EFFICIENT MACRO DIVERSITY HANDOVER TECHNIQUES FOR MULTIHOP CELLULAR NETWORKS

GAMIL SULTAN ABDULAZIZ

THESIS SUBMITTED IN FULFILMENT FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

FACULTY OF ENGINEERING AND BUILT ENVIRONMENT UNIVERSITI KEBANGSAAN MALAYSIA

BANGI

2011

TEKNIK-TEKNIK KECEKAPAN LEPAS TANGAN KEPELBAGAIAN MAKRO UNTUK RANGKAIAN SELULAR BERBILANG HOP

GAMIL SULTAN ABDULAZIZ

TESIS YANG DIKEMUKAKAN UNTUK MEMPEROLEH IJAZAH DOKTOR FALSAFAH

FAKULTI KEJURUTERAAN DAN ALAM BINA UNIVERSITI KEBANGSAAN MALAYSIA

BANGI

2011

iii

DECLARATION

I hereby declare that the work in this thesis is my own except for quotations and

summaries which have been duly acknowledged.

8 August 2011 GAMIL SULTAN ABDULAZIZ P41401

iv

ACKNOWLEDGEMENTS

First and foremost, I am grateful to Allah, the Almighty, Whose blessings and guidance have helped me throughout the course of my life.

It is my pleasure to express my sincere gratitude and deepest thanks to my supervisor Assoc. Prof. Dr. Norbahiah Misran for her encouragement, endless support and guidance she generously provided during the course of my PhD in UKM. Her valuable discussions, comments and suggestions have greatly improved the content and the presentation of this thesis. I would like also to express my deepest gratitude to my co-supervisor Prof. Dr. Mahamod Ismail for his encouragement, continuous support and guidance during the period of my PhD. His valuable comments, discussions and feedback have improved this thesis tremendously. I would like also to extend my special thanks to my co-supervisor Assoc. Prof. Dr. Mohammad Tariqul Islam for his help, guidance and encouragement.

I would like to appreciate the kind cooperation of the staff in the department of electrical, electronic and systems engineering and the faculty of engineering and built environment during the years in UKM. I also thank all my colleagues and friends in the network lab, especially, my friends Redwan Abdulkader and Abduljalil Radman. The Financial supports of the MIMOS Berhad under Grant No. PKT 3/2008 and the Science Fund under Grant No. 01-01-02-SF0376 are also gratefully acknowledged.

I wish to thank my uncle Sadiq Al-ameri, my aunt Nazirah Al-ameri, Prof. Dr. Abdulsamad Hazza’a, Khalid Abdu Qasim and my friends Azzam Al-nahari, Mubarak Saif, Mohammed Nusari and Walid Al-kadasi for their continuous encouragement.

I am deeply indebted to my beloved mother, Mannon, for her endless love, patience and prayers. Her prayers and love gave me the strength and motivation to achieve this milestone. I would like also to extend my special thanks and appreciation to my brothers, Adel and Abdulilah, and my sisters, Mona, Iman, Aminah and Amirah, for their continuous support and encouragement during my study.

To my wife, Elham, and our beloved daughters, Roa’a and Raghad, goes my deepest gratitude for their love, understanding and patience and for always lifting my spirits through this period. Without their existence beside me, this thesis would have never been completed.

Finally, may Allah shows mercy to my father, Sultan, who left this world before he could see this moment that he had always hoped to. To his soul, to my mother and to my family, I would like to pass and dedicate this success.

v

ABSTRACT

Macro diversity handover (MDHO) is the process by which the mobile station (MS) maintains connection with two or more access stations called a diversity set. In the downlink (DL) of the conventional MDHO, the MS receives only the simultaneous transmissions of the diversity set members either from a base station (BS) and a relay station (RS), two RSs, or two BSs. Whenever the diversity set members of the MS are a BS and an RS, the BS transmits to the RS during the first phase, whereas during the second phase both the BS and RS transmit simultaneously to the MS by using the same radio resource. Hence, the signal transmitted by the BS in the first phase is not received by the MS even though the MS is idle in this phase. In contrast, in the uplink (UL) of the conventional MDHO, the signal-to-interference-and-noise ratio (SINR)-based selection combining (SC) among the received signals is performed. This SINR-based SC (SSC) scheme does not necessarily achieve the best performance in multihop cellular networks. These raise the need to investigate more efficient DL and UL MDHO techniques for multihop cellular networks. The objective of this research is to propose a topology-aware MDHO technique with efficient selection and combining schemes for TDD-OFDMA-based interference-limited multihop cellular networks. In the proposed topology-aware DL MDHO technique, the MS receives all the data signals transmitted by the diversity set members. It ensures that the topology of the diversity set members is always fully exploited. In the first proposed UL joint maximal-ratio combining (MRC)-SC scheme, the signals received from the MS → BS and RS → BS links; or the signals received from two RSs, are combined at the BS using MRC to achieve higher spatial diversity gain in case of intra-cell MDHO scenarios. However, in case of inter-cell MDHO scenarios, SC is performed. The second proposed UL scheme combines the advantages of the end-to-end (e2e) throughput-based SC (ETSC) with the benefits of using UL power control at the RSs. In the third proposed UL e2e bit error rate (BER)-based SC (EBSC) scheme, the e2e BER is used as the selection metric to decide on the appropriate link. The superiority of the proposed MDHO techniques is validated using both mathematical and simulation models. The DL performance evaluation shows that the proposed topology-aware MDHO significantly outperforms the conventional MDHO in terms of the average DL SINR, the average DL e2e BER, the average DL spectral efficiency and the outage probability. Over the MDHO regions in which the diversity set members are a BS and an RS, the proposed MDHO achieves a SINR gain and a spectral efficiency gain up to 5.32 dB and 79% (1.07 to 1.92 bps/Hz) respectively compared with the conventional MDHO. Meanwhile, the UL performance evaluation shows that the proposed joint MRC-SC scheme obtains a SINR gain up to 1.33 dB whereas the proposed ETSC scheme with power control at the RS achieves an average throughput gain up to 49% (0.83 to 1.24 bps/Hz) compared with the conventional SSC scheme. Finally, the proposed EBSC scheme significantly outperforms the conventional SSC scheme in terms of the UL e2e BER.

vi

ABSTRAK

Lepas tangan kepelbagaian makro (MDHO) merupakan proses apabila stesen mobil (MS) mengekalkan sambungan dengan dua atau lebih stesen capaian yang dipanggil set kepelbagaian. Dalam pautan bawah (DL) bagi MDHO konvensional, MS hanya menerima penghantaran serentak dari ahli set kepelbagaian sama ada dari satu stesen tapak (BS) dan satu stesen geganti (RS), dua RS atau dua BS. Apabila ahli set kepelbagaian MS ialah satu BS dan satu RS, BS menghantar kepada RS semasa fasa pertamanya, manakala semasa fasa kedua kedua-dua BS dan RS menghantar secara serentak kepada MS dengan menggunakan sumber radio yang sama. Maka, penghantaran isyarat oleh BS dalam fasa pertama tidak akan diterima oleh MS walaupun MS dalam keadaan melahu dalam fasa ini. Sebaliknya, dalam pautan bawah (UL) bagi MDHO konvensional, penggabungan pemilihan berasaskan nisbah isyarat-ke-gangguan-dan-hingar (SINR) di antara isyarat terima dilaksanakan. Skema SC berasaskan SINR ini tidak semestinya mencapai prestasi terbaik dalam rangkaian selular berbilang loncatan. Hal ini meningkatkan keperluan untuk mengkaji teknik MDHO bagi DL dan UL yang lebih cekap untuk rangkaian selular berbilang hop. Objektif utama kajian ini adalah untuk mencadangkan suatu teknik sedar-topologi MDHO dengan skema pemilihan dan penggabungan cekap untuk rangkaian selular TDD-OFDMA berbilang loncatan berasaskan gangguan terhad. Dalam teknik DL MDHO sedar-topologi yang dicadangkan, MS menerima semua isyarat data yang dihantar oleh ahli set kepelbagaian. Teknik ini akan memastikan bahawa topologi bagi ahli set kepelbagaian sentiasa diekploitasi sepenuhnya. Dalam skema SC bersama nisbah-maksimum penggabungan (MRC) UL yang dicadangkan pertamanya, isyarat yang diterima dari pautan MS → BS dan RS → BS; atau isyarat diterima dari dua RS, akan digabung di BS menggunakan MRC bagi menghasilkan gandaan kepelbagaian ruang yang lebih tinggi bagi kes senario MDHO intra-sel. Walaupun demikian, dalam kes senario MDHO antara-sel, SC telah dilaksanakan. Skema UL kedua yang dicadangkan menggabungkan kelebihan truput hujung-ke-hujung (e2e) berasaskan SC (ETSC) dengan kelebihan menggunakan kawalan kuasa UL di RS. Dalam cadangan ketiga bagi skema UL e2e kadar ralat bit (BER) berasaskan SC (EBSC), BER e2e digunakan sebagai metrik pilihan bagi menentukan pautan yang sesuai. Keunggulan teknik MDHO yang dicadangkan telah ditentusahkan menggunakan kedua-dua model matematik dan simulasi. Penilaian prestasi bagi DL menunjukkan bahawa MDHO sedar-topologi yang dicadangkan ternyata secara signifikan mengatasi MDHO konvensional dari segi purata DL SINR, purata DL e2e BER, purata kecekapan spektrum DL dan kebarangkalian keluaran. Bagi kawasan-kawasan MDHO apabila ahli set kepelbagaian ialah satu BS dan satu RS, MDHO yang dicadangkan mencapai gandaan SINR dan gandaan kecekapan spektrum masing-masing sehingga 5.32 dB dan 79% (1.07 ke 1.92 bps/Hz) berbanding dengan MDHO konvensional. Sementara itu, keputusan penilaian UL menunjukkan bahawa skema MRC-SC bersama memperolehi gandaan SINR sehingga 1.33 dB manakala skema ETSC dengan kawalan kuasa di RS yang dicadangkan memperolehi gandaan truput purata sehingga 49% (0.83 ke 1.24 bps/Hz) berbanding dengan skema SSC konvensional. Akhirnya, skema EBSC yang dicadangkan secara signifikannya mengatasi skema SSC dari segi e2e BER bagi UL.

vii

TABLE OF CONTENTS

Page

DECLARATION iii

ACKNOWLEDGEMENTS iv

ABSTRACT v

ABSTRAK vi

TABLE OF CONTENTS vii

LIST OF TABLES xi

LIST OF FIGURES xii

LIST OF ABBREVIATIONS xvii

LIST OF SYMBOLS xxiii

CHAPTER I INTRODUCTION

1.1 Background 1

1.2 Motivations and Problem Statement 4

1.3 Objectives and Scope of the Research 8

1.4 Thesis Contributions 9

1.5 Thesis Organization 11

CHAPTER II LITERATURE REVIEW

2.1 Introduction 13

2.2 Evolution of Wireless Access Networks towards Fourth

Generation

13

2.2.1 Mobile WiMAX 16 2.2.2 OFDMA Basics 22

2.3 Multihop Relay Networks 24

2.3.1 Basic Relaying Concepts 25

2.4 Wireless Radio Channel 28

2.4.1 Path Loss 30 2.4.2 Long-term Fading 31 2.4.3 Short-Term Fading 32

2.5 Diversity Techniques 33

2.5.1 Temporal Diversity 33

viii

2.5.2 Frequency Diversity 34 2.5.3 Spatial Diversity 34

2.6 Diversity Combining Techniques 37

2.6.1 Selection Combining 37 2.6.2 Switched Combining 38 2.6.3 Maximal Ratio Combining 38 2.6.4 Equal Ratio Combining 39

2.7 Handover 39

2.7.1 Handover Types 41 2.7.2 MDHO Algorithm 46 2.7.3 MDHO Procedures 48 2.7.4 Comparison between the Handover Techniques in

Multihop Cellular Networks 51

2.8 Related Studies on Handover in Multihop Cellular

Networks

54

2.9 Summary 62

CHAPTER III ANALYTICAL AND SIMULATION MODELS

3.1 Introduction 66

3.2 Theoretical and Conceptual Background 67

3.2.1 Hard Handover (HHO) 67 3.2.2 Fast Access Station Switching (FASS) 67 3.2.3 Macro Diversity Handover (MDHO) 68

3.3 Baseband Channel and Signal Models 73

3.3.1 Baseband Channel 73 3.3.2 Case 1 of the Conventional MDHO 74 3.3.3 Case 1 of the Proposed MDHO 75 3.3.4 Case 2 of the Proposed MDHO and the

Conventional MDHO 77

3.3.5 Fast Access Station Switching 79 3.3.6 Hard Handover 79

3.4 Derivation and Analysis of the Bit Error Rate 79

3.4.1 Derivation of the Bit Error Rate for the Proposed MDHO

81

3.4.2 Probability of Diversity Error for the Proposed MDHO

81

3.4.3 Probability of Error Propagation for the Proposed MDHO

83

3.4.4 Derivation of the Bit Error Rate for the Conventional MDHO

86

ix

3.4.5 Probability of Diversity Error for the Conventional MDHO

86

3.4.6 Probability of Error Propagation for the Conventional MDHO

86

3.5 Simulation Model 87

3.5.1 Network Model 88 3.5.2 Adaptive Modulation and Coding (AMC) 94 3.5.3 MDHO Algorithm 94 3.5.4 Propagation Model 96 3.5.5 Interference Model 98 3.5.6 Simulation Output 104 3.5.7 Simulation Flowchart 108

3.6 Implementation Aspects for the Proposed DL MDHO 112

3.7 Summary 113

CHAPTER IV MDHO DOWNLINK PERFORMANCE

4.1 Introduction 115

4.2 Analytical Results for the Average DL SINR for the

Proposed MDHO and the Conventional MDHO

115

4.3 Analytical Results for the Average DL e2e BER for the

Proposed MDHO and the Conventional MDHO

119

4.4 DL Simulation Results and Discussions 126

4.4.1 The Effect of the MS Mobility Speed on the Performance of the Various Handover Techniques

127

4.4.2 The Impact of the RS Transmitted Power on the Performance of the Various Handover Techniques

135

4.4.3 The Impact of the RS Location on the Performance of the Various Handover Techniques

140

4.5 Summary 145

CHAPTER V MDHO UPLINK PERFORMANCE

5.1 Introduction 148

5.2 The UL MDHO Schemes 149

5.2.1 Conventional SINR-based SC Scheme 149 5.2.2 Joint MRC-SC Scheme 149 5.2.3 End-to-End Throughput-Based SC Scheme 152 5.2.4 End-to-End BER-based SC Scheme 156

5.3 Results and Discussions 158

x

5.3.1 Performance Evaluation of the Proposed Joint MRC-SC Scheme

159

5.3.2 Performance Evaluation of the Proposed ETSC Scheme with Power Control at the RS

160

5.3.3 Performance Evaluation of the Proposed EBSC Scheme

169

5.4 Summary 175

CHAPTER VI CONCLUSIONS AND FUTURE WORK

6.1 Conclusions and Research Findings 178

6.1.1 DL MDHO Technique 179 6.1.2 UL MDHO Technique 181

6.2 Future Work 183

REFERENCES 186

LIST OF PUBLICATIONS 199

xi

LIST OF TABLES

Table No. Page

2.1 Comparison between EV-DO, HSPA, 3GPP-LTE, IMT-Advanced, IEEE 802.16m and mobile WiMAX

21

2.2 Brief comparison of the various handover techniques 54

2.3 Summary of characteristics of handover techniques in multihop cellular networks

64

3.1 Description of intra-cell and inter-cell scenarios of MDHO technique

69

3.2 OFDMA parameters 92

3.3 Simulation parameters 93

3.4 MCSs’ parameters in AMC 95

4.1 Summary of performance analysis of the average DL e2e BER

126

4.2 Maximum performance gains achieved by the proposed DL MDHO studied at different RS transmitted powers

139

4.3 Maximum performance gains achieved by the proposed DL MDHO investigated at different relative RS locations

145

xii

LIST OF FIGURES

Figure No. Page

1.1 Macro Diversity Handover 3

2.1 Evolutionary path of cellular technology 14

2.2 Heterogeneous network with interworking access systems for next generation

16

2.3 IEEE 802.16 relevant standards evolution 18

2.4 Different usage models of mobile WiMAX in the same network

20

2.5 OFDMA sub-carrier structure 22

2.6 DL PUSC sub-carrier permutation scheme 23

2.7 Usage scenarios for the fixed, nomadic and mobile relay stations

26

2.8 The mechanisms of radio wave propagation 29

2.9 Path loss, shadowing and multipath effects versus distance 29

2.10 Hard handover scenario 42

2.11 Fast access station switching with diversity set size of 4 43

2.12 Comparison between HHO and MDHO 45

2.13 MDHO algorithm 47

2.14 Timing diagram of MAC management messages for the MDHO scenario in which the diversity set members are two RSs in two different cells

51

3.1 MDHO scenarios in multihop cellular networks 68

3.2 Transmission sequence of case 1 of the proposed MDHO 70

3.3 Transmission sequence of case 1 of the conventional MDHO 70

3.4 Transmission sequence of case 2 of the conventional MDHO and the proposed MDHO when the diversity set members are RS1 and RS2

71

3.5 Transmission sequence of case 2 of the conventional MDHO and the proposed MDHO when the diversity set members are BS1 and BS2

72

3.6 Comparison between the conventional MDHO and the proposed MDHO techniques. A → B denotes data communications between terminals A and B

72

3.7 The positions of the BS and the RSs in one cell 88

xiii

3.8 Simulated network layout for the MDHO DL performance 89

3.9 Mobility model for macrocellular environment 90

3.10 Antenna pattern for the 6-sector antenna used for each RS 91

3.11 Interference received by cell-interior user MS1 from the neighboring cells when MS1 is in MDHO and the diversity set members are BS1 and RS3

101

3.12 Interference received by cell-interior user MS1 from the neighboring cells when MS1 is in FASS and the anchor station is BS1

102

3.13 Interference received by cell-interior user MS1 from the neighboring cells when MS1 is in HHO and the serving station is BS1

102

3.14 Interference received by cell-edge user MS4 from the neighboring cells when MS4 is in MDHO and the diversity set members are RS3 and RS17

103

3.15 Interference received by cell-edge user MS4 from the neighboring cells when MS4 is in FASS and the anchor station is RS3

103

3.16 Interference received by cell-edge user MS4 from the neighboring cells when MS4 is in HHO and the serving station is RS3

104

3.17 Flowchart for the system-level simulation 109

3.18 Flowchart for the MDHO algorithm 111

4.1 Average post-processing DL SINR for the proposed MDHO and the conventional MDHO at =2,SDγ 15 dB and =ρ 0.5 as a function of 2,RDγ

116

4.2 Average post-processing DL SINR for the proposed MDHO and the conventional MDHO at =2,RDγ 15 dB and =ρ 0.5 as a function of 2,SDγ

117

4.3 Average post-processing DL SINR for the proposed MDHO and the conventional MDHO at =1,SDγ 20 dB and =2,RDγ 11 dB as a function of ρ

118

4.4 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 3 dB, =2,RDγ 23 dB and =ρ 0.5 as a function of 2,SDγ

119

4.5

Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 3 dB, =2,RDγ 4 dB and =ρ 0.5 and 0.9 as a function of 2,SDγ

121

xiv

4.6 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 30 dB, =2,RDγ 15 dB and =ρ0.5 as a function of 2,SDγ

121

4.7 Average e2e BER of the proposed MDHO and the conventional MDHO at =2,SDγ 3 dB, =2,RDγ 7 dB and =ρ 0.5 as a function of 1,SRγ

122

4.8 Average e2e BER of the proposed MDHO and the conventional MDHO at =2,SDγ 15 dB, =2,RDγ 14 dB and =ρ0.5 as a function of 1,SRγ

123

4.9 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 30 dB, =2,SDγ 5 dB and =ρ 0.5 as a function of 2,RDγ

124

4.10 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 3 dB, =2,SDγ 15 dB and =ρ 0.5 as a function of 2,RDγ

125

4.11 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 3 dB, =2,SDγ 24 dB and =ρ 0.5 as a function of 2,RDγ

125

4.12 Percentages of users being in case 1 and case 2 of MDHO from the total number of users at different MS speeds

128

4.13 CDF of the average DL SINR at a pedestrian MS speed of 3 km/hr

129

4.14 CDF of the average DL spectral efficiency at a pedestrian MS speed of 3 km/hr

130

4.15 CDF of the average DL SINR at a vehicular MS speed of 120 km/hr

131

4.16 CDF of the average DL spectral efficiency at a vehicular MS speed of 120 km/hr

132

4.17 Outage probability against MS speeds 133

4.18 Selection probability of the different MCSs at MS speed of (a) 3 km/hr (b) 120 km/hr

134

4.19 Percentages of users being in case 1 and case 2 of MDHO at different RS transmitted powers

135

4.20 Average DL SINR as a function of the RS transmitted power 137

4.21 Average DL spectral efficiency at different RS transmitted powers

138

4.22 Outage probability against RS transmitted power 139

xv

4.23 Percentage of users for each MDHO scenario at an RS location of halfway between the BS and the cell boundary

141

4.24 Total MDHO probability as a function of the relative location of RS that is located on the straight line connecting the BS and the cell vertices

141

4.25 Average DL SINR at different relative RS locations 142

4.26 Average DL spectral efficiency as a function of the relative RS location

143

4.27 Outage probability as a function of the relative RS location 144

5.1 UL transmission sequence for the joint MRC-SC scheme when the diversity set members are a BS and an RS within the same cell

150

5.2 UL transmission sequence for the joint MRC-SC scheme when the diversity set members are two RSs within the same cell

151

5.3 Simulated network layout for the MDHO UL performance 158

5.4 Average SINR of the joint MRC-SC scheme and the conventional SSC scheme as a function of 2,RDγ

159

5.5 Average e2e throughput achieved with the ETSC scheme with power control as compared to the other considered UL schemes at =1,SRγ 11 dB and =2,RDγ 30 dB as a function of 1,SDγ . PC denotes power control

161

5.6 Average e2e throughput of the ETSC scheme with power control as compared to the other considered UL schemes at

=1,SRγ 8 dB and =2,RDγ 15 dB as a function of 1,SDγ

162

5.7 Average e2e throughput of the ETSC scheme with power control as compared to the other considered UL schemes at

=1,SRγ 11 dB, =2,RDγ 30 dB and =1,SDγ 20 dB as a function of the difference between 1,SDγ and 2,SDγ

163

5.8 Average e2e throughput of the ETSC scheme with power control as compared to the other considered UL schemes at

=1,SDγ 20 dB and =2,RDγ 30 dB as a function of 1,SRγ

164

5.9 Average e2e throughput of the ETSC scheme with power control as compared to the other considered UL schemes at

=1,SDγ 11 dB and =2,RDγ 30 dB as a function of 1,SRγ

165

5.10 Interference experienced by the MS during the first and second phases when the power control at the RS is not used and used

166

5.11 CDF of the average e2e throughput of the proposed ETSC scheme with power control as compared to the other considered UL schemes

167

xvi

5.12 Average e2e throughput of the proposed ETSC scheme with power control as compared to the other considered UL schemes as a function of tγ

168

5.13 e2e BER for the proposed EBSC scheme, the BSC scheme and the SSC scheme with =1,SDγ 13 dB, =2,RDγ 25 dB, =1M 16-QAM and =2M 64-QAM at different values of 1,SRγ

170

5.14 e2e BER for the proposed EBSC scheme, the BSC scheme and the SSC scheme with =1,SDγ 13 dB, =2,RDγ 17 dB, =1M 16-QAM and =2M 64-QAM at different values of 1,SRγ

171

5.15 e2e BER for the proposed EBSC scheme as compared to the BSC and SSC schemes at =1,SRγ 5 dB, =1,SDγ 12 dB, =1M 16-QAM and =2M 64-QAM as a function of 2,RDγ

172

5.16 e2e BER for the proposed EBSC scheme as compared to the BSC and SCC schemes at =1,SRγ 5 dB, =1,SDγ 12 dB, =1M16-QAM and =2M 16-QAM as a function of 2,RDγ

173

5.17 e2e BER for the proposed EBSC scheme as compared to the BSC and SSC schemes at =1,SRγ 5 dB, =2,RDγ 25 dB, =1M64-QAM and =2M 64-QAM as a function of 1,SDγ

174

5.18 e2e BER for the proposed EBSC scheme as compared to the BSC and SSC schemes at =1,SRγ 5 dB, =2,RDγ 25 dB, =1M16-QAM and =2M 64-QAM as a function of 1,SDγ

175

xvii

LIST OF ABBREVIATIONS

16-QAM 16-Quadrature Amplitude Modulation

1G First Generation

1xEV-DO 1 times Evolution-Data Optimized

2.5G 2.5 Generation

2G Second Generation

3.99G 3.99 Generation

3G Third Generation

3GPP 3G Partnership Project

3GPP2 3G Partnership Project 2

4G Fourth Generation

64-QAM 64-Quadrature Amplitude Modulation

8PSK 8 Phase Shift Keying

AAA Authentication, Authorization, Accounting

AdDF Adaptive Decode and Forward

AF Amplify-and-Forward

AMC Adaptive Modulation and Coding

AMPS Advanced Mobile Phone System

AS Access Station

AWGN Additive White Gaussian Noise

B3G Beyond Third Generation

BER Bit Error Rate

BLER Block Error Rate

BPSK Binary Phase Shift Keying

BRAN Broadband Radio Access Network

BS Base Station

BSC BER-based SC

BSID BS Identifier

CDD Cyclic Delay Diversity

CDF Cumulative Distribution Function

CDM Code Division Multiplexing

CDMA Code Division Multiple Access

xviii

CF Compress-and-Forward

CID Connection Identifier

CMDHO Conventional Macro Diversity Handover

COST-231 Cooperation in the Field of Scientific and Research 231

CPE Customer Premises Equipment

CQICH Channel Quality Indicator Channel

CRC Cyclic Redundancy Check

CSI Channel State Information

DAB Digital Audio Broadcasting

D-AMPS Digital AMPS

DCD Downlink Channel Descriptor

DF Decode-and-Forward

DL Downlink

DL-MAP DL Medium Access Protocol

DSL Digital Subscriber Line

DVB Digital Video Broadcasting

e2e end-to-end

EBSC e2e BER-based Selection Combining

EDGE Enhanced Data Rates for GSM Evolution

EIRP Effective Isotropic Radiated Power

ETSC e2e Throughput-based Selection Combining

FASS Fast Access Station Switching

FCH Frame Control Header

FDD Frequency Division Duplex

FEC Forward Error Correction

FFT Fast Fourier Transform

FRS Fixed Relay Station

FUSC Full Usage of Sub-carriers

GPRS General Packet Radio Service

GSM Global System for Mobile Communications

H-ARQ Hybrid Automatic Repeat Request

HHO Hard Handover

HiperLAN 2 High Performance Radio Local Area Network type 2

xix

HO Handover

HSDPA High Speed Downlink Packet Access

HSPA High Speed Packet Access

HSUPA High Speed Uplink Packet Access

i.i.d. Independent Identically Distributed

ID Identifier

IE Information Element

IM Instant Messaging

IMS Internet-Protocol Multimedia Subsystem

IMT-2000 International Mobile Telecommunications-2000

IP Internet Protocol

IS Interim Standard

ISP Internet Service Provider

ITU International Telecommunication Union

ITU-R ITU-Radiocommunication Sector

LOS Line-of-Sight

LTE Long Term Evolution

MAC Medium Access Control

MCS Modulation and Coding Scheme

MDHO Macro Diversity Handover

MIB Management Information Base

MIH Media Independent Handover

MIMO Multiple-Input Multiple-Output

MISO Multiple-Input Single-Output

MMR-BS Mobile Multihop Relay Base Station

MOB_BSHO-REQ Mobile BS Handover Request

MOB_BSHO-RSP Mobile BS Handover Response

MOB_HO-IND Mobile Handover Indication

MOB_MSHO-REQ Mobile MS Handover Request

MOB_NBR_ADV Mobile Neighbor Advertisement

MOB_SCN-REQ Mobile Scanning interval allocation Request

MOB_SCN-RSP Mobile Scanning interval allocation Response

M-PSK M-ary Phase Shift Keying

xx

M-QAM M-ary QAM

MR Multihop Relay

MR_HO-IND MR Handover Indication

MR_HOINFO_REQ MR Handover Information Request

MR_HOINFO_RSP MR Handover Information Response

MR_MSINFO_REQ MR MS Information Request

MR_MSINFO_RSP MR MS Information Response

MR_SCN-REQ MR Scanning interval allocation Request

MR_SCN-RSP MR Scanning interval allocation Response

MRAN Multihop Radio Access Networks

MRC Maximal-Ratio Combining

MRS Mobile Relay Station

MS Mobile Station

NLOS Non Line-of-Sight

NRS Nomadic Relay Station

OFDM Orthogonal Frequency Division Multiplexing

OFDMA Orthogonal Frequency Division Multiple Access

PAR Project Authorization Request

PC Power Control

PDC Personal Digital Cellular

PDF Probability Density Function

PDU Protocol Data Unit

PHY Physical

PMDHO Proposed MDHO

PTT Push-to-Talk

PUSC Partial Usage of Sub-carriers

QoS Quality of Service

QPSK Quadrature Phase Shift Keying

Rel. Release

Rev. Revision

RNG_REQ Ranging Request

RNG_RSP Ranging Response

RS Relay Station

xxi

RSID RS Identifier

SC Selection Combining

SC-FDMA Single Carrier Frequency Division Multiple Access

SFID Service Flow Identifiers

SIMO Single-Input Multiple-Output

SINR Signal-to-Interference-and-Noise Ratio

SIP Session Initiation Protocol

SM Spatial Multiplexing

SNR Signal-to-Noise Ratio

SOFDMA Scalable OFDMA

SoHo Small Office/Home Office

SSC SINR-based Selection Combining

TDD Time Division Duplex

TDM Time Division Multiplexing

TDMA Time Division Multiple Access

TLV Type/Length/Value

UCD Uplink Channel Descriptor

UL Uplink

UL-MAP UL Medium Access Protocol

UMB Ultra Mobile Broadband

UMD Ultra Mobile Device

UMTS Universal Mobile Telecommunications System

UTRA UMTS Terrestrial Radio Access

UTRAN UTRA Network

VLAN Virtual Local Area Network

VoD Video on Demand

VoIP Voice-over-IP

WAN Wide Area Network

WCDMA Wideband CDMA

WiFi Wireless Fidelity

WiMAX Worldwide Interoperability for Microwave Access

WINNER Wireless World Initiative New Radio

WLAN Wireless Local Area Network

xxii

WMAN Wireless Metropolitan Area Network

WPAN Wireless Personal Area Network

xxiii

LIST OF SYMBOLS

PL Path loss

Rd Distance between the transmitter and the receiver

cf Carrier frequency

α Path loss exponent

λ Wavelength

fPLΔ Frequency correction factor

htPLΔ RS height correction factor

a A vector

A A matrix

][⋅ Expectation operator

T][⋅ Transpose operation *][⋅ Conjugate operation

2|||| F⋅ Squared Frobenius norm

( )⋅tr Trace of matrix

S Source terminal

R Relay terminal

D Destination terminal j Phase index

jx complex-valued constellation points transmitted by the terminal during phase j

y Received signal

SP Source transmit power

RP DL relay transmit power

jSRh , Channel coefficient of the S → R link

jSDh , Channel coefficient of the S → D link

jRDh , Channel coefficient of the R → D link

2, jSRσ Variance of the channel coefficient of the S → R link

2, jSDσ Variance of the channel coefficient of the S → D link

xxiv

2, jRDσ Variance of the channel coefficient of the R → D link

jDn , Noise-plus-interference component at the destination

jRn , Noise-plus-interference component at the relay

jDI , Variance of the noise-plus-interference at the destination

jRI , Variance of the noise-plus-interference at the relay

γ Instantaneous SINR γ Average SINR

( )γf Probability density function of γ

ρ Interference ratio

( )⋅Q Q Function

e Error event

SRe Error event in the S → R link

SDe Error event in the S → D link

RDe Error event in the R → D link

prope Error propagation event

dive Diversity error event

)|( γe Probability of error event e conditioned on the instantaneous SINR

)(e Probability of error event e conditioned on the average SINR

Mα , Mβ Modulation dependent parameters for BER expressions

exp Exponential function 1

,2CMDHO

avgeeBER Average DL end-to-end BER for case 1 of the conventional MDHO

1,2

PMDHOavgeeBER Average DL end-to-end BER for case 1 of the proposed MDHO

γψ Characteristic function of a random variable γ

rd Cell radius

SRd Distance between the BS and the RS

( )θA Antenna gain in the direction of θ

θ Angle between the direction of interest and the steering direction of the antenna

dB3θ The 3 dB beamwidth for the directional antenna

xxv

mA Maximum attenuation for the directional antenna

cR Code rate

σX Shadow fading with standard deviation of σ

δ Shadow fading correlation coefficient between two locations

dΔ Distance moved between two adjacent locations

cord Shadow fading de-correlation distance

ε Probability that subcarrier is allocated to MS served by BS η Probability that subcarrier is allocated to MS served by RS

ξ Probability that subcarrier is allocated to MS served by BS and RS

ϕ Probability that subcarrier is allocated to MS served by two RSs

u Index of user terminals

k Index of sub-carriers

iΦ Set of the interfering cells

DSΦ Set of the cells of the diversity set access stations

ukjiI ,

, Average total interference of each sub-carrier k for user u being in handover technique i during phase j

NP Noise power

bK Boltzmann’s constant

T System temperature

W Transmission bandwidth

F Noise figure

recP Received power of the desired or interfering signal

serP Received power of the desired signal

tP Transmitted power of the desired or interfering signal

tG Gain of the transmit antenna

rG Gain of the receive antenna

SE Spectral efficiency

( )γR Nominal rate in bps/Hz for the selected MCS mode based on γ

d Hamming distance of each path from the all-zero path of the state diagram of the convolutional encoder

fd Minimum free distance of the code

xxvi

dP Pairwise error probability

dw Sum of bit errors for error events of distance d

outP Outage probability

ieeThr 2 Average end-to-end throughput for the UL MDHO scheme i

maxRP Maximum UL RS transmitted power min

RP Minimum UL RS transmitted power

tγ Target SINR of the link between the RS and the BS

iinsteeBER ,2 Instantaneous UL end-to-end BER for the UL MDHO scheme i

CHAPTER I

INTRODUCTION

1.1 BACKGROUND

As wireless communication becomes more prevalent, the demand for an ubiquitous

high data rate coverage is increasing. Future wireless systems are provisioned to meet

the increasing demands for higher data rate and better quality of service (QoS) that are

much higher than those of the currently deployed cellular systems. Therefore, there

are excessive expectations put on certain communication resources such as scarce

radio spectrum and link budget. It is necessary for the deployed system architectures

to realize these objectives to be economically feasible to ensure attractive business

opportunities for service providers and equipment manufacturers (Oyman & Laneman

2007).

