2f1t 0

4 - cos 2f1t + 1

5
cos 2f1t

0 + 1

0 + 1

0 0

Phase diff sk

It is assumed that two consecutive binary bit 1 are transmitted and A=1. The phase shift is zero. The input of the low pass filter is
q (t ) = A cos(2f1t ) * A cos(2f1t )

1 1 + cos(2f 1 2t ) 2 2

(3.32)

28 Since the q(t) is filtered, only the dc terms remains and output of the low-pass filter is r (t ) = 1 2 (3.33)

The BER for differential PSK is

A2 1 exp 2N p 2

Pe =

(3.34)

The filtered noise power is

Np =

2No Tb

(3.35)

where Tb is the bit-duration, and N0 is the power of the additive white Gaussian noise. The BER for differential PSK is
A2Tb 1 Pe = exp 4N 2 o

(3.36)

29
3.8 Robustness To Phase Synchronization Error in DPSK Detection

Given that 1 is the phase delay present in the received signal. It is assume that A=1. The input of the low pass filter is
q (t ) = cos(2f1t + 1 ) * cos(2f1t + 1 )

= =

1 1 cos(1 1 ) + cos(2f1 2t + (1 + 1 ) ) 2 2 1 1 + cos(2f1 2t + 21 ) 2 2 (3.37)

The high frequency component in q(t)is filtered by the filter. Then the output signal is only the dc term of the signal.

r (t ) =

1 2

(3.95)

The result shows that the phase is not present in DPSK detection. Thus, the DPSK detection is robust to phase synchronization error.

CHAPTER IV

DESIGN OF DIFFERENTIAL MULTIPLE PSK MODULATION

4.1 Introduction

This chapter describes the design of differential multiple PSK modulation techniques proposed in this study, which are DQPSK, D8PSK and D16PSK modulation techniques. The theoretical BER performance, robustness to phase synchronization error in differential detection and signal representation of the modulated signals were discussed in detail in this chapter.

31 4.2 Differential Multiple Phase Shift Keying

Multiple differential PSK modulations are very important to increase the transmission rate without changing the baud rate. In DQPSK detection, two bits of data represents one symbol of information. D8PSK is capable to transmit 3-bit per symbol, and if D16PSK modulation is used, the transmission rate is four times greater than DPSK modulation. The modulation technique represents 4-bit per symbol for 100 bauds data transmission rate. Similar to DPSK modulation technique, the problem phase synchronization can be resolved by introducing differential symbol coding in differential multiple PSK modulation. Detection can be made simpler since no carrier recovery detection is required.

For the general case of differential M-ary PSK, the various signals are given by Equation (4.1) [Roden, 1988][Rodger, 2000] si (t ) = A cos 2f1t + i

(4.1)

where the index, i, takes on values from 0 to M-1. The angles are given by

i =

(2i + 1)
M (4.2)

M is the number of phases used in the modulation.

32 The signal space diagram in differential multiple PSK is shown in Figure 3.9.

Figure 4.1

Constellations diagram for differential multiple PSK

The waveform can be generated from the binary signal by combining groups of bits and then performing a digital-to-analog conversion. Thus, for example, if M=8, the combination of bits are in triplets. A serial to parallel converter converts each three input bits to a 3-bit binary number, which forms the input to the D/A converter [Roden, 1988]. This is shown in Figure 4.2.

d(t) Serial to parallel 3-bit D/A Phase Modulator

s(t)

Figure 4.2

Basic differential multiple PSK modulator

The corresponding receiver is shown in Figure 4.3. The phase demodulators are then processed by an analog-to-digital converter to reconstruct the data sequence. The parallel bit streams are converted into a series bit stream for detection of the data sequence [Roden, 1988].

33

s(t) Phase demodulator 3-bit A/D Parallel to serial

d(t)

Figure 4.3

Basic differential multiple PSK demodulator

Code search for multi level differential PSK modulation require excessive search time, which a wide range of coefficient need to be check to demodulate signal accurately.

4.3 Union Bound On Probability of Error

When the geometry of the signal set is difficult to analyze, union bound is used to find the probability of error. The additive noise will cause an error if the phase angle of the received signal varies from the transmitted phase angle by more than /M in either direction, or in other words, error will occur if the symbol, sj varies more than the distance, dj between adjacent phase states [Roden, 1988][Proakis, 1995].

34 Figure 4.4 views the union bound signal space plane for differential M-ary PSK.

Figure 4.4

Union bound signal space diagram for differential M-ary PSK

The error distance, dj is expressed by

d j = Q MAX sin M 2

(4.3)

where M is the number of phase states. For case in differential detection, dj can be written as
1 A 2Tb sin d j = exp 4N 2 M o

(4.4)

35 For M=4, the error Pe,11 is the union for the probability of error events e10 and e01. Pe,11 can be written as

Pe ,11 = P[e10 + e01 ] = P[e10 ] + P[e01 ] P[e10 , e01 ] P[e10 ] + P[e01 ]

(4.5)

For M-ary, the union bound is ( M 1) Q(d j / 2 N 0 ) M

Pe , j P[eij ]
i =1

i j,

Pe , j

i j

(4.6)

where dj is the distance between symbol sj and its adjacent symbols. The BER performance for differential multiple PSK will be discussed later in the next section for DQPSK, D8PSK and D16PSK respectively.