The recent developments in the fields of antenna technologies (such as smart

antenna and multiple-input multiple-output (MIMO) systems) and advanced coding

and signal processing techniques (such as low-density parity-check codes and

interference-cancellation algorithms) have enabled significant improvements to

system performance. These developments dramatically increase spectral efficiency,

capacity, coverage and reliability, and have become part of the latest wireless

standards, such as long term evolution (LTE)-Advanced, IEEE 802.11 and IEEE

802.16. These technologies alone cannot satisfy future demands of wireless systems

without further fundamental modifications in the wireless network architecture itself

in the way systems are designed and deployed as well as novel signal processing

techniques. Even though increasing the density of base stations (BSs) is a potential

approach to achieve the above mentioned objectives, it is economically not feasible.

2

One of the promising approaches is the integration of multihop capability in the

current wireless networks. This is believed to be the most feasible cost-effective

network architecture towards providing the ubiquitous high data rate coverage. The

cost effectiveness of this approach comes from the fact that the additional

infrastructure devices, namely the relay stations (RSs), do not have any wired

connection to the backhaul but rather transmit and receive in a completely wireless

manner. Due to its cost-effectiveness, there has recently been increasing interest in the

multihop concept to be developed in networks such as beyond third generation (B3G),

fourth generation (4G), IEEE 802.11/WiFi, high performance radio local area network

type 2 (HiperLAN2), IEEE 802.16/WiMAX (worldwide interoperability for

microwave access). (Pabast et al. 2004; Yanikomeroglu 2006; Salem et al. 2010).

However, the limitation in the signal processing hardware prevents the

wireless terminals from transmitting and receiving simultaneously at the same time

and frequency (Laneman et al. 2004). Consequently, the relay communication is

organized in two phases (two time slots). During the first phase the source transmits to

the relay, whereas during the second phase the relay forwards the received

information to the destination.

Handover is an essential component of mobile cellular communication systems

to allow full user mobility in the coverage areas. Mobility causes dynamic variations

in link quality and interference levels in cellular systems, sometimes requires a

particular user to change its serving station. This change is known as a handover. The

main target of handover is to provide continuity in service when a mobile station (MS)

migrates from the air-interface of one BS/RS to the air-interface provided by another

BS/RS. There are several possible causes that might trigger the handover procedure:

the quality of the signal in terms of bit error rate (BER) or block error rate (BLER);

the received signal level; the distance to the serving station; a change of service; or it

may even result from a decision of load balancing strategies (Perez-Romero et al.

2005; Makelainen 2007). Signal-to-interference-and-noise ratio (SINR) is a major and

fast metric of handover target selection in IEEE 802.16j (Shen et al. 2006; Sun et al.

2008) as well as in B3G systems employing orthogonal frequency division multiple

access (OFDMA) (Feng et al. 2005). Three handover methods are supported within

3

the IEEE 802.16j standard, namely hard handover (HHO), fast access station

switching (FASS) and macro diversity handover (MDHO). The implementation of

HHO is mandatory, while FASS and MDHO are two optional modes.

MDHO is the process by which the MS maintains connection with two or more

access stations called a diversity set, as shown in Figure 1.1. The access station can be

a BS (BS1) or an RS (RS1, RS2 and RS3). In the downlink (DL), multiple copies of

the transmitted signal are received at the MS and the diversity combining is

performed. In the uplink (UL), on the other hand, the MS transmits the data to the

diversity set members such that selection diversity is performed to select the best

signal. Compared with the HHO and due to maintaining of the diversity set, MDHO

has the advantages of smoother transmission (there is no break point in connection)

and less ping-pong effects. Apart from handling the continuity of wireless services,

MDHO also brings macro-diversity gain to the systems. MDHO is also used as an

interference-reduction mechanism particularly for those users at the cell edge, which

increase the capacity and coverage.

Figure 1.1 Macro diversity handover

Active RS1

MS1

Active BS1

Area of Neighbor Stations

Diversity Set

Active RS3

Neighbor RS4

Neighbor BS2 Anchor RS2

Uplink and downlink communication Including traffic

Only signal strength measurement No traffic

4

The anchor station (RS2) is the access station where the MS is registered,

synchronized, performs ranging and monitors DL for control information. The active

stations (BS1, RS1 and RS3) are informed of the MS capabilities, security parameters,

service flows and full medium access control (MAC) context information. The term

neighbor stations refers to the access stations that can be reached by the MS but their

signal strengths are not strong enough to be added to the diversity set, and hence they

are kept outside the diversity set. The set of the neighbor stations are also called

monitored set. Generally, the serving station is the BS/RS with which the MS has

most recently completed registration at initial network entry or during handover. The

target station is the BS/RS that the MS intends to be registered with at the end of a

handover (IEEE 2006).

In single-hop cellular networks, handover basically occurs between BSs in

different cells or different sectors of the same cell. In contrast, due to the introduction

of RSs in the cellular network architecture, additional handovers occur between the

BS and the RSs (that can be within the same or different cells) or between RSs (that

can also be within the same or different cells). Hence, different intra-cell and inter-cell

MDHO scenarios occur in multihop cellular networks. Assuming diversity set size of

two, the intra-cell MDHO scenarios occur within the same cell and include the

scenarios in which the diversity set members are a BS and an RS, or two RSs. The

inter-cell MDHO scenarios occur between different cells and include the scenarios in

which the diversity set members are a BS and an RS, two RSs, or two BSs. In this

research, the MDHO scenarios can be further categorized into two cases. The first

case comprises the MDHO scenarios in which the diversity set members are two

different-topology access stations, for instance a BS and an RS, and it is called as case

1. The second case corresponds to the MDHO scenarios in which the diversity set

members are two similar-topology access stations, for instance two RSs or two BSs,

and it is called as case 2.

1.2 MOTIVATIONS AND PROBLEM STATEMENT

Because different combining schemes are employed in the DL and UL and the

different features of interference for the DL and UL, the performance of the MDHO

5

needs to be analyzed separately in each direction. Unfortunately, handover issues in

multihop cellular networks have not been studied extensively in the literature yet.

Even though relay-handover problems are investigated in Yap et al. (2002),

Ghassemian et al. (2005) and Khadivi et al. (2008), the main focus in these works is

on ad hoc networks and not on cellular networks. In multihop cellular networks, on

the other hand, most of the previous literature on handover has been devoted to

evaluate the performance (Nourizadeh et al. 2006; Becvar 2007; Kim et al. 2008;

Chen et al. 2010), to propose the required handover procedure changes due to the RSs’

involvement (Zhao et al. 2008; Sun et al. 2008; Becvar et al. 2008) and to reduce the

overhead, latency and service interruption time (Park et al. 2007; Yang et al. 2008;

Yoo et al. 2009; Cho et al. 2008; Kim & Cho 2009; Cho et al. 2009; Becvar & Mach

2010) of HHO, FASS, and conventional MDHO. In the DL of the conventional

MDHO, the MS receives only the simultaneous transmissions of the diversity set

members whether the diversity set members are a BS and an RS, two RSs, or two BSs.

In fact, in the DL of the conventional MDHO where the diversity set members are two

different-topology access stations, that is a BS and an RS, only RS receives the

transmission of the BS during the first phase. During the second phase, both BS and

RS transmit simultaneously to the MS by using the same radio resource. Hence, the

signal transmitted by the BS in the first phase is not received by the MS even though

the MS is idle in this phase. The topology of the diversity set members is not fully

exploited. This constitutes an inefficient utilization of radio resources and poor

performance of the MSs in the MDHO regions.

However, cooperative diversity has emerged as a new form of diversity in

wireless networks where some terminals serve as relays for another terminal’s

transmission (Sendonaris et al. 2003a; Laneman et al. 2004; Nabar at al. 2004;

Herhold et al. 2005; Onat et al. 2008; Ding et al. 2009; Ikki & Ahmed 2010; Wang et

al. 2011). In Nabar et al. (2004), the destination terminal overhears transmission from

a source to a relay terminal to achieve higher spatial diversity gain or higher spatial

multiplexing gain. In these works, however, the main interest of using the multihop

and diversity concepts is in ad hoc networks and not in cellular networks. To the best

of our knowledge, most of the previous works on the cooperative diversity have been

carried out in a single cell noise-limited or ad hoc topology environment and for fixed

6

users with the diversity set members are always a BS and RSs. Furthermore, since the

cooperative diversity is not limited to the overlapped coverage areas of the access

stations, cooperation may increase the resource consumptions. More detailed on the

cooperative diversity will be discussed in Chapter 2.

Therefore, an efficient topology-aware DL MDHO technique for time division

duplex (TDD)-OFDMA-based interference-limited multihop cellular networks is

essential. The proposed DL MDHO technique should receive all the data signals

transmitted by the diversity set members. It should ensure that the topology of the

diversity set members is always fully exploited. This constitutes an efficient utilization

of the radio resources and is expected to increase the spatial diversity gain and spatial

multiplexing gain, and decrease the BER, which are important for enhancing the

performance of the MSs in the MDHO regions.

On the other hand, in the UL of the conventional MDHO, the selection

combining (SC) among the received signals is performed for both intra-cell and inter-

cell MDHO scenarios (Becvar 2007; Zhao et al. 2008). In the conventional SC

scheme, the link that has the maximum SINR is selected (Simon & Alouini 2000;

IEEE 2006; Sediq & Yanikomeroglu 2009) and is referred to in this thesis as SINR-

based SC (SSC) scheme. On account of the power limitation of MS, the UL is

considered as the limiting link for the performance of the cellular networks. Hence,

ensuring a good UL performance is of utmost importance and very beneficial to the

cellular network. In multihop cellular networks, the SSC scheme does not necessarily

achieve the best performance in terms of the average SINR. This is because in case of

intra-cell MDHO scenarios where the diversity set members are either a BS and an RS

or two RSs, the signals received at the BS can be diversity combined (using maximal-

ratio combining (MRC)) to increase the spatial diversity gain. Therefore, it is essential

to develop an UL MDHO scheme that uses MRC in case of intra-cell MDHO

scenarios, and uses the conventional SSC scheme in case of inter-cell MDHO

scenarios.

In addition, the SSC scheme also may not essentially offer the best

performance in multihop cellular network since SINR does not really capture the

7

actual happening in the data sense such as the end-to-end (e2e) throughput and BER.

The reasons behind the likely inferior e2e throughput performance of the SSC scheme

are explained in the following. For instance, if the SINR of the direct link can support

adequately high modulation and coding scheme (MCS) mode, then direct transmission

might outperform relay based transmission even if the SINR of the relay-to-

destination (R → D) link is higher than that of the direct source-to-destination

(S → D) link. This is due to the facts that the relay based transmission requires

additional radio resources as compared to transmission using the direct S → D link

(Can et al. 2007). Besides, in the UL scenarios of the multihop cellular networks, the

e2e throughput of the source-to-relay (S → R) and R → D links is limited by the

performance of the S → R link. Finally, the resource allocated to the S → D link

during the first phase can also be used to transmit new data during the second phase.

Moreover, in the UL scenarios of the interference-limited environment, the

desired MS signal may be interfered by the transmissions of the co-channel MSs

during the first phase, whereas it may be interfered by the transmissions of the RSs or

MSs during the second phase. During the second phase, however, if the interference

comes from the other cells co-channel RSs, then the desired MS signal may be

seriously degraded by the transmission of the interfering RSs. Hence, UL power

control is needed at the RSs in order to minimize the interference comes from the RSs,

thereby allowing the MS to transmit during the second phase using an MCS with

spectral efficiency that is near or similar to that of the first phase. Therefore, by taking

all the above facts into consideration, an efficient UL scheme that combines the

advantages of the e2e throughput-based selection with the benefits of using the UL

power control at the RSs is essential.

BER-based SC (BSC) scheme is introduced in Sediq and Yanikomeroglu

(2009), as an alternative to the conventional SSC scheme, to be used in cooperative

communications when a relay may use a modulation scheme different than that of the

source. The reason behind their proposal is that when different modulation schemes

are used on each link, the link that has the maximum SINR may not be necessarily the

most reliable link. This is because of the different error-resistance capabilities of the

different modulation schemes. The DL is considered in Sediq and Yanikomeroglu

8

(2009) and the S → R link is assumed to be reliable and error-free. However, in the

UL of the multihop cellular networks, the link between the MS and the RS, namely

S → R link, is normally in non line-of-sight (NLOS) and thus cannot assume to be

reliable and error-free. In fact, even if the BER of the R → D link is lower than that of

the S → D link, the BER of the relay based transmission might still be limited by the

BER at the RS. Consequently, the probability of error at the RS should be taken into

account when using the BER as the selection metric to decide on the appropriate link.

Some of the limitations of the previous studies can be summarized as follows:

(i) In the DL of the conventional MDHO for multihop cellular networks, the

topology of the diversity set members is not fully exploited since the MS

receives only the simultaneous transmissions of the diversity set members.

(ii) The environment in the cooperative diversity is limited to single-cell noise-

limited or ad hoc topology and for fixed users with the diversity set members

are always a BS and RSs. In addition, cooperation may increase the resource

consumptions because the cooperative diversity is not limited to the

overlapped coverage areas of the access stations.

(iii) Inefficient conventional UL MDHO schemes for multihop cellular networks

and hence new efficient UL MDHO schemes need to be proposed.

1.3 OBJECTIVES AND SCOPE OF THE RESEARCH

The main objective of this research is to develop efficient MDHO techniques for

TDD-OFDMA-based interference-limited multihop cellular networks. The specific

objectives of this research are:

(i) To propose a topology-aware DL MDHO technique for TDD-OFDMA-based

interference-limited multihop cellular networks.

(ii) To formulate and derive the average post-processing SINR and the average

e2e BER for the proposed DL MDHO technique.

9

(iii) To propose efficient UL schemes for MDHO technique of TDD-OFDMA-

based interference-limited multihop cellular networks.

The proposed DL MDHO technique should receive all the data signals

transmitted by the diversity set members. It should also ensure that the topology of the

diversity set members is always fully exploited. In contrast, the proposed UL MDHO

schemes should be more efficient than the conventional UL schemes. However, the

DL and UL MDHO schemes to be developed in this thesis are applied for the IEEE

802.16j multihop cellular networks even though the developed schemes can be applied

for the multihop cellular networks in general. The developed DL MDHO technique

will be validated by analytical and simulation results. On the other hand, the

developed UL MDHO schemes will be validated by analytical and simulation results

except for the e2e BER-based SC (EBSC) scheme which will be validated by

analytical results only. The work in this research assumes transparent RSs operated on

centralized scheduling mode where the MSs are within the coverage area of the BS

and can receive BS’ control information. In addition, the work in this research can

also be applied to the non-transparent RSs operated on centralized scheduling mode.

1.4 THESIS CONTRIBUTIONS

The key contributions of this research are the following:

(i) Proposing a new efficient topology-aware DL MDHO technique for TDD-

OFDMA-based interference-limited multihop cellular networks. The proposed

MDHO receives all the data signals transmitted by the diversity set members

whether the diversity set members are a BS and an RS, two RSs, or two BSs. It

ensures that the topology of the diversity set members is always fully

exploited. In the proposed MDHO and whenever the diversity set members of

the MS are two different-topology access stations, that is a BS and an RS, the

MS receives the signal transmitted by the BS during the first phase in addition

to the simultaneous transmissions of the BS and the RS during the second

phase. On the other hand, when the diversity set members of the MS are two

similar-topology access stations, that is two RSs or two BSs, the proposed

10

MDHO performs similarly to the conventional MDHO where only the

simultaneous transmissions of the diversity set members are received by the

MS.

(ii) Formulating and deriving the average post-processing DL SINR for the

proposed DL MDHO technique. The derived equations express the average

DL SINR as a function of the average SINRs of the S → D links and R → D

link as well as the interference ratio (the ratio of the noise-plus-interference

variance during the first phase to the noise-plus-interference variance during

the second phase). Therefore, the derived equations can be directly used to

study the effect of these different parameters on the average post-processing

DL SINR. The average DL SINR for the proposed DL MDHO in the noise-

limited environment can be obtained from the derived equations for the

interference-limited environment. Furthermore, the average DL SINRs of the

conventional MDHO, FASS and HHO can be obtained from the derived SINR

of the proposed MDHO.

(iii) Formulating and deriving closed-form expressions for the average DL e2e

BER for the proposed DL MDHO. The derived equations express the average

DL e2e BER as a function of the average SINRs of the S → R link, S → D

links and the R → D link as well as the interference ratio, and therefore can be

directly used to investigate the effect of these different parameters on the

average DL e2e BER. In order to assure the accuracy of the derived equations,

the BER performance curves predicted by the derived equations are compared

to those predicted by the Monte Carlo simulations.

(iv) Developing the simulation model using MATLAB platform that is used to

investigate the impact of the MS mobility speed, the RS transmitted power and

the relative RS location on the performance of the proposed DL MDHO. The

performance evaluation is carried out in multi-cell interference-limited

environments and for users with high speeds.

11

(v) Proposing a new UL MDHO scheme that uses MRC in case of intra-cell

MDHO scenarios and uses the conventional SSC scheme in case of inter-cell

MDHO scenarios. The intra-cell MDHO scenarios comprise the scenarios in

which the diversity set members are a BS and an RS, or two RSs within the

same cell. This scheme is referred to as joint MRC-SC scheme.

(vi) Proposing a new efficient UL MDHO scheme that combines the advantages of

the e2e throughput-based SC (ETSC) with the benefits of using UL power

control at the RSs.

(vii) Proposing a new efficient UL EBSC scheme which takes the probability of

error at the RS into account when using the BER as the selection metric to

decide on the appropriate diversity branch.

1.5 THESIS ORGANIZATION

The rest of this thesis is organized as follows. Chapter 2 provides a brief background

on the evolution of the wireless access networks towards 4G. The multihop cellular

networks and some relaying concepts are described. A brief overview of the wireless

channels, diversity techniques and diversity combining techniques are also introduced.

The handover triggering parameters, handover types, MDHO algorithm and MDHO

procedures are then described. After that, a comprehensive review on the literature of

the handover in multihop relay (MR) cellular networks and their relation to the work

presented in this research are presented. Chapter 3 presents the DL analytical and

simulation models. In the analytical model, the average DL SINR is derived and the

input-output relations and the MCS selection criterion are described for each of the

handover techniques. The average DL e2e BER for the proposed MDHO and the

conventional MDHO are also derived. The simulation model developed for the

performance evaluation of the various handover techniques of IEE 802.16j multihop

cellular networks is described in details. The implementation aspects for the proposed

DL MDHO technique in the IEEE 802.16j multihop cellular networks are also

described. Chapter 4 illustrates the DL analytical and simulation results based on the

developed models in Chapter 3. The first part of Chapter 4 illustrates and discusses the

12

DL analytical results for the average post-processing DL SINR and the average DL

e2e BER of the proposed MDHO and the conventional MDHO. The BER

performance results predicted by the developed analytical model are compared to

those predicted by the Monte Carlo simulation to assure the validity of the developed

model. The second part of Chapter 4 presents and discusses the DL simulation results

for performance evaluation of the proposed MDHO, conventional MDHO, FASS, and

HHO techniques. In the first part of Chapter 5, the conventional and proposed UL

schemes are described. The selection criterion, the average SINR, the average e2e

throughput and/or the e2e BER are also presented for each of the UL schemes. In the

second part of Chapter 5, the UL results for the performance evaluation and

comparison of the proposed schemes and the conventional schemes are illustrated and

discussed. Finally, Chapter 6 gives the conclusions drawn from the work presented in

this thesis. This chapter summarizes the main results obtained in this thesis and

suggests the potential directions for future work.

CHAPTER II

LITERATURE REVIEW

2.1 INTRODUCTION

In this chapter, background information on the evolution of the wireless access

networks towards 4G is first presented. The multihop cellular networks and some

relaying concepts are described. A brief review of the wireless channels and their

impairments on the data signal transmissions are also introduced. Diversity techniques

used to combat the effects of fading and diversity combining techniques used to

improve the received SINRs are briefly discussed. The handover triggering

parameters, handover types, MDHO algorithm and MDHO procedures are then

described. Finally, the literature on the handover in the multihop cellular networks and

their relation to this research are comprehensively reviewed.

2.2 EVOLUTION OF WIRELESS ACCESS NETWORKS TOWARDS FOURTH GENERATION

The rapid growth of wireless communication and its pervasive use in all walks of life

are changing the way we communicate in all fundamental ways. It is one of the most

vibrant areas in the communication field today (Prasad & Velez 2010). Evolution of

wireless access technologies is about to reach its 4G. Looking past, wireless access

technologies have followed different evolutionary paths aimed at unified target:

performance and efficiency in high mobile environment (Ergen 2009). Figure 2.1

illustrates the evolutionary path of cellular technologies towards 4G. In the following,

a brief overview on each generation is described.

14

Figure 2.1 Evolutionary path of cellular technology

Source: Ergen 2009

The first generations (1G) used the analog transmissions and were introduced

to fulfill the basic mobile voice. The most notable example of such standards is the

Advanced Mobile Phone System (AMPS) operated in the United States. True wireless

communications have gained a momentum with the worldwide success of the second

generation (2G) that is based on digital cellular technology. Low bit rates data services

up to 14.4 kbps were supported as well as the traditional voice services. There are four

main 2G standards: Global System for Mobile Communications (GSM), time division

multiple access (TDMA) or Digital AMPS (D-AMPS), code division multiple access

(CDMA)-based Interim Standard-95 (IS-95) and Personal Digital Cellular (PDC). The

2G has evolved to offer packet-switched data services with higher bit rates up to

384 kbps, and all advanced upgrades for the 2G systems are commonly referred to as

2.5 generation (2.5G). General Packet Radio Service (GPRS) and Enhanced Data

Rates for GSM Evolution (EDGE) are evolutions for GSM systems, whereas IS-95B

is an evolution for CDMA-based IS-95 systems (Garg 2002).

The process of defining the standard for third generation (3G) systems was

started by the International Telecommunication Union (ITU) and referred to as

International Mobile Telecommunications-2000 (IMT-2000). The 3G standard has

aimed for data at higher speeds up to 2 Mbps to open the ways for truly “mobile

broadband” experience. The two significant 3G standards are the wideband CDMA

(WCDMA) managed by the 3G Partnership Project (3GPP), also referred to as

15

Universal Mobile Telecommunications System (UMTS) or UMTS Terrestrial Radio

Access (UTRA), and the CDMA2000 introduced by the 3GPP2 (Holma & Toskala

2004). Extensions to both WCDMA and CDMA2000 have been defined by the 3GPP

and 3GPP2 with the objective of adding network capacity and features to enable

operators to offer new data-oriented services over their existing networks. The

extensions to the CDMA2000 include 1xEV-DO (Evolution-Data Optimized)

Revision-0 (Rev-0), 1xEV-DO Rev-A and 1xEV-DO Rev-B (3GPP2 2001; 3GPP2

2006; 3GPP2 2009). On the other hand, the extensions to the WCDMA include

Release-5 (Rel-5) High Speed Downlink Packet Access (HSDPA) and Rel-6 High

Speed Uplink Packet Access (HSUPA) (3GPP 2010a; 3GPP 2010b). The combination

of HSDPA and HSUPA is called HSPA. Rel-7 with new enhancements and evolutions

to HSPA is also introduced by 3GPP and is commonly known as HSPA+. The 3GPP

envisions additional Rel-8 long-term WCDMA enhancements leading to UTRA

network (UTRAN) LTE known as 3GPP-LTE, 3.99G or evolved UTRAN. The

objectives of the LTE are to constitute a refactoring of UMTS as an entirely Internet

protocol (IP)-based 4G network and to enable to support a peak data rate of 100 Mbps

in the DL and 50 Mbps in the UL with spectral efficiency that is three to four times

that of Rel-6 HSPA (3GPP 2010c). However, the work on developing the Ultra

Mobile Broadband (UMB), which is the planned 3GPP2 solution for the 4G radio

access technology, was discontinued in November 2008 in favor of LTE (Agilent

Technologies 2009).

Meanwhile, the ITU proposed a new concept called IMT-Advanced, which is

responsible to set the vision of the characteristics of the 4G mobile networks.

Although there is no clear definition as of now, the ITU-Radiocommunication Sector

(ITU-R) M.1645 foresees 4G as a radio interface(s) that need to support up to

100 Mbps for high mobility and up to 1 Gbps for low mobility (ITU-R 2003). The

future infrastructures of 4G will comprise a set of heterogeneous networks using IP as

a common protocol and seamless interworking will be required between them. Figure

2.2 shows an example of the concept of All-IP heterogeneous networks. In the All-IP

heterogeneous 4G networks, the following different technologies might coexist (Glisic

2006):

(i) Cellular networks that include 2G, 2.5G, 3G and B3G.

16

(ii) Broadband radio access networks (BRANs)/HiperLAN2 and wireless local

area networks (WLANs)/IEEE 802.11x.

(iii) Wireless personal area networks (WPANs)/IEEE 802.15.x and wireless

metropolitan area networks (WMANs)/IEEE 802.16x/WiMAX.

(iv) Digital audio broadcasting (DAB), digital video broadcasting (DVB) and

satellite communications.

(v) Ad hoc and sensor networks with emerging applications.

Figure 2.2 Heterogeneous network with interworking access systems for next generation

Source: Park & Adachi 2007

2.2.1 Mobile WiMAX

The growing demand for mobile Internet and wireless multimedia applications and

services has created new interest among existing and emerging operators to develop

new broadband wireless access technologies and network architectures to offer such

services at low cost to operators and end users (Etemad 2008). Although the existing

Billing

Heterogeneous Inter-Working - User convenient network - User service continuity

A single user equipment - Multi-functions - Portable

Convergence over IP - Wireless and Wired - Mobile & Broadcasting

VoD VoIP/SIP Proxoes AAA

IM PTT

Signaling Gateway

ISP

Context-aware Information Centre

SoHo

Backhauls

WCDMA CDMA2000

HSPA 1x EV-DO

IEEE 802.16e

WMAN

Enterprise

17

WLAN and the 3G technologies have successfully provided broadband access for the

last several years, they have their specific disadvantages, inhibiting their full-fledged

growth. The short range and the restricted scalability are the main shortcomings of

WLANs. On the other hand, the 3G systems have such limitations as low bandwidth

and high infrastructural cost. The development of the recent IEEE 802.16-based

WiMAX for WMAN has filled this gap between LAN and wide area network (WAN)

technologies. Developed as a truly broadband access solution, the WiMAX

technology offers promising features in terms of high bandwidth, extended coverage

area and low cost. This has led to its rapid rise as one of the most popular last mile

broadband access technologies and a likely component in future 4G networks. The

WiMAX Forum is an industry-led consortium formed to promote the IEEE 802.16

family of standards for broadband wireless access systems. It develops the end-to-end

WiMAX network architecture and protocols and decides on the commonly agreed

system profile to certify and promote the compatibility and interoperability among

equipment vendors and system operators (WiMAX Forum 2010; Wang et al. 2008;

WiMAX Forum 2008b). The evolution of the relevant IEEE 802.16 standards is

illustrated in Figure 2.3.

The fixed WiMAX is based on the IEEE 802.16-2004 (IEEE 2004) orthogonal

frequency division multiplexing (OFDM) air interface and provides fixed broadband

access from anywhere within a metropolitan area networks. On the other hand, the

mobile WiMAX was the first mobile broadband wireless access solution based on the

IEEE 802.16e standard (IEEE 2006) that enabled convergence of mobile and fixed

broadband networks through a common wide area radio access technology and

flexible network architecture. The mobile WiMAX air interface adopted scalable

OFDMA (SOFDMA) as the preferred multiple access method in the DL and UL for

improved multipath performance and bandwidth scalability (Yagoobi 2004). Since the

mobile WiMAX has evolved from fixed wireless access, it inherits its features for

optimized broadband data services.

The mobile WiMAX provides higher data rates with OFDMA support and

introduces several salient features required for delivering mobility at vehicular speeds

with QoS comparable to broadband access alternatives. Several features that are

18

employed to improve data throughput are common to EV-DO and HSPA, such as

adaptive modulation and coding (AMC), hybrid automatic repeat request (H-ARQ),

fast scheduling, and bandwidth efficient handover. The main difference is in

duplexing where EV-DO and HSPA are frequency division duplexing (FDD)

operating on a carrier frequency of 2.0 GHz, whereas mobile WiMAX is currently

supporting spectrally-efficient TDD and operating at 2.5 GHz. Additionally, mobile

WiMAX has higher tolerance to multipath and self-interference, has scalable channel

bandwidth and provides orthogonal uplink multiple access with frequency selective

scheduling and fractional frequency reuse. Unlike EV-DO and HSPA, mobile

WiMAX is also capable of utilizing 2×2 MIMO in addition to 1×2 single-input

multiple-output (SIMO) (WiMAX Forum 2006a).

Figure 2.3 IEEE 802.16 relevant standards evolution

Source: Puthenkulam 2010

As previously shown in Figure 2.1, 3GPP-LTE can be considered as the

strongest potential competitor to the mobile WiMAX technology for mobile

broadband access. In fact, all mobile broadband access technologies that have been

developed lately exploit, enhance, and expand fundamental concepts that were

802.16-2004 Base Standard Fixed Application Licensed/Unlicensed Non-LOS: < 10 GHz & LOS: 10-66 GHz

802.16e Fixed and Mobile Licensed: Non-LOS: < 6 GHz

802.16 /Corrigendum 2

802.16f Fixed MIBs

802.16i Mobile MIBs

802.16g Fixed and Mobile management

802.16-2009 Revised Base Standard (Consolidated book on 802.16-2004, 802.16e, 802.16f,

802.16/Cor2, 802.16g, 802.16i)

5.8 GHz Non-exclusive licensed bands (US: FCC 3650-3700 MHz)

802.16h License Exempt Co-existence Procedures

(Unlicensed)

802.16j MMR Mobile Multihop Relay

802.16m Advanced Air Interface

(IMT-Advanced)

OFDM – 256 Fixed WiMAX July 2004

OFDMA (512, 1024) Mobile WiMAX

Dec 2005 Q3 2007 June 2010

Dec 2005 Q1 2008 May 2009

Sep 2007

May 2009

IEEE Approved IEEE Draft Standard Merged

802.16n Higher Reliability Networks

802.16p Machine to Machine Communication

19

originally utilized in mobile WiMAX. The LTE is being designed with the same

OFDMA air interface as mobile WiMAX. OFDMA selection is driven by the demand

for higher spectral efficiency and low cost per bit. This is because the basic problem

for a service provider is to deliver more data to users, quicker and cheaper. OFDMA is

also selected because WCDMA has a constraint to scale in bandwidth. OFDMA

resolves this problem by dividing the high speed input data stream into several lower

speed data streams and transmitting the lower speed streams on individual frequency

channels. In the receiver side, the user recombines these lower streams to construct a

high speed data stream (Nee & Prasad 2000). In addition to OFDMA technology, both

WiMAX and LTE are IP-based services and are not backward compatible with circuit-

switched services. This is another real breakthrough in technologies when moving

towards 4G since it gives a significant advantage to technologies that are coming out

of blue like WiMAX. This has resulted in the rapid development and deployment of a

large number of operator-managed as well as new open Internet applications, such as,

email, messaging, gaming and content distribution services (Park & Adachi 2007;

Etemad 2008). The major drawback of LTE in comparison to mobile WiMAX is its

delayed commercialization which will only be available in 4-5 years time (Prasad &

Velez 2010). Hence, WiMAX is seen as the only player that can offer a unified fixed-

mobile solution in broadband wireless as well as mobile broadband markets.

The WiMAX network can be deployed as a green field network without

support to legacy circuit-switched system or as an overlay to existing fixed or mobile

access networks such as 2.5G/3G cellular systems or cable/digital subscriber line

(DSL) networks by supporting different levels of interworking to ensure the continuity

of service. As shown in Figure 2.4, the same WiMAX network can be employed for

different usage models such as wireless backhaul to WiFi hot spots, fixed/nomadic

access to customer premises equipment (CPE) and residential gateways, and mobile

access to notebooks, smart phones, and next-generation WiMAX embedded ultra-

mobile devices (UMD) (Etemad 2008).

The mobile WiMAX based on IEEE 802.16e would not qualify as a 4G IMT-

Advanced standard since its data rates even under ideal conditions are much lower

than those predicted for the 4G systems. However, IEEE 802.16m, which is

20

considered as the next-generation mobile WiMAX technology, is being developed as

an advanced air interface to meet the requirements of the IMT-Advanced for 4G

systems as well as the requirements for the next-generation mobile network operator

(Cudak 2010; Bacioccola et al. 2010). Based on the available bandwidth and multi-

antenna mode, the IEEE 802.16m will be capable of over-the-air data transfer rates in

excess of 1 Gbps and of supporting a wide range of high-quality and high-capacity IP-

based services and applications while maintaining full backward compatibility with

the existing mobile WiMAX systems to preserve investments and continuing to

support first-generation products. It will enable roaming and seamless connectivity

across IMT-Advanced and IMT-2000 systems through the use of appropriate

interworking functions (Ahmadi 2009).

Figure 2.4 Different usage models of mobile WiMAX in the same network

Source: Etemad 2008

Table 2.1 presents a comparison of the mobile WiMAX based on IEEE

802.16e and its evolution IEEE 802.16m with the EV-DO Rev. A, HSDPA/HSUPA

(HSPA), 3GPP-LTE and IMT-Advanced (4G). Mobile WiMAX has clear

performance edge in terms of data rate over HSPA and EV-DO. It is also shown that

3GPP is projecting LTE as being more powerful than the existing versions of mobile

WiMAX. However, LTE will face a strong challenge from the future IEEE 802.16m

which is being currently standardized and can qualify as an IMT-Advanced

technology.

21

Table 2.1 Comparison between EV-DO, HSPA, 3GPP-LTE, IMT-Advanced, IEEE 802.16m and mobile WiMAX

Source: Prasad & Velez 2010; Cudak 2010; Ergen 2009

Feature 3.5 G

3GPP-LTE (3.99G)

IMT-Advanced (4G)

WiMAX EV-DO Rev. A

HSDPA/HSUPA (HSPA) IEEE 802.16e IEEE 802.16m

Duplex scheme

FDD FDD FDD/TDD Not specified TDD (FDD optional)

TDD, FDD, Half-duplex FDD

Access method

DL: TDM UL: CDMA

DL: CDM-TDM UL: CDMA

DL: OFDMA UL: SC-FDMA

DL: OFDMA, (?) UL: (?)

DL: SOFDMA UL: SOFDMA

DL: SOFDMA UL: SOFDMA

Channel bandwidth

1.25 MHz 5 MHz 1.25, 1.6, 2.5, 5, 10, 15, 20 MHz

Up to 100 MHz (with band aggregation)

3.5, 5, 7, 8.75, 10, 20 MHz

5-20 MHz (up to 100 MHz through band aggregation)

Modulation BPSK/QPSK/8PSK/ 16-QAM

BSPK/ QPSK/ 16-QAM

QPSK/ 16-QAM/ 64-QAM

QPSK/ 16-QAM/ 64-QAM

QPSK/ 16-QAM/ 64-QAM

QPSK/ 16-QAM/ 64-QAM

Data rate (max.)

DL:3.1 Mbps UL:1.8 Mpbs

DL: 14 Mbps UL: 5.8 Mbps

DL: 100 Mbps UL: 50 Mbps (20 MHz)

DL: 100 Mbps – 1 Gbps UL: > 50 Mbps

DL: 46 Mbps UL: 14 Mbps (10 MHz TDD)

DL: > 130 Mbps UL: > 56 Mbps (20 MHz)

Mobility/ vehicular speed

Up to 120 km/hr

Up to 120 km/hr Up to 500 km/hr depending on frequency

Up to 350 km/hr Up to 120 km/hr Up to 500 km/hr depending on operating frequency

Transmit diversity and MIMO models

Simple open loop diversity

Simple open & closed loop diversity

Spatial multiplexing (SM), Alamouti, CDD, beamforming DL: 2x2, 3x2, 4x2, 4x4 UL: 1x2, 2x2

Not specified SM, Alamouti, CDD – R 1.5, beamforming, Collaborative UL SM DL: 2x2 UL: 1x2

SM, Alamouti, CDD – R 1.5, beamforming, Collaborative UL SM DL: 2x2, 2x4, 4x2, 4x4, 8x8 UL: 1x2, 1x4, 2x4, 4x4

22

2.2.2 OFDMA Basics

OFDMA is a multiple-access/multiplexing scheme that provides multiplexing

operation of data streams from multiple users onto the DL sub-channels, and UL

multiple access by means of UL sub-channels (WiMAX Forum 2006b). OFDMA

symbol structure consists of three types of sub-carriers as shown in Figure 2.5:

(i) Data sub-carriers for data transmissions

(ii) Pilot sub-carriers for channel estimation and synchronization purposes

(iii) Null sub-carriers for no transmission; used for guard bands and DC carriers

Figure 2.5 OFDMA sub-carrier structure

Source: Andrew et al. 2007

Active (data and pilot) sub-carriers are grouped into subsets of sub-carriers

called sub-channels. Mobile WiMAX based on OFDMA physical (PHY) (Yagoobi

2004) allows sub-channelization in both DL and UL. The minimum frequency-time

resource unit of sub-channelization is one slot, which is equal to 48 data sub-carriers.

Sub-channels may be constituted using either contiguous permutation or diversity

permutation. The contiguous permutation (which is also called band AMC) groups a

block of contiguous sub-carriers to form a sub-channel. The contiguous permutations

include DL AMC and UL AMC, and have the same structure. AMC permutation

exploits multi-user diversity and it is well suited for fixed, portable, and low mobility

environments. In diversity permutations, on the other hand, sub-carriers are distributed

pseudo-randomly across the frequency spectrum to provide frequency diversity and

inter-cell interference averaging, which is particularly useful for mobile applications.

Mobile WiMAX defines several sub-channelization schemes based on distributed sub-

carriers for both the UL and DL, such as DL FUSC (full usage of sub-carriers), DL

Data Sub-carriers Pilot Sub-carriers

Guard Sub-carriers DC Sub-carriers Guard Sub-carriers

23

PUSC (partial usage of sub-carriers), UL PUSC and additional optional permutations

(Andrew et al. 2007; WiMAX Forum 2006b).

(a) Downlink Partial Usage of Sub-carriers (DL PUSC)

In the case of DL PUSC, all the active sub-carriers are first arranged into clusters.

Each cluster consists of 14 adjacent sub-carriers over two OFDMA symbols, as shown

in Figure 2.6. In each cluster, the sub-carriers are divided into 24 data sub-carriers and

4 pilot sub-carriers. The clusters are then renumbered using a pseudo-random

numbering scheme. After renumbering, a re-arranging scheme is used to form six

groups of clusters such that each group is made up of clusters that are distributed

throughout the sub-carrier space. A sub-channel in a group contains 2 clusters and is

made up of 48 data sub-carriers and 8 pilot sub-carriers. The data sub-carriers in each

group are further permutated to generate sub-channels within the group. The data sub-

carriers in the cluster are distributed to multiple sub-channels (Andrew et al. 2007).

Figure 2.6 DL PUSC sub-carrier permutation scheme

Source: Andrew et al. 2007

(b) Uplink Partial Usage of Sub-carriers (UL PUSC)

In UL PUSC, the available sub-carriers are first split into tiles. Each tile consists of 4

sub-carriers over 3 OFDMA symbols and the sub-carriers within a tile are divided into

Tim

e

Frequency

Cluster Cluster

Sub-channel (2 clusters from a group)

Even Symbol

Group 6 Group 1

Odd Symbol

Pilot sub-carrier

24

8 data sub-carriers and 4 pilot sub-carriers. Six tiles, chosen from across the entire

spectrum by means of a re-arranging/permutation scheme, are grouped together to

form a slot. The slot comprises 48 data sub-carriers and 24 pilot sub-carriers in 3

OFDMA symbols (WiMAX Forum 2006b).

2.3 MULTIHOP RELAY NETWORKS

The ubiquitous high data-rate coverage predicted for the future generation of wireless

systems do not seem to be feasible with the conventional cellular architecture. The

envisioned transmission rates, coverage and QoS for these systems are much higher

than those of the conventional cellular systems. It is advantageous for network service

providers to distribute system capacity across the network area, reaching MSs in the

most cost-effective way. In the traditional cellular architecture, increasing the capacity

along with the coverage requires the deployment of a large number of BSs. This

approach is not cost-efficient to network service providers. Even the recent

developments in the fields of antenna technologies and advanced coding and signal

processing techniques alone cannot satisfy future demands of wireless systems.

Hence, fundamental modifications in the current wireless network architecture are

necessary in the way systems are designed and deployed as well as novel signal

processing techniques. One of the promising approaches is relaying technique which

is expected to alleviate this coverage problem because the RS with less functionality

than the BS can forward high data rates to remote areas of the cell while reducing

infrastructure cost. Hence, multihop relaying is a cost-effective approach to extend the

coverage, to significantly enhance the throughput and capacity of cellular networks

and to relax the link budget. The idea is to split the distance between a source and a

destination node into several hops; the nonlinear relation between propagation loss

and distance helps in reducing the end-to-end attenuation and thus in relaxing the link

budgets (Yanikomeroglu 2002; Pabast et al. 2004; Oyman & Laneman 2007).

Towards that end, there has recently been increasing interest in both academia

and industry on developing the multihop relaying and mesh-enabled networks. Among

these research and standardization efforts are IEEE 802.11s/WLANs, IEEE

802.16j/m/WMANs, IEEE 802.15.4/WPANs, HiperLAN2, LTE-Advanced and

25

Wireless World Initiative New Radio (WINNER) project (Pabast et al. 2004;

Yanikomeroglu 2006; Salem et al. 2010). However, the focus of the work in this

thesis is on the multihop relaying of the IEEE 802.16j standard even though it can be

applicable to the multihop cellular networks in general.

The MR study group was formed in July 2005 to evaluate merits of multihop

relaying technologies for future 802.16-based cellular wide area networks. The project

authorization request (PAR) was approved in the March 2006 IEEE standards meeting

to initiate the 802.16j relay task group; and the standard is completed and approved in

May 2009. The 802.16j relay task group specified OFDMA PHY layer and MAC

layer enhancements to the IEEE 802.16e standard for licensed bands to enable the

operation of RSs. The first phase of IEEE 802.16j is confined to infrastructure RSs

that extend the coverage of IEEE 802.16e BSs without impacting the specifications of

the subscriber station. The RSs are fully backward compatible in the sense that they

operate seamlessly with existing IEEE 802.16e subscriber stations. Main technical

issues discussed in the IEEE 802.16j task group include general relay concepts, frame

structures, network entry, security, measurement and reporting, bandwidth request,

construction and transmission of MAC protocol data units (PDUs), routing,

scheduling, interference control, and mobility management (Oyman & Laneman 2007;

IEEE 2009).

2.3.1 Basic Relaying Concepts

(a) Relay Station Type

RSs have been discussed in IEEE 802.16j technical reports as fixed, nomadic or

mobile RSs (Sydir 2006). Fixed RS (FRS) is permanently installed at a fixed location

to improve coverage, capacity, or per user throughput in areas not sufficiently covered

(e.g., indoor, in shadow, tunnels, or underground), or provide access for clusters of

users outside the coverage area of the BS. Nomadic RS (NRS) is intended to

temporarily provide additional coverage or capacity in an area where BSs and/or FRSs

do not provide good coverage or capacity. For instance, the temporary coverage may

be needed in emergency/disaster recovery situations and events such as sporting

26

occasions or fairs, where coverage is needed only for the period of that particular

event. A mobile RS (MRS) is intended to be mounted on a vehicle and connected to a

BS or RS through a wireless link. In this case, the RS provides a fixed access link to

terminals travelling on the mobile vehicle. The different types of RSs and examples of

the most important usage scenarios in which the RSs can be deployed are illustrated in

Figure 2.7.

Figure 2.7 Usage scenarios for the fixed, nomadic and mobile relay stations

Source: Soldani & Dixit 2008

(b) Relaying Schemes

According to the processing at the relay, the wireless relaying schemes can be

classified into three types, namely decode-and-forward (DF), amplify-and-forward

(AF) and compress-and-forward (CF) (Herhold et al. 2005; Kramer et al. 2006). A DF

relaying, sometimes referred to as digital or regenerative relaying, is more robust. It

decodes, re-encodes and modulates the received signal before retransmission. In this

case, the forwarded signal does not contain additional degradation, but the decoding

and re-encoding operations require more processing and add more delay. In addition,

FRS

FRS

FRS

FRS

FRS

FRS

MRS

NRS

Valley between buildings

Emergency/disaster recovery

Coverage extension at cell edge

Coverage on mobile vehicles

Indoor coverage

MMR-BS

Multihop relay for rural areas

Shadows of buildings

Coverage hole

27

if there are decoding errors at the RS, this causes error propagation. However, the

error propagation can be avoided simply by detecting the packets received with errors

via cyclic redundancy check (CRC) or similar measures and forwarding only when the

packets are correctly received. Such a scheme is referred to as simple-adaptive

decode-and-forward (AdDF) based relaying (Herhold et al. 2004; Lin et al. 2005;

Herhold et al. 2005). Other techniques to mitigate the error propagation are called

selection relaying; an example is the selection relaying based on the signal-to-noise

ratio (SNR) of the S → R link where the RS uses a threshold to decide when to

retransmit, and it retransmits only if the SNR of the S → R link is above this threshold

(Onat et al. 2008).

On the other hand, an AF relaying, sometimes referred to as analog or non-

regenerative relaying, is a less complex system that just amplifies and retransmits the

received signal without performing any decoding. AF relays have the advantage of

introducing a minimum delay but have the drawback of amplifying the noise. In

addition, AF relays may need higher transmission overhead for transmitting the S → R

channel information to the destination.

A CF relaying, also known as estimate-and-forward or quantize-and-forward,

encodes (using source coding) a quantized version of the received signal and transmits

it to the destination. In this case, the forwarded signal contains possible estimation

errors. The destination uses the relay estimation as side information when decoding

the actual direct link signal. In this thesis, the DF relaying is considered because it is

more viable with respect to implementation and it is the focus of most of the next

generation wireless networks standards, such as IEEE 802.16j/m and LTE-Advanced

(Pabast et al. 2004; Wang et al. 2007; Upase & Hunukumbure 2008; Sediq 2008).

(c) Full-Duplex and Half-Duplex

The RS can operate in full-duplex or half-duplex mode. In full-duplex mode, the RS

can transmit and receive simultaneously at the same time on the same frequency band.

However, the limitations in the radio implementation preclude the terminals from full-

duplex operation mode. Due to severe attenuation over the wireless channel and

28

inadequate electrical isolation between transmit and receive circuitry, the terminal’s

strong transmitted signal drowns out the weak signals at its receiver input. Therefore,

in the near future, RSs are expected to operate in half-duplex mode only. The half-

duplexity constraint requires the use of orthogonal channels for transmission and

reception. For instance, the relay can use different time slots or different frequency

bands to receive and transmit (Laneman et al. 2004; Wang et al. 2007). It is assumed

throughout this thesis that the orthogonality is maintained in the time domain using

TDMA, and the relay communication occurs in two phases (namely two time slots). In

the first phase, the source transmits to the relay, whereas in the second phase, the relay

forwards the received information to the destination. With TDMA, relaying can be

easily integrated to wireless networks.

2.4 WIRELESS RADIO CHANNEL

In wireless channels, the transmitted signal is assumed to arrive at the receiver after

propagating through several different paths. The signal propagates from the

transmitter to the receiver through different mechanisms such as reflection, diffraction

and scattering (Vanghan & Andersen 2003; Oestges & Clerckx 2007), as shown in

Figure 2.8. The existence of these paths results in receiving several versions of the

same transmitted signal. At the receiver, all the received signal versions are added up

together constructively or destructively resulting in a fluctuating received signal. This

phenomenon is referred to as multipath. The power of the transmitted signal drops off

due to three effects: path loss, long-term fading, also referred to as macroscopic fading

or shadowing, and short-term fading, also known as microscopic fading or multipath

fading. Figure 2.9 shows the received signal power against the separation distance

between the transmitter and the receiver. The path loss and shadowing determine the

average operating SINR, whereas the path loss, shadowing and multipath fading

determine the instantaneous SINR.

29

Figure 2.8 The mechanisms of radio wave propagation

Source: Oestges & Clerckx 2007

Figure 2.9 Path loss, shadowing and multipath effects versus distance

Source: Andrews et al. 2007

receiver scattering

transmitter

line-of-sight diffraction

specular reflection

Shadowing + Path loss

Includes multipath fading around shadowing + Path loss

Path loss

Transmit – receive separation distance

Rec

eive

d po

wer

(dB

m)

30

2.4.1 Path Loss

Path loss is the phenomenon of decreasing the received signal power with the

separation distance between the transmitter and the receiver. Path loss arises from the

inverse square law power loss, absorption by water and other objects and also the

effect of ground reflections. Several path loss models are described in Sizun (2005)

that provides theoretical models, empirical models using statistical analysis of very

extensive experimental measurements or semi-empirical models using statistical

analysis of experiments and combining signals reflections, scattering and other loss

properties, in order to determine (predict) signal path loss calculations with respect to

transmitter and receiver antenna heights, separation distance between transmitter and

receiver, frequency of cellular system and others. The widely used path loss models

for broadband wireless access networks are Cooperation in the field of Scientific and

Research-231 (COST-231) Hata model, COST-231 Walfisch-Ikegami model and

Erceg model (Andrews et al. 2007). Once the path loss is determined, the received

power versus distance could be obtained.

The Hata model is one of the most widely used models for estimating median

path loss in macrocellular systems. The Hata model is valid only for frequency

between 150 MHz and 1500 MHz. The Hata model was modified by the European

COST group, and the extended path loss model is often referred to as the COST-231

Hata model. The extended model supports frequency range between 150 MHz and

2000 MHz. The WiMAX Forum recommends using this COST-231 Hata model for

system simulations and network planning of macrocellular systems in both urban and

suburban areas for mobility applications (WiMAX Forum 2008a). The IEEE 802.16j

relay task group (Ikeda et. al. 2006) also recommends using the COST-231 Hata

model for modeling the path loss between the BS/RS and the MS.

The Erceg model (Erceg et al. 1999) is based on extensive experimental data

collected by AT&T at 1.95 GHz in 95 macrocells across the United States, and is

applicable mostly for fixed wireless deployment with the subscriber stations installed

under the eave/window or on the rooftop. The model is extended and adopted by the

IEEE 802.16 group (Erceg et al. 2003) as the recommended model for fixed

31

broadband applications, and it is referred to as the IEEE 802.16 model. The IEEE

802.16 model is also modified and recommended by the IEEE 802.16j relay task

group for path loss modeling between the BS/RS and the RS, and it is referred to as

the modified IEEE 802.16 model (Senarath et al. 2007). The Erceg model has three

variants, based on terrain type:

(i) Terrain type A: hilly terrain with moderate to heavy tree density.

(ii) Terrain type B: hilly terrain with light tree density or flat terrain with moderate

to heavy tree density.

(iii) Terrain type C: flat terrain with light tree density.

2.4.2 Long-term Fading

In addition to the path loss experienced by the propagating signal, the received signal

power suffers from a slow fluctuation. This slow fluctuation or shadowing is caused

by the existence of obstacles, such as buildings, trees, hills and foliage, in the

propagation path between the transmitter and the receiver. Long-term fading occurs

over relatively large distance of several meters, and is determined by the local mean of

a short-term fading signal. Long-term fading may be modeled by a log-normal

distribution which is described by the following probability density function (PDF)

(Paulraj et al. 2003):

( )( )

⎥⎥⎦

⎤

⎢⎢⎣

⎡ −−

=2

2

2

21 σ

μ

πσ

x

exf (2.1)

where x is a random variable representing the long-term signal power fluctuation and

μ and σ are respectively the mean and the standard deviation of x . μ is equal to

the path loss described in the previous section and all the random variables in the

model above are expressed in dB.

32

2.4.3 Short-Term Fading

Short-term fading is the rapid fluctuations of the received signal in time, frequency

and spatial dimensions. These rapid fluctuations are caused by the constructive and

destructive addition of the received multipath components which experience different

path losses and phases. These rapid fluctuations occur usually over very short

distances of about half a wavelength or over short time duration. Assuming the

existence of the direct path and a large number of scatterers in the channel, the

received signal complex amplitude gain in the presence of L paths can be modeled as

(Biglieri & Taricco 2004):

∑=

+=L

i

jidr

ieAAA1

θ (2.2)

where dA is a constant representing the amplitude of direct path’s signal and iA and

iθ are random variables representing respectively the amplitude and phase of the

signal propagating through the thi path.

Under the assumption of a large number of scatterers, the signals can be

assumed to be independent zero mean Gaussian processes. iθ is modeled as uniformly

distributed over ]2,0[ π , while in the absence of the direct path, iA can be modeled by

the Rayleigh PDF as (Proakis & Salehi 2008):

( ) ( )xuexxfx

⎟⎟⎠

⎞⎜⎜⎝

⎛

Ω−

Ω=

2

2 (2.3)

where Ω is the average received power and ( )xu is the unit step function defined as:

( )⎩⎨⎧

∈<∈≥

=RxxifRxxif

xu,00,01

(2.4)

33

If there is a direct path between the transmitter and the receiver, the signal

envelop is no longer Rayleigh distributed. The signal envelop in such situation can be

modeled as Ricean with PDF given by (Proakis & Salehi 2008):

( ) ( ) ( )

( )xuKKxIeKxxfxKK

⎟⎟⎠

⎞⎜⎜⎝

⎛

Ω+

Ω+=

⎟⎟⎠

⎞⎜⎜⎝

⎛

Ω+

+− )1(2120

1 2

(2.5)

where K is the Ricean factor, Ω is the mean received power and ( )xI 0 is the zero

order modified Bessel function of the first kind.

When there is no direct path, meaning that K = 0, the Ricean PDF in Equation

(2.5) reduces to the Rayleigh PDF given in Equation (2.3), given that ( )00I = 1.

2.5 DIVERSITY TECHNIQUES

As discussed in the previous section, fading results in fluctuations in the received

signal power leading to degradation in the reliability of the wireless channel. Diversity

is one of the most powerful techniques to combat the effects of fading by finding

independently fading paths in the wireless channel (Wornell 1998; Rappaport 2002).

The basic idea of diversity is to provide the receiver with multiple replicas of the same

information bearing signals, where the replicas are affected by uncorrelated fading.

This implies that the probability of all replicas being simultaneously in deep fades is

much lower than the probability of being one replica in a deep fade. Consequently, the

quality of the signal is considerably improved. Diversity techniques are classified

according to the domain where they applied into three classes: temporal, frequency

and spatial diversity techniques.

2.5.1 Temporal Diversity

Temporal diversity is achieved through transmitting the same information bearing

signal at different time slots. The main idea is that, at least one of these replicas

arrives at the receiver with a good SINR. To achieve a diversity gain, the separation

34

between the time slots must be at least equal to the coherence time of the channel.

Some modern implementations of time diversity is the use of RAKE receiver for

spread spectrum CDMA, and interleaving and coding. One of the drawbacks of this

diversity scheme is the reduction in bandwidth efficiency since several redundant

signals are transmitted over the same frequency spectrum.

2.5.2 Frequency Diversity

In this scheme, the diversity is achieved by transmitting the same information bearing

signal over several carrier frequencies. These frequencies should be separated enough

by more than the coherence bandwidth of the channel so that they fade independently,

thereby obtaining a good diversity gain. Examples of systems that exploit frequency

diversity are direct-sequence or frequency-hopped spread-spectrum communication

systems and OFDM. Like the temporal diversity, frequency diversity induces a loss in

bandwidth efficiency due to the redundancy introduced in the frequency domain.

2.5.3 Spatial Diversity

Space diversity, also known as antenna diversity, is one of the most popular forms of

diversity used in wireless systems. In this scheme, several signals having the same

information contents are provided across multiple antennas at the transmitter and/or

the receiver. These antennas are physically separated in space to assure the individual

signals are uncorrelated. For this diversity scheme to be effective, the separation

distance between different antennas must be larger than the coherence distance of the

channel. Typically, a separation distance of a few wavelengths is enough to obtain

uncorrelated signals. Unlike temporal and frequency diversity schemes, spatial

diversity scheme does not include any loss in bandwidth efficiency. This property

makes this scheme very attractive for the future high data rate wireless communication

systems. Spatial diversity can be classified into two schemes depending on whether

the multiple antennas are at the transmitter or the receiver. These are the transmit

diversity (also referred to as multiple-input single-output (MISO)) and the receive

diversity (also referred to as SIMO). Using multiple transmit and multiple receive

antennas is referred to as MIMO which provides even more potential.

35

However, macro diversity is a space diversity technique used to mitigate the

effects of shadowing from building and objects and to reduce the outage probability

(Panajotovic et al. 2009). In this scheme, the signals received from multiple access

stations are combined at the mobile terminal to increase the received SINR.

Cooperative diversity is another diversity method that achieves the benefits of spatial

diversity without requiring the use of physical antenna arrays (Boyer et al. 2004; Sun

et al. 2010), which will be briefly reviewed in the following subsection.

(a) Cooperative Diversity

Relaying, which is the use of intermediate nodes to help transmission from a source to

a destination, has been used to enhance the coverage area, throughput and capacity

and to relax the link budget. While such conventional relaying has long been known

for ad hoc networks, it was only until recently that these concepts have received

interest for cellular networks. Cooperative diversity goes one step further. By

combining the transmissions from various nodes, one can explicitly exploit two

benefits that are inherently offered by relaying systems. First, one can make use of the

broadcast nature of the wireless medium: a signal transmitted by a node propagates

not only to the intended final destination, but it can be received at multiple nodes.

Second, viewing the individual nodes of relaying systems as distributed antennas leads

to regarding cooperative diversity networks as a generalization of multiple-antenna

systems. The related main advantages, namely spatial diversity, spatial multiplexing,

and power saving, are well-known (Laneman et al. 2004). In this sense, cooperative

diversity brings together the worlds of conventional relaying and MIMO systems.

Explicit cooperation for the mutual benefit of neighboring nodes was first

considered by Sendonaris et al. (1998; 2003a; 2003b). They showed that cooperative

diversity increases the channel capacity over the non-cooperative transmission for

ergodic fading. The authors also showed that cooperative diversity improves the

outage performance for non-ergodic fading and decreases the sensitivity of the

achievable data rate to the variations of the channels. The main drawback is that this

work, by assuming the channel state information (CSI) available at the transmitter,

36

requires considerable modifications to the existing hardware and software of the

transmitter and receiver terminals.

In their work, Laneman et al. (2004) assumed no CSI available at the source

and proposed the analysis of cooperative diversity protocols under the framework of

diversity-multiplexing tradeoffs. Their basic setup included a source, a destination and

a relay. Both analog and digital relaying were considered. Subsequently, the diversity-

multiplexing tradeoff of cooperative diversity protocols with multiple relays was

studied in Laneman and Wornell (2003) and Azarian et al. (2005). While Laneman

and Wornell (2003) considered the case of orthogonal transmission between the

source and relays, Azarian et al. (2005) considered the case where the source and

relays could transmit simultaneously. It was shown in Azarian et al. (2005) that by

relaxing the orthogonality constraint, a significant improvement in performance could

be obtained, although at a higher complexity at the decoder. These approaches were

however information theoretic in nature and the design of practical codes that

approach these limits was left for further investigation. Such a code design is difficult

in practice and an open area of research (Beletsas et al. 2006).

Nabar et al. (2004) considered three different TDMA-based cooperative

protocols that vary the degree of broadcasting and receive collision. In their proposed

protocols, the destination overhears transmissions from the source to the relay to

achieve higher spatial diversity gain or higher spatial multiplexing gain. For each

protocol, the authors studied the ergodic and outage capacity behavior under the AF

and DF modes of relaying, and found that the achievable rates with the proposed

protocols are better than that of the existing protocols. The authors also analyzed the

spatial diversity performance of the various protocols and found that full spatial

diversity is achieved by certain protocols provided that appropriate power control is

employed.

The average error probability of a two-hop cooperative system is analyzed in

Hasana and Alouini (2003) and Ikki and Ahmed (2007) for the Rayleigh and

Nakagami-m fading channels, respectively. Onat et al. (2008) proposed selective

relaying schemes based on SNR in cooperative digital relaying systems using uncoded

37

BPSK modulation. In the SNR-based selective relaying, the relay either retransmits or

remains silent depending on the SNRs of the S → R, R → D, and S → D links.

Different models assuming the availability of different sets of instantaneous and

average SNR information at the relay were studied. For all of the models, the optimal

threshold for the S → R SNR below which the relay must remain silent depends on the

SNRs (average or instantaneous) of the R → D and S → D links. The authors showed

that the proposed SNR-based selection relaying significantly reduced the e2e BER

compared to the simple digital relaying in which the RS always transmits, such as the

work in Boyer et al. (2004). More recently, a variety of results on the current research

trends in cooperative diversity have appeared in the work of Adinoyi, Ikki and others

(Adinoyi et al. 2009; Ding et al. 2009; Ikki & Ahmed 2010; Sun et al. 2010; Seyfi et

al. 2011; Wang et al. 2011).

2.6 DIVERSITY COMBINING TECHNIQUES

As have been seen in the previous section, the key feature of the diversity scheme is

that the received power is more stable and therefore the probability of received signal

being in a deep fade is greatly reduced. Here in this section, diversity combining

techniques used to improve the SINR at the output of the receiver are discussed.

According to the implementation complexity and the level of CSI required by the

combining method at the receiver, there are four diversity combining techniques

(Goldsmith 2005). These techniques are: SC, switched combining, MRC, and equal

gain combining. These techniques are briefly reviewed in the following subsections.

However, the focus in this research is on the SC and the MRC since they are used in

the UL and DL of the MDHO technique, respectively.

2.6.1 Selection Combining

In SC, the diversity branch with the highest SINR is chosen as the output. Since only

one branch output is used, this technique does not require knowledge of CSI and

therefore coherent demodulation of the received signal is not required.

38

2.6.2 Switched Combining

In this technique, the receiver scans the receive antennas in a sequential order and the

first antenna with an SINR above a predetermined threshold is selected to be the

output. This technique is simpler to implement than the SC technique since the

receiver does not necessarily scan all the antennas. On the other hand, the

performance of this technique is inferior to that of the SC technique. Similar to the

SC, this technique does not require knowledge of CSI.

2.6.3 Maximal Ratio Combining

MRC is a linear combining technique in which antenna outputs are weighted by their

respective complex gain and then added up together to produce the output signal. The

output signal y can be represented as:

∑=

=rM

iii yay

1 (2.6)

where rM is the number of receive antennas, iy is the received signal at the output of

thi antenna and ia is the weighting factor for the receive antenna i . The weighting

factor of each receive antenna is proportional to the SINR at its outputs. The

weighting factor ia can be represented as:

ijii eAa φ−= (2.7)

where iA and iφ are respectively the amplitude and phase of the signal iy .

Note that using the weighting factor in Equation (2.7) results in an optimum

MRC performance only if the noise-plus-interference variances of the diversity

branches are the same. However, if the noise-plus-interference variances of the

diversity branches are not the same and the weighting factors of the MRC do not take

this fact into account, the MRC is not the optimum combining scheme in this case and

39

it is called imperfect MRC. On the other hand, if the noise-plus-interference variance

is taken into account in this case by normalizing the weighting factor of each diversity

branch by its noise-plus-interference variance, the combining scheme is called

optimum combining or perfect MRC (Shah & Haimovich 2000; Ko et al. 2003).

Hence, in this technique, each branch’s output is co-phased, weighted by its

corresponding weight factor and then summed up so that the output SINR is maximal.

This technique requires knowledge of the CSI and hence the signals are demodulated

coherently.

2.6.4 Equal Ratio Combining

This technique is simpler to implement than MRC. In this technique, the signals of the

receive antennas are co-phased and then combined together to produce the output

signal. The amplitude of the received signal at each antenna is not required to be

estimated. Hence, the weighting factor is represented as ( iA is set to 1):

iji ea φ−= (2.8)

where iφ is similar to previous.

2.7 HANDOVER

Handover is needed in cellular systems to allow full users mobility in the coverage

area. The main target of handover is to provide the continuity of wireless services

when a MS moves from the air-interface of one BS/RS to that provided by another

BS/RS. There exists also intra-frequency handover which basically means changing

from one frequency to another while the serving station remains unchanged. This

feature could be exploited in a femtocell scenario where a user moves from outdoors

to indoors (Hytonen 2009). The intra-frequency handover also occurs between sectors

within the same cell where each sector has different carrier frequency. From wide

perspective handover may be divided into two groups; vertical handover and

40

horizontal handover. The vertical handover takes place between access stations

belonging to different technologies, whereas the network technology remains the same

in the latter. Media independent handover (MIH) defined in IEEE 802.21

specifications is an example of the vertical handover that makes the handover possible

between IEEE 802.3 (basic Ethernet), IEEE 802.11, IEEE 802.16 and different

cellular 2G, 2.5G, 3G, B3G and 4G networks (IEEE 2008). However, the focus in this

research is only on the horizontal handover between different access stations.

The application of the multihop concept to cellular networks raises many

technical challenges, such as the best positions for the BSs and the RSs, the number of

RSs, radio resource allocation and multiplexing between the BSs and the RSs,

scheduling and handover (Cho et al. 2009). Unlike single hop cellular networks where

handover occurs between BSs in different cells or different sectors of the same cell,

the introduction of RSs into cellular networks creates additional handover scenarios

between the BS and the RSs within the same or different cells or between different

RSs that can also be within the same or different cells. The IEEE 802.16j also

introduces support for RS´s movement. These MRSs are able to move from the air-

interface of one access station (BS or RS) to the air-interface of another access station.

The handover process for a MRS is performed similarly to that of a MS (IEEE 2009).

This is because the network only needs to consider the MRS itself and there is no need

to consider its relayed MSs. On the other hand, from handover point of view, NRS is

similar to FRS because it is fixed when it is operated and it is switched off when it is

moving. However, the RSs in this research are FRSs and hence the handovers of RSs

are not considered.

The triggering parameters that can be used for the initiation of the handover

process (Perez-Romero et al. 2005; Makelainen 2007; Shen et al. 2006; Lin & Chen

2008; Yang & Tseng 2008) can be summarized as follows:

(i) The received signal strength from the current serving station is not enough for

maintaining proper connections and there is better signal quality from one or

more neighboring access stations. This is the main and fast triggering

parameter which is also considered in this research.

41

(ii) QoS requirement.

(iii) Lack of access station’s capacity and more traffic is pending (load balancing).

(iv) Disturbing co-channel interference from the neighboring cells.

(v) Availability of another network that is faster, cheaper and/or that offers lower

MS battery power consumption (in case that vertical handovers are supported).

2.7.1 Handover Types

There are three types of handover supported within the IEEE 802.16j multihop

cellular networks, namely HHO, FASS and MDHO (IEEE 2006; IEEE 2009).

However, the support for HHO is mandatory, while FASS and MDHO are two

optional modes. The MDHO and FASS are also referred to as soft handover.

(a) Hard Handover

In HHO, the MS communicates with only one BS/RS in each time. The HHO is a

break-before-make switching method where the connection with the old serving

BS/RS is broken before the connection to the new target BS/RS is established. Hence,

the MS experiences a connection drop between its termination from previously

connected station and the reconnection to the new target station. This handover is the

simplest scheme for the practical operation since it is based on received signal

strength measurement from different BSs/RSs. The HHO is executed (Becvar &

Zelenka 2007) after the signal strength from the target station exceeds the signal

strength from the current serving station.

Figure 2.10 demonstrates a HHO scenario contains the received signal

strengths, in terms of SINR, of two different access stations (ASs), for instance AS1

and AS2. The access station can be a BS or an RS. In this HHO scenario, two

adjustable parameters are considered, namely HHO threshold hysteresis (HHO_Th)

and time-to-trigger timer )( TΔ . The former determines how much the signal strength

of the target station should exceed the signal strength of the current serving station,

whereas the latter determines how long such a condition remains before triggering the

handover (Hytonen 2009; De Sanctis et al. 2009; Kawai et al. 2010).

42

Figure 2.10 Hard handover scenario

The hysteresis margin needs to be introduced to prevent the ping-pong effect,

the phenomenon that when the user hands over forth and back between the access

stations and hence frequent handover occurs. The ping-pong effect increases the

network signaling and overhead. Aside from the mobility of user, fading effects of the

radio channel can also make the ping-pong effect more severe. When the hysteresis

margin is introduced, the ping-pong effect is alleviated since the user does not

handover immediately to the most suitable access station. When the margin is large,

on the one hand the ping-pong effect is reduced and on the other hand the delay is

increased. However, more delay would increase the probability of dropped packets

and degrade the system throughput performance since the MS stays for a longer

period connected to the suboptimal access station.