4.4 Differential Quadrature Phase Shift Keying

Differential quadrature phase shift keying (DQPSK) is a phase modulation, which it uses four different phase angles to transmit a symbol of binary data. In DQPSK, the four angles are out of phase by 90 degrees. The DQPSK system uses both sine and cosine at the carrier frequency to transmit two separate message signals, referred as the Quadrature and Inphase signals [Kolimbiris, 2000][Rodger, 2000]. Both signals are allowing the system to transmit 2 bits per symbol at one time. To increase spectrum efficiency, two bits of information represent a symbol for DQPSK. Within a symbol duration, Inphase order bit transmit at cos(2f1t) and Quadrature order bit as sin(2f1t). DQPSK uses two bits differential encoding as a start bit to encode the data at the modulation part.

36

01

11

00

10

Figure 4.5

DQPSK constellations diagram

The constellation diagram in Figure 4.5 shows the different four angles used in DQPSK modulation technique. The transmitted signal has one of four possible forms as shown in Table 4.1.

Table 4.1:

Combination for DQPSK Transmitted Signal x(t) Inphase (I) Quadrature (Q)

Symbol (s0,s1)

11 10 01 00

+Acos2f1t + Asin2f1t +Acos2f1t - Asin2f1t -Acos2f1t + Asin2f1t -Acos2f1t - Asin2f1t

The combination of DQPSK transmitted signal shown in Table 4.1 can be expressed as [Kolimbiris, 2000] x(t) = +(Acos2f1t + Asin2f1t)

(4.7)

37
Table 4.2: Encoded sequence for Inphase channel of DQPSK transmitter

k sk ak xI(t)

1 1

2 1 1
cos 2f1t

3 0 0 - cos 2f1t

4 1 0 - cos 2f1t

5 0 1
cos 2f1t

1
cos 2f1t

1
cos 2f1t

Table 4.3:

Encoded sequence for Quadrature channel of DQPSK transmitter

k+1 sk+1 ak+1 xQ(t)

1 1

2 1 1
sin 2f 1t

3 0 0 - sin 2f 1t

4 1 0 - sin 2f 1t

5 0 1
sin 2f 1t

1
sin 2f 1t

1
sin 2f 1t

Table 4.4:

DQPSK detected sequence for Inphase channel

k
xI(t)

0
cos 2f1t

1
cos 2f1t

2
cos 2f1t

3 - cos 2f1t 0

4 - cos 2f1t + 1

5
cos 2f1t

0 + 1

0 + 1

0 0

Phase diff sk

Table 4.5:

DQPSK detected sequence for Quadrature channel

k+1
xQ(t)

0
sin 2f 1t

1
sin 2f 1t

2
sin 2f 1t

3 - sin 2f 1t 0

4 - sin 2f 1t + 1

5
sin 2f 1t

0 + 1

0 + 1

0 0

Phase diff sk+1

38 Table 4.2 - 4.3 show the example of encoded sequence for Inphase and Quadrature channel for DQPSK transmitted signals. The process for detecting the transmitted sequence is shown in Table 4.4 - 4.5, respectively. The block diagram for the DQPSK receiver is shown in Figure 4.6.

Figure 4.6

DQPSK receiver structure

It is assumed that two consecutive symbols 11 are transmitted and the phase shift is zero. Using differential encoding, the demodulator signal for I channel is
cos(2f 1t ) . By /2 phase shifter, the demodulator signal for Q channel is sin( 2f 1t ) .

Input of the low pass filter is,

Inphase channel:
q I (t ) = (cos(2f 1t ) + sin( 2f 1t ) ) * cos(2f 1t )

= cos 2 (2f 1t ) + cos(2f 1t ) sin( 2f 1t ) = 1 (1 + cos(2f1 2t )) + 1 sin(2f 1 2t ) 2 2

(4.8)

39 The high frequency components are filtered by the low-pass filter, only the dc terms remains and output of the low-pass filter is

rI (t ) =

1 2

(4.9)

Quardrature channel: qQ (t ) = (cos(2f 1t ) + sin(2f 1t ) ) * sin(2f 1t )

= cos(2f1t ) sin( 2f1t ) + sin 2 (2f1t ) = 1 (1 sin (2f1 2t )) + 1 sin(2f1 2t ) 2 2

(4.10)

Since the qQ(t) is filtered, only the dc terms remains and output of the lowpass filter is

rQ (t ) =

1 2

(4.11)

The start bit from differential encoder is removed by correlating with the previous signal at the output of the low-pass filter. The + dc levels from Inphase and Quadrature channels representing the two parallel bit streams, are converted into a series bit stream for detection.

40
4.4.1 DQPSK BER Performance

BER performance for DQPSK detection scheme is evaluated using union bound as discussed in previous section. The signal space plane for DQPSK union bound is shown in Figure 4.7.

Figure 4.7

Union bound for DQPSK BER performance

Given that s=11. Error Pe,11 will occur if s=01 or s=10. The probability is written as
max ) + Q( max ) 2 2 max ) 2

Pe,11 = Q(

= 2Q(

(4.12)

The BER performance can be written as Pe = Pe ,11 P( s = 11) + Pe ,10 P( s = 10) + Pe , 01 P( s = 01) + Pe , 00 P( s = 00)

(4.13)

41 Since Pe,11 = Pe,10 = Pe,01= Pe,00, then the BER performance is Pe = 4[Pe ,11 P( s = 11)]
1 = 4 2Q( max ) 2 4 = 2Q( max ) 2

(4.14)

The BER performance for DQPSK can be expressed as 1 A 2Tb Pe = 2 exp 4N o 2

(4.15)

Therefore,
A 2Tb Pe = exp 4N o

(4.16)

42
4.5 Differential 8 Phase Shift Keying

Differential 8 phase shift keying (D8PSK) uses eight different angles, which every angle will represents one symbol of data in triplet bits. The constellation diagram in Figure 4.8 shows the different eight angles used in D8PSK modulation technique.

011

111

010

110

000

100

001

101

Figure 4.8

D8PSK constellations diagram

From Equation (4.1 4.2), the transmitted signal of D8PSK is: s (t ) = A cos(2f1t i )

i =

(2 + 1)i 8

(4.17)

where index, i, takes on values from 0 to 7. The block diagram for the D8PSK receiver is shown in Figure 4.9. The receiver consists of Inphase (I) channel, Quadrature (Q) channel and Control (C) channel.

43

x(t) - signal BPF fc=f1 fBW=2/Tb Inphase signal LPF fc f1 Tb s(t) - detection LPF Quadrature signal fc f1 Tb

Phase shifter

/2

/4

LPF fc f1 Control signal Tb

Figure 4.9

Demodulation structure for D8PSK detection

It is assumed that consecutive symbols 110 are transmitted and the phase shift is zero. The received signal is x(t ) = A cos 2f 1t

(4.18)

It is assume that A=1. Input of low-pass filter is:

Inphase channel: q I (t ) = cos 2f 1t = =

)* cos(2f t )
1

1 1 cos 2f 1 2t + cos 8 8 2 2

) ( )]
(4.19)

1 cos(2f 1 2t ) cos + sin (2f 1 2t )sin + 0.46 8 8 2

( )

44 The high frequency components are removed by the low-pass filter, output of the lowpass filter is only the dc term of the signal which it equals to
rI (t ) = 0.46

(4.20)

Quadrature channel: qQ (t ) = cos 2f 1t = =

)* sin(2f t )
1

1 1 sin 2f 1 2t sin 8 8 2 2

) ( )]
(4.21)

1 sin (2f1 2t ) cos cos(2f1 2t )sin + 0.19 8 8 2

( )

The high frequency components are removed by the low-pass filter, only the dc terms remains and output of the low-pass filter is rQ (t ) = 0.19

(4.22)

Control channel: Using /4 phase shifter differential encoder, two demodulator signals for control channel are xc (t ) = cos(2f 1t ) 4
1

xc (t ) = sin(2f 1t ) 4
2

(4.23)

45 For c1 channel, input of the low-pass filter is qC (t ) = cos 2f 1t

1

)* cos(2f t 4 )
1

= =

1 1 cos 2f 1 2t 3 + cos 8 8 2 2

( ) ( )]
(4.24)

1 cos(2f 1 2t ) cos + sin (2f 1 2t )sin + 0.46 8 8 2

( )

For c2 channel, input of the low-pass filter is qC (t ) = cos 2f 1t

2

)* sin (2f t 4 )
1

= =

1 1 sin 2f 1 2t 3 sin 8 8 2 2

( )
( )]
(4.25)

1 sin (2f 1 2t ) cos cos(2f 1 2t )sin 0.19 8 8 2

( )

Output of the control channel is the correlation of the output of c1 and c2. The high frequency components are filtered by the low-pass filter, only the dc terms remains and output of the low-pass filter is rC (t ) = rC (t ) * rC (t )
1 2

= 0.1

(4.26)

The start bit from differential encoding is removed by correlating with the previous signal at the output of the low-pass filter. The + dc levels from every channels, representing the three parallel bit streams, are converted into a series bit stream for detection.

46 Table 4.6 shows the lookup table for D8PSK that expressed how the modulated signals are demodulated at the detection part with the detected bits at the output of the receiver.

Table 4.6:

Lookup table for D8PSK detection

Demodulation signals Signal phases x(t) Bit 1 cos (2f1t) 1 1 0 0 0 0 1 1 Bit 2 sin (2f1t) 1 1 1 1 0 0 0 0 Bit 3 cos (2f1t-/4) 1 1 1 0 0 0 0 1 sin (2f1t-/4) 0 1 1 1 1 0 0 0 Bits 110 111 011 010 000 001 101 100

cos (2f1t-/8) sin (2f1t-/8) cos (2f1t-3/8) sin (2f1t-3/8) -cos (2f1t-/8) -sin (2f1t-/8) -cos(2f1t-3/8) -sin(2f1t-3/8)

4.5.1

D8PSK BER Performance

BER performance for D8PSK detection scheme is evaluated using union bound, as discussed in Section 4.3. The signal space plane for D8PSK union bound is shown in Figure 4.10.