There are four important action points in Figure 2.10, namely 1t - 4t . At times

1t and 3t , the SINR of AS2 exceeds the SINR of AS1 by HHO_Th and time-to-trigger

timer TΔ is started. However, at time 2t , the MS notices that the SINR difference is

lower than HHO_Th, and hence the timer is stopped. On the other hand, at time 4t , the

HHO trigger timer is expired and thus the MS performs a handover to AS2 which then

becomes the new serving station. Therefore, the value of the hysteresis margin and the

time-to-trigger timer are fairly important in the handover algorithm.

TΔ

HHO_Th

AS1

AS2 SINR

HHO_Th

Time 1t 2t 3t 4t

43

(b) Fast Access Station Switching

For an MS and a BS that support FASS, the MS and BS maintain a list of the access

stations that are involved in FASS with the MS, as can be seen in Figure 2.11. This list

is called a diversity set, and it is maintained for every MS in the handover regions.

The access station can be a BS or an RS. The MS continuously monitors the access

stations in the diversity set and defines an anchor station. The anchor station is the

only BS/RS of the diversity set that MS communicates with for all UL and DL

messages including management and traffic connections. The anchor station can be

changed from frame to frame depending on access station selection scheme. This

means that every frame can be transmitted by different BS/RS in the diversity set. The

change from one anchor station to another, namely access station switching, is done

without invocation of explicit handover signaling messages. An important requirement

of FASS is that the data is simultaneously transmitted to all members of the diversity

set that are able to serve the MS (WiMAX Forum 2006b). The difference between the

FASS shown in Figure 2.11 and the MDHO shown previously in Figure 1.1 is that the

MS communicates with only the anchor station of the diversity set in FASS, whereas

the MS communicates with all access stations in the diversity set in MDHO.

Figure 2.11 Fast access station switching with diversity set size of 4

Active RS1

MS1

Active BS1

Area of Neighbor Stations

Diversity Set

Active RS3

Neighbor RS4

Neighbor BS2 Anchor RS2

Data are transmitted and received but not processed in BS/RS (MS)

UL and DL communication Including traffic

Only signal strength measurement No traffic

44

(c) Macro Diversity Handover

If an MS and a BS support MDHO, the diversity set is maintained by the BS and the

MS similarly to FASS. The MS communicates on the UL and the DL with all the

access stations in the diversity set. The normal mode of operation is a special case of

the MDHO when there is only one access station in the diversity set. In the DL,

multiple copies of the transmitted signal are received at the MS and the diversity

combining is performed. In the UL, however, the MS transmits the data to the

diversity set members such that selection diversity is performed to pick the best link.

Figure 2.12 shows the basic handover process of the HHO and MDHO

techniques. It is assumed in Figure 2.12 that there is an MS moving from AS1 towards

AS2, where AS1 is the original serving station of the MS. While moving, it is also

assumed that the MS continuously measures the received signal strength from the

neighbor access stations. With HHO shown as (a) in Figure 2.12, a definite decision is

made whether to handover or not and the MS communicates with only one access

station at a time. In case of MDHO, shown as (b) in Figure 2.12, before the SINR of

AS2 goes beyond the SINR of AS1 and as long as the MDHO trigger condition is

fulfilled, the MS enters the MDHO state and a new link is set up. Before AS1 is

dropped from the diversity set (that occurs when the MDHO dropping condition is

fulfilled), the MS communicates with both AS1 and AS2 simultaneously. Hence,

unlike HHO, MDHO is a break-before-make process in which the connection with the

new target station is established before the connection with the old serving station is

broken. Clearly, MDHO do not experience any drop in the ongoing communication

and the MS remains connected to multiple access stations simultaneously.

The following conditions should be satisfied in order for the MDHO or FASS

to be feasible (IEEE 2006):

(i) The access stations involved in MDHO or FASS are synchronized, based on a

common timing source.

(ii) The DL frames transmitted from the access stations involved in MDHO or

FASS should arrive at the MS within the cyclic prefix interval.

45

(iii) Access stations involved in MDHO or FASS must have synchronized frame

structures in the DL and the UL and must have the same frequency

assignment.

(iv) Access stations involved in MDHO should use the same set of connection

identifiers (CIDs) for the connections established with the MS.

(v) The same MAC/PHY PDUs should be transmitted to the MS by the access

stations involved in MDHO.

(vi) Access stations involved in MDHO or FASS are also required to share all

information that MS and BS normally exchange during network entry.

(vii) Access stations involved in MDHO must share all information, such as service

flow identifiers (SFIDs), encryption and authentication keys.

Figure 2.12 Comparison between HHO and MDHO

HHO

SINR of AS1

AS2 AS1

AS2 AS1

SINR of AS2

MS MS MS

(a) MS direction of movement

MDHO

AS1 AS2 MS MS MS

(b)

46

2.7.2 MDHO Algorithm

The performance of the MDHO is closely related to the implemented algorithm.

Several algorithms have been proposed to support soft handover in previous systems

and different criteria are used in different algorithms. The soft handover algorithm

proposed for IS-95A is based on the fixed absolute thresholds, whereas the soft

handover algorithm proposed for IS-95B is based on the dynamic absolute thresholds

(Laiho-Steffens et al. 1999; Li et al. 2005; Homnan et al. 2000). In contrast, the soft

handover algorithm proposed for WCDMA is based on the relative thresholds (3GPP

2002; Perez-Romero et al. 2005).

However, in the MDHO of the IEEE 802.16e and thus IEEE 802.16j

specifications, the diversity set is maintained or updated based on the absolute

H_ADD and H_DELETE thresholds contained in the downlink channel descriptors

(DCDs) broadcasted by the BSs or RSs. If the long-term SINR of the access station

currently in the diversity set is less than H_DELETE, then this access station is

removed from the diversity set. On the other hand, if the long-term SINR of the access

station currently in the monitored set is high than the H_ADD threshold, then this

access station is added to the diversity set. However, in reality, with the load of an

access station changing at every moment, relative thresholds rather than absolute

thresholds seem to be more realistic for accurate maintaining or updating the diversity

set (Ulvan et al. 2007; Ray et al. 2010). Although this method provides a more

accurate way for maintaining or updating the diversity set, it is more complicated to

implement. Based on the previous discussion, the MDHO algorithm implemented in

this research is based on the relative thresholds WCDMA soft handover algorithm

(3GPP 2002) that will be described in the rest of this subsection.

In order to decide on the access stations to be included in the diversity set, the

MS measures the long-term SINR of the neighboring access stations. The MS then

uses these performed measurements together with the long-term SINR of the current

serving station and the pre-assigned thresholds to apply the soft handover algorithm.

Figure 2.13 shows the most relevant aspects of the considered soft handover

algorithm. Particularly, Figure 2.13 considers the long-term average SINRs of three

47

access stations, for instance AS1, AS2 and AS3, vary in time and limits the maximum

dimension of the diversity set to two. In Figure 2.13, Add_Th, Del_Th, Rep_Th and

TΔ denote the addition threshold, deletion threshold, replacement threshold and time-

to-trigger timer, respectively. The soft handover algorithm works as follows:

(i) If the average SINR difference between strongest access station (AS1) in the

diversity set and best access station (AS2) not currently in the diversity set is

less than Add_Th for a period of time TΔ , then AS2 is added to the diversity

set if it is not full. This event is called radio link addition and occurs at time 2t .

(ii) If the diversity set is full and the average SINR of best access station (AS3)

not currently in the diversity set is superior than the average SINR of weakest

access station (AS1) of the diversity set by Rep_Th for a time period TΔ , then

AS1 is replaced by AS3. This event is called radio link addition and removal

and occurs at time 3t .

(iii) If the average SINR of weakest access station (AS3) in the diversity set is

lower than the average SINR of strongest access stations (AS2) in the diversity

set by Del_Th for a period of time TΔ , then AS3 is removed from the

diversity set. This event is called radio link removal and occurs at time 4t .

Figure 2.13 MDHO algorithm

AS1

AS2

AS3

SINR

Time

Add_Th

Del_Th

Rep_Th

TΔ TΔ TΔ

4t 3t 1t 2t

Add AS2 Connect to AS1 Replace AS1 with AS3

Remove AS3

48

2.7.3 MDHO Procedures

The general handover method performs network topology advertisement, allocation of

scanning intervals to the MSs, handover decision and initiation and handover

execution and termination. With MDHO enabled, the MS performs the following

stages, namely diversity set selection/updates and anchor station selection/update

(IEEE 2006). In the system with transparent RSs, only the BS transmits all the

broadcast control messages, and the RSs in the same cell forward the same broadcast

control messages to the MSs. On the other hand, the non-transparent RS can transmit

its own preamble, frame control header (FCH), DL and UL medium access protocol

(DL-MAP and UL-MAP) messages and DCD and UL channel descriptor (UCD)

messages (IEEE 2009). The non-transparent RSs can compose the handover signals

under the direction of BS.

(a) Network Topology Advertisement

A BS and RS periodically broadcast a neighbor advertisement (MOB_NBR-ADV)

message to all MSs that are in its cell. This message gives the MS information about

the access link channel of the neighbor stations for possible handover or initial

network entry. This information can be obtained over the backbone as well as over the

wireless relay links.

(b) MS Scanning

The scan request (MOB_SCN-REQ) message may be transmitted by the MS to the

access BS or RS in order to request an allocation of scanning intervals and a certain

type of association with each potential target access stations. The allocated scanning

intervals are used by the MS for seeking the neighbor stations and determining their

suitability as targets for handover. When the access station receives the MOB_SCN-

REQ message, it responds with a scan response (MOB_SCN-RSP) message. The

coordination between stations can be realized over the backbone as well as over the

wireless relay link (Zhao et al. 2008).

49

(c) MDHO Decision and Initiation

A MDHO begins with a decision by an MS to simultaneously receive from and

transmit to multiple BSs and/or RSs. The MDHO can start with either a MS handover

request (MOB_MSHO-REQ) message by the MS or a BS handover request

(MOB_BSHO-REQ) message by the anchor station. The BS or RS supporting MDHO

broadcasts the DCD message that includes the MDHO thresholds. These thresholds

are used by the MDHO capable MS to determine whether to transmit the

MOB_MSHO-REQ message or not.

(d) Diversity Set Update

When the MOB_MSHO-REQ message is transmitted by an MS, the MS may indicate

a possible list of BSs and/or RSs to be incorporated in the diversity set of the MS. The

MS may evaluate the possible list of BSs and/or RSs through the received

MOB_NBR-ADV message and previously performed signal strength measurement,

propagation delay measurement, scanning, ranging, and association activity. The

anchor station responds with the BS handover response (MOB_BSHO-RSP) message

in which the BSs may provide a list of BSs and/or RSs recommended for inclusion in

the diversity set of the MS. The information included in the MOB_BSHO-REQ and

MOB_BSHO-RSP messages can be obtained over the backbone as well as over the

wireless relay links (Zhao et al. 2008). After the diversity set update is initiated by a

BS or MS with MOB_BSHO/MSHO-REQ, the MS can cancel the diversity set update

at any time by transmitting a handover indication (MOB_HO-IND) message with

proper parameter.

(e) Anchor Station Selection/Update

The MS is required to continuously monitor the signal strength of the diversity set

members and select one station from its current diversity set to be the anchor station.

The MS then reports the selected anchor station on the MOB_MSHO-REQ message

or on the fast feedback channel quality indicator channel (CQICH). However, the

anchor station is the one with the best received signal strength.

50

(f) Handover Execution and Termination

After the handover is initialized, the MS synchronizes with the DL and UL

transmissions of the potential station by obtaining the required parameters, such as

DL-MAP, UL-MAP, DCD, and UCD. The MS and the potential stations in the

diversity set then perform ranging by exchanging ranging request (RNG-REQ) and

ranging response (RNG-RSP) messages. The MS can indicate a handover attempt by

sending a RNG-REQ message, which includes a station identifier (ID)

type/length/value (TLV) and ranging purpose indication TLV with Bit #0 set to 1.

When the potential station receives the RNG-REQ message, it may request the MS

information if the station has not obtained the MS information yet. The MS

information may be obtained over the backbone as well as over the wireless relay link.

After successful registration with the potential BS/RS, the MS sends

MOB_HO-IND message to the serving station to indicate that the handover is

completed. The previous anchor station is informed of the successful MS network

attachment at the diversity set. In the MR networks, this successful attachment may be

informed over the wireless relay links as well as over the backbone

Figure 2.14 illustrates an example of the timing diagram of MAC management

messages with non-transparent RSs for the MDHO scenario in which the diversity set

members are two RSs in two different cells. The anchor station is also updated in this

MDHO scenario. In this figure, the arrows illustrate the direction of messages.

However, the signaling between MS and access stations occur over the wireless access

link and is shown as solid arrows in Figure 2.14, whereas the signaling between RS

and BS occur over the wireless relay link and is illustrated as dashed arrows in Figure

2.14. The signaling between BSs takes place over the wired backbone or network

infrastructure and is shown as dashed-dotted arrows in Figure 2.14.

51

Figure 2.14 Timing diagram of MAC management messages for the MDHO

scenario in which the diversity set members are two RSs in two different cells

Source: Zhao et al. 2008

2.7.4 Comparison between the Handover Techniques in Multihop Cellular Networks

HHO: The HHO mechanism in IEEE 802.16j multihop cellular networks is based on

the HHO in IEEE 802.16e/mobile WiMAX. Compared to the HHO used in its cellular

competitors B3G technologies like EV-DO and HSDPA, the HHO scheme in mobile

WiMAX and IEEE 802.16j is more bandwidth efficient, fast, smooth and nearly

glitch-free. The WiMAX Forum developed a network optimized HHO mechanism

52

that reduces handover overheads and achieves a Layer-2 handover delay of less than

50 ms (WiMAX Forum 2006b). The HHO is the simplest handover technique to

ensure efficient support for the provisioning of different high-speed real-time

applications without significant interruptions and degradations of QoS.

In the HHO of multihop cellular networks as well as in any other system, an

MS assumes that there are always sufficient resources available in the target access

station to support the MS’ handover to it, which increases the chances of call drops

and delays. In the sectorized deployment scenario where each sector has different

carrier frequency but a fixed frequency reuse pattern, lossless handover can be

achieved using HHO (Das et al. 2006). In the heterogeneous networks containing

different systems operating on different frequency bands, the only possible inter-

frequency handover technique between the different systems is the HHO. In the IEEE

802.16, the PHY and MAC layers provide support for dynamic and accurate

measurements of UL and DL signal strengths of the neighbor access stations by the

MS and the serving access stations, as well as efficient support for broadcast-related

features. This results in minimizing resource wastages and handover delays (Ray et al.

2010). Finally, the real advantage of HHO technique is its low complexity and its low

deployment cost.

MDHO and FASS: In order to manage voice-centric applications with high-speed

mobility users, the HHO technique is not very attractive due to its high latency. In

contrast, MDHO and FASS are introduced to support full seamless mobility at much

higher speeds up to 120 km/hr. Some of the design features of MDHO and FASS

include very low packet loss (less than 1%), low handover latency (less than 50 ms)

and very fast switching, which give them the ability to support high-speed real-time

voice-centric applications like voice-over-IP (VoIP) (Ray et al. 2010). The MDHO

and FASS minimize the ping-pong effect leading to reduced load on the network

signaling and overhead. The MDHO and FASS also provide lower time constraints on

the network. In other words, there is a longer mean queuing time to obtain a new

channel from the target access station, and this helps to reduce the blocking

probability and dropping probability of connections.

53

Both the MDHO and FASS techniques have the potential to further minimize

the handover delays and the handover signaling overheads when switching the anchor

station within the diversity set. This is due to the fact that the switching does not

require invocation of explicit handover signaling messages and, in addition, the

network re-entry procedures do not need to be performed every time when switching

between anchor stations (IEEE 2006). Within sectors having the same carrier

frequency, MDHO and FASS can be performed due to their employing universal

frequency reuse concept (Das et al. 2006). Compared with the soft handover in

CDMA systems, both the MDHO and FASS techniques used in mobile WiMAX and

multihop cellular networks are designed to provide better performance with respect to

multiple access interference, flexibility, coverage, and system capacity. The

applications of both OFDMA FUSC and PUSC techniques in mobile WiMAX (Das et

al. 2006) MDHO mechanisms have improved the range and cell coverage. For a cell-

edge user having diversity set members from different cells, on one hand the main DL

interference sources are eliminated and on the other hand this user will not cause any

UL interference to other users in the diversity set members’ cells that otherwise would

seriously degrade the performance of the user using the same radio resources.

Consequently, the cell-edge users’ performance, the cell range and coverage, and the

system capacity are improved. Moreover, unlike soft handover in CDMA systems that

have static mode of operation, an MS can dynamically activate and deactivate the

MDHO or FASS when required depending on the radio channel condition

encountered by the MS. This conserves the radio resources and thus increases the

overall system capacity (Gage et al. 2005).

However, the MDHO and FASS advantages come at the cost of increased

implementation complexity and cost compared to HHO technique. Additional radio

resources are also consumed in the DL direction (sub-channel and power resources).

Finally, in the intra-cell MDHO scenarios of the multihop cellular networks and when

the diversity set members are a BS and an RS or two RSs, the simultaneous

transmissions of the diversity set members increases the co-channel interference

sources compared to the scenarios in which either the BS or the RS transmits. Table

2.2 provides a brief comparison of the HHO, FASS and MDHO techniques with

respect to the mobile WiMAX handover scenario.

54

Table 2.2 Brief comparison of the various handover techniques

Parameter HHO FASS MDHO

Latency High Medium Low

Complexity Low Medium High

Reliability Low Medium High

Packet loss High Low Low

Cost Low Medium High

Support for delay sensitive applications Low High High

Speed Low Medium High

Link quality Low Medium High

Source: Ray et al. 2010

2.8 RELATED STUDIES ON HANDOVER IN MULTIHOP CELLULAR NETWORKS

Unfortunately, handover issues in multihop cellular networks have not been studied

extensively in the literature yet. Yap et al. (2002) proposed a position assisted relaying

and handover algorithm for hybrid ad hoc and cellular networks. The MS is assumed

to be able to estimate its geo-location and establish direct connections with nearby

MSs to form a temporary wireless relay network. The position information provides

the mobility profile for each MS, allowing MS to avoid unnecessary handover. The

position information also assists the MS in selecting the best nearby MS for relaying

to avoid call drop due to sudden channel degradation.

Ghassemian et al. (2005) classified and studied the performance of different

kinds of handovers in multihop radio access networks (MRANs). In Ghassemian et al.

(2005), the multihop handover schemes are classified into forced handover and route

optimization-based handover, depending on the reason for the handover initiation.

Furthermore, the generic signaling mechanisms for these handover scenarios are

proposed, and handover delay and signaling overhead are also investigated.

Khadivi et al. (2008) proposed to use the ad hoc relaying for reducing the

dropping probability during vertical handover in hybrid WLAN and cellular systems.

55

This scheme starts when an active MS, that initiates a WLAN-to-Cell handover, uses

non-active MSs as RSs in handover regions. If there is no channel available in the

cellular system, the MS establishes a routing algorithm to build up a path through a

number of non-active MSs. If the MS finds a relaying path, it can use this path to relay

its connection back to the original access point.

Even though relay-handover problems are investigated in Yap et al. (2002),

Ghassemian et al. (2005) and Khadivi et al. (2008), the focus was on ad hoc networks

and not on cellular networks.

Nourizadeh et al. (2006) studied how frequent the inter-relay HHO takes place

and its effect on the performance of the multihop cellular network. In their study,

different algorithms to decide when an MS should inter-relay handover were proposed

and evaluated through a dynamic system-level simulator. Compared to the

conventional cellular networks using the UMTS FDD mode, their simulation results

demonstrated that the UL capacity gain can be enhanced by 20% to 50% depending on

the employed inter-relay handover scheme.

In Becvar (2007), the author investigated the impact of the RS implementation

on the handover in mobile WiMAX networks. The considered handover techniques

are MDHO and FASS. The effects of the number of deployed RSs on the number of

initialized handovers and on the diversity set size are investigated. Another work by

Becvar et al. (2009) introduced a new approach in triggering of HHO procedure in

relay-enhanced WiMAX networks. In order to increase the overall network

throughput, the proposed method initializes the handover procedure based on

evaluation of maximal network throughput in UL or DL direction.

Kim et al. (2008) studied the effect of using different absolute thresholds

corresponding to different service types in the HHO algorithm for the multihop

cellular networks. The absolute threshold value is used to initiate a handover

procedure in handover algorithm, and in this way, the MS could decide whether to use

RS or not by adjusting the absolute threshold value. Their results showed that using

56

low absolute threshold results in reducing the probability of handover to the RSs at the

cell boundary since the BSs can still maintain the connection in this case.

Cho et al. (2009) introduced various HHO scenarios in multihop cellular

networks. In addition, Cho et al. (2009) also presented the detailed handover operation

and signaling where the RSs are deployed either inside a cell or on the boundary

between two adjacent cells, and investigated the effects of the RSs’ deployment

position on handover performance. Their simulation results showed that the

throughput for both deployment scenarios of multihop cellular networks increases by

90% compared to that of single-hop cellular networks. The inter-cell handover latency

in multihop cellular networks is increased by 20% to 56% compared with that in

single-hop cellular networks. Their results also illustrated that the overall throughput

of the multihop cellular networks with RSs deployed on the boundary between two

adjacent cells is lower than for those with RSs deployed inside a cell, whereas the

opposite is correct for the throughput of cell-edge users. In the scenarios where the

RSs are deployed on the boundary between two adjacent cells, the handover process is

simplified and the handover signaling overhead is reduced compared to the scenarios

where the RSs are deployed inside a cell, and the service interruption time is

significantly reduced with respect to single-hop cellular networks.

In Chen et al. (2010), the authors proposed a HHO algorithm for cellular

relaying networks to choose a more accurate access stations for handover target,

which was based on the MS movement state (direction of movement and velocity) in

addition to the signal strength. In this proposed algorithm, the access stations towards

which the MS is moving are given higher priority for handover. From those access

stations, the ones located far away from the high speed MS is given further priority for

handover, whereas the opposite is performed for the low speed MS. However, after

doing these processes, the conventional HHO algorithm is applied. Simulation results

showed that compared to the conventional HHO algorithm, the proposed algorithm

reduces the signaling cost, leading to optimized network performance.

Park et al. (2007) proposed a handover approach that reduces inter-cell

handover but increases intra-cell handover for the multihop cellular networks. The aim

57

was to reduce the high handover signaling overhead and latency caused by inter-cell

handover compared to intra-cell handover. The inter-cell handover was reduced at the

level of the MAC-layer using the MOB_NBR-ADV message including BS identifiers

(BSIDs), RSIDs and preamble IDs of neighbor BSs and RSs. The RSID can be

defined as a subset of the BSID since the RS has a sub-ordinate relationship with the

BS. To achieve their aim, the authors use adaptive hysteresis margin where the inter-

cell handover is performed based on higher hysteresis level than that of the intra-cell

handover. The disadvantages with this approach are that it is not easy to simplify the

scanning procedure.

In Yoo et al. (2009), a new approach of distinguishing an inter-cell handover

from an intra-cell handover was proposed at the level of the PHY layer using a

hierarchically designed preamble. In this PHY layer approach, the same Cell ID is

assigned to a BS and RSs within the same cell, and a different Subcell ID is assigned

to the BS and each RS in the cell. The Subcell IDs can be reused in neighboring cells

since their Cell IDs are normally different. The decision regarding the use of either an

inter-cell handover or an intra-cell handover is performed based on the measurement

of the signal quality of the BSs and RSs, provided by the hierarchical preamble.

Unlike the MAC-layer approach in Park et al. (2007), the MS in this approach is not

required to decode data to obtain information on an inter-cell handover or intra-cell

handover. Not only this approach differentiates between an inter-cell handover and an

intra-cell handover, but also realizes a significant reduction in the scanning procedure.

Yang et al. (2008) proposed a handover protocol for de-centralized MR

networks in which some management functions are delegated and performed by high

capability RSs with the objective to reduce handover signaling delay and overhead.

The authors also defined the MAC handover procedure and the corresponding MAC

management messages over MR links so that an IEEE 802.16e MS can handover

seamlessly without noticing that it is attached to an MR networks. Kim and Cho

(2009) proposed a pre-buffering scheme in which the BS performs multicast

transmission to the handover candidate RSs in order to reduce the packet loss and the

service interruption time. After the MS detects a nearby RS and considers it as a target

for the handover, the MS requests the BS to multicast the traffic to the candidate RS.

58

Becvar et al. (2008) developed an optimization of management messages and

their exchange during the MS’ scanning procedure of neighbor access stations in

WiMAX based networks with RSs. The reason behind this optimization was that

when the RSs are taking into account, a new wireless interface among the BSs and

RSs comes out. Consequently, a new communication scheme over the radio interface

has to be proposed and optimized in order to reduce the management information

overhead and to maximize the user data throughput. Another work by Becvar and

Mach (2010) proposed a new method for reporting of scanning results to the serving

BS when the RSs are deployed in the network. The main objective of the proposal was

to design a reporting technique that generates minimum management overhead during

reporting. This was achieved by collecting of individual MSs’ scanning results into

one message in the access RS and retransmitting this single message to the serving

BS. The results showed a reduction up to 30% in the scanning reporting management

overhead.

Sun et al. (2008) proposed a mechanism to transform the network topology and

performance metrics, such as number of hops, antenna configurations and/or mobile

channel conditions, into power indications to enable fast handover. This is done

according to the fact that the handover procedure based on the signal strength from

different access stations is the most efficient method to support fast handover. The

power indications are the actual power levels transformed to reflect network

topologies and link level qualities among BSs and RSs. The power levels can be

detected by the MS during handover procedure. The power levels might be different

from the transmit power level for a BS or a RS on its data transmissions.

Cho et al. (2008) developed a relay-assisted soft handover in multihop cellular

networks. The RS is deployed on the boundary between two adjacent cells and

assumed to be synchronized and controlled by both BSs of the adjacent cells. In this

work, a handover ranging is not performed and an association process can be

significantly simplified since it is not necessary to change the serving node during

handover process. The handover latency and the service interruption time are reduced

by 21%, whereas the capacity loss is 5.33% compared to the case in which HHO is

performed between two RSs where each RS is deployed inside a cell and controlled by

59

only one BS. However, the work considered in this research focused on the scenarios

in which the RSs are deployed inside cells.

The required changes in the MDHO and FASS handover procedures and new

MAC management messages due to the deployment of RSs in the network

infrastructure were proposed in Zhao et al. (2008). The handover procedures for these

two types of handover and the MAC messages are described for nine main classes of

network topologies that are further classified into intra-BS, where and when the

handover is within the same MR cell, and inter-BS handovers, where and when the

handover is between different MR cells. The handover phases where the new MAC

messages are proposed include network topology advertisement, MS scanning,

MDHO or FASS decision and initiation, handover execution and handover

termination phases.

In summary, most of the previous literature on the handover in multihop

cellular networks has been devoted to reduce the handover overhead, handover latency

and service interruption time, to study the required handover procedure changes and

propose new handover MAC management messages due to the RS involvement, and

to evaluate the performance of the HHO, FASS and conventional MDHO. The focus

in these works has been mostly on the HHO. In the DL of the conventional MDHO,

the MS receives only the simultaneous transmissions of the diversity set members

whether the diversity set members are a BS and an RS, two RSs, or two BSs. In fact,

in the DL of the conventional MDHO where the diversity set members are two

different-topology access stations, namely a BS and an RS, the signal transmitted by

the BS in the first phase is not received by the MS even though the MS is idle in this

phase. Thus, the topology of the diversity set members is not fully exploited. This

results in an inefficient utilization of radio resources and poor performance of the MSs

in the MDHO regions.

However, cooperative diversity has emerged as a new form of space diversity

in wireless networks and has been studied extensively in the literature as described

previously in Subsection 2.5.3. Even though these works studied the diversity in

multihop networks, the main interest of using multihop and diversity concepts is not in

60

cellular networks but in ad hoc networks. To the best of our knowledge, most of the

previous works on the cooperative diversity have been carried out in a single cell or ad

hoc topology noise-limited environment and for fixed users with the diversity set

members are always a BS and RSs. Furthermore, since the cooperative diversity is not

limited to the overlapped coverage areas of the access stations, cooperation may

increase the resource consumptions.

Therefore, in the first part of this research, a new efficient topology-aware DL

MDHO technique for TDD-OFDMA-based interference-limited multihop cellular

networks is proposed. In the proposed MDHO, the MS receives all the data signals

transmitted by the diversity members whether the diversity set members are a BS and

an RS, two RSs, or two BSs. It ensures that the topology of the diversity set members

is always fully exploited. In the proposed MDHO and when the diversity set members

of the MS are two different-topology access stations, that is a BS and an RS, the MS

receives the data signal transmitted by the BS in the first phase; and at the same time it

also receives the simultaneous transmissions of the BS and the RS during the second

phase. On the other hand, when the diversity set members are two similar-topology

access stations, namely two RSs or two BSs, the proposed MDHO performs similarly

to the conventional MDHO where only the simultaneous transmissions of the diversity

set members are received by the MS. This constitutes an efficient utilization of the

radio resources and is expected to increase the spatial diversity gain and spatial

multiplexing gain, and decrease the BER, which are important for enhancing the

performance of the MSs in the MDHO regions.

On the other hand, in the UL of the conventional MDHO, the SC among the

received signals is performed for both intra-cell and inter-cell MDHO scenarios. In the

conventional SC scheme, the link that has the maximum SINR is selected (IEEE 2006;

Sediq & Yanikomeroglu 2009; Simon & Alouini 2000) and it is referred to in this

research as SSC scheme. In multihop cellular networks, the SSC scheme might not

necessarily achieve the best performance in terms of average SINR, average e2e

throughput and e2e BER. In order to ensure a good UL performance, three efficient

UL schemes for MDHO technique are proposed in the second part of this research.

61

It is firstly proposed a new UL MDHO scheme that uses MRC to combine the

signals received from the different diversity branches in case of intra-cell MDHO

scenarios and uses the conventional SSC scheme in case of inter-cell MDHO

scenarios. The intra-cell MDHO scenarios comprise the scenarios in which the

diversity set members are a BS and an RS or two RSs within the same cell. On the

other hand, the inter-cell MDHO scenarios comprise the MDHO scenarios in which

the diversity set members are a BS and an RS, two RSs, or two BSs in different cells.

This proposed UL scheme is called joint MRC-SC scheme.

The end-to-end optimization of certain QoS measures such as throughput,

reliability and latency plays a key role in designing broadband multihop cellular

networks (Oyman 2007). The SSC scheme may not essentially attain the best

performance in terms of the e2e throughput. For instance, if the SINR of the direct

link can support adequately high MCS mode, then direct transmission might

outperform relay based transmission even if the SINR of the R → D link is higher than

that of the direct S → D link. This is due to the facts that the relay based transmissions

require additional radio resources as compared to transmission using the direct S → D

link (Can et al. 2007). Besides, in the UL scenarios of the multihop cellular networks,

because of the mobility of users and the fact that the links between the users and the

RSs are normally in NLOS, the e2e throughput of the S → R and R → D links is

limited by the performance of the S → R link. Finally, the resource allocated to the

S → D link during the first phase can also be used to transmit new data during the

second phase.

Moreover, in the UL scenarios of the interference-limited environment, the

desired MS signal may be interfered by the transmission of the co-channel MSs during

the first phase, whereas it may be interfered by the transmissions of the RSs or the

MSs during the second phase. During the second phase, however, if the interference

comes from the other cells co-channel RSs, then the desired MS signal may be drown

out by the transmission of the interfering RSs. Hence, using UL power control at the

RSs is essential in order to significantly minimize the interference comes from the

RSs, thereby allowing the MS to transmit during the second phase using an MCS with

spectral efficiency that is near or similar to that of the first phase. Therefore, by taking

62

all the above facts into consideration, it is secondly proposed a new efficient UL

scheme that combines the advantages of the e2e throughput-based selection with the

benefits of using the UL power control at the RSs. This proposed UL scheme is called

ETSC scheme with power control at the RSs.

Sediq and Yanikomeroglu (2009) introduced BSC scheme, as an alternative to

the conventional SSC scheme, to be used in cooperative communications when a relay

may use a modulation scheme different than that of the source. The authors provided

BER performance analysis for the SSC and BSC schemes and analytically quantified

the gain achieved by the BSC scheme over the SSC scheme. The DL is considered and

the S → R link is assumed to be reliable and error-free in the BSC scheme. However,

in the UL of the multihop cellular networks, the link between the MS and the RS (that

is S → R link) is normally in NLOS and thus cannot assume to be reliable and error

free. In fact, even if the BER of the R → D link is lower than that of the S → D link,

the BER of the relay based transmissions might still be limited by the BER at the RS.

Therefore, the probability of error at the RS should be taken into account when using

BER as the selection metric to decide on the appropriate link. Thus, an EBSC scheme

is thirdly proposed that takes into account the probability of error at the RS and uses

the e2e BER as the selection metric to decide on the suitable diversity branch.

2.9 SUMMARY

In this chapter, background information on the evolution of the wireless access

networks towards 4G was presented. The multihop relay networks and some relaying

concepts, such as RS usage models, RS forwarding schemes and half-duplex and full-

duplex operation modes, were described. The wireless channel and its impairments on

the signal transmissions were reviewed. The diversity techniques to overcome the

signal fluctuations in the receiver and the diversity combining techniques to improve

the received SINR were discussed. The handover triggering parameters, handover

types, MDHO algorithm, MDHO procedures, and comparison between the various

handover techniques were also discussed. Finally, the literature on the handover in

multihop cellular networks and its relation to the work presented in this research were

comprehensively reviewed.

63

However, the previous studies related to the work presented in this research

were categorized into groups. Relay handover issues were studied in some literature

with the focus on ad hoc networks rather than on cellular networks. Studies on

handover in multihop cellular networks have been mostly devoted to reduce the

handover overhead, handover latency and service interruption time, to propose the

required handover procedure changes and new handover MAC management messages

due to the introduction of RS, and to evaluate the performance of the HHO, FASS,

and conventional MDHO. The focus in these works has been mostly on the HHO.

However, cooperative diversity has emerged as a new form of diversity in wireless

networks and has been studied extensively in the literature. Most of the previous

works on the cooperative diversity have been carried out in a single cell or ad hoc

topology noise-limited environment and for fixed users with the diversity set members

are always a BS and RSs. Cooperation might also increase the resource consumptions.