47

Figure 4.10

D8PSK union bound signal space plane

Using trigonometric function, error distance, d can be expressed as, d = A sin

(4.27)

From Equation 4.4, error distance, dj for a given symbol, sj is written as

d j = Q MAX sin 8 2

(4.28)

Given that s=111. Error Pe,111 will occur if s=011 or s=110. The probability is written as
max sin ) + Q( max sin ) 2 8 2 8

Pe,111 = Q(

48
max sin ) 2 8

= 2Q(

(4.29)

The BER performance can be written as Pe = Pe ,111 P( s = 111) + Pe ,110 P( s = 110) + Pe ,101 P( s = 101) + Pe ,100 P( s = 100) + Pe , 011 P( s = 011) + Pe , 010 P( s = 010) + Pe , 001 P( s = 001) + Pe , 000 P ( s = 000) (4.30)

Since Pe,111 = Pe,101 = Pe,101= Pe,100 = Pe,011 = Pe,001 = Pe,001= Pe,000, then the BER performance is Pe = 8[Pe ,111 P( s = 111)]
1 = 8 2Q( max sin ) 2 8 8 = 2Q(

max sin ) 2 8

(4.29)

The BER performance for D8PSK can be expressed as 1 A 2Tb Pe = 2 exp sin 4N 2 8 o

(4.30)

Therefore,
A 2Tb Pe = exp sin 4N 8 o

(4.31)

49
4.6 Differential 16 Phase Shift Keying

Differential 16PSK uses sixteen different angles, which every angle represents four bits per symbol of information. D16PSK technique can transmits four bits per symbol for every transmission. Similar to D8PSK, D16PSK uses differential coding at the modulation part to overcome phase synchronization error at the receiver side. The constellation diagram in Figure 4.11 shows the different angles used in D16PSK modulation technique.

0111 0110 0100 0101

1111 1110 1100 1101

0001 0000 1000 0010 0011 1011 1010

1001

Figure 4.11

D16PSK constellations diagram

The transmitted signal of D16PSK can be written as x(t ) = A cos(2f 1t i )

i =

(2 + 1)i
16

(4.32)

50 where index, i, takes on values from 0 to 15. The block diagram for the D16PSK receiver is shown in Figure 4.12.

Figure 4.12

D16PSK receiver structure

It is assumed that consecutive symbols 1101 are transmitted with assumption that the phase shift is zero. The received signal is x(t ) = A cos(2f 1t

16

(4.33)

51 It is assume that A=1. Input of the low-pass filter is

Inphase channel: q I (t ) = cos 2f 1t = =

16

)* cos(2f t )
1

1 1 cos 2f 1 2t + cos 16 16 2 2

) ( )]
(4.34)

1 cos(2f1 2t ) cos + sin (2f1 2t )sin + 0.49 8 8 2

( )

The high frequency components are filtered by the low-pass filter, only the dc terms remains and output of the low-pass filter is
rI (t ) = 0.49

(4.35)

Quadrature channel: qQ (t ) = cos 2f 1t = =

)* sin(2f t )
1

1 1 sin 2f 1 2t sin 8 8 2 2

1 sin (2f1 2t ) cos cos(2f1 2t )sin + 0.1 8 8 2

( )

( )]

(4.36)

The high frequency components are filtered by the low-pass filter, only the dc terms remains and output of the low-pass filter is rQ (t ) = 0.1

(4.37)

52 Control channel for 3rd bit, using /4 phase shifter, two demodulator signals are shown as below xc (t ) = cos(2f 1t ) 4
1

xc (t ) = sin(2f 1t ) 4
2

(4.38)

c1 channel: c1 (t ) = cos 2f 1t = =

16

)* cos(2f t 4 )
1

1 1 cos 2f 1 2t 5 + cos 3 16 2 16 2

) ( )]
(4.39)

1 cos(2f 1 2t ) cos 5 + sin (2f 1 2t )sin 5 + 0.42 16 16 2

c2 channel: c 2 (t ) = cos 2f 1t =

16

)* sin(2f t 4 )
1

1 1 sin 2f 1 2t 5 sin 3 16 2 16 2

) ( )]

1 = sin 2f1 2t 5 0.28 16 2 = 1 sin (2f 1 2t ) cos 5 cos(2f 1 2t )sin 5 0.28 16 16 2

(4.40)

Input of low-pass filter is the correlation of c1(t) and c2(t) of the channel and can be shown as q c (t ) = c1 (t ) * c 2 (t )

(4.41)

53 The high frequency components are filtered by the low-pass filter, only the dc terms remains and output of the low-pass filter is rC (t ) = 0.1

(4.42)

Control channel for 4th bit, using /4 phase shifter, the demodulator signals for the channel are x d (t ) = cos(2f 1t ) 8
1

x d (t ) = sin(2f 1t ) 8
2

x d (t ) = cos(2f 1t
3

3 ) 8 3 ) 8 (4.43)

x d (t ) = sin( 2f 1t
4

d1 channel: d 1 (t ) = cos 2f 1t = =

16

)* cos(2f t 8 )
1

1 1 cos 2f 1 2t 3 + cos 16 16 2 2

( ) ( )]
(4.44)

1 cos(2f 1 2t ) cos 3 + sin (2f 1 2t )sin 3 + 0.49 16 16 2

54 d2 channel: d 2 (t ) = cos 2f 1t = =

16

)* sin(2f t 8 )
1

1 1 sin 2f 1 2t 3 sin 16 16 2 2

( ) ( )]
(4.45)

1 sin (2f 1 2t ) cos 3 cos(2f 1 2t )sin 3 0.1 16 16 2

d3 channel: d 3 (t ) = cos 2f1t = =

16

)* cos(2f t 3 8 )
1

1 1 cos 2f 1 2t 7 + cos 5 16 16 2 2

) ( )]
(4.46)

1 cos(2f1 2t ) cos 7 + sin (2f1 2t )sin 7 + 0.28 16 16 2

d4 channel: d 4 (t ) = cos 2f1t = =

16

)* sin(2f t 3 8 )
1

1 1 sin 2f 1 2t 7 sin 5 16 2 16 2

) ( )]
(4.47)

1 sin (2f1 2t ) cos 7 cos(2f1 2t )sin 7 0.42 16 16 2

55 Input of low-pass filter for 4th bit channel is the correlation of d1(t), d2(t), d3(t) and d4(t). It can be expressed by: q d (t ) = d1 (t ) * d 2 (t ) * d 3 (t ) * d 4 (t )

The high frequency components are filtered by the low-pass filter, only the dc term will remain at the output of the filter. The output is rd (t ) = 0.006

The start bit from the differential encoding is removed by correlating with previous signal at the output of low-pass filter. The + dc levels from every channel, representing the four parallel bit streams are converted into a series bit stream for detection.

Table 4.7 shows the lookup table for D16PSK that expressed how the modulated signals are demodulated at the detection part of the receiver.

Table 4.7:
Signal phases

Lookup table for D16PSK detection

Demodulation signals
Bit 1 cos (2 f1t) 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 Bit 2 sin (2 f1t) 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 Bit 3 cos (2 f1t- /4) 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 sin (2 f1t- /4) 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 cos (2 f1t- /8) 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 sin (2 f1t- /8) 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 Bit 4 cos (2 f1t-3 /8) 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 sin (2 f1t-3 /8) 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 1101 1100 1110 1111 0111 0110 0100 0101 0001 0000 0010 0011 1011 1010 1000 1001

x(t)
cos (2f1t-/16) cos (2f1t-3/16) cos (2f1t-5/16) cos (2f1t-7/16) cos (2f1t-9/16) cos (2f1t-11/16) cos (2f1t-13/16) cos (2f1t-15/16) cos (2f1t-17/16) cos (2f1t-19/16) cos (2f1t-21/16) cos (2f1t-23/16) cos (2f1t-25/16) cos (2f1t-27/16) cos (2f1t-29/16) cos (2f1t-31/16)

Bits

56

57 4.6.1 D16PSK BER Performance

BER performance for D16PSK detection scheme is evaluated using union bound, as discussed in Section 4.3. From D16PSK constellation diagram in Figure 4.8, error distance, d is evaluated using trigonometric function, and can be expressed as, d = A sin

16

(4.50)

From Equation (4.4), error distance, dj for a given symbol, sj is written as

d j = Q MAX sin 16 2

(4.51)

Given that s=1111. Error Pe,1111 will occur if s=0111 or s=1110. The probability is written as
max sin ) + Q( max sin ) 2 16 2 16 max sin ) 2 16

Pe,1111 = Q( = 2Q(

(4.52)

The BER performance can be written as

Pe = Pe ,1111 P( s = 1111) + Pe ,1110 P( s = 1110) + .... + Pe , 0000 P( s = 0000)

(4.53)

58 Since the priory probability is same, then the BER performance can be written as Pe = 16[Pe ,111 P( s = 111)]
1 ) = 16 2Q( max sin 2 16 16 = 2Q( max sin ) 2 16

1 A 2Tb = 2 exp sin 16 4N o 2

(4.55)

Therefore, the BER performance of D16PSK detection is

A 2Tb Pe = exp 4 N sin 16 o

(4.56)

4.7

Robustness to Phase Synchronization Error in Differential Multiple PSK Detection

This section will display the effect of phase delay presented in received signal for differential multiple PSK modulations. It is to show that the phase is not present in the detection, and differential phase modulation is robust to phase synchronization error.

59
4.7.1 Case for DQPSK Detection

Given that a received signal, x(t) is

x(t ) = A cos(2f 1t + 1 ) + A sin (2f 1t + 1 )

0 t Tb

(4.57)

where 1 is the phase delay present in the signal. It is assume that A=1. Input of lowpass filter is the correlation of the received signal and the previous signal from differential encoding. Input of the low-pass filter is presents as

Inphase channel:
q I (t ) = (cos(2f 1t + 1 ) + sin( 2f 1t + 1 ) ) * cos(2f 1t + 1 ) = cos 2 (2f1t + 1 ) + cos(2f1t + 1 ) sin( 2f1t + 1 )

1 1 1 + cos(2f 1 2t + 21 ) + sin(2f 1 2t + 21 ) 2 2 2

(4.58)

Since the high frequency component is filtered, the output of low-pass filter is only the dc term of the signal. Output of the low-pass filter is

rI (t ) =

1 2

(4.59)

Quadrature channel: qQ (t ) = (cos(2f 1t + 1 ) + sin(2f 1t + 1 ) ) * sin(2f 1t + 1 )

= cos(2f1t + 1 ) sin( 2f1t + 1 ) + sin 2 (2f1t + 1 )

60 = 1 1 1 sin (2f 1 2t + 21 ) + sin(2f 1 2t + 21 ) 2 2 2

(4.60)

High frequency component of the signal is removed by the low-pass filter. Only the dc term of the signal remains at the output. Output of the low-pass filter is

rQ (t ) =

1 2

(4.61)

The result proofs that the phase is not presented in the detection, and the DQPSK detection is robust to phase error.

4.7.2

Case for Differential Multiple Phase Detection

For multiple phase detection, the example of D8PSK will be used to calculate the effect of phase delay. The result of phase effect in any differential M-ary PSK is same, since it uses the same method to demodulate the data.