In the DL of the conventional MDHO, the MS receives only the simultaneous

transmissions of the diversity set members either from a BS and an RS, two RSs, or

two BSs.

On the other hand, in the UL of the conventional MDHO, the SSC scheme

among the received signals is performed for both intra-cell and inter-cell MDHO

scenarios. In the conventional SSC scheme, the link that has the maximum SINR is

selected. The UL performance of the ETSC scheme is limited by the sever co-channel

interference comes from the RSs during the second phase. On the other hand, the BSC

scheme performs better than SSC scheme when different modulation levels are used

on each diversity branch and when the S → R link is reliable. However, BSC scheme

does not take the probability of error at the RS into account. The previous studies

related to the work presented in this thesis which are reviewed in this chapter are

summarized in Table 2.3. In Chapter 3, a new efficient topology-aware DL MDHO

technique for TDD-OFDMA-based interference-limited multihop cellular networks is

proposed. The average DL SINR and the average DL e2e BER for the proposed DL

MDHO technique are formulated and derived in Chapter 3. The simulation model

developed to evaluate the performance of the various handover techniques is also

described in details. In Chapter 5, three new efficient UL schemes for MDHO

technique for TDD-OFDMA-based multihop cellular networks are proposed.

64

Table 2.3 Summary of characteristics of handover techniques in multihop cellular networks

Technique Category Characteristics

Yap et al. (2002); Khedivi et al. (2008)

HO in hybrid networks

• Relaying and HO algorithm for hybrid ad hoc and cellular networks, and hybrid WLAN and cellular networks.

• Decrease unnecessary HO and call drop. • Focus on ad hoc networks.

Ghassemian et al. (2005)

HO in MRAN

• Classifies and studies performance of different kind of HO in MRAN.

• Focuses on ad hoc networks.

Cho et al. (2009) HHO • Introduces various HHO scenarios in multihop cellular networks.

• Presents detailed HO operation and signaling where RSs are deployed either inside a cell or on the boundary between two cells.

Chen et al. (2010)

HHO • HO target is chosen based on MS direction of movement and speed in addition to signal strength.

• Reduces signaling cost, leading to optimized network performance.

Park et al. (2007); Yoo et al. (2009)

HHO • Reduces inter-cell HO and increases intra-cell HO at the MAC-layer and PHY-layer level, respectively.

• Reduces HO signaling overhead and latency. • Yoo. et al. significantly simplifies scanning

procedure than Park et al.

Yang et al. (2008)

HHO • HO protocol for de-centralized MR networks. • High-capability RSs perform some

management functions. • Reduces HO signaling delay and overhead.

Becvar et al. (2008); Becvar and Mach (2010)

HHO, FASS, MDHO

• Proposes and optimizes management messages during HO scanning procedure.

• New method for reporting of scanning results to BS.

• Reduces management information overhead and maximizes the user data throughput.

Sun et al. (2008)

HHO • Mechanism to transform the network topology and performance metrics into power indications.

• Enables fast HO.

continue...

65

continued…

Kim and Cho (2009)

HHO • Pre-buffering scheme where BS performs multicast transmissions to HO candidate RSs.

• Reduces packet loss and service interruption time.

• Might utilize more radio resources for multicasting data to multiple RSs.

Cho et al. (2008) MDHO • RSs are deployed on boundary between two adjacent cells.

• Minimizes HO latency and service interruption time.

• Slightly increases capacity loss.

Zhao et al. (2008)

MDHO, FASS

• Proposes the required changes in MDHO and FASS procedure and new MAC management messages due to introduction of RSs.

• Non-transparent RSs.

Sendonaris et al. (2003a); Laneman et al. (2004); Nabar et al. (2004); Herhold et al. (2005); Onat et al. (2008); Ding et al. 2009; Ikki and Ahmed (2010); Wang et al. (2011)

Cooperative diversity

• Spatial diversity gain, spatial multiplexing gain and power saving, are well-known advantages.

• Cooperative diversity brings together the worlds of conventional relaying and MIMO systems.

• Environment is limited to a single cell or ad hoc topology noise-limited.

• Fixed users with diversity set members are always a BS and RSs.

• Increases resource consumptions.

Oyman (2007); Can et al. (2007)

Relaying; Cooperative diversity

• Uses e2e throughput rather than SNR as a selection metric.

• Single cell or ad hoc topology noise-limited environment.

• Poor UL performance in multi-cell interference-limited environment.

Sediq and Yanikomeroglu 2009

BER-based SC scheme

• Uses BER rather than SNR as a selection metric.

• Performs better than SSC scheme when different modulation levels are used on each diversity branch and when the S → R link is reliable.

• The probability of error at the RS is not taken into account.

CHAPTER III

ANALYTICAL AND SIMULATION MODELS

3.1 INTRODUCTION

In this chapter, the analytical model as well as the simulation model for the proposed

DL MDHO technique for TDD-OFDMA-based interference-limited multihop cellular

networks is presented. First of all, the conceptual and theoretical background on the

related handover techniques of the IEEE 802.16j multihop cellular networks is

described. Next, the average DL SINRs are derived for the various handover

techniques. The MCS’ selection criterion is also provided. After that, the average DL

e2e BERs for the proposed MDHO and the conventional MDHO are derived. The

simulation model employed in the performance evaluation of the various handover

techniques of IEEE 802.16j multihop cellular networks is then presented in details.

Finally, the implementation aspects for the proposed DL MDHO technique in IEEE

802.16j multihop cellular networks are described.

In this chapter, a new efficient topology-aware DL MDHO technique for TDD-

OFDMA-based interference-limited multihop cellular networks is proposed. In the

proposed DL MDHO, the MS receives all the data signals transmitted by the diversity

set members. It ensures that the topology of the diversity set members is always fully

exploited. In addition, the average post-processing DL SINR for the proposed MDHO

technique is formulated and derived. The derived equations express the average DL

SINR as a function of the average SINRs of the S → D links and R → D link as well

as the interference ratio. Therefore, the derived equation can be directly used to

investigate the effect of these different parameters on the average post-processing DL

SINR. The average DL SINR for the proposed MDHO in the noise-limited

67

environment can be obtained from the derived equation for the interference-limited

environment. However, the average DL SINRs of the conventional MDHO, FASS and

HHO can be obtained from the derived SINR of the proposed MDHO. Furthermore, a

closed-form expression for the average DL e2e BER for the proposed MDHO is

derived. The derived equations express the average DL e2e BER as a function of the

average SINRs of the S → R link, S → D links and the R → D link as well as the

interference ratio, and therefore can be directly used to investigate the effect of these

different parameters on the average DL e2e BER. Finally, the various modules of the

developed simulation model are described in details. The developed simulation model

is used to investigate the performance of the various handover techniques in multi-cell

interference-limited environments and for users with high speeds. The implementation

aspects for the proposed DL MDHO technique in the IEEE 802.16j multihop cellular

networks are also described.

3.2 THEORETICAL AND CONCEPTUAL BACKGROUND

This section briefly describes the theoretical and conceptual background on the

considered handover techniques of IEEE 802.16j multihop cellular networks. The

transmission sequences and the MCS selection criterion are also discussed.

3.2.1 Hard Handover (HHO)

In HHO, the MS communicates only with one access station at any one time, which is

called the serving station. If the serving station is an RS, the RS receives the

transmission of the BS during the first phase, and the RS transmits to the MS during

the second phase. Otherwise, the MS relies solely on the signal transmitted by the BS.

3.2.2 Fast Access Station Switching (FASS)

In FASS, the MS and BS maintain a list of the access stations that are involved in

FASS with the MS. This list is called a diversity set. The MS communicates with the

anchor station of the diversity set for all UL and DL messages including management

and traffic connections. The anchor station can be a BS or RS and can be changed

68

from frame to frame depending on the access station selection mechanism. In this

research, the anchor station is the diversity set member with the highest average

SINR. The transmission sequences in FASS are similar to that for HHO.

3.2.3 Macro Diversity Handover (MDHO)

In MDHO, the diversity set is maintained by the MS and BS just like to FASS. The

MS communicates with all access stations in the diversity set. Due to the introduction

of RSs in the cellular networks, different intra-cell and inter-cell MDHO scenarios

occur as shown in Figure 3.1. In this study, the MDHO scenarios are further classified

into two main cases. The first case comprises the MDHO scenarios 1 and 2 in which

two different-topology access stations are included into the diversity set of the MS, for

instance BS and RS, and it is called as case 1. On the other hand, the second case

corresponds to the MDHO scenarios 3, 4 and 5 in which the diversity set members of

the MS are two similar-topology access stations, for instance two RSs or two BSs, and

it is called as case 2. Table 3.1 describes the different intra-cell and inter-cell scenarios

of the MDHO technique. In this research, the topology of the access station refers to

the number of hops, and the diversity set size is assumed to be two.

Figure 3.1 MDHO scenarios in multihop cellular networks

BS2

MS2 MS1 MS5

MS4

Cell 1 Cell 2

Scenario 1 Scenario 5

RS3

RS1

RS4

RS8

RS5 MS3

BS1

RS6

RS7 RS2

Scenario 3

Scenario 2

Scenario 4

69

Table 3.1 Description of intra-cell and inter-cell scenarios of MDHO technique

Case Scenario Characteristics

Case 1 Scenario 1 Intra-cell MDHO scenario. The diversity set members are a BS and an RS within the same cell, for instance BS2 and RS7 shown in Figure 3.1. The BS can easily control the handover process because the RS is under its own control. Inter-BS information or signaling is not required.

Case 1 Scenario 2 Inter-cell MDHO. The diversity set members are a BS and an RS in different cells, for instance BS2 and RS3 shown in Figure 3.1. The handover process is controlled by the two BSs of the two cells. Inter-BS information and signaling is required in addition to the signaling between the BS1 and its associated RS3.

Case 2 Scenario 3 Intra-cell MDHO scenario. The diversity set members are two RSs within the same cell, for instance RS1 and RS2 shown in Figure 3.1. The BS can easily control the handover process because the two RSs are under its own control. Inter-BS information and signaling is not required.

Case 2 Scenario 4 Inter-cell MDHO scenario. The diversity set members are two RSs in two different cells, for instance RS4 and RS5 shown in Figure 3.1. The handover process is controlled by the two BSs of the two cells. Inter-BS information and signaling is required in addition to the signaling between each RS and its controlling BS.

Case 2 Scenario 5 Inter-cell MDHO scenario. The diversity set members are two BSs in different cells, for instance BS1 and BS2 shown in Figure 3.1. Inter-BS information and signaling is required.

(a) Case 1 of the Proposed MDHO

The diversity set members in this case are two different-topology access stations, that

is a BS and an RS. The transmission sequence in this case is illustrated in Figure 3.2.

In the first phase, the MS and the RS receive and buffer the transmission of the BS. In

the second phase, both the BS and the RS transmit synchronously to the MS by using

the same radio resource. At the end of the two phases, the MS combines the signals

received during the first phase and the second phase using MRC. Since the MS

receives during both phases in this case, the same MCS should be used over the two

phases.

70

Figure 3.2 Transmission sequence of case 1 of the proposed MDHO

(b) Case 1 of the Conventional MDHO

Case 1 of the conventional MDHO can be viewed as a subset of case 1 of the

proposed MDHO wherein the MS does not exploit the signal transmitted by the BS in

the first phase even though the MS is idle in this phase. Hence, only the RS receives

the transmission of the BS during the first phase, as shown in Figure 3.3. During the

second phase, both the BS and the RS transmit synchronously to the MS by using the

same radio resource. At the end of the two phases, the MS combines the signals

received during the second phase using MRC. Since the MS does not receive any

signal during the first phase, the MCS in each phase can be adjusted independently.

Figure 3.3 Transmission sequence of case 1 of the conventional MDHO

BS2

RS7

MS5

1,SDh2,SDh

2,RDh 1,SRh

First phase

Second phase

RS7

BS2

MS5

2,SDh

1,SRh

2,RDh

First phase

Second phase

71

(c) Case 2 of the Proposed MDHO and the Conventional MDHO

The diversity set members in this case are two similar-topology access stations, that is

either two RSs, for instance RS1 and RS2, or two BSs, for instance BS1 and BS2. In

this case, the proposed MDHO and the conventional MDHO perform similarly. On the

one hand; if the diversity set members are RS1 and RS2, the transmission sequences

are as follows, as shown in Figure 3.4. Both RS1 and RS2 receive the transmission of

the BS in the first phase, whereas in the second phase, both RS1 and RS2 transmit

synchronously to the MS by using the same radio resource. At the end of the two

phases, the MS combines the signals received from RS1 and RS2 using MRC. As the

MS does not receive any signal during the first phase, the MCSs in the first and

second phases are adjusted independently according to the average received SINR at

the RSs and the MS, respectively. On the other hand, if the diversity set members are

BS1 and BS2, both BS1 and BS2 transmit synchronously to the MS by using the same

radio resource, as shown in Figure 3.5. The MS combines the signals received from

BS1 and BS2 using MRC. It is to be noted here that both BSs can transmit to the MS

during the first phase and/or the second phase. However, since the focus in the DL of

this research is on the spatial diversity gain, it is assumed in this research that the BSs’

transmissions occur in the second phase.

Figure 3.4 Transmission sequence of case 2 of the conventional MDHO and the proposed MDHO when the diversity set members are RS1 and RS2

BS1

RS2

MS1

RS1

1,1SRh

1,2SRh

2,1DRh

2,2DRh

First phase

Second phase

72

Figure 3.5 Transmission sequence of case 2 of the conventional MDHO and the proposed MDHO when the diversity set members are BS1 and BS2

Figure 3.6 summarizes the transmission sequences of the conventional MDHO

and the proposed MDHO during the two phases.

Figure 3.6 Comparison between the conventional MDHO and the proposed

MDHO techniques. A → B denotes data communications between terminals A and B

Con

vent

iona

l MD

HO

Prop

osed

MD

HO

First phase

Second phase

jDSh ,1

BS1

MS2

BS2

jDSh ,2

73

Notation: The term ][⋅ denotes the expectation operator. The superscripts T and *

stand for transpose and conjugate operations, respectively. Bold uppercase letters

represent matrices while bold lowercase letters represent vectors. The subscript j

where }2,1{∈j , represents the phase index. 2|||| FH is the squared Frobenius norm of

H , and ( )Htr is the trace of matrix H . The term ( )a,0 represents a circularly

symmetric complex Gaussian random variable with zero mean and variance a .

3.3 BASEBAND CHANNEL AND SIGNAL MODELS

In this section, the average post-processing DL SINR is derived for the handover

techniques. The RS uses DF scheme where the signal received from the source

terminal is demodulated and decoded before retransmission. Note that this study can

be extended to the case where forwarding schemes other than DF are used by the RS.

The source is a BS, whereas the destination is an MS. The diversity set members are

assumed to be perfectly synchronized and the MS is assumed to be equipped with

multiple receive antennas (Andrews et al. 2007). The complex-valued constellation

points transmitted by the source terminal during the first and second phases at a given

sub-carrier are denoted as 1x and 2x , respectively. It is assumed that the mean and

the variance of jx are given by ][ jx 0 and =]|[| 2jx 1, respectively, for =j 1, 2.

3.3.1 Baseband Channel

Let jSRh , , jSDh , and jRDh , denote the channel coefficients at a given sub-carrier during

phase j for S → R, S → D and R → D links, with variances 2, jSRσ , 2

, jSDσ and 2, jRDσ

respectively. jSRh , , jSDh , and jRDh , are modeled as independent identically

distributed (i.i.d.) Rayleigh flat fading random variable, and thus 2

, jSRh , 2

, jSDh and

2

, jRDh are exponentially distributed random variables. However, some of the analysis

in this research is general and not limited to Rayleigh distribution. The term 2, jABσ

accounts for the distance-dependent path loss and lognormal shadow fading of the

74

A → B link during phase j . Since the shadow fading changes very slowly, the

average channel coefficients 2, jABσ are assumed to remain fixed during the first phase

and the second phase and hence the phase index can be dropped. At a given sub-

carrier, jRn , ~ ( )jRI ,,0 and jDn , ~ ( )jDI ,,0 capture the effects of the additive

white Gaussian noise (AWGN)-plus-interference samples observed during phase j at

the relay and destination terminals, respectively. SP and RP denote the fixed transmit

signal power at a given sub-carrier of the source and relay terminals, respectively.

Since the handover is based on the large-scale fading and because of the mobility of

users, only average SINR is of particular interest. Hence, the instantaneous SINRs at a

given sub-carrier of S → R, S → D and R → D links are given by

1,2

1,1, RSSRSR IPh=γ , jDSjSDjSD IPh ,

2

,, =γ and 2,2

2,2, DRRDRD IPh=γ ,

respectively. In addition, at a given sub-carrier, the average SINRs of S → R, S → D

and R → D links are given by 1,SRγ 1,2

1, RSSR IPh 1,2

RSSR IPσ , jSD,γ

jDSjSD IPh ,

2

, jDSSD IP ,2σ and 2,RDγ 2,

22, DRRD IPh 2,

2DRRD IPσ ,

respectively. The PDF of ABγ is then given by ( ) ( ) ( )ABABABABf γγγγ −= exp1 for

0≥ABγ .

3.3.2 Case 1 of the Conventional MDHO

In this case, the diversity set members of the MS are a BS and an RS. The signal

received at the relay terminal in the first phase is given by:

1,11,1, RSRSR nxhPy += (3.1)

Assuming that the relay terminal correctly decodes the signals transmitted by the

source terminal during the first phase, the destination terminal receives a superposition

of the relay transmission and the source transmission during the second phase

according to:

2,12,22,2, DRDRSDSD nxhPxhPy ++= (3.2)

75

The effective input-output relations for case 1 of the conventional MDHO may

now be written as:

=1casey gx n+ (3.3)

where 2,1 Dcase yy = is the received signal, x Txx ][ 21= is the transmitted signal

vector, 2,Dnn = is the noise-plus-interference sample and g is the effective channel

gain vector given by:

g [ ]2,2, SDSRDR hPhP= (3.4)

Note that the variance of the noise-plus-interference is the same for both

diversity branches in this case. From Equation (3.4), it is clear that knowledge of 1,SRh

and 1,SDh is not required at the destination terminal. Assuming the MS has perfect

knowledge of channel coefficients 2,SDh and 2,RDh , at a given sub-carrier the average

post-processing SINR achieved at the MS after MRC is given by:

=1CMDHOpostγ

g F2,DI 2,2, RDSD γγ += (3.5)

In this case, the MCS is selected based on 1,SRγ in the first phase, whereas it is

decided based on 1CMDHOpostγ in the second phase.

3.3.3 Case 1 of the Proposed MDHO

In this case, the diversity set members of the MS are also a BS and an RS. Case 1 of

the proposed MDHO is similar to case 1 of the conventional MDHO except that the

MS receives the signal transmitted by the BS during the first phase in addition to the

simultaneous transmissions of the BS and RS during the second phase. Hence, the

signal received at the relay terminal in the first phase is given by Equation (3.1),

76

whereas the simultaneous transmission signals received at the destination terminal in

the second phase are given by Equation (3.2). On the other hand, the signal received at

the destination terminal in the first phase is given by:

1,11,1, DSDSD nxhPy += (3.6)

In a matrix notation, the effective input-output relations for case 1 of the

proposed MDHO may now be written as:

ycase1 Hx n (3.7)

where ycase1T

DD yy ][ 2,1, is the received signal vector, H is the effective channel

gain matrix given by:

H⎥⎥⎦

⎤

⎢⎢⎣

⎡

2,2,

1, 0

SDSRDR

SDS

hPhPhP (3.8)

x Txx ][ 21 is the transmitted signal vector and n TDD nn ][ 2,1, is the noise-

plus-interference sample.

Note that the noise-plus-interference variances for the diversity branches may

not be the same in this case. From Equation (3.8) it is also obvious that knowledge of

1,SRh is not required at the destination terminal. Assuming that the MS has perfect

knowledge of channel coefficients jSDh , and 2,RDh and assuming perfect MRC in

which the weight of each diversity branch is the conjugate of the branch channel

coefficient normalized to the noise-plus-interference variance of that branch (Ko et al.

2003), the MS linearly combines the received signal vector ycase1 by the receive

weight matrix Hw, that is

Hwycase1 HwHx n (3.9)

77

where Hw is defined as:

Hw⎥⎥⎦

⎤

⎢⎢⎣

⎡

2,*

2,

2,*

2,1,*

1,

0 DSDS

DRDRDSDS

IhPIhPIhP (3.10)

and n~ Hwn.

Hence, at a given sub-carrier, the average post-processing SINR obtained at

the MS after MRC can be derived from Equation (3.8), (3.9) and (3.10) as:

=1PMDHOpostγ tr HwH 2|n|2

2,2,1, RDSDSD γγγ ++= (3.11)

It should be noted that the relationship between 1,SDγ and 2,SDγ can be given by:

1,2, SDSD γργ = (3.12)

where 2,1, DD II=ρ is the ratio of the variance of the noise-plus-interference during

the first phase to the variance of the noise-plus-interference during the second phase.

Furthermore, the SINR in the noise-limited environment is a special case of that of the

interference-limited environment wherein 1=ρ .

In this case, for both first phase and second phase the MCS is determined

based on { }11, ,min PMDHO

postSR γγ .

3.3.4 Case 2 of the Proposed MDHO and the Conventional MDHO

If the diversity set members of the MS are two RSs, that is RS1 and RS2, the signals

received at the relay terminals in the first phase are identical to that for case 1 and are,

thus, calculated by Equation (3.1). Assuming that RS1 and RS2 correctly decode the

78

signals transmitted by the source terminal during the first phase, the signal received at

the destination terminal during the second phase is given by:

2,12,212,12, DDRRDRRD nxhPxhPy ++= (3.13)

The effective input-output relations for case 2 of the proposed and

conventional MDHO can be summarized as:

2casey h nx+ (3.14)

where 2,2 Dcase yy = is the received signal, h is the effective channel gain vector given

by:

h [ ]2,22,1 DRRDRR hPhP (3.15)

1xx = is the transmitted signal and 2,Dnn = is the noise-plus-interference sample.

Note that the noise-plus-interference variance is identical for both diversity

branches in this case. At a given sub-carrier, the average post-processing SINR

achieved at the MS after MRC can be derived as:

=rcasepost

2γh F

2

2,DI 2,22,1 DRDR γγ += (3.16)

where 2,RiDγ is the average SINR of RSi → MS link.

In this case, the MCS is decided based on { }1,21,1 ,min SRSR γγ in the first phase,

whereas it is chosen based on rcasepost

2γ in the second phase, where 1,SRiγ is the average

SINR of the BS → RSi link.

79

Similarly, if the diversity set members of the MS are two BSs, that is BS1 and

BS2, the average post-processing SINR at the MS after MRC can be derived as:

2,22,12

DSDSbcase

post γγγ += (3.17)

where 2,SiDγ is the average SINR of the BSi → MS link. In this case, the MCS is

adjusted based on bcasepost

2γ .

3.3.5 Fast Access Station Switching

In FASS, if the anchor station is a BS, the average SINR at the MS is equal to 2,SDγ .

Otherwise; the anchor station is a RS and the average SINR at the MS is equal to

2,RDγ .

3.3.6 Hard Handover

During HHO, if the MS is connected to the BS, the average SINR at the MS is equal

to 2,SDγ . Otherwise; the MS relies solely on the signal transmitted by the RS and the

average SINR at the MS is equal to 2,RDγ .

3.4 DERIVATION AND ANALYSIS OF THE BIT ERROR RATE

In this section, closed-form expressions for the BER of the proposed MDHO and the

conventional MDHO are derived. Since the conventional MDHO and the proposed

MDHO differ mainly in case 1 while they perform similarly in case 2, the BER

analysis in the rest of this section will be conducted for case 1 only in which the

diversity set members of the MS are a BS and an RS. For simplicity and convenience

of presentation, it is assumed that all the links use binary phase shift keying (BPSK)

modulation. However, the analysis can be extended to M-ary phase shift keying

(MPSK) modulations. It is assumed that the CSI is available at the receiver terminal

for all the links and the signals are demodulated coherently. Note that the BER

80

analysis in this section is based on the methodology used in Onat et al. (2008) and Ikki

and Ahmed (2007) for the BER analysis of the cooperative diversity in which the RS

always transmits, that is quite similar to the BER of the conventional MDHO.

However, different approach is used in this thesis to derive the probability of error

propagation for the conventional MDHO and proposed MDHO.

The error event in the A → B link is denoted by ABe . The probability of error

conditioned on the instantaneous link SINR and the average link SINR are denoted by

( )γ|e and ( )e , respectively. Note that for most Gray bit-mapped modulation

schemes employed in practical systems, the instantaneous bit error probability can be

expressed as ( ) ( )γβαγ MM Qe =| , where Mα and Mβ are constants that depend

on the modulation type, and )( xQ is the Q-function defined as

( ) ( ) ( ) dttxQx∫∞

−= 2exp21 2π . However, Mα depends on the number of

nearest neighbors to a constellation at the minimum distance and Mβ depends on the

minimum distance in the constellation.

Hence, based on ( )γ|e general expression, the average bit error probability

in point-to-point links under Rayleigh fading can be calculated as (Goldsmith 2005):

( )e γ ( )[ ]⎥⎥⎦

⎤

⎢⎢⎣

⎡

+−=

γβγβαγβαM

MMMM Q

21

2 (3.18)

For BPSK modulation which is considered in this analysis, ( ) =MM βα , (1, 2)

and the exact expression is:

( ) ( )ABABAB Qe γγ 2| = (3.19)

( ) ⎟⎟⎠

⎞⎜⎜⎝

⎛

+−=

AB

ABABe

γγ

11

21 (3.20)

81

The average e2e BER of the conventional MDHO and the proposed MDHO

are denoted by 1,2

CMDHOavgeeBER and 1

,2PMDHO

avgeeBER , respectively.

3.4.1 Derivation of the Bit Error Rate for the Proposed MDHO

The average e2e BER for case 1 of the proposed MDHO can be expressed using the

law of the total probability as:

1,2

PMDHOavgeeBER ( )SRe ( )1PMDHO

prope ( −1 ( ))SRe ( )1PMDHOdive (3.21)

where ( )SRe is the average probability of error at the RS that is given by Equation

(3.20), ( )1PMDHOprope is the average probability that an error occurs in the diversity

transmissions from the source and the relay to the destination given that the relay

decoded unsuccessfully which is referred to as error propagation, and ( )1PMDHOdive is

the average probability that an error happens in the diversity transmissions from the

source and the relay given that the RS decoded correctly which is referred to as

diversity error.

3.4.2 Probability of Diversity Error for the Proposed MDHO

Since the destination employs perfect MRC, the SINR after MRC is the sum of the

SINRs of the S → D signals, received during the first and second phases, and the

R → D signal, received during the second phase. Hence, the instantaneous probability

of diversity error conditioned on the SINR at the combiner’s output can be written as:

( ) ( )equequPMDHOdiv Qe γγ 2|1 = (3.22)

where equγ is the equivalent instantaneous SINR at the combiner’s output which is

given by 2,2,1, RDSDSDequ γγγγ ++= .

82

The average probability of diversity error is derived by determining the PDF of

equγ and then averaging the conditional error probability in Equation (3.22) over this

PDF. Hence, the average probability of diversity error can be written as:

( )1PMDHOdive γequ

( )equPMDHOdive γ|1

( ) ( ) equequequ dfQ γγγ∫∞

=0

2 (3.23)

where ( )equf γ is the PDF of equγ .

Actually, the average probability of diversity error in Equation (3.23) is equal

to the BER of a 3-branch MRC receiver in Rayleigh fading, which is given as (Proakis

& Salehi 2008):

( )( ) ( ) ( )

⎪⎪⎪

⎩

⎪⎪⎪

⎨

⎧

≠≠

⎥⎥⎦

⎤

⎢⎢⎣

⎡

⎟⎟⎠

⎞⎜⎜⎝

⎛

+−+⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−+⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−

==⎥⎦⎤

⎢⎣⎡ ++++⎥⎦

⎤⎢⎣⎡ −

=

2,2,1,

2,

2,3

2,

2,2

1,

1,1

2,2,1,2

3

1

,1

11

11

121

,1231

2311

21

RDSDSD

RD

RD

SD

SD

SD

SD

RDSDSD

PMDHOdive

γγγ

γγ

πγ

γπ

γγ

π

γγγμμμ

(3.24)

where μ , 1π , 2π and 3π are given by:

1,

1,

1 SD

SD

γγ

μ+

=

⎟⎟⎠

⎞⎜⎜⎝

⎛−⎟

⎟⎠

⎞⎜⎜⎝

⎛−

=

1,

2,

1,

2,1

11

1

SD

RD

SD

SD

γγ

γγ

π

⎟⎟⎠

⎞⎜⎜⎝

⎛−⎟

⎟⎠

⎞⎜⎜⎝

⎛−

=

2,

2,

2,

1,2

11

1

SD

RD

SD

SD

γγ

γγ

π

⎟⎟⎠

⎞⎜⎜⎝

⎛−⎟

⎟⎠

⎞⎜⎜⎝

⎛−

=

2,

2,

2,

1,3

11

1

RD

SD

RD

SD

γγ

γγ

π (3.25)

83

3.4.3 Probability of Error Propagation for the Proposed MDHO

Since BPSK is assumed; without loss of generality, it is assumed that the source sends

the symbol 1, +=jsx and the relay sends the symbol 12, −=rx . The error occurs if the

destination decides that 1− was sent by the source. For convenience of presentation,

Equation (3.2) and (3.6) are rewritten to express all signals received at the destination

terminal. Hence, the signals received from the source and the relay are expressed as:

1,1,1,1, DsSDSSD nxhPy +=

2,2,2,2, DsSDSSD nxhPy +=

2,2,2,2, DrRDRRD nxhPy += (3.26)

Note in this case that jsr xx ,2, −= . After combining the received signals using perfect

MRC, the following decision variable can be used:

2,2,

*2,

2,2,

*2,

1,1,

*1,1

RDD

RDRSD

D

SDSSD

D

SDSPMDHOMRC y

IhP

yI

hPy

IhP

y ++=

2,2,

*2,

2,2,

*2,

1,1,

*1,

1,2,

22,

2,

22,

1,

21,

DD

RDRD

D

SDSD

D

SDS

sD

RRD

D

SSD

D

SSD

nI

hPn

IhP

nI

hP

xI

PhI

PhI

Ph

+++

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛−+=

( ) nRDSDSD~

2,2,1, +−+= γγγ (3.27)

where n~ is the effective noise-plus-interference, which is also a Gaussian random

variable with zero mean and variance equal to )(21]|~[| 2,2,1,

2RDSDSDn γγγ ++= .

Since the destination assumes that both the source and the relay send the same

symbol, the decision rule at the destination is to declare 1− if 01 <PMDHOMRCy .

84

The probability of error propagation conditioned the instantaneous SINRs of

the R → D and S → D links can be given by:

( ) =2,2,1,1 ,,| RDSDSD

PMDHOprope γγγ ( )2,2,1,

1 ,,|0 RDSDSDPMDHOMRCy γγγ<

( )( )2,2,1,2,2,1, ,,|~RDSDSDRDSDSDn γγγγγγ −+>

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

++

−+=

2)( 2,2,1,

2,2,1,

RDSDSD

RDSDSDQγγγγγγ

(3.28)

Then, the average probability of error propagation is given by:

( ) =1PMDHOprope

2,2,1, ,, RDSDSD γγγ ( )2,2,1,1 ,,| RDSDSD

PMDHOprope γγγ

( ) ( ) ( ) 2,2,1,2,2,1,

0 0 02,2,1,

2,2,1,

2)(

RDSDSDRDSDSD

RDSDSD

RDSDSD

dddfff

Q

γγγγγγ

γγγγγγ

∫ ∫ ∫∞ ∞ ∞

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

−+−+

= (3.29)

Owing to the difficulty of finding the exact expression given in Equation

(3.29), an approximate expression for calculating the average probability of error

propagation in Equation (3.29) is provided. Assuming that the incorrect relay signal

and not the noise-plus-interference term is the dominant factor that causes the decision

variable 1PMDHOMRCy to be negative, the probability of error propagation is approximated

by the probability of ( ){ }02,2,1, <−+= RDSDSD γγγγ .

The characteristic function of a random variable aγ is defined as

( ) ( ) ( ) aajj dfejeE a

a

a γγυψ υγγ

υγ ∫∞

∞−=≡ (Proakis & Salehi 2008). Accordingly,

the characteristic function of a single branch SINR, aγ , with Rayleigh fading channel

can be expressed as ( ))1(

1

ajj

a γυυψ γ −

= . Since γ is the sum of 3 statistically

independent components, the characteristic function of γ is:

85

( ) ( )( )( )2,2,1, 1111

RDSDSD jjjj

γυγυγυυψ γ +−−

= (3.30)

and thus the PDF of γ can be found by taking the inverse Fourier transform to the

characteristic function in Equation (3.30). This can be achieved by first resolving

Equation (3.30) into partial fractions, applying inverse Fourier transform and then

using Gradshteyn and Ryzhik (2007) to evaluate the integral parts. Afterwards, the

PDF of γ can be expressed as:

( ) 2,2,1,

2,

1

2,

1

1,

1 RDSDSD eCeBeAfRDSDSD

γγγγγγ

γγγγ ++= −− (3.31)

where 1A , 1B and 1C are given by:

⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟

⎟⎠

⎞⎜⎜⎝

⎛−

=

1,

2,

1,

2,1

11

1

SD

RD

SD

SD

A

γγ

γγ

⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟

⎟⎠

⎞⎜⎜⎝

⎛−

=

2,

2,

2,

1,1

11

1

SD

RD

SD

SD

B

γγ

γγ

⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟

⎟⎠

⎞⎜⎜⎝

⎛+

=

2,

2,

2,

1,1

11

1

RD

SD

RD

SD

C

γγ

γγ

(3.32)

When the error probability ( )γ2Q is averaged over the PDF given in

Equation (3.31), the approximate expression for the average probability of error

propagation can be expressed as ( 12, >>RDγ ):

( )⎥⎥⎦

⎤

⎢⎢⎣

⎡

⎟⎟⎠

⎞⎜⎜⎝

⎛

−++⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−+⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−=

11

11

11

21

2,

2,1

2,

2,1

1,

1,1

1

RD

RD

SD

SD

SD

SDPMDHOprop CBAe

γγ

γγ

γγ

(3.33)

86

Thus, the average e2e BER for case 1 of the proposed MDHO can be directly

calculated by substituting Equation (3.20), (3.24) and (3.33) into Equation (3.21).