The received signal is given by x(t ) = A cos 2f 1t +

+ 1

0 t Tb

(4.62)

61 where 1 is the phase delay present in the signal. It is assume that A=1. By using differential encoding from the previous start bit signal, and by using phase shifter, the demodulators can be shown as
xI (t ) = cos(2f1t + 1 )

xQ (t ) = sin (2f1t + 1 ) xc1 (t ) = cos 2f1t + + 1 4 xc 2

1 1

( ) (t ) = sin (2f t + + ) 4

(4.62)

Input of low-pass filter is the correlation of the received signal and the demodulator of channels. Input of the low-pass filter is calculated as below.

Inphase channel: q I (t ) = cos 2f 1t = =

+ 1 * cos(2f 1t + 1 )

1 1 cos 2f 1 2t + 2 1 + cos 8 8 2 2 1 cos 2f 1 2t + 2 1 + 0.46 8 2

)
(4.63)

The high frequency component in the signal is filtered by the low-pass filter. Then the output signal is only the dc term of the signal.
rI (t ) = 0.46

(4.64)

62 Quadrature channel: qQ (t ) = cos 2f 1t = =

+ 1 * sin(2f 1t + 1 )

1 1 sin 2f 1 2t + 2 1 sin 8 8 2 2 1 sin 2f 1 2t + 2 1 + 0.19 8 2

)
(4.65)

Since the qQ(t) is filtered, only the dc terms remains and output of the low-pass filter is rQ (t ) = 0.19

(4.66)

C channel: c1 (t ) = cos 2f 1t = =

+ 1 * cos 2f 1t

+ 1

1 1 cos 2f1 2t 3 + 21 + cos 8 8 2 2 1 cos 2f 1 2t 3 + 2 1 + 0.46 8 2

( )
(4.67)

c 2 (t ) = cos 2f 1t = =

+ 1 * sin 2f 1t

) (

+ 1

1 1 sin 2f 1 2t 3 + 2 1 sin 8 8 2 2 1 sin 2f 1 2t 3 + 2 1 0.19 8 2

( (

( )
(4.68)

Input of low-pass filter for C channel is q c (t ) = c1 (t ) * c 2 (t )

(4.69)

63 Since the qQ(t) is filtered, only the dc terms remains and output of the lowpass filter is rc (t ) = 0.1

The result proofs that the phase is not presented in the detection. Thus, the differential multiple phase detection is robust to phase synchronization error.

4.8

Signal Representation in Time and Frequency Domain

Figure 4.13 shows the example of 8-bit data sequence {1 0 1 1 0 1 0 0} have been transmitted using all DPSK, DQPSK, D8PSK and D16PSK techniques discussed in this chapter. The signals are shown in time and frequency domain. The frequency representation shows the data is placed at the carrier frequency of 1000Hz, and the time domain representation observe the modulated signals that carry symbol of data for 100 baud transmission rate.

64

(a)

(b)

(c)

(d)

Figure 4.13

Signals representation in time and frequency domain. (a) DPSK

modulated signal (b) DQPSK modulated signal (c) D8PSK modulated signal (d) D16PSK modulated signal

CHAPTER V

RESULTS AND DISCUSSIONS

5.1

Introduction

The chapter will display the results of the BER and PER performance analysis by using four modulation techniques in differential phase modulation. The results are discussed on the theoretical and simulation value to compare the performance of DPSK, DQPSK, D8PSK and D16PSK detection. Simulation is performed in the present of additive white Gaussian noise based on Equation (3.1) in chapter III. A random phase terms is included in the simulation that is based on a uniform distribution where the phase range is 0 5 / 2 . This is to determine the effect of phase synchronization error on the detection method.

66 5.2 Analysis of BER Performance

The BER performance of DPSK, DQPSK, D8PSK and D16PSK detection has been analyzed theoretically and by simulation. It is proved theoretically and also by simulation that, as the number of phases increases, the SNR ratio also increases for a fixed bit error rate.

Figure 5.1

Theoretical BER performance

The theoretical BER performance for DPSK, DQPSK, D8PSK and D16PSK detection are shown in Figure 5.1. The result shows that the DPSK detection gives the best performance amongst the other detection techniques. For a given BER of 10-4, the performance of DQPSK detection downgrades by only 0.4 dB compared to DPSK

67 detection. For d8PSK detection, the reduction of performance is 4 dB compared to DQPSK detection and better than D16PSK detection by 3dB. The theoretical BER performance for each of the modulation is summarized in Table 5.1.

Table 5.1:

Theoretical BER Performance for a fixed BER of 10-4 BER 0.0001 0.0001 0.0001 0.0001 SNR (dB) 12.2 12.6 16.8 19.7

Type of detection DPSK DQPSK D8PSK D16PSK

The result is also true by simulation. It shows that the DPSK detection gives the best performance. The reduction in performance for DQPSK detection in terms of the fixed BER of 10-4 is about 2dB. Beyond D8PSK detection, the performance downgrades 6dB for every doubling of phases. Figure 5.2 shows the simulation result of DPSK, DQPSK, D8PK and D16PSK detection. The value of SNR for BER performance of 10-4 for every type of the simulation is shown in Table 5.2.

68

Figure 5.2

Simulation result of BER performance

Table 5.2:

BER performance of simulation result for a fixed BER of 10-4 BER 0.0001 0.0001 0.0001 0.0001 SNR (dB) 13.8 16.3 22.0 27.7

Type of detection DPSK DQPSK D8PSK D16PSK

69 5.3 Analysis of PER Performance

The PER performance are analyzed theoretically and also by simulation. The results of the performance are shown in Figure 5.3 - 5.4, respectively.

Figure 5.3

Theoretical PER performance

Figure 5.3 shows the theoretical PER performance of DPSK, DQPSK, D8PSK and D16PSK derived in previous chapter. The result shows that for a given PER of 10-2 the performance based on SNR is almost the same with the theoretical BER performance discussed in Section 5.2. DPSK gives the lowest packet error rate, followed by DQPSK, which the reduction in performance is only 0.4 dB. The performance of D8PSK detection reduced 4 dB than DQPSK detection, and is better

70 than D16PSK detection by 3dB. Table 5.3 illustrates the theoretical performance for a fixed PER of 10-2. Note that, the SNR value is almost the same with BER performance of 10-4 discussed in previous section.

Table 5.3:

Theoretical PER Performance for a fixed PER of 10-2 BER 0.01 0.01 0.01 0.01 SNR (dB) 12.2 12.6 16.7 19.6

Type of detection DPSK DQPSK D8PSK D16PSK

Figure 5.4 shows the PER performance of the simulation result. In the simulation, the packet size is 80 bits, which the data is transmitted using PACTOR format as discussed in previous chapter. For a fixed PER of 10-2, DPSK gives the lowest PER, followed by DQPSK, D8PSK and D16PSK. The reduction in performance for DQPSK detection is 2dB. Beyond D8PSK detection, the performance downgrades 6dB for every doubling of phases. For the simulation of PER, it is also follow the result of BER performance for a fixed BER of 10-4. Table 5.4 shows that it gives similar SNR value with the simulation of BER performance.

71

Figure 5.4

Simulation result of PER performance

Table 5.4:

Simulation PER Performance for a fixed PER of 10-2 BER 0.01 0.01 0.01 0.01 SNR (dB) 12.2 12.6 16.7 19.6

Type of detection DPSK DQPSK D8PSK D16PSK

From the results, it is shown that for every doubling of phases, the BER performance decreases, but the main important thing is the capability to increase the performance of data transmission rate. It is also shown that even the random phase delay is applied in the simulations, the error performance is not effected by this phase

72 delay. This is because the differential encoding is used in the modulation, so the phase synchronization is not critical such in coherent PSK detection.

From the comparison of the simulation and the theoretical results, it is clearly shows that the BER and PER of the simulation is higher than the theoretical result. It is also shows that, as the SNR increases, the performance of simulation degrade faster than the theoretical performance. This is expected since the theoretical performance is calculated in ideal case. By simulation, the performance of modulations are depending on the design of filters and detection structures, carrier frequency and the number of samples used in modulation, and also how big the random noise power will effect the change in phases. All these factors are contributing bigger error in the detection. The graph of BER and PER comparison performances for DPSK, DQPSK, D8PSK and D16PSK detections are shown in Appendix A-B, respectively.

CHAPTER VI

CONCLUSIONS

6.1

Conclusions

In this study, the differential multiple PSK modulation techniques are evaluated to analyze the performance of the modulation in term of the BER and PER. Differential multiple PSK modulation techniques proofs to be the suitable method because of it capability of achieving higher data transmission rate, its robustness to phase synchronization error and it also capable in providing lower error rate compared to many other possible advance modulation techniques.

Modulation techniques based on DPSK is proofed in capability to provide lower error rate than modulation based on FSK techniques [Martin, 1988]. In addition, DQPSK modulation technique provides better performance than noncoherent FSK, and the transmission rate is two times faster than the FSK technique [Shaameri-Jaswar, 2003] (Appendix C).

74 In general, the BER and PER performance for differential multiple phase detection decrease for every doubling of phases, but the main advantage is the capability in achieving higher transmission rate, and implementation can be made simpler compared to coherent PSK detection since the phase synchronization error is not critical.

6.2

Suggestions

The suggestions of this study are as follow:

1. For future study, it is suggested to implement the error control mechanism in differential multiple PSK detection to detect errors that occur in packets of data transmission.

2. Since the performance analysis of this study was limited in presence of additive white Gaussian noise and phase delay, it is suggested to analyze the performance of differential multiple PSK in presence of multipath fading, such as Rayleigh and Rician fading channel.

3. The complexity of differential multiple PSK is simple and reliable, so it is suggested to implement the design of modulation technique on hardware for transmission using HF.

4. To have more accurate performance results, it is suggested to explore other advance modulation techniques in PSK, such as differential in Quadrature Amplitude Modulation (QAM). QAM possesses the potential for providing

75 better performance since the symbol is separated as widely as possible in the signal space diagram [Roden, 1988].

76

REFERENCES

[Abdullah-Husni et al, 2003]

Abdullah, M.A.; Husni, E.M.; S. Hassan, S.I.,

Investigation of a Rural Telecommunication System using VSAT Technology in Malaysia. 9th Asia Pacific Conference on Communications, 2003 (APCC), Penang, 21-24 September 2003.

[Charles-Tri, 2003]

Charles, W.V.; Tri, T.H., Performance Analysis of Direct

Sequence Differential Phase Shift Keying (DS-DPSK) with Self-Normalization and L-Fold Diversity in a Fading Channel. California: Naval Postgraduate School, 2003.