3.4.4 Derivation of the Bit Error Rate for the Conventional MDHO

The average e2e BER for case 1 of the conventional MDHO can be expressed using

the law of the total probability as:

=1,2

CMDHOavgeeBER ( )SRe ( ) ( −+ 11CMDHO

prope ( ))SRe ( )1CMDHOdive (3.34)

3.4.5 Probability of Diversity Error for the Conventional MDHO

Similarly, the average probability of diversity error for case 1 of the conventional

MDHO is equal to the BER of a 2-branch MRC receiver in Rayleigh fading, which is

given as (Proakis & Salehi 2008):

( )

⎪⎪⎪

⎩

⎪⎪⎪

⎨

⎧

≠

⎥⎥⎦

⎤

⎢⎢⎣

⎡

⎟⎟⎠

⎞⎜⎜⎝

⎛

+−

−+⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−

−

=⎟⎟⎠

⎞⎜⎜⎝

⎛

++⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−

=

2,2,

2,

2,

2,2,

2,

2,

2,

2,2,

2,

2,2,2,

2,

2

2,

2,

1

,1

11

121

,12

111

121

RDSD

RD

RD

SDRD

RD

SD

SD

RDSD

SD

RDSDSD

SD

SD

SD

CMDHOdive

γγ

γγ

γγγ

γγ

γγγ

γγγ

γγ

γ

(3.35)

3.4.6 Probability of Error Propagation for the Conventional MDHO

Using similar derivation method to that for case 1 of the proposed MDHO, the

approximate expression for the average probability of error propagation for case 1 of

the conventional MDHO can be derived as ( 12, >>RDγ ):

( )⎥⎥⎦

⎤

⎢⎢⎣

⎡

⎟⎟⎠

⎞⎜⎜⎝

⎛

−++⎟

⎟⎠

⎞⎜⎜⎝

⎛

+−=

11

11

21

2,

2,2

2,

2,2

1

RD

RD

SD

SDCMDHOprop CBe

γγ

γγ (3.36)

87

where 2B and 2C are given by:

2,

2,2

1

1

SD

RDB

γγ

+=

2,

2,2

1

1

RD

SDC

γγ

+= (3.37)

Hence, the average e2e BER for case 1 of the conventional MDHO can be

calculated directly by substituting Equation (3.20), (3.35) and (3.36) into Equation

(3.34). It is to be mentioned here that there are only two diversity branches in case of

the conventional MDHO, whereas there are three diversity branches in case of the

proposed MDHO.

3.5 SIMULATION MODEL

In this section, the simulation model employed in the performance evaluation of the

various handover techniques of IEEE 802.16j multihop cellular networks is presented

in details. The purpose of performing the simulation is to validate the superiority of

the proposed MDHO in a multi-cell interference-limited environment and to

investigate the impacts of varying some system parameters on the performance of the

various handover techniques. The system-level simulation model is used to investigate

the impacts of the MS mobility speed, RS transmitted power and the relative RS

location on the performance of the various handover techniques as will be shown in

Chapter 4. In this section, the network model, the AMC and the MDHO algorithm are

first described. After that, the propagation model and the interference model are

presented. The performance evaluation metrics that are the outputs of the simulation

are then pointed out. Finally, the simulation flowchart is explained.

88

3.5.1 Network Model

The DL of IEEE 802.16j TDD-OFDMA-based interference-limited multihop wireless

relay network that consists of seven hexagonal cells is considered. Each cell has one

BS located at its centre and six FRSs. Each RS is located on the line that connects the

centre of the cell to one of the six cell vertices. Figure 3.7 shows the positions of the

BS and the RSs in one cell, where the RS is located at a 2/3 position ( ,32 rd where

rd is the cell radius) between the BS and the cell boundary. Using these RSs positions

design, the BSs and the RSs are spread out evenly over the hexagonal layout.

However, due to the finite number of cells when considering the 7-cell hexagonal

network only, accurate level of inter-cell interference cannot be captured in the model

(Park et al. 2009). In fact, when considering the 7-cell hexagonal network only, the

centre cell has 6 first-tier interfering sources, whereas each of the other 6 cells has 3

first-tier interfering sources. Hence, data can be collected from the centre cell only. It

is desirable to collect data from cells other than the centre cell to speed up data

collection and hence speed up the simulation. It is also necessary to account for the

mobility of users particularly on the boundary between two adjacent cells which is

called a boundary effect. In order to account for all these needs, the so-called wrap-

around technique is considered. The wrap-around technique is done by extending the

network to a cluster of network consisting of 7 copies of the original hexagonal

network, with the original hexagonal network in the middle while the other 6 copies

are attached to it symmetrically on 6 sides (Huo 2005), as shown in Figure 3.8. The

link between the BS and RS is called a relay link, whereas the links between the BS

and MS and between the RS and MS are called access links. It is assumed that the RS

uses DF forwarding scheme.

Figure 3.7 The positions of the BS and the RSs in one cell

3/2 rd 3/rd

89

Figure 3.8 Simulated network layout for the MDHO DL performance

It is assumed that 30 MSs (WiMAX Forum 2006a) are uniformly distributed

throughout each cell and move along a direction randomly selected in each frame

using the modified random direction mobility model (Camp et al. 2002). The initial

direction of each MS in degrees is generated randomly by the uniform distribution in

the range [0, 45, 90, 135, 180, 225, 270, 315] degrees. The new direction of each MS

is selected randomly in the range [-45, 0, 45] degrees related to the previous direction.

The mobility model is illustrated in Figure 3.9. Meanwhile, the full-buffer traffic

model is considered in which each MS always has data to send or receive in the buffer

(Senarath et al. 2007).

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

x 104

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1x 104

BS-RS-MS x-location (m)

BS

-RS

-MS

y-lo

catio

n (m

)

BSRSMS

90

Position update with direction change

dirC : Current direction

New direction o45−dirC

o45+dirC

Position update with no direction change

Figure 3.9 Mobility model for macrocellular environment

It is assumed that the RS is equipped with a 6-sector directional antenna to

communicate with the BS, whereas it is equipped with an omni-directional antenna to

communicate with the MSs (Lee & Cho 2007; Lin et al. 2007). However, the BSs and

the MSs are assumed to be equipped with omni-directional antennas. The antenna

pattern for the 6-sector directional antenna is specified as (Senarath et al. 2007):

( )⎥⎥⎦

⎤

⎢⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛⋅−= m

dB

AA ,12min2

3θθθ dBi (3.38)

where ( )θA is the antenna gain in dBi in the direction of θ , °≤≤°− 180180 θ , θ is

the angle between the direction of interest and the steering direction of the antenna, o353 =dBθ is the 3 dB beamwidth for the 6-sector antenna and =mA 23 dB is the

maximum attenuation (front-to-back ratio) for the 6-sector antenna. Figure 3.10 shows

the antenna pattern for the 6-sector antenna used for each RS.

91

Figure 3.10 Antenna pattern for the 6-sector antenna used for each RS

The system operates at a carrier frequency of 2.5 GHz and frames have 5 ms

duration. The scalable OFDMA mode with 1024 sub-carriers and a system bandwidth

of 10 MHz are considered. A sub-carrier spacing of 11.2 MHz/1024 = 10.94 kHz is

assumed. Each frame is composed of 48 symbols. A DL/UL ratio of 3:1 is assumed,

that is, 28 symbols for the DL subframe and 9 symbols for the UL subframe (WiMAX

Forum 2006b). The remaining symbols are used for preamble and control information.

Each user is allocated one sub-channel, which is defined as a set of 48 sub-carriers.

The PUSC sub-carrier permutation scheme is considered. In this scheme, the sub-

carriers belonging to a sub-channel are distributed randomly over the entire bandwidth

to provide frequency diversity and average the inter-cell interference. In this case, the

SINR of each sub-channel is determined by taking the average of all corresponding

sub-carriers. There are 420 sub-channels available for allocation in each DL subframe

(e.g., 720 sub-carrier/symbol × 1 sub-channel/48 sub-carriers × 28 symbols/frame =

420 sub-channels/frame). On the other hand, the number of sub-channels available for

-200 -150 -100 -50 0 50 100 150 200-25

-20

-15

-10

-5

0

Angle, θ (degree)

Gai

n (d

B)

-3 dB

Am

θ3dB

92

allocation in each UL subframe is 105 sub-channels (560 sub-carrier/symbol × 1 sub-

channel/48 sub-carriers × 9 symbols/frame = 105 sub-channels/frame). The OFDMA

parameters are listed in Table 3.2 below. The frequency reuse factor is assumed to be

unity. Perfect time and frequency synchronizations are assumed. The main simulation

parameters are listed in Table 3.3. They are typical data widely considered in

simulating cellular networks. Note that the mobile WiMAX system profile defined by

WiMAX Forum is used as a reference (WiMAX Forum 2006b; WiMAX Forum

2006c).

Table 3.2 OFDMA parameters

Parameter Value

Channel bandwidth 10 MHz

Sampling frequency 11.2 MHz

FFT size 1024

Sub-carrier frequency spacing 10.94 kHz

Useful symbol time 91.4 µsec

Guard time 11.4 µsec

OFDMA symbol duration 102.9 µsec

Frame duration 5 msec

Number of OFDMA symbols 48

DL PUSC

Null sub-carriers 184

Pilot sub-carriers 120

Data sub-carriers 720

UL PUSC

Null sub-carriers 184

Pilot sub-carriers 280

Data sub-carriers 560

Source: WiMAX Forum 2006b

93

Table 3.3 Simulation parameters

Parameter Value

Cell radius 1400 m

Number of hops 2

Duplex mode TDD

Multiple access scheme OFDMA

Frequency reuse factor 1

Carrier frequency 2.5 GHz

HHO threshold 3 dB

MDHO add threshold 3 dB

Diversity set size 2

BS transmitted power 43 dBm

MS transmitted power 23 dBm

BS antenna height 32 m

RS antenna height 15 m

MS antenna height 1.5 m

BS antenna gain 14 dBi (omni)

RS antenna gain 12 dBi (omni), 18 dBi (directional)

MS antenna gain 0 dBi (omni)

Traffic model Full buffer

Relay links propagation model Desired links: Free space LOS model Interfering links: Modified IEEE 802.16 model terrain type C

Access links propagation model COST 231 Hata model

Shadowing standard deviation Relay links= 3.4 dB; Access links=8 dB

Shadowing de-correlation distance

20 m

BS/RS noise figure 4 dB

MS noise figure 7 dB

Simulation trials 2000

94

3.5.2 Adaptive Modulation and Coding (AMC)

AMC is used for each sub-channel and for each frame that requires CSI at the BS. The

AMC is done at the BS in order to have centralized control. Owing to the mobility of

users with high speeds, the channel changes rapidly, and thus using the instantaneous

SINR as CSI is not feasible. Unless otherwise stated, the average SINR is used in this

research as CSI to decide on the appropriate MCS. For each sub-channel, however, it

is assumed that the CSI is accurately estimated at the MS and the RS and fed back to

the BS at the end of each UL subframe using the fast feedback CQICH. Based on the

received CSI, the BS selects the suitable MCS for each of S → R, S → D and R → D

links. It is also assumed that no delay or transmission errors can occur in the feedback

channel. Since AMC is used, the DL power control does not improve the performance

significantly (Hu et al. 2004; Ahmed & Yanikomeroglu 2009). Thus, the DL power

control at the transmitters of the BS and the RS is not considered and the transmit

signal power per sub-carrier is fixed. The modulation schemes that are considered in

this research are: BPSK, quadrature phase shift keying (QPSK), 16-quadrature

amplitude modulation (16-QAM) and 64-QAM. The forward error correction (FEC) is

considered in the form of convolutional coding with code rates ( )cR : 1 (no coding),

1/2, 2/3, 3/4 and 5/6. As shown in Table 3.4, the 9 MCSs that employed in the AMC

are: BPSK with =cR 1/2, QPSK with =cR 1/2, 3/4, 16-QAM with =cR 1/2, 3/4,

and 64-QAM with =cR 2/3, 3/4, 5/6, 1 (no coding). The required SINR to achieve a

target BER of 10-6 for the simulated MCSs is also given in Table 3.4. This target BER

is considered as specified in IEEE 802.16e standard (IEEE 2006) and WiMAX Forum

(WiMAX Forum 2006c) for delivering data services in mobile WiMAX networks.

3.5.3 MDHO Algorithm

The MDHO algorithm is implemented using the soft handover algorithm proposed in

3GPP TR 25.922 (3GPP 2002) and described previously in Section 2.7.2 with a

diversity set size of 2. Accordingly, the access station can be added to, removed from

or replaced in the diversity set. The algorithm comprises of the following conditions:

95

(i) If (weakest_DS_signal_strength < strongest_DS_signal_strength – Del_Th)

for a period TΔ , then the weakest station is removed from the diversity set;

(ii) If (best_monitored_signal_strength > strongest_DS_signal_strength –

Add_Th) for a period TΔ and the diversity set is not full, then the best station

in the monitored set is added to the diversity set; and

(iii) If the diversity set is full and (best_monitored_signal_strength >

weakest_DS_signal_strength + Rep_Th) for a period TΔ , then the weakest

station in the diversity set is replaced by the best station in the monitored set.

The weakest_DS_signal_strength is the measured signal strength of the

weakest station in the diversity set, strongest_DS_signal_strength is the measured

signal strength of the best station in the diversity set and

best_monitored_signal_strength is the measured signal strength of the best station in

the monitored set. The Del_Th is the deletion threshold, Add_Th is the addition

threshold and Rep_Th is the replacement threshold. The TΔ is a short delay between

the time when the handover conditions are met and the time when the handover

initialization is started and the monitored set includes the access stations that the MS

measures and reports but that are not included in the diversity set.

Table 3.4 MCSs’ parameters in AMC

MCS Modulation Code rate Nominal rate

(bps/Hz)

Required

SINR (dB)

1 BPSK 1/2 0.5 3.0

2 QPSK 1/2 1 6.0

3 QPSK 3/4 1.5 8.5

4 16-QAM 1/2 2 11.5

5 16-QAM 3/4 3 15.0

6 64-QAM 2/3 4 19.0

7 64-QAM 3/4 4.5 21.0

8 64-QAM 5/6 5 23.0

9 64-QAM 1 6 28.0

Source: IEEE 2006

96

3.5.4 Propagation Model

(a) Path Loss Models

The suburban macrocellualr environment is assumed. The relay link between the BS

and RS is assumed to be reliable and in line-of-sight (LOS), while the access links

between the BS and MS and between the RS and MS are in NLOS. The LOS

assumption can be practically realized by placing RSs at a carefully selected location,

such as on the roof of a building and/or by using directional antenna at the RSs. The

free space LOS propagation loss model and the modified IEEE 802.16 model terrain

type C are considered for the desired and interfering relay links, respectively (Liu et

al. 2006; Senarath et al. 2007). On the other hand, for the NLOS access links between

the BS and MS and between the RS and MS, the COST-231 Hata model is considered

(Baum et al. 2005; Ikeda et al. 2006; Lee & Cho 2007).

The free space LOS propagation loss when there is no obstruction between the

transmitter and the receiver is defined as:

( ) ( )Rc dfPL log20log2044.32 ++= (3.39)

where cf is the operating carrier frequency in MHz, and Rd is the distance between

the transmitter and receiver in km.

The median path loss formula for the modified IEEE 802.16 version of the

Erceg model is expressed as:

⎪⎪⎩

⎪⎪⎨

⎧

≤⎟⎠⎞

⎜⎝⎛

>Δ+Δ+⎟⎟⎠

⎞⎜⎜⎝

⎛+

='0

'0

0

,4log20

,log10

ddford

ddforPLPLddA

PLR

htfR

λπ

α (3.40)

where,

97

⎟⎟⎠

⎞⎜⎜⎝

⎛=

λπ '

04log20 dA ,

0d = 100 m,

⎟⎟⎠

⎞⎜⎜⎝

⎛ Δ+Δ−

= α100

'0 10

htf PLPL

dd

bb h

cbha −−=α , is the path loss exponent

⎟⎠⎞

⎜⎝⎛=Δ

2000log6 c

ffPL , is the frequency correction factor

⎪⎪⎩

⎪⎪⎨

⎧

>⎟⎠⎞

⎜⎝⎛−

≤⎟⎠⎞

⎜⎝⎛−

=Δmhforh

mhforh

PLt

t

tt

ht

3,3

log20

3,3

log10, is the RS height correction factor

Rd is the distance between the transmitter and the receiver in m

λ is the wavelength in m

bh is the BS/RS antenna height in m

th is the received RS antenna height in m

cf is the carrier frequency in MHz and

the constants =a 3.6, =b 0.005 and =c 20 for terrain type C.

On the other hand, the median path loss in dB for the COST-231 Hata model is

given by:

( )( ) ( ) ( ) ( )

( ) Frt

crRt

ChhfhdhPL

++−−++−=

7.0log82.13log1.146.355.45loglog55.69.44

(3.41)

where th is the BS/RS antenna height in m, rh is the MS antenna height in m, Rd is

the distance between the transmitter and the receiver in km, cf is the carrier

frequency in MHz and the parameter FC is the correction factor that is defined as

3 dB for urban areas and defined as 0 dB for medium sized city and suburban areas.

98

(b) Lognormal Shadow Fading Model

Large-scale fading is modeled as a lognormal random variable with zero mean and

standard deviations of 8 dB for the access links and 3.4 dB for the relay links. The

temporal correlation of the large-scale fading of the access links is modeled as follows

(Senarath et al. 2007):

1,1 1,2

,1, ≥−+= ++ lUXX lll σσσ δδ (3.42)

where lX ,σ is the mean fading level in dB that is experienced at location l , 1, +lUσ is a

normal random variable with a standard deviation of 8 dB which is independent of

lX ,σ and δ is the correlation coefficient between locations l and 1+l which is

spatially determined as follows:

( ) 2lncordde Δ−=δ (3.43)

where dΔ is the distance moved between two adjacent observations and cord denotes

the de-correlation distance which is set to 20 m in the current simulation.

3.5.5 Interference Model

In this research, only the co-channel interference and the first tier of co-channel

interference are taken into account in the performance evaluation. The orthogonal

allocation scheme is considered in which no sub-channel can be shared among the

MSs directly served by BS and those served by RSs. Thus, there is no intra-cell

interference and only inter-cell interference exists. Here the worst case scenario in

calculating the interference is considered. In other words, it is assumed in this research

that the system is fully loaded which means that all sub-carriers are assigned in every

cell at the same time. However, if the system is not assumed to be fully loaded,

sometimes some sub-carriers in some cells will not be used in which case the

interference from those sub-carriers need not to be considered.

99

During the first phase of all handover techniques, the interference at a given

sub-carrier k is because of the transmissions of BSs only. During the second phase of

MDHO and at a given sub-carrier ,k the interference is because of either the

simultaneous transmissions of the BS and one of the RSs, simultaneous transmissions

of two RSs, transmission of the BS only or transmission of the RS only; depending on

whether sub-carrier k in the other cell is allocated to MS served by the BS and the

RS, two RSs, the BS only or the RS only, respectively. The interference in FASS is

similar to that for MDHO except that there are no simultaneous transmissions during

the second phase; and the interference comes from either the BS or the RS. For HHO,

the interference is caused by either the BS or the RS from all surrounding cells. On the

other hand, the relay link inter-cell interference comes from the co-channel BSs only.

It should be noted that for MDHO and FASS, the cells whose access stations

are included into the diversity set of the user u do not cause any interference to the

desired signal of the user u . This is especially important for those users at the edge of

the cell where the interfering signal is strong and can be comparable to the desired

signal. Hence, for MDHO and FASS, the cell-edge user which is communicating with

two access stations from two different cells can convert the interfering signal of the

main interferer into a desired signal which is not the case in HHO. In the latter, the

interference comes from all the neighboring cells. Consequently, the performance of

the cell-edge users can be significantly improved by employing MDHO or FASS

rather than HHO. In order to model the interference for MDHO, FASS, and HHO, the

interference model developed in Park et al. (2009) is adapted.

Figure 3.11, 3.12 and 3.13 show examples of the potential interference sources

for the user of interest MS1 being in MDHO, FASS and HHO, respectively, where the

MS1 is located at the interior of cell 1. In case of that MS1 is in MDHO and FASS, it

is assumed the diversity set members of the MS1 are BS1 and RS3. On the other hand,

BS1 is assumed to be the anchor station and the serving station in case of FASS and

HHO, respectively. Note that in Figure 3.11, 3.12 and 3.13, the inter-cell interferences

at the MS1 come from all neighboring cells. Note also that the number of interference

sources is increased in case of MDHO compared to FASS and HHO. Figure 3.14, 3.15

and 3.16 show the potential interference sources for the user of interest MS4 being in

100

MDHO, FASS and HHO, respectively, where the MS4 is located at the edge of cell 1.

It is assumed in these scenarios that the diversity set members are RS3 and RS17 in

case of MDHO and FASS. In addition, RS3 is considered as the anchor station and the

serving station in case of FASS and HHO, respectively. It should be noted that the

interference calculations in these scenarios are different from the scenarios considered

in Figure 3.11, 3.12 and 3.13. In fact, in case of MDHO and FASS, the inter-cell

interferences at the MS4 come from all neighboring cells except from cell 3 since its

access station RS17 is included into the diversity set of the MS4 as described

previously. However, for HHO, the inter-cell interferences come from all neighboring

cells including cell 3.

Let kBi ,ε , k

ri,η , krBi +,ξ and k

rri 21, +ϕ denote the probability that subcarrier k is

allocated to MS served by BS of cell i , the probability that subcarrier k is allocated to

MS served by RS r associated with BS of cell i , the probability that subcarrier k is

assigned to MS served by BS of cell i and RS r associated with this BS and the

probability that subcarrier k is assigned to MS served by two RSs, for instance 1r and

2r , associated with BS of cell i , respectively. The level of interference caused by the

RS in the other cell depends on which RS is transmitting using the same radio

resources allocated to the interfered MS. In the current analysis, the average of the

potential interference from all RSs in the neighboring cells is simply taken, assuming

that the resources are randomly assigned among all RSs, to reduce the simulation

complexity.

Thus, the average total interference of each sub-carrier k for an HHO user u

can be written as:

( )∑Φ∈

⋅+⋅=ii

krirec

kri

kBirec

kBi

ukHHO PPI ,,,,,,

,2, ηε (3.44)

where kBirecP ,, is the received signal power for sub-carrier k served by BS of cell i ,

krirecP ,, is the received signal power for sub-carrier k served by RS r associated with

BS of cell i and iΦ is the set of interfering cells.

101

Conversely, when user u is in FASS, the average total interference of each

sub-carrier k can be expressed as:

( ){ }∑

Φ−Φ∈

⋅+⋅=DSii

krirec

kri

kBirec

kBi

ukFASS PPI ,,,,,,

,2, ηε (3.45)

where DSΦ denotes the cells of the diversity set members of user u .

However, when user u is in MDHO, the average total interference of each

sub-carrier k can be given by:

({ }∑

Φ−Φ∈+ +⋅+⋅+⋅=

DSii

krirec

kBirec

krBi

krirec

kri

kBirec

kBi

ukMDHO PPPPI )( ,,,,,,,,,,,

,2, ξηε

))( 2,,1,,21,k

rireck

rireck

rri PP +⋅+ +ϕ (3.46)

Figure 3.11 Interference received by cell-interior user MS1 from the neighboring

cells when MS1 is in MDHO and the diversity set members are BS1 and RS3

BS1

BS2

BS3

BS4

BS5

BS6

BS7

RS1

RS3

RS4

RS2

RS6

RS5

RS23

RS24

RS19 RS20

RS21

RS22

RS16

RS13

RS18

RS17

RS15

RS14

RS10

RS8

RS9

RS11

RS12

RS7

RS38

RS39

RS40 RS41

RS42

RS37

RS34

RS33

RS32 RS31

RS36

RS35

RS29

RS30

RS25 RS26

RS27

RS28

MS1

Cell 1

Cell 2

Cell 3

Cell 4

Cell 5

Cell 6

Cell 7

102

Figure 3.12 Interference received by cell-interior user MS1 from the neighboring

cells when MS1 is in FASS and the anchor station is BS1

Figure 3.13 Interference received by cell-interior user MS1 from the neighboring

cells when MS1 is in HHO and the serving station is BS1

Cell 1

Cell 2

Cell 3

Cell 4

Cell 7

Cell 5

Cell 6 BS1

BS2

BS3

BS4

BS5

BS6

BS7

RS1

RS3

RS4

RS2

RS6

RS5

RS23

RS24

RS19 RS20

RS21

RS22

RS16

RS13

RS18

RS17

RS15

RS14

RS10

RS8

RS9

RS11

RS12

RS7

RS38

RS39

RS40 RS41

RS42

RS37

RS34

RS33

RS32 RS31

RS36

RS35

RS29

RS30

RS25 RS26

RS27

RS28

MS1

RS41

Cell 1

Cell 2

Cell 3

Cell 4

Cell 7

Cell 5

Cell 6 BS1

BS2

BS3

BS4

BS5

BS6

BS7

RS1

RS3

RS4

RS2

RS6

RS5

RS23

RS24

RS19 RS20

RS21

RS22

RS16

RS13

RS18

RS17

RS15

RS14

RS10

RS8

RS9

RS11

RS12

RS7

RS38

RS39

RS40

RS42

RS37

RS34

RS33

RS32 RS31

RS36

RS35

RS29

RS30

RS25 RS26

RS27

RS28

MS1

103

Figure 3.14 Interference received by cell-edge user MS4 from the neighboring cells

when MS4 is in MDHO and the diversity set members are RS3 and RS17

Figure 3.15 Interference received by cell-edge user MS4 from the neighboring cells when MS4 is in FASS and the anchor station is RS3

Cell 2

Cell 3

Cell 4

Cell 7

Cell 6

Cell 5

Cell 1

BS1

BS2

BS3

BS4

BS5

BS6

BS7

RS1

RS3

RS4

RS2

RS6

RS5

RS23

RS24

RS19 RS20

RS21

RS22

RS16

RS13

RS18

RS17

RS15

RS14

RS10

RS8

RS9

RS11

RS12

RS7

RS38

RS39

RS40 RS41

RS42

RS37

RS34

RS33

RS32 RS31

RS36

RS35

RS29

RS30

RS25 RS26

RS27

RS28 MS4

Cell 1

Cell 2

Cell 3

Cell 4

Cell 7

Cell 5

Cell 6 BS1

BS2

BS3

BS4

BS5

BS6

BS7

RS1

RS3

RS4

RS2

RS6

RS5

RS23

RS24

RS19 RS20

RS21

RS22

RS16

RS13

RS18

RS17

RS15

RS14

RS10

RS8

RS9

RS11

RS12

RS7

RS38

RS39

RS40 RS41

RS42

RS37

RS34

RS33

RS32 RS31

RS36

RS35

RS29

RS30

RS25 RS26

RS27

RS28 MS4

104

Figure 3.16 Interference received by cell-edge user MS4 from the neighboring cells when MS4 is in HHO and the serving station is RS3

3.5.6 Simulation Output

The performance evaluation metrics used to evaluate the DL system performance and

that are the outputs to the system-level simulation are the following:

(i) Average DL SINR;

(ii) Average DL spectral efficiency;

(iii) Outage probability;

(iv) MDHO probability;

(v) Selection probability of the different MCSs.

Cell 1

Cell 2

Cell 3

Cell 4

Cell 7

Cell 5

Cell 6 BS1

BS2

BS3

BS4

BS5

BS6

BS7

RS1

RS3

RS4

RS2

RS6

RS5

RS23

RS24

RS19 RS20

RS21

RS22

RS16

RS13

RS18

RS17

RS15

RS14

RS10

RS8

RS9

RS11

RS12

RS7

RS38

RS39

RS40RS41

RS42

RS37

RS34

RS33

RS32 RS31

RS36

RS35

RS29

RS30

RS25 RS26

RS27

RS28 MS4

105

(a) Average DL SINR

The average DL SINR of each sub-carrier k for an HHO user u can be written as:

N

ukHHO

ukserHHO

uk PIP

+= ,

2,

,2,

,γ (3.47)

where ukserP ,

2, is the received power of the desired signal at sub-carrier k taking into

account the path loss and the shadow fading between the serving station and the

destination terminal, the subscript ser stands for the serving cell and NP is the

receiver noise that is calculated according to the following formula (Sklar 2001):

FWTKP bN ×××= (3.48)

where bK is the Boltzmann’s constant (1.38 10-23 Joules/Kelvin or Watt/Kelvin-Hz),

T is the system temperature (290 Kelvin), W is the transmission bandwidth in Hz and

F is the noise figure.

When user u is in FASS, the average DL SINR of each sub-carrier k can be

expressed as:

( )

Nuk

FASS

ukserserFASS

uk PI

PDS

+= Φ∈

,2,

,2,

,

maxγ (3.49)

When user u is in MDHO, the average DL SINR of each sub-carrier k can be

given by:

∑Φ∈ ⎥

⎥⎦

⎤

⎢⎢⎣

⎡

+=

DSser Nuk

jMDHO

ukjserMDHO

uk PIP

,,

,,

,γ (3.50)

where }2,1{∈j denotes the phase index.

106

However, the received signal power, whether a desired or an interfering signal,

is calculated as:

][][][][][][ dBXdBPLdBGdBGdBmPdBmP rttrec σ+−++= (3.51)

where recP is the received signal power at the receiver antenna in dBm, tP is the

transmitted signal power of the serving antenna or the interfering antenna in dBm, tG

is the transmitting antenna gain in dB (or in dBi), rG is the receiving antenna gain in

dB (or in dBi), PL is the path loss between the transmitting antenna and the receiving

antenna in dB and σX is a random variable, normally distributed in dB, that accounts

for the large-scale variation of the channel, namely shadowing.

(b) Average DL Spectral Efficiency

The spectral efficiency is measured as the average modulation efficiency, which is

defined as the average correctly received information-bits/sec/Hz (bps/Hz). The

spectral efficiency can be calculated as ( ) ( ) ( )( )γγγ BERRSE −= 1 . The term ( )γR

denotes the nominal rate in bps/Hz for the selected MCS mode based on γ and

( )γBER is the probability of error with the selected MCS mode based on γ . For

example, if the BS selects 64-QAM without coding to match the current DL channel

condition that is fed back by the MS, then the nominal rate is 6 information-bits/sec in

one Hz of the transmission bandwidth; that is 6 bps/Hz. Convolutional coding with

several code rates may be combined with the adaptive modulation to improve the

spectral efficiency, as described in Subsection 3.5.2. For instance, if the BS chooses a

combination of 64-QAM and convolutional coding with code rate 32=cR , then the

nominal rate is ( ) 4326 =× information-bits/sec in one Hz of the transmission

bandwidth; namely 4 bps/Hz. However, the probability of error ( )γBER can be upper

bounded by (Frenger et al. 1999; Glavieux 2007):

( ) ∑∞

=

=fdd

dd PwBER γ (3.52)

107

where d denotes the Hamming distance of the sequence of output bits corresponding

to each path from the sequence of output bits corresponding to the all-zero path of the

state diagram of the convolutional encoder, fd is the minimum free distance of the

code, dw is the sum of bit errors for error events of distance d and dP is the pairwise

error probability that is given by:

( )md

m

dd q

mmd

qP −⎟⎟⎠

⎞⎜⎜⎝

⎛ +−= ∑

−

=

111

0 (3.53)

for a Rayleigh fading channel with

⎟⎟⎠

⎞⎜⎜⎝

⎛

+−=

γβγβα

cM

cMM

RR

q2

12

(3.54)

where Mα and Mβ are defined in Section 3.4.

Note that the average spectral efficiency gain of scheme Y with respect to

scheme Z is defined as:

( ) 100, ×−

=Z

ZYgain SE

SESEZYSE (3.55)

where YSE and ZSE are the average spectral efficiency of scheme Y and scheme Z ,

respectively.

(c) Outage probability

The outage probability is defined as a probability that the received SINR does not

meet the minimum SINR requirement for the receiver to obtain services. Thus, the

outage probability can be expressed by:

[ ]0γγ <= PPout (3.56)

108

where γ is the average SINR and 0γ denotes the minimum SINR required for the

receiver to obtain service. However, in this developed system-level simulation, the

outage probability is calculated as the percentage of users for which the average

received SINR is lower than the required SINR (3 dB from Table 3.4) to support the

minimum MCS level (BPSK with 1/2 code rate).

(d) MDHO Probability

The MDHO probability is defined as the percentage of users being in MDHO from the

total number of users. The MDHO probability includes the percentage of users whose

diversity set members are a BS and an RS, the percentage of users whose diversity set

members are two RSs and the percentage of users whose diversity set members are

two BSs.

(e) Selection Probability of the Different MCSs

The selection probability of an MCS is defined as the percentage of users using a

specific MCS with a specific spectral efficiency from the total number of users.

3.5.7 Simulation Flowchart

The performance evaluation is carried out using a system-level simulation developed

in MATLAB software. In this system-level simulation, the performance of the various

handover techniques is evaluated and compared at different environments, such as at

various MS mobility speeds, RS transmitted powers and relative RS locations. Figure

3.17 illustrates the flowchart for the developed system-level simulation.

109

Figure 3.17 Flowchart for the system-level simulation

110

At the beginning of the simulation, the simulation parameters are defined.

After that, the hexagonal cellular layout is generated and the MSs are uniformly

distributed throughout the coverage area of each cell. The receiver noise and the

distances between the BSs and RSs and the path losses between the BSs and RSs are

calculated. Initially, each MS is connected to the nearest access station and keeps the

second closest access station as a diversity set member in case of MDHO and FASS.

At the beginning of each simulation trial, the MSs’ directions of movement,

the distances between the MSs and BSs and between the MSs and RSs and the path

losses between the MSs and BSs and between the MSs and RSs are calculated. In

addition, the lognormal shadow fading is calculated between each RS and each BS

and between each MS and each access station taking into account the temporal

correlation of the shadow fading between the current MS location and the previous

MS location using Equation (3.42) and (3.43). The receiver of interest then calculates

the received signal powers, the sum of the received interference powers and the

average DL SINR.