[Couch, 1997]

Couch, L.W. II, Digital and Analog Communication Systems.

Prentice-Hall International, Inc. 5th edition, 1997.

[Goodman, 1992]

Goodman, J.M., HF Communications Science and Technology,

Van Nostrand Reinhold, 1992.

[Haykin, 1988] Inc., 1988

Haykin, S., Digital Communications. John Wiley and Sons,

[Jaswar-Shaameri, 2003]

Jaswar, F.D.; Shaameri, A.Z., FPGA Implementation of

CPFSK Modulation Techniques for HF Data Communication. Malaysia: Faculty of Elect. Engr., Univ. Teknologi Malaysia, 2003.

[John-Mohamed et al, 2002] Recovery in

John, D.B.; Mohamed, K.N.; Mary, E.D., Data Encoded Quadrature Phase Shift Keying

Differentially

(DEQPSK). Mnemonics, Inc., 2002.

77 [Kasser, 1991] Kasser, J.E., Applied digital communications via HF radio.

Fifth International Conference on HF Radio Systems and Techniques, 1991, pp. 44 47.

[Kolimbiris, 2000]

Kolimbiris, H., Digital Communications Systems with Satellite

and Fiber Optics Applications. New Jersey: Prentice Hall, Inc., 2000.

[McNamara, 1991]

McNamara, L.F., Ionosphere: Communications, Surveillence,

and Direction Finding. Krieger Publishing Company, Malabar, Florida, 1991.

[NATO]

NATO Standardization Agreement: Profile For High Frequency (HF)

Radio Data Communicatios STANAG 5066.

[Nilsson-Giles, 1997]

Nilsson, J.E.M.; Giles, T.C., Wideband multi-carrier

transmission for military HF communication, MILCOM 97 Proceedings ,Volume: 2, pp.1046 1051, 1997. Proakis, J.G., Digital Communications. Mc Graw-Hill, Inc., 3rd

[Proakis, 1995] Edition, 1995.

[Henry, 1992] Bill Henry, K9GWT. Getting Started in Digital Communications-Part 4-AMTOR. American Radio Relay, Inc. QST June 1992, pp. 34-35, 1992.

[Raos et al, 2003]

Raos; Del Cacho, A.; Perez-Alvarez, I.; Zazo, S.; Mendieta-

Otero, E.; Santana-Sosa, H.; Paez-Borralo, J.M., Advanced OFDM-CDMA HF Modem With Self-interference Cancellation. Spain: Universidad de Las Palmas de Gran Canaria, 2003.

[Reynolds-Gillespie, 1997]

Reynolds, P.D.; Gillespie, A.F.R., Interim Profile For

Maritime HF Data Communications. IEE 7th Int. Conf. on HF Radio Systems and Techniques, 1997, pp. 265-270, 1997.

78 [Roden, 1988] Roden, M.S., Digital Communication Systems Design.

Prentice-Hall International Inc., 1988.

[Rodger, 2000]

Rodger,

E.Z.;

Roger,

L.P.,

Introduction

to

Digital

Communication. New Jersey: Prentice-Hall Inc., 2000.

[Shaameri-Jaswar, 2003]

Shaameri, A.Z.; Jaswar, F.D., Detection of Binary Data

for FSK Digital Modulation Signals using Spectrum Estimation Techniques. 4th National Conference On Telecommunication Technology (NCTT 2003), 2003.

[Shanmugan, 1988]

Shanmugan, K.S.;Breipohl, A.M., Random Signals: Detection,

Estimation and Data Analysis, John Wiley & Sons Inc. 1988.

[Stalings, 2000]

Stalings W., Data and Computer Communications. Prentice-

Hall Internal Editions 6th edition, 2000.

[Tomasi, 1987]

Tomasi;

Wayne,

Advanced

Electronic

Communications

Systems. Prentice Hall, Inc., Englewood Cliffs, New Jersey, 1987.

[Trinder-Brown, 1999]

Trinder, S.E.; Brown, D.D., Algorthms For The

Adaption Of Data Rate Using STANAG 5066. IEEE, Savoy Place, London, 1999.

[Willink et al, 1996] Willink, T.J.; Davies, N.C.; Clarke, J.; Jorgenson, M.B.; Validation of HF Channel Simulators. Frequency Selection and Management Techniques for HF Communications, IEE Colloquium on, 1996, Page(s): 13/113/6.

79 [Yao-Teng et al, 2002] Yao, M.; Teng, J.L.; Subbarayan, P., Error Probability

for Coherent and Differential PSK Over Arbitrary Rician Fading Channels With Multiple Co-channel Interferers. IEEE Transactions on Communications, Volume: 50. No. 3. 2002, pp 428-431.

APPENDICES

BER Performance

BER DPSK

BER D8PSK

BER DQPSK

BER D16PSK

PER Performance

PER DPSK

PER D8PSK

PER DQPSK

PER D16PSK

Detection based on FSK

1.00E+00
0 5 10 15 20

1.00E-01

1.00E-02

1.00E-03

BER

1.00E-04

coherent noncoherent psd_sinusoid psd_squarewave

Type of detection Noncoherent FSK

BER 10-3

SNR (dB) 16.9

1.00E-05

SNRdB

[Shaameri-Jaswar, 2003] on: 4th National Conference on Telecommunication Technology (NCTT 2003), 2003.