The HHO, FASS and MDHO algorithms are then executed. The FASS and

MDHO are implemented using the same algorithm described previously in Subsection

3.5.3. The flowchart for the implemented MDHO and FASS algorithm is shown in

Figure 3.18. Now, for each handover technique, each MS is connected to the access

station(s) with the best received SINR. In MDHO, the MS communicates with all the

diversity set members, whereas in FASS, the MS communicates with the anchor

station. In HHO, the MS communicates with the serving station. In both FASS and

HHO, the MS communicates with a single station at every simulation trial. The

difference is that in the former, the anchor station can be changed from frame to

frame, whereas in the latter, the serving station is changed only when the HHO

conditions are met that do not necessarily occur at each frame. It is worth mentioning

that, at each simulation trial, either case 1 or case 2 of the proposed MDHO and the

conventional MDHO is performed depending on the diversity set members of the MS.

Note that the diversity set members in case 1 and case 2 of the proposed MDHO are

similar to those in case 1 and case 2 of the conventional MDHO.

111

Figure 3.18 Flowchart for the MDHO algorithm

Moreover, at each simulation trial and for each MS and RS, the achieved SINR

γ is calculated for each handover technique based on the derived equations in Section

3.3 and Subsection 3.5.6. Then, the MCS is adapted based on the methods described

in Section 3.3. The MCS that gives the highest spectral efficiency with the achieved

SINR > required SINR value is selected. For instance, if the achieved SINR is 12 dB,

112

16-QAM-1/2 scheme is selected according to Table 3.4. The method described in

Subsection 3.5.6 is used to find out the spectral efficiency for each MS based on the

selected MCS for that MS. This scheme assigns different spectral efficiency levels to

different users based on their channel conditions.

At the end of each simulation trial, the average DL SINR, the average DL

spectral efficiency, the outage probability, the MDHO probability, the selection

probability for each MCS are recorded taking into account all users, users being in

case 1 and/or users being in the MDHO regions.

The aforementioned processes are repeated until the end of the prescribed

simulation trials that is 2000 trials. At the end of the simulation trials, the recorded

values of the performance evaluation metrics are averaged over specific region and/or

overall simulation trials. Note that after so many times of averaging, the randomness

of the data is averaged out (Hu 2003). In fact, the average DL SINR and the average

DL spectral efficiency are averaged by 2000×uN ( uN is the number of users for

specific region; 2000 is the number of simulation trials). On the other hand, the

MDHO probability, the selection probability for each MCS and the outage probability

are averaged by 2000 simulation trials.

3.6 IMPLEMENTATION ASPECTS FOR THE PROPOSED DL MDHO

The proposed DL MDHO is mainly operated on the BS. Thus, all BSs should be fully

aware of the topology (number of hops) of the access stations constituting the

diversity set of the MS in order to schedule the diversity set members and the MS

accordingly. The topology information may be exchanged within the RS network

entry procedures using RNG-REQ/RSP messages. The BS is also aware of the

topology information update due to events such as mobility. However, the BS might

obtain or update the topology information of its associated RSs, directly or indirectly,

through wireless relay links. In contrast, for BS to BS communications, the topology

information might be obtained over the backbone network (IEEE 2009). The BS

scheduler then allocates the radio resources (symbol in the time domain and sub-

channel in the frequency domain) to the diversity set members and the MS depending

113

on the MDHO scenario. Both the RS and the MS are notified of the allocated

resources.

In order to maintain the current MS configurations so that the IEEE 802.16e

compliant MSs can handover seamlessly, existing IEEE 802.16e standard procedures

are used to inform the MS about its allocated data regions during the first phase and/or

the second phase. In fact, in case 1 of the proposed MDHO, in order to notify the MS

of its allocated data regions, the BS can use the DL-MAP information elements (IEs)

considered in the IEEE 802.16e standard for MDHO and MIMO which is denoted as

Macro_MIMO operation (IEEE 2006). Thus, the BS uses the

Macro_MIMO_DL_Basic_IE ( ) and MIMO_in_another_BS_IE ( ) defined for

Macro_MIMO operation to notifying the MS. However, in case 2 of the proposed

MDHO as well as in case 1 and case 2 of the conventional MDHO, the BS uses the

DL-MAP IEs considered in the standard for the MDHO to notify the MS of the

allocated data regions. In fact, the HO_Anchor_Active_DL_MAP_IE ( ) and

HO_Active_Anchor_DL_MAP_IE ( ) messages are used for the MS notification.

3.7 SUMMARY

In this chapter, a new efficient topology-aware DL MDHO technique for TDD-

OFDMA-based interference-limited multihop cellular networks was proposed. As

opposed to the conventional MDHO, in the proposed MDHO, the MS received all the

data signals transmitted by the diversity set members either from a BS and an RS, two

RSs, or two BSs. It ensured that the topology of the diversity set members is always

fully exploited. In the proposed MDHO and whenever the diversity set members are a

BS and an RS, the MS receives the signal transmitted by the BS during the first phase

in addition to the simultaneous transmissions of the BS and the RS during the second

phase. On the other hand, when the diversity set members are two RSs or two BSs, the

proposed MDHO performs similarly to the conventional MDHO where only the

simultaneous transmissions of the diversity set members are received by the MS.

The average post-processing DL SINR for the proposed DL MDHO technique

were formulated and derived. The derived equations expresses the average DL SINR

114

as a function of the average SINRs of the S → D links and R → D link as well as the

interference ratio. Therefore, these derived equations can be directly used to

investigate the effect of these different parameters on the average post-processing DL

SINR. The average DL SINR for the proposed DL MDHO in the noise-limited

environment can be obtained from the derived equation for the interference-limited

environment. Furthermore, the average DL SINRs of the conventional MDHO, FASS

and HHO can obtained from the derived SINR of the proposed MDHO.

Furthermore, closed-form expression for the average DL e2e BER for the

proposed DL MDHO was derived. The derived equations express the average DL e2e

BER as a function of the average SINRs of the S → R link, S → D links and the

R → D link as well as the interference ratio, and therefore can be directly used to

study the effect of these different parameters on the average DL e2e BER.

Finally, the different modules of the developed simulation model were

described in details. The developed simulation model is used to validate the

superiority of the proposed MDHO in a multi-cell interference-limited environment

and to investigate the impacts of varying some system parameters on the performance

of the various handover techniques. The system-level simulation model will be used to

investigate the impacts of the MS mobility speed, RS transmitted power and the

relative RS location on the performance of the various handover techniques. The

implementation aspects for the proposed DL MDHO technique in the IEEE 802.16j

multihop cellular networks were also described. Next chapter will present the DL

analytical and simulation results for the performance evaluation and comparison of the

various handover techniques.

CHAPTER IV

MDHO DOWNLINK PERFORMANCE

4.1 INTRODUCTION

In the previous chapter, the analytical and simulation models are presented. In this

chapter, the DL analytical and simulation results are illustrated and discussed. First of

all, the analytical results for the comparison of the average post-processing DL SINR

for the proposed MDHO and the conventional MDHO are illustrated and discussed.

The analytical results for evaluating and comparing the average DL e2e BER of the

proposed MDHO and the conventional MDHO are then illustrated and discussed.

Finally, the simulation results for evaluating the DL performance of the proposed

MDHO, conventional MDHO, FASS and HHO techniques are presented and

discussed. The simulation is carried out for different environments, such as for various

MS mobility speeds, RS transmitted powers and relative RS locations. However, the

uplink results will be covered in the next chapter.

4.2 ANALYTICAL RESULTS FOR THE AVERAGE DL SINR FOR THE PROPOSED MDHO AND THE CONVENTIONAL MDHO

This section concludes the analytical discussions of the average post-processing DL

SINR provided in Chapter 3 with some analytical results to compare the performance

of the proposed MDHO and the conventional MDHO. Since the proposed MDHO and

the conventional MDHO differ mainly in case 1 while they perform similarly in case

2, the performance comparison in this section is conducted for case 1 only using

Equation (3.5), (3.11) and (3.12) derived in Section 3.3. It should be mentioned that

the exact values of the parameters is not of particular interest. However, the

116

performance trends of the proposed MDHO and the conventional MDHO curves and

the relative performance difference between them are of particular interest.

Figure 4.1 illustrates the average post-processing DL SINR for the proposed

MDHO and the conventional MDHO with =2,SDγ 15 dB and =ρ 0.5 for different

values of 2,RDγ . It is clear that the proposed MDHO significantly outperforms the

conventional MDHO. The proposed MDHO offers a SINR gain of as much as 4.68 dB

over the conventional MDHO at lowest values of 2,RDγ compared to 2,SDγ . However,

when 2,2, SDRD γγ = , the proposed MDHO achieves a SINR gain of 3 dB compared to

the conventional MDHO. An interesting observation from Figure 4.1 is that as 2,RDγ

increases, the difference between the proposed MDHO and the conventional MDHO

decreases. In fact, at high values of 2,RDγ , the R → D link becomes the dominant link,

while the effects of the S → D links are marginal and thus the gain of the proposed

MDHO over the conventional MDHO decreases.

Figure 4.1 Average post-processing DL SINR for the proposed MDHO and the conventional MDHO at =2,SDγ 15 dB and =ρ 0.5 as a function of

2,RDγ

0 5 10 15 20 25 3014

16

18

20

22

24

26

28

30

32

Average SINR in the R → D link, γRD,2 (dB)

Ave

rage

pos

t-pro

cess

ing

SIN

R (

dB)

Proposed MDHOConventional MDHO

_

117

Figure 4.2 plots the average post-processing DL SINR as a function of 2,SDγ

with =2,RDγ 15 dB and =ρ 0.5. The results in this figure show that the proposed

MDHO has better average post-processing SINR compared to the conventional

MDHO. The proposed MDHO offers a significant SINR gain of as much as 4.68 dB

over the conventional MDHO. The maximum SINR gain of the proposed MDHO over

the conventional MDHO is achieved when 2,SDγ is much higher than 2,RDγ . This is

due to the fact that when 2,SDγ is much higher than 2,RDγ , 1,SDγ becomes the dominant

link according to Equation (3.11) and (3.12) which, as a result, increases the gain of

the proposed MDHO over the conventional MDHO. Note that even though the

average post-processing SINR is linearly proportional to 2,RDγ and 2,SDγ in Equation

(3.5) and (3.11), the average post-processing SINR exponentially increases as 2,RDγ or

2,SDγ increases, as shown in Figure 4.1 and 4.2, respectively. The reason is that the

terms in Equation (3.5) and (3.11) are in linear scale, whereas Figure 4.1 and 4.2 show

the results in logarithmic scale and hence the exponential increase.

Figure 4.2 Average post-processing DL SINR for the proposed MDHO and the

conventional MDHO at =2,RDγ 15 dB and =ρ 0.5 as a function of

2,SDγ

0 5 10 15 20 25 3015

20

25

30

35

Average SINR in the S → D link, γSD,2 (dB)

Ave

rage

pos

t-pro

cess

ing

SIN

R (

dB)

Proposed MDHOConventional MDHO

_

118

Figure 4.3 presents the average post-processing DL SINR for the proposed

MDHO and the conventional MDHO with =1,SDγ 20 dB and =2,RDγ 11 dB for

different values of ρ . The results in Figure 4.3 show that the proposed MDHO

significantly improves the average post-processing DL SINR compared to the

conventional MDHO. However, the performance difference between the proposed

MDHO and the conventional MDHO decreases as ρ increases. This is because the

interference level in the first phase is getting closer to the interference level in the

second phase and thus the difference between 1,SDγ and 2,SDγ decreases as ρ

increases. Furthermore, the proposed MDHO achieves the lowest SINR gain over the

conventional MDHO at =ρ 1 which also corresponds to the noise-limited

environment.

Figure 4.3 Average post-processing DL SINR for the proposed MDHO and the

conventional MDHO at =1,SDγ 20 dB and =2,RDγ 11 dB as a function of ρ

0.4 0.5 0.6 0.7 0.8 0.9 117

18

19

20

21

22

23

24

ρ = I D,1 / I D,2

Ave

rage

pos

t-pro

cess

ing

SIN

R (

dB)

Proposed MDHOConventional MDHO

119

4.3 ANALYTICAL RESULTS FOR THE AVERAGE DL E2E BER FOR THE PROPOSED MDHO AND THE CONVENTIONAL MDHO

In this section, the average DL e2e BER performance of the proposed MDHO and the

conventional MDHO are evaluated using Equation (3.21) and (3.34) derived in

Section 3.4. In order to verify the accuracy of the BER analysis given in Section 3.4, a

Monte Carlo simulation is also carried out.

Figure 4.4 plots the average DL e2e BER of the conventional MDHO and the

proposed MDHO at =1,SRγ 3 dB, =2,RDγ 23 dB and =ρ 0.5 as a function of 2,SDγ . It

can be seen from Figure 4.4 that the simulation results match the theoretical results

very well. This proves that the theoretical BER expressions are almost exact. It is also

observed that when 1,SRγ is small and 2,RDγ is high, the BER of the conventional

MDHO and the proposed MDHO is high. In this case, increasing 2,SDγ and, as a

result, 1,SDγ result in a small improvement in the BER of both MDHO techniques.

This is because when the S → R link is not reliable and 2,RDγ is high, the error

propagation is the dominant factor in this case.

Figure 4.4 Average e2e BER of the proposed MDHO and the conventional

MDHO at =1,SRγ 3 dB, =2,RDγ 23 dB and =ρ 0.5 as a function of 2,SDγ

0 5 10 15 20 25 3010-3

10-2

10-1

Average SINR in the S → D link, γSD,2 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

120

Figure 4.5 shows the average DL e2e BER of the conventional MDHO and the

proposed MDHO for the same scenario as in Figure 4.4 but with =2,RDγ 4 dB. Figure

4.5 also illustrates the effect of the interference ratio ρ on the performance of the

proposed MDHO. It is clear from Figure 4.5 that the proposed MDHO significantly

improves the BER performance in comparison with the conventional MDHO. In this

case, the benefit of increasing 2,SDγ compared to the previous scenario becomes

important. Thus, increasing 2,SDγ would decrease the BER of both proposed MDHO

and conventional MDHO. In addition, as 2,SDγ increases, the BER performance

difference between the proposed MDHO and the conventional MDHO increases. This

is due to the fact that when 1,SRγ and 2,RDγ are small, the effect of the error

propagation is negligible while the effect of the diversity error is dominant. The

proposed MDHO gets higher number of diversity branches than the conventional

MDHO; and the diversity error gets higher benefits from increasing the average SINR

of the S → D links compared to that of the R → D link. Hence, the BER performance

difference between the two MDHO techniques increases as 2,SDγ increases. It is also

clear from Figure 4.5 that when ρ increases from 0.5 to 0.9, the BER of the proposed

MDHO slightly increases. This is because 1,SDγ decreases as ρ increases according to

Equation (3.12), which results in increasing the BER of the proposed MDHO.

Figure 4.6 shows the average DL e2e BER of the proposed MDHO and the

conventional MDHO at =1,SRγ 30 dB, =2,RDγ 15 dB and =ρ 0.5 at different values of

2,SDγ . It is evident from the results of Figure 4.6 that the proposed MDHO

significantly outperforms the conventional MDHO. When 1,SRγ is high, the effect of

the error propagation is marginal, and the benefit of increasing 2,SDγ becomes more

apparent compared to previous scenarios. Thus, in this case, increasing 2,SDγ would

significantly decrease the BER of both proposed MDHO and conventional MDHO.

For example if the targeted BER 410 −= , then 2,SDγ is required to be 11.5 dB in case

of the proposed MDHO, whereas it should be 20.5 dB in case of the conventional

MDHO. Thus, the proposed MDHO achieves 9 dB SINR gain compared with the

conventional MDHO.

121

Figure 4.5 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 3 dB, =2,RDγ 4 dB and =ρ 0.5 and 0.9 as a function of 2,SDγ

Figure 4.6 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 30 dB, =2,RDγ 15 dB and =ρ 0.5 as a function of

2,SDγ

0 5 10 15 20 25 3010-7

10-6

10-5

10-4

10-3

10-2

Average SINR in the S → D link, γSD,2 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

0 5 10 15 20 25 3010-7

10-6

10-5

10-4

10-3

10-2

10-1

100

Average SINR in the S → D link, γSD,2 (dB)

Ave

rage

BE

R e2e

Proposed MDHO(Analytical), ρ=0.5Proposed MDHO (Simulation), ρ=0.5Proposed MDHO(Analytical), ρ=0.9Proposed MDHO (Simulation), ρ=0.9Conventional MDHO(Analytical)Conventional MDHO (Simulation)

_

122

In Figure 4.7, the effect of the average SINR of the S → R link on the average

DL e2e BER is shown. In Figure 4.7, the average DL e2e BER is plotted as a function

of 1,SRγ with =2,SDγ 3 dB, =2,RDγ 7 dB and =ρ 0.5. At high values of 1,SRγ , the BER

of both MDHO techniques is lower than that at low values of 1,SRγ . On the other hand,

increasing 1,SRγ further will only slightly decrease the BER. This is attributed to the

small values of 1,SDγ , 2,SDγ and 2,RDγ .

Figure 4.7 Average e2e BER of the proposed MDHO and the conventional MDHO at =2,SDγ 3 dB, =2,RDγ 7 dB and =ρ 0.5 as a function of 1,SRγ

Figure 4.8 presents the average DL e2e BER as a function of 1,SRγ with

=2,SDγ 15 dB, =2,RDγ 14 dB and =ρ 0.5. In this case, the values of 1,SDγ , 2,SDγ and

2,RDγ are high compared to those in Figure 4.7. Hence, as 1,SRγ increases, the BERs of

both MDHO techniques decrease. The BER performance difference between the

proposed MDHO and the conventional MDHO are almost fixed for most of 1,SRγ

values. Such a difference can be seen over the regions where the received SINR of the

S → D and R → D links are high and comparable to each other.

0 5 10 15 20 25 3010-3

10-2

10-1

100

Average SINR in the S → R link, γSR,1 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

123

Figure 4.8 Average e2e BER of the proposed MDHO and the conventional

MDHO at =2,SDγ 15 dB, =2,RDγ 14 dB and =ρ 0.5 as a function of

1,SRγ

Figure 4.9 illustrates the effect of the average SINR of the R → D link on the

average DL e2e BER at =1,SRγ 30 dB, =2,SDγ 5 dB and =ρ 0.5. As 2,RDγ increases,

on one hand the probability of error propagation increases. In contrast, the probability

of diversity error decreases. In Figure 4.9, 1,SRγ is high and the probability of diversity

error dominates over the probability of error propagation. Consequently, as 2,RDγ

increases, the average BERs of both MDHO techniques decrease. Moreover, as 2,RDγ

increases, the difference in the BER performance between the proposed MDHO and

the conventional MDHO decreases. This is due to the fact that as 2,RDγ becomes

higher than the average SINR in the S → D links, the R → D link becomes the

dominant link which decreases the BER performance difference between the proposed

MDHO and the conventional MDHO.

0 5 10 15 20 25 3010-5

10-4

10-3

10-2

10-1

Average SINR in the S → R link, γSR,1 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

124

Figure 4.9 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 30 dB, =2,SDγ 5 dB and =ρ 0.5 as a function of

2,RDγ

Figure 4.10 presents the average DL e2e BER of both MDHO techniques at

=1,SRγ 3 dB, =2,SDγ 15 dB and =ρ 0.5 at different values of 2,RDγ . In this case, the

S → R link is not reliable and the probability of error propagation dominates over the

probability of diversity error. Hence, increasing 2,RDγ results in increasing the BERs

of both MDHO techniques. At =2,RDγ 6.5 dB, for instance, the proposed MDHO

achieves BER of approximately 310 − , whereas the conventional MDHO achieves BER

of approximately 210− .

Figure 4.11 plots the average DL e2e BER of both MDHO techniques for the

same scenario as in Figure 4.10 but with higher value of 2,SDγ , that is 24 dB.

Comparing the results in this figure with those in Figure 4.10, it is clear that the BER

is lower in this case. Moreover, the BER increases as 2,RDγ increases. At =2,RDγ 6.5 dB, for example, the proposed MDHO achieves BER of around 510 − , whereas the

conventional MDHO achieves BER of around 310 − .

0 5 10 15 20 25 3010-4

10-3

10-2

10-1

Average SINR in the R → D link, γRD,2 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

125

Figure 4.10 Average e2e BER of the proposed MDHO and the conventional

MDHO at =1,SRγ 3 dB, =2,SDγ 15 dB and =ρ 0.5 as a function of 2,RDγ

Figure 4.11 Average e2e BER of the proposed MDHO and the conventional MDHO at =1,SRγ 3 dB, =2,SDγ 24 dB and =ρ 0.5 as a function of 2,RDγ

0 5 10 15 20 25 3010-4

10-3

10-2

10-1

Average SINR in the R → D link, γRD,2 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

0 5 10 15 20 25 3010-6

10-5

10-4

10-3

10-2

10-1

Average SINR in the R → D link, γRD,2 (dB)

Ave

rage

BE

R e2e

Proposed MDHO (Analytical)Proposed MDHO (Simulation)Conventional MDHO (Analytical)Conventional MDHO (Simulation)

_

126

Table 4.1 summarizes the effect of the different parameters on the average DL

e2e BER performance of the proposed MDHO and the conventional MDHO.

Table 4.1 Summary of performance analysis of the average DL e2e BER

Scenario eeBER 2 as the variable of interest increases

BER difference 1,SRγ 2,RDγ 2,SDγ ρ

small 3 dB

high 23 dB

varies 0.5 slightly decreases slightly increases

small 3 dB

small 4 dB

varies 0.5, 0.9

decreases, and BER of proposed MDHO is better at 5.0=ρ

increases

high 30 dB

medium 15 dB

varies 0.5 decreases and lower than previous scenario

increases

varies small 7 dB

small 3 dB

0.5 decreases but slightly at higher 1,SRγ

increases

varies medium 14 dB

medium 15 dB

0.5 decreases almost fixed

high 30 dB

varies small 5 dB

0.5 decreases decreases

small 3 dB

varies medium 15 dB

0.5 increases decreases

small 3 dB

varies high 24 dB

0.5 increases and better than previous scenario

decreases

4.4 DL SIMULATION RESULTS AND DISCUSSIONS

This section presents the DL simulation results for the performance evaluation and

comparison of the proposed MDHO, conventional MDHO, FASS and HHO

techniques. The performance evaluation is carried out in multi-cell interference-

limited environments. The effects of the MS speed, the RS transmitted power and the

relative RS location on the performance of the various handover techniques are

investigated. The performance evaluation metrics are the average DL SINR, the

average DL spectral efficiency and the outage probability. The MDHO probability and

the MCSs’ selection probability are also illustrated. Note that the MDHO regions refer

to the overlapping coverage areas of the BSs, the overlapping coverage areas of BSs

and RSs (whether within the same or different cells), and the overlapping coverage

areas of RSs (whether within the same or different cells). The users in the MDHO

127

regions are the users exists in these overlapping areas. However, case 1 regions refer

to the overlapping coverage areas of the BSs and RSs whether within the same or

different cells. Thus, the users being in case 1 only are those users exist in the case 1

regions. On the other hand, case 2 regions refer to the overlapping coverage areas of

the BSs (in different cells), and the overlapping coverage areas of RSs (whether within

the same or different cells). Thus, the users being in case 2 only are those users exist

in the case 2 regions.

4.4.1 The Effect of the MS Mobility Speed on the Performance of the Various Handover Techniques

In this subsection, the performance of the proposed MDHO, conventional MDHO,

FASS and HHO techniques are compared at different MS mobility speeds, namely a

pedestrian MS speed of 3 km/hr and typical vehicular MS speeds of 30, 60 and 120

km/hr. All MSs in each simulation scenario move at the same fixed speed. The RS

transmitted power is 33 dBm, and the RS location is 2/3 with respect to the cell radius.

Figure 4.12 illustrates the average percentage of users being in case 1 of

MDHO, denoted as case 1 probability, and the average percentage of users being in

case 2 of MDHO, denoted as case 2 probability, from the total number of users at

different MS speeds. The combination of case 1 and case 2 percentages represents the

percentage of users being in the MDHO regions or the total MDHO probability. As

can be seen from Figure 4.12 the total MDHO probability increases as MS speed

increases. In addition, the percentage of case 1 is always much higher than the

percentage of case 2. This could be explained by noting that the effective isotropic

radiated power (EIRP) of the BS is much higher than that of the RS. However, EIRP

values are the same at the RSs. Consequently, the BS has the highest priority to be

included into the diversity set of the MS. However, as the MS speed increases, the MS

crosses the overlapping access stations coverage areas more frequently which

increases the overall MDHO probability. Since case 1 represents most of the MDHO

cases, any enhancement proposed to case 1 will result in significant improvements to

the performance of MDHO.

128

Figure 4.12 Percentages of users being in case 1 and case 2 of MDHO from the

total number of users at different MS speeds

Figure 4.13 presents the cumulative distribution function (CDF) of the average

DL SINR for the proposed MDHO, conventional MDHO, FASS and HHO techniques

at a pedestrian MS speed of 3 km/hr. The results presented in Figure 4.13 are taken for

the users in the MDHO regions only. In Figure 4.13, it is clear that the average DL

SINR of the proposed MDHO is better than that of the conventional MDHO, FASS

and HHO. The median DL SINR for the proposed MDHO, conventional MDHO,

FASS and HHO are 11.64, 9.36, 8.12 and 5.26 dB, respectively. The SINR gains of

the proposed MDHO over the conventional MDHO, FASS and HHO are 2.28, 3.52

and 6.38 dB, respectively. In order to have a fair comparison with the theoretical

results presented in Figure 4.1 and 4.2, for the users being in case 1 of MDHO, the

SINR gain of the proposed MDHO over the conventional MDHO is 3.11 dB.

Comparing these simulation results with the theoretical results presented in Figure 4.1

and 4.2, it is clear that in both simulation and theoretical results the proposed MDHO

outperforms the conventional MDHO. However, the achieved simulation gain differs

3 30 60 1200

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

MS speed (km/hr)

Pro

babi

lity

Case 1 (BS+RS)Case 2 (2BS/2RS)

129

from the maximum achieved theoretical gain because in the simulation the highest

SINR access station in the diversity set and the interference ratio ρ all depend on the

locations of the MSs in the cells.

Figure 4.13 CDF of the average DL SINR at a pedestrian MS speed of 3 km/hr

The CDF of the average DL spectral efficiency for the proposed MDHO,

conventional MDHO, FASS and HHO techniques at a pedestrian MS speed of 3 km/hr

is illustrated in the results of Figure 4.14. The results presented in Figure 4.14 are

taken for the users in the MDHO regions only. From this figure it is clear that the

proposed MDHO provides the highest spectral efficiency among the considered

handover techniques. In fact, the median DL spectral efficiency of the proposed

MDHO, conventional MDHO, FASS and HHO are 1.99, 1.56, 1.37 and 0.94 bps/Hz,

respectively. The proposed MDHO offers spectral efficiency gains of as much as 28%,

45% and 112% over the conventional MDHO, FASS and HHO, respectively. On the

other hand, for the MDHO users whose diversity set members are a BS and an RS, the

proposed MDHO achieves spectral efficiency gains up to 39% (1.61 to 2.23 bps/Hz),

2 4 6 8 10 12 140

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Average SINR (dB)

Pro

babi

lity

( Ave

rage

SIN

R <

Abs

ciss

a)

Proposed MDHOConventional MDHOFASSHHO

130

60% (1.39 to 2.23 bps/Hz) and 106% (1.08 to 2.23 bps/Hz) over the conventional

MDHO, FASS and HHO, respectively. It is interesting to note that the FASS

outperforms the HHO even though the MS receives from a single access station in

both handover techniques. This is because the FASS changes the anchor station at

every frame which provides diversity gain against shadowing. In addition, for the cell-

edge users having diversity set members from two different cells, the FASS eliminates

the interference from the dominant interference source and therefore the cell-edge

performance can be remarkably improved.

Figure 4.14 CDF of the average DL spectral efficiency at a pedestrian MS speed of

3 km/hr

Figure 4.15 shows CDF of the average DL SINR for the proposed MDHO,

conventional MDHO, FASS and HHO techniques at a vehicular MS speed of 120

km/hr for the users in the MDHO regions only. The median DL SINR of the proposed

MDHO, conventional MDHO, FASS and HHO are 9.34, 7.24, 5.16 and 2.22 dB,

respectively. It is obvious that the average DL SINR of the proposed MDHO is better

than that of the conventional MDHO, FASS and HHO. The proposed MDHO offers

0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Spectral efficiency (bps/Hz)

Pro

babi

lity

(Spe

ctra

l effi

cien

cy <

Abs

ciss

a)

Proposed MDHOConventional MDHOFASSHHO

131

SINR gains of 2.1, 4.18 and 7.12 dB compared to the conventional MDHO, FASS and

HHO techniques, respectively. Compared with the results presented in Figure 4.13,

the average DL SINR obtained at a vehicular MS speed is lower than that achieved at

a pedestrian MS speed.

Figure 4.15 CDF of the average DL SINR at a vehicular MS speed of 120 km/hr

Figure 4.16 plots the CDF of the average DL spectral efficiency for the

proposed MDHO, conventional MDHO, FASS and HHO techniques at a vehicular

MS speed of 120 km/hr for the users in the MDHO regions only. The median DL

spectral efficiency for the proposed MDHO, conventional MDHO, FASS and HHO

are respectively 1.75, 1.42, 1.21 and 0.95 bps/Hz, respectively. The proposed MDHO

provides the highest spectral efficiency among the considered handover techniques.

The spectral efficiency gains of the proposed MDHO compared with the conventional

MDHO, FASS and HHO are respectively 23%, 45% and 84%. Note that the average

DL spectral efficiency shown in this figure at a vehicular MS speed is lower than that

presented in Figure 4.14 at a pedestrian MS speed.

-2 0 2 4 6 8 10 12 140

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Average SINR (dB)

Pro

babi

lity

(Ave

rage

SIN

R <

Abs

ciss

a)

Proposed MDHOConventional MDHOFASSHHO

132

Figure 4.16 CDF of the average DL spectral efficiency at a vehicular MS speed of

120 km/hr

Figure 4.17 shows the outage probability of the various handover techniques at

MS speeds of 3, 30, 60, and 120 km/hr, taking into account all MSs. The results in

Figure 4.17 show that the proposed MDHO has the lowest outage probability, whereas

the HHO has the highest outage probability. In addition, the outage probability

increases as the MS speed increases. At a pedestrian MS speed of 3 km/hr and for a

total number of 210 users, approximately 38 users, 25 users, 21 users and 17 users are

in outage in case of HHO, FASS, conventional MDHO, and proposed MDHO,

respectively. On the other hand, at a vehicular MS speed of 120 km/hr and for a total

number 210 users, approximately 77 users, 66 users, 58 users and 51 users are in

outage in case of HHO, FASS, conventional MDHO, and proposed MDHO,

respectively.

0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Spectral efficiency (bps/Hz)

Pro

babi

lity

(Spe

ctra

l effi

cien

cy <

Abs

ciss

a)

Proposed MDHOConventional MDHOFASSHHO

133

Figure 4.17 Outage probability against MS speeds

Figure 4.18 (a) and (b) illustrate the average percentages of MSs using the

various combinations of MCSs at a pedestrian MS speed of 3 km/hr and at a vehicular

MS speed of 120 km/hr, respectively. The results in these figures are taken for all

MSs. Figure 4.18 (a) and (b) reveal that because of the various channel impairments

experienced by the users in the interference-limited environment, the lower spectral

efficiency MCSs are selected more often than the higher spectral efficiency MCSs for

the various handover techniques. It is to be noted here that at the other vehicular MS

mobility speeds of 30 and 60 km/hr, the usage of the lower spectral efficiency MCSs

is also higher than that of the higher spectral efficiency MCSs. However, these results

are not shown here to avoid repetition.

0 20 40 60 80 100 1200.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

MS speed (km/hr)

Out

age

prob

abili

ty

Proposed MDHOConventional MDHOFASSHHO

134

(a)

(b) Figure 4.18 Selection probability of the different MCSs at MS speed of (a) 3 km/hr

and (b) 120 km/hr

1 2 3 4 5 6 7 8 90

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

MCS mode

Pro

babi

lity

of s

elec

tion

Proposed MDHOConventional MDHOFASSHHO

1 2 3 4 5 6 7 8 90

0.02

0.04

0.06

0.08

0.1

0.12

0.14

MCS mode

Pro

babi

lity

of s

elec

tion

Proposed MDHOConventional MDHOFASSHHO

135

4.4.2 The Impact of the RS Transmitted Power on the Performance of the Various Handover Techniques

In this subsection, the performance of the proposed MDHO, conventional MDHO,

FASS and HHO techniques are compared at different RS transmitted powers, namely

23, 26, 30 and 36 dBm. In each simulation scenario, all RSs transmit at the same fixed

transmitted power. The RS location is 2/3 with respect to the cell radius, and the MS

speed is 30 km/hr.

Figure 4.19 indicates the average percentage of users being in case 1 of

MDHO, being in case 2 of MDHO and being in the MDHO regions from the total

number of users at different RS transmitted powers. As can be seen from Figure 4.19,

the percentage of users being in case 1 is higher than the percentage of users being in

case 2 except for an RS transmitted power of 23 dBm where case 2 has slightly higher

percentage than case 1. In addition, the total MDHO probability increases as the RS

transmitted power increases. This is because as the RS transmitted power increases,

the overlapping access stations coverage areas increase which hence increases the

total MDHO probability.

Figure 4.19 Percentages of users being in case 1 and case 2 of MDHO at different RS transmitted powers

23 26 30 360

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

RS transmitted power (dBm)

Pro

babi

lity

Case 1 (BS+RS)Case 2 (2BS/2RS)

136

Figure 4.20 presents the average DL SINR as a function of the RS transmitted

power for the proposed MDHO, conventional MDHO, FASS and HHO techniques.

The results presented in Figure 4.20 are taken for the users in the MDHO regions only.

It is clearly shown in Figure 4.20 that the average DL SINR performance of the

proposed MDHO is better than that of the conventional MDHO, FASS and HHO at

the different RS transmitted powers. Furthermore, as the RS transmitted power

increases, the average DL SINRs for all handover techniques increase. This is because

as the RS transmitted power increases, 2,RDγ increases, which as a result increases the

average DL SINRs for all handover techniques. When the RS transmitted power is

increased from 23 dBm to 36 dBm, the average SINR is increased by 2.73, 3.31, 3.97

and 5.21 dB for the proposed MDHO, conventional MDHO, FASS and HHO,

respectively. The proposed MDHO achieves SINR gains up to 2.51, 4.66 and 8.69 dB

compared to the conventional MDHO, FASS and HHO, respectively. For the MDHO

users being in case 1 only, the proposed MDHO obtains maximum SINR gains of

5.32, 7.2 and 9.67 dB compared to the conventional MDHO, FASS and HHO,

respectively.

The effect of the RS transmitted power on the average DL spectral efficiency

for the users in the MDHO regions is illustrated in the results of Figure 4.21. From

these results it is obvious that the proposed MDHO provides the highest spectral

efficiency at the considered RS transmitted powers. Moreover, for the various

handover techniques, the average DL spectral efficiency increases as the RS

transmitted power is increased. By increasing the RS transmitted power from 23 dBm

to 36 dBm, the average DL spectral efficiency for the proposed MDHO, conventional

MDHO, FASS and HHO is improved by 32% (1.48 to 1.96 bps/Hz), 50% (1.08 to

1.62 bps/Hz), 66% (0.87 to 1.44 bps/Hz) and 118% (0.5 to 1.09 bps/Hz), respectively.

Note also that an average spectral efficiency of 1.62 bps/Hz can only be achieved in

the conventional MDHO with an RS transmitted power of 36 dBm; but it can be

achieved with an RS transmitted power of about 28 dBm in the proposed MDHO.

This means that with the proposed MDHO, the requirement on the RS transmitted

power is largely reduced. Similarly, the HHO achieves an average spectral efficiency

of 1.1 bps/Hz with an RS transmitted power of 36 dBm, whereas the FASS achieves

the same average spectral efficiency with a lower RS transmitted power, that is,

137

30 dBm. However, the proposed MDHO offers spectral efficiency gains of as much as

37% (1.08 to 1.48 bps/Hz), 70% (0.87 to 1.48 bps/Hz) and 196% (0.50 to 1.48

bps/Hz) over the conventional MDHO, FASS and HHO, respectively. On the other

hand, over the MDHO regions in which the diversity set members are a BS and an RS,

the proposed MDHO achieves spectral efficiency gains up to 79% (1.07 to 1.92

bps/Hz), 116% (0.89 to 1.92 bps/Hz) and 191% (0.66 to 1.92 bps/Hz) compared with

the conventional MDHO, FASS and HHO, respectively.

Figure 4.20 Average DL SINR as a function of the RS transmitted power

22 24 26 28 30 32 34 36-2

0

2

4

6

8

10

12

RS transmitted power (dBm)

Ave

rage

DL

SIN

R (

dB)

Proposed MDHOConventional MDHOFASSHHO

138

Figure 4.21 Average DL spectral efficiency at different RS transmitted powers

Figure 4.22 shows the outage probability of the various handover techniques at

different RS transmitted powers, taking into account all MSs. The results in Figure

4.22 show that the proposed MDHO has the lowest outage probability, whereas the

HHO has the highest outage probability. In addition, the outage probability decreases

as the RS transmitted power is increased. When RS transmitted power is 23 dBm and

for a total number of 210 users, there are approximately 88 users, 73 users, 64 users

and 56 users in outage in case of HHO, FASS, conventional MDHO and proposed

MDHO, respectively. In contrast, at an RS transmitted power of 36 dBm and for a

total number of 210 users, there are approximately 52 users, 41 users, 36 users and

31 users in outage in case of HHO, FASS, conventional MDHO and proposed

MDHO, respectively. It should also be noted that the outage probability of around

25% can be achieved at an RS transmitted power of 26 dBm with only the proposed

MDHO, but it can be achieved in the conventional MDHO, FASS and HHO with RS

transmitted powers of 30, 32 and 36 dBm, respectively.

22 24 26 28 30 32 34 360.5

1

1.5

2

RS transmitted power (dBm)

Ave

rage

DL

spec

tral e

ffici

ency

(bp

s/H

z)

Proposed MDHOConventional MDHOFASSHHO

139

Figure 4.22 Outage probability against RS transmitted power

Table 4.2 summarizes the maximum achieved SINR and spectral efficiency

gains of the proposed MDHO over the conventional MDHO, FASS and HHO

techniques when the performance is investigated at different RS transmitted powers.

Table 4.2 Maximum performance gains achieved by the proposed DL MDHO studied at different RS transmitted powers

Maximum SINR gain in dB over

Maximum spectral efficiency gain in % over

HHO FASS CMDHO HHO FASS CMDHO

MDHO regions 8.69 4.66 2.51 196 70 37

Case 1 regions 9.67 7.2 5.32 191 116 79

22 24 26 28 30 32 34 360.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

RS transmitted power (dBm)

Out

age

prob

abili

tyProposed MDHOConventional MDHOFASSHHO

140

4.4.3 The Impact of the RS Location on the Performance of the Various Handover Techniques

In this subsection, the simulation results studying the impact of the RS location on the

performance of the various handover techniques are illustrated and discussed. In fact,

the location of the RS is varied along the straight line connecting the BS to the cell

vertices. The parameter of interest is therefore the relative RS location from the BS,

that is rSR dd , where SRd is the distance between the BS and the RS and rd is the

distance between the BS and the cell boundary or the cell radius. The considered

relative RS locations are 0.3, 0.4, 0.5, 0.67 and 0.9. In addition, the RS transmitted

power is 33 dBm, and the MS speed is 30 km/hr.

Figure 4.23 shows the percentage of users for each MDHO scenario, shown

previously in Figure 3.2 and described in Table 3.1, at an RS location of halfway

between the BS and the cell boundary. It should be noted that the percentage of users

in this figure is taken from the MDHO users only. It is clear from Figure 4.23 that the

percentage of users being in scenario 1, in which the diversity set members are a BS

and an RS within the same cell, is higher than that in the other MDHO scenarios. In

fact, scenario 1 represents the majority of the MDHO scenarios.

Figure 4.24 illustrates the case 1 probability, case 2 probability and total

MDHO probability as a function of the relative RS locations. As can be seen from

Figure 4.24 the percentage of users being in case 1 is higher than the percentage of

users being in case 2 at the different RS locations. In addition, the total MDHO

probability decreases as the relative RS location is increased. In fact, at a relative RS

location of 0.9, the MDHO probability is decreased by 18.2% compared to that at a

relative RS location of 0.3. The reason is that as the RS is located away from the BS,

the overlapping access stations coverage areas are decreased which, hence, decreases

the total MDHO probability. However, as the percentage of users being in MDHO is

decreased, the handover signaling overhead and resource consumptions are decreased.

141

Figure 4.23 Percentage of users for each MDHO scenario at an RS location of halfway between the BS and the cell boundary

Figure 4.24 Total MDHO probability as a function of the relative location of RS that is located on the straight line connecting the BS and the cell vertices

1 2 3 4 50

5

10

15

20

25

30

35

Per

cent

age

of u

sers

(%)

M DHO scenario

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Relative RS location, dSR / dr

Pro

babi

lity

Case 1 (BS+RS)Case 2 (2BS/2RS)

142

Figure 4.25 presents the average DL SINR at different RS locations for the

proposed MDHO, conventional MDHO, FASS and HHO techniques. The results

presented in Figure 4.25 are taken for the users in the MDHO regions only. It is

obvious that the proposed MDHO significantly outperforms the conventional MDHO,

FASS and HHO at the different RS locations. Furthermore, as the relative RS location

is changed from 0.3 to 0.9, the average DL SINR is decreased by 1.26, 1.6, 2.2 and

2.81 dB in case of the proposed MDHO, conventional MDHO, FASS and HHO,

respectively. Over the MDHO regions, the proposed MDHO can provide maximum

SINR gains of about 2.53, 4.96 and 8.07 dB compared to the conventional MDHO,

FASS and HHO, respectively. When the diversity set members are a BS and an RS,

the proposed MDHO brings SINR gains of around 4.13, 6.39 and 8.73 dB over the

conventional MDHO, FASS and HHO, respectively.

Figure 4.25 Average DL SINR at different relative RS locations

Figure 4.26 illustrates the effect of the RS location on the average DL spectral

efficiency for the users in the MDHO regions. It is clear from the results of Figure

4.26 that for the various handover techniques, the DL spectral efficiency decreases as

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 11

2

3

4

5

6

7

8

9

10

11

Relative RS location, dSR / dr

Ave

rage

DL

SIN

R (

dB)

Proposed MDHOConventional MDHOFASSHHO

143

the RS is placed away from the BS. Moreover, the proposed MDHO provides the

highest spectral efficiency at the considered RS locations. However, increasing the RS

location from 0.3 to 0.9 leads to decreasing the average spectral efficiency of the

proposed MDHO, conventional MDHO, FASS and HHO by 18.5% (1.92 to 1.62

bps/Hz), 28% (1.54 to 1.2 bps/Hz), 45% (1.36 to 0.94 bps/Hz) and 77% (1.1 to 0.62

bps/Hz), respectively. The proposed MDHO offers maximum spectral efficiency gains

of 35% (1.2 to 1.62 bps/Hz), 72% (0.94 to 1.62 bps/Hz) and 161% (0.62 to 1.62

bps/Hz) compared to the conventional MDHO, FASS and HHO, respectively. For the

users being in case 1 only of the MDHO, the proposed MDHO achieves spectral

efficiency gains up to 64% (1.09 to 1.79 bps/Hz), 106% (0.87 to 1.79 bps/Hz) and

171% (0.66 to 1.79 bps/Hz) with respect to the conventional MDHO, FASS and HHO,

respectively. It can be inferred from the results of Figure 4.25 and 4.26 that the

performance of the proposed MDHO and the conventional MDHO is less sensitive to

the change of the RS location compared to that for FASS and HHO.

Figure 4.26 Average DL spectral efficiency as a function of the relative RS location

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0.8

1

1.2

1.4

1.6

1.8

2

Relative RS location, dSR / dr

Ave

rage

DL

spec

tral e

ffici

ency

(bp

s/H

z)

Proposed MDHOConventional MDHOFASSHHO

144

Figure 4.27 shows the outage probability of the various handover techniques at

different RS locations, taking into account all MSs. It is clear from the results of

Figure 4.27 that the proposed MDHO has the lowest outage probability among the

considered handover techniques. However, deploying RS at a relative location of 2/3

between the BS and the cell boundary or near to this location results in a lower outage

probability compared to the other RS locations. The results in Figure 4.27 show also

that this is a good choice for all of the considered handover techniques. In contrast, the

highest outage probability occurs when the RS is placed near to the BS, that is, at a

relative location of 0.3. At a RS location of 2/3 between the BS and the cell boundary

and for a total number of 210 users, there are approximately 61 users, 51 users,

44 users and 38 users in outage in case of HHO, FASS, conventional MDHO and

proposed MDHO, respectively. Conversely, at a RS location of 0.3 and for a total

number of 210 users, there are around 82 users, 70 users, 61 users and 53 users in

outage in case of HHO, FASS, conventional MDHO and proposed MDHO,

respectively.

Figure 4.27 Outage probability as a function of the relative RS location

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0.2

0.25

0.3

0.35

0.4

Relative RS location, dSR / dr

Out

age

prob

abili

ty

Proposed MDHOConventional MDHOFASSHHO

145

Generally, the maximum SINR and spectral efficiency gains of the proposed

MDHO compared to the conventional MDHO, FASS and HHO techniques when the

performance is investigated at different relative RS locations can be summarized in

Table 4.3 below.

Table 4.3 Maximum performance gains achieved by the proposed DL MDHO investigated at different relative RS locations

Maximum SINR gain in dB over

Maximum spectral efficiency gain in % over

HHO FASS CMDHO HHO FASS CMDHO

MDHO regions 8.07 4.96 2.53 161 72 35

Case 1 regions 8.73 6.39 4.13 171 106 64

4.5 SUMMARY

In this chapter, the superiority of the proposed DL MDHO was validated using

analytical and simulation results. The first part of this chapter illustrated the analytical

results for the average post-processing DL SINR and the average DL e2e BER

performance of the proposed topology-aware MDHO and the conventional MDHO.

The analytical evaluation results showed that the proposed MDHO significantly

outperforms the conventional MDHO in terms of the average DL SINR and the

average DL e2e BER. For instance, the proposed MDHO achieved SINR gain of as

much as 4.68 dB over the conventional MDHO. In order to verify the accuracy of the

BER analysis provided in Section 3.4, the BER performance predicted by the derived

equations was compared to those predicted by the Monte Carlo simulation. Analytical

results showed that the BER performance curves predicted by the derived equations

have a good match with those predicted by the Monte Carlo simulation. This proved

that the analytical BER expressions are almost exact. The performance gain of the

proposed MDHO over the conventional MDHO decreases as the average SINR of the

R → D link or as the interference ratio ρ increases. On the other hand, the

performance gain of the proposed MDHO with respect to the conventional MDHO

increases as the average SINRs of the S → D links increase or the interference ratio ρ

146

decreases. However, if the RS has a decoding error, this has a strong impact on the

average e2e BER of the proposed MDHO and the conventional MDHO and this effect

is called error propagation. In other words, the average e2e BER performance of the

proposed MDHO and the conventional MDHO in which the RS always transmits is

limited by the error propagation.

The second part of this chapter presented the DL simulation results for the

performance evaluation and comparison of the proposed MDHO, conventional

MDHO, FASS and HHO techniques. The impacts of the MS speed, the RS transmitted

power and the relative RS location on the performance of the various handover

techniques were studied. The performance evaluation metrics are the average DL

SINR, the average DL spectral efficiency and the outage probability. In the different

simulation environments, evaluation results showed that the proposed MDHO

significantly outperforms the conventional MDHO, FASS and HHO in terms of the

average DL SINR, the average DL spectral efficiency and the outage probability. For

instance, over the MDHO regions in which the diversity set members are BSs and RSs

(case 1 regions), the proposed MDHO achieved SINR gains of as much as 5.32, 7.2

and 9.67 dB compared to the conventional MDHO, FASS and HHO, respectively.

Over the same regions, the proposed MDHO obtained spectral efficiency gains up to

79% (1.07 to 1.92 bps/Hz), 116% (0.89 to 1.92 bps/Hz) and 191% (0.66 to 1.92

bps/Hz) compared to the conventional MDHO, FASS and HHO, respectively. Thus,

simulation results validated the superiority of the proposed MDHO over the

conventional MDHO. However, the maximum achieved simulation SINR gain

differed from the maximum obtained theoretical SINR gain. This was because of the

fact that in the simulation, the highest SINR access station in the diversity set and the

interference ratio ρ all depend on the locations of the MSs in the cells.

Evaluation results showed also that the MS speed, the RS transmitted power

and the RS location had impacts on the MDHO probability. Moreover, in the proposed

MDHO, the requirement on the RS transmitted power was largely reduced. For

instance, an average spectral efficiency of 1.62 bps/Hz can only be achieved in the

conventional MDHO with an RS transmitted power of 36 dBm; but it can be achieved

with an RS transmitted power of about 28 dBm in the proposed MDHO. However,

147

this average spectral efficiency cannot be achieved using FASS or HHO even with 36

dBm as the maximum considered RS transmitted power. Moreover, an outage

probability of around 25% can be achieved at an RS transmitted power of 26 dBm

with only the proposed MDHO, but it can be achieved in the conventional MDHO,

FASS and HHO with RS transmitted powers of 30, 32 and 36 dBm, respectively. This

means that with the proposed MDHO, the RS transmitted power is reduced by about 4

to 8 dBm. Furthermore, the performance of the proposed MDHO and the conventional

MDHO is less sensitive to the RS location compared to that for FASS and HHO.

Deploying RS at a relative location of 2/3 between the BS and cell boundary or

around this location resulted in the lowest outage probability among the considered

RS locations. This was a good choice for all the considered handover techniques.

Finally, due to the various channel impairments experienced by the MSs in the

interference-limited environment, the lower spectral efficiency MCSs were selected

more often than the higher spectral efficiency MCSs. It was also found that the FASS

outperforms the HHO even though the MS receives from a single access station in

both handover techniques. Next chapter discusses the results for the performance

evaluation of the proposed and conventional UL MDHO schemes.

CHAPTER V

MDHO UPLINK PERFORMANCE

5.1 INTRODUCTION

In the previous chapter, the DL analytical and simulation results for the various

handover techniques are presented and discussed. This chapter focuses on the UL of

the MDHO technique for TDD-OFDMA-based interference-limited multihop cellular

networks. The conventional and the proposed UL MDHO schemes are first described.

The selection criterion for each of the considered schemes is also provided. The

results of the performance evaluation and comparison for the proposed and the

conventional UL schemes are then illustrated and discussed. The performance

evaluation metrics are the average SINR, the average e2e throughput and the e2e

BER.

In this chapter, three new efficient UL schemes for MDHO technique for

TDD-OFDMA-based interference-limited multihop cellular networks are proposed.

The first proposed UL scheme uses the MRC to combine the diversity branches

signals in case of the intra-cell MDHO scenarios, whereas it uses the conventional

SSC to select the appropriate diversity branch in case of inter-cell MDHO scenarios.

The second proposed UL scheme combines the advantages of the e2e throughput-

based selection with the benefits of using the UL power control at the RSs. The third

proposed UL scheme takes the probability of decoding error at the RS into account

when using the BER as a selection metric to decide on the appropriate diversity

branch.

149

5.2 THE UL MDHO SCHEMES

This section describes the considered UL schemes for the MDHO technique of TDD-

OFDMA-based interference-limited IEEE 802.16j multihop cellular networks. The

selection criterion, the average SINR, the average e2e throughput and the e2e BER for

the considered schemes are provided. The source is a MS, and the destination is a BS.

5.2.1 Conventional SINR-based SC Scheme

Due to the introduction of RSs in the cellular networks, different intra-cell and inter-

cell MDHO scenarios occur, as shown previously in Figure 3.1. In the UL of the

conventional MDHO, the SC among the received signals is performed in both intra-

cell and inter-cell MDHO scenarios. In the conventional SC, the receiver decodes the

signals only from the link which has the maximum SINR; and in this thesis, this

scheme is called as SSC scheme. The selection criterion for the SSC scheme can be

written as:

Select branch i , where ii SINRi maxarg=

where iSINR is the SINR of the diversity branch i .

5.2.2 Joint MRC-SC Scheme

In the proposed joint MRC-SC scheme and in case of intra-cell MDHO scenarios, the

signals received at the destination terminal can be diversity combined (using MRC) in

order to increase the spatial diversity gain. On the other hand, in case of inter-cell

MDHO scenarios, the conventional SSC among the received signals is performed. The

intra-cell MDHO scenarios occur within the same cell and include the scenarios in

which the diversity set members are a BS and an RS, or two RSs. The inter-cell

MDHO scenarios occur between different cells and comprise the scenarios in which

the diversity set members are a BS and an RS, two RSs, or two BSs.

150

In the intra-cell MDHO scenarios and when the diversity set members of the

MS are a BS and an RS, both the BS and the RS receive the transmission of the MS

during the first phase as illustrated in Figure 5.1. During the second phase, only the

RS transmits to the BS. At the end of the two phases, the BS combines the signals

received during the two phases using MRC. The average UL SINR achieved at the BS

after MRC is given by:

2,1,, RDSDaMRC γγγ += (5.1)

Figure 5.1 UL transmission sequence for the joint MRC-SC scheme when the diversity set members are a BS and an RS within the same cell

In order for the BS to be able to diversity combine the signals received during

the two phases, the same MCS should be used over the two phases. The average e2e

throughput is then given by:

( ) ( ){ }aMRCSRjo

aee SESEThr ,1,int,2 ,min

21 γγ⋅= (5.2)

where )( 1,SRSE γ is the spectral efficiency in bps/Hz of the selected MCS for the

S → R link and )( ,aMRCSE γ is the spectral efficiency in bps/Hz of the selected MCS

for the diversity combined signals. The factor 21 accounts for the fact that two time

phases with equal duration is required. Note that the spectral efficiency can be

BS2

RS7

MS5

Combined signals

First phase

Second phase

151

calculated as ))(1)(()( γγγ BERRSE −= . The terms )(γR and )(γBER are

previously defined in Subsection 3.5.6. Note also that the spectral efficiency in bps/Hz

of the A → B link is theoretically upper bounded by the well-known Shannon capacity

formula as:

)1(log)( 2 ABABSE γγ += (5.3)

However, the first method described in Subsection 3.5.6 to calculate the

spectral efficiency is considered in this chapter since it returns the achievable

throughput in practice.

On the other hand; when the diversity set members of the MS are two RSs

within the same cell, for instance, RS1 and RS2, the UL transmission sequences are as

follows, as shown in Figure 5.2. During the first phase, both the RS1 and the RS2

receive the transmission of the MS, whereas during the second phase, both the RS1

and the RS2 transmit simultaneously to the BS by using the same radio resource. At

the end of the two phases, the BS combines the signals received from the RS1 and the

RS2 using MRC. The average UL SINR obtained at the BS after MRC can be written

as:

2,22,1, DRDRbMRC γγγ += (5.4)

Figure 5.2 UL transmission sequence for the joint MRC-SC scheme when the

diversity set members are two RSs within the same cell

BS1 RS2

MS1 RS1 Combined signals

First phase

Second phase

152

In order for the RS1 and the RS2 to be able to correctly decode the signal

received from the MS, the MCS is adjusted based on },min{ 1,21,1 SRSR γγ during the first

phase. However, the MCS is chosen based on bMRC,γ during the second phase. In this

scenario, the average e2e throughput can be expressed as:

)(}),(min{)(}),(min{

,1,21,1

,1,21,1int2

bMRCSRSR

bMRCSRSRjoee RR

SESEThr

γγγγγγ

+×

= (5.5)

5.2.3 End-to-End Throughput-Based SC Scheme

In the multihop cellular networks, different number of hops is used to deliver the data

to the destination terminal in case of the direct and relay based transmissions. Hence,

the e2e throughput of direct transmission may be different from that of the relay based

transmission. The proposed ETSC scheme with UL power control at the RSs

combines the advantages of the e2e throughput-based selection with the benefits of

using the UL power control at the RSs. In this proposed scheme, the UL power control

is used at the RSs in order to significantly minimize the interference caused by the

RSs to the MS during the second phase, thereby allowing the MS to transmit during

the second phase using an MCS with spectral efficiency that is near or similar to that

of the first phase. In addition, this proposed scheme then selects either the direct

transmission or the relay based transmission depends on which diversity branch

provides the best e2e throughput. The selection criterion for the ETSC scheme can be

written as:

Select branch i , where ieei Thri ,2maxarg=

where ieeThr ,2 is the e2e throughput of the diversity branch i .

When the power control is used at the RS, the UL RS transmitted power is

adjusted so that all RSs achieve the target SINR according to their channel conditions

and encountered interference levels. Hence, the transmitted power of the RS r , rP ,

can be updated as follows:

153

( )( )( )( )( )

( )( )⎪⎩

⎪⎨

⎧

<>

=+otherwisenPX

PnPXifPPnPXifP

nP

r

RrR

RrR

r

,,,

1 maxmin

maxmax

(5.6)

where n is the frame index, maxRP is the maximum RS transmitted power, min

RP is the

minimum RS transmitted power and ))(( nPX r is given by:

)(

)())((2, nnPnPX

RD

trr γ

γ= (5.7)

where )(nPr is the transmitted power of the RS at the thn frame, tγ is the target SINR

of the R → D link and )(2, nRDγ is the average SINR of the R → D link at the thn

frame that can be given by:

N

ciic

rrcRD PnI

nPGn+

=∑

≠

)()()(2,γ (5.8)

where rcG is the link gain that captures the effects of the path loss, the shadow fading

and the transmitting and receiving antenna gains between RS r and serving BS of cell

c , icI is the average interference comes from cell i to the BS of cell c and NP is the

thermal noise of the BS of cell c .

It should be noted that not only do the users being in MDHO benefit from

using the UL power control at the RSs, but also the users not being in MDHO do

benefit as well. The maximum and minimum UL RS transmitted powers ( maxRP and

minRP ) are set according to 50 dB dynamic range at 33 dBm and -17 dBm, respectively

(IEEE 2006; IEEE 2009).

The average e2e throughput of the relay based transmission can be given by

the harmonic mean formula (Oyman 2007):

154

)()()()(

2,1,

2,1,,2

RDSR

RDSRRee RR

SESEThr

γγγγ

+×

= (5.9)

On the other hand, the average e2e throughput of the direct S → D links is

given by:

( ))()(21

2,1,,2 SDSDBee SESEThr γγ +⋅= (5.10)

where the factor 21

accounts for the fact that two phases with equal duration are

needed in this case.

Hence, the average e2e throughput at the output of the e2e throughput-based

selection combiner can be expressed as:

⎩⎨⎧

≥<

=ReeBeeBee

ReeBeeReeETSCee ThrThrifThr

ThrThrifThrThr

,2,2,2

,2,2,22 ,

, (5.11)

In contrast, the average e2e throughput at the output of the SINR-based

selection combiner can be given by:

⎩⎨⎧

≥<

=2,1,,2

2,1,,22 ,

,

RDSDBee

RDSDReeSSCee ifThr

ifThrThr

γγγγ

(5.12)

When the diversity set members are two RSs, that is RS1 and RS2, the average

e2e throughput at the output of the e2e throughput-based selection combiner can be

expressed as:

⎩⎨⎧

<≥

=2,21,22,2

2,21,21,22 ,

,

ReeReeRee

ReeReeReeETSCee ThrThrifThr

ThrThrifThrThr (5.13)

155

where RieeThr ,2 is the average e2e throughput of the diversity branch with RSi that is

given by:

)()()()(

2,1,

2,1,,2

RiDSRi

RiDSRiRiee RR

SESEThr

γγγγ

+×

= (5.14)

In this case, the average e2e throughput at the output of the SINR-based

selection combiner can be expressed as:

⎩⎨⎧

<≥

=2,22,12,2

2,22,11,22 ,

,

DRDRRee

DRDRReeSSCee ifThr

ifThrThr

γγγγ

(5.15)

Finally, when the diversity set members are two BSs, that is BS1 and BS2, the

average e2e throughput at the output of the e2e throughput-based selection combiner

can be written as:

⎩⎨⎧

<≥

=2,21,22,2

2,21,21,22 ,

,

BeeBeeBee

BeeBeeBeeETSCee ThrThrifThr

ThrThrifThrThr (5.16)

where BieeThr ,2 is the average e2e throughput of the diversity branch with BSi that can

be given by:

( ))()(21

2,1,,2 SiDSiDBiee SESEThr γγ +⋅= (5.17)

In this case, the average e2e throughput at the output of the SINR-based

selection combiner can be expressed as:

⎩⎨⎧

<≥

=1,21,12,2

1,21,11,22 ,

,

DSDSBee

DSDSBeeSSCee ifThr

ifThrThr

γγγγ

(5.18)

156

5.2.4 End-to-End BER-based SC Scheme

In the SSC scheme described previously, the receiver decodes the signals from the

link that has the maximum SINR. On the other hand, in the BSC scheme introduced in

Sediq and Yanikomeroglu (2009), the receiver decodes the signals from either the

S → D link or the R → D link depends on which link has the minimum BER. In

addition, the DL is considered in the BSC scheme and the S → R link is assumed to be

reliable and error free. The proposed EBSC scheme takes into account the probability

of error at the RS. The proposed EBSC scheme selects either the direct transmission

or the relay based transmission depends on which branch has the minimum e2e BER.

The selection criterion for the EBSC scheme can be expressed as:

Select branch i , where ieei BERi ,2minarg=

where ieeBER ,2 is the e2e BER of the diversity branch i .

The analysis in this subsection is carried out for the MDHO scenarios in which

the diversity set members of the MS are a BS and an RS. However, this analysis can

be easily extended to the MDHO scenarios in which the diversity set members of the

MS are two RSs. In this subsection, low mobility users are also assumed where the

channel changes slowly and using the instantaneous SINR as CSI is feasible. Thus, it

is assumed in this subsection that the BS knows the instantaneous SINR conditions

1,SRγ , 1,SDγ and 2,RDγ , and adapts the modulation schemes for both the RS and the MS

such that the spectral efficiency is maximized while keeping the instantaneous BER

below the targeted BER.

Hence, the instantaneous e2e BER at the output of the e2e BER-based

selection combiner can be written as:

=EBSCinsteeBER ,2

),|( 1,SDSDe γ if )|( 1,SDSDe γ ),|( 2,1, RDSRe γγ

ife RDSR ),,|( 2,1, γγ )|( 1,SDSDe γ ),|( 2,1, RDSRe γγ (5.19)

157

where )|( 1,SDSDe γ is the instantaneous probability of error for the direct S → D link

conditioned on 1,SDγ and ),|( 2,1, RDSRe γγ is the instantaneous e2e probability of error

for both the S → R link and the R → D link conditioned on 1,SRγ and 2,RDγ that is

given by:

( ) =2,1, ,| RDSRe γγ ( )( −1| 1,SRSRe γ ( ))+2,| RDRDe γ ( )( −1| 2,RDRDe γ ( ))1,| SRSRe γ (5.20)

Assuming square M-QAM modulation schemes, the approximate

instantaneous probability of bit error for A → B link can be given by (Simon &

Alouini 2000; Goldsmith 2005):

( )ABMMABAB Qe γβαγ =)|( (5.21)

where ( )

i

iM M

M

2log112 −

=α and )1(

log3 2

−=

i

iM M

Mβ .

Note that even though the focus of the BER analysis in this chapter is on the

square M-QAM modulation schemes, the analysis is applicable to any modulation

scheme that has the form ( )ABMMABAB Qe γβαγ =)|( .

The instantaneous e2e BER at the output of the BER-based selection combiner

can be expressed as:

BSCinsteeBER ,2 =

),|( 1,SDSDe γ if ≤)|( 1,SDSDe γ )|( 2,RDRDe γ

),,|( 2,1, RDSRe γγ if >)|( 1,SDSDe γ )|( 2,RDRDe γ (5.22)

However, the instantaneous e2e BER at the output of the SINR-based selection

combiner can be given by:

158

=SSCinsteeBER ,2

2,1,1, ),|( RDSDSDSD ife γγγ ≥

2,1,2,1, ),,|( RDSDRDSR ife γγγγ < (5.23)

5.3 RESULTS AND DISCUSSIONS

This section illustrates the results for the performance evaluation of the proposed joint

MRC-SC scheme, the proposed ETSC scheme with power control at the RS and the

proposed EBSC scheme. The simulation model described in Chapter 3 is considered

in this section. It should be noted in the following simulation scenarios that users are

uniformly distributed in the centre cell, the first tier cells and the second tier cells as

illustrated in Figure 5.3. The users in the second tier cells are fixed and the purpose of

introducing them is just to generate interference to the users of the first tier cells. In

addition to the simulation parameters listed in Table 3.2 and 3.3, the following

parameters are also considered. The DL RS transmitted power is fixed at 33 dBm,

whereas the maximum and minimum UL RS transmitted powers are set at 33 dBm

and -17 dBm, respectively. The relative RS location is 2/3 with respect to the cell

radius. The MS transmitted power is 23 dBm, and the MS speed is 30 km/hr.

Figure 5.3 Simulated network layout for the MDHO UL performance

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

x 104

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1x 10

4

BS-RS-MS x-location (m)

BS

-RS

-MS

y-lo

catio

n (m

)

BSRSMS

159

5.3.1 Performance Evaluation of the Proposed Joint MRC-SC Scheme

This subsection presents the simulation result for the performance evaluation of the

proposed joint MRC-SC scheme. The performance metric is the average UL SINR.

Note that UL power control at the RSs is not used in this case.

Figure 5.4 plots the average SINR of the joint MRC-SC scheme and the

conventional SSC scheme as a function of the average SINR of the R → D link, 2,RDγ .

It can be seen from the results of Figure 5.4 that the joint MRC-SC scheme

outperforms the conventional SSC scheme. The average SINRs of both schemes

increase as 2,RDγ increases. The maximum SINR gain achieved by the joint MRC-SC

scheme compared to the SSC scheme is 1.33 dB. This maximum gain is obtained at

=2,RDγ 3 dB. However, the SINR gain of the joint MRC-SC scheme compared to the

SSC scheme decreases as 2,RDγ increases. This is owing to the fact that as 2,RDγ

increases, the R → D link dominates over the S → D link.

Figure 5.4 Average SINR of the joint MRC-SC scheme and the conventional SSC scheme as a function of 2,RDγ

0 5 10 15 20 25 305

10

15

20

25

30

Average SINR in the R → D link, γRD,2 (dB)

Ave

rage

UL

SIN

R (

dB)

Conventional SSCJoint MRC-SC

_

160

5.3.2 Performance Evaluation of the Proposed ETSC Scheme with Power Control at the RS

In this subsection, the analytical and simulation results for the performance evaluation

of the proposed ETSC scheme with power control at the RS are illustrated and

discussed. The considered schemes that the proposed scheme is compared to are the

ETSC scheme without power control at the RS, the SSC scheme and the joint MRC-

SC scheme. The performance metric is the average e2e throughput. For the analytical

results, it is to be mentioned here that when the power control at the RS is not used,

and due to the sever co-channel interference during the second phase, it is assumed

that 2,SDγ does not allow supporting any of the considered MCSs. On the other hand,

when the power control at the RS is used and unless otherwise stated, it is assumed

that 2,SDγ is lower than 1,SDγ by 1 dB. These assumptions are in accordance with the

simulation result that will be shown in the next subsection.

(a) Analytical Results

Figure 5.5 illustrates the average e2e throughput for the proposed ETSC scheme with

power control at the RS as compared to the ETSC scheme without power control at

the RS, the SSC scheme and the joint MRC-SC scheme at =1,SRγ 11 dB and

=2,RDγ 30 dB as a function of 1,SDγ . At low values of 1,SDγ , the proposed ETSC

scheme with power control at the RS shows similar performance with the ETSC

scheme without power control at the RS and the SSC scheme. Nevertheless, the

performance difference between the proposed scheme and the other considered

schemes increases as 1,SDγ is increased. At =1,SDγ 16.5 dB, for instance, the proposed

ETSC scheme with power control at the RS achieves average e2e throughput gains of

100% (1.5 to 3 bps/Hz), 150% (1.2 to 3 bps/Hz) and 300% (0.75 to 3 bps/Hz)

compared to the ETSC scheme without power control at the RS, the SSC scheme and

the joint MRC-SC scheme, respectively. Note that the SSC scheme always selects the

relay link since 2,RDγ is always higher than 1,SDγ . The ETSC scheme without power

control at the RS shows similar performance with the SSC scheme until reaching a

point, that is to say when ≥1,SDγ 15 dB, wherein the throughput of the S → D link

161

starts to get higher than that of the relay based transmission and hence is chosen by

this scheme. The joint MRC-SC scheme has the lowest average e2e throughput

performance among the considered schemes as the performance of this scheme is

limited by the performance of the S → R link according to Equation (5.2).

Figure 5.5 Average e2e throughput achieved with the ETSC scheme with power control as compared to the other considered UL schemes at =1,SRγ 11 dB and =2,RDγ 30 dB as a function of 1,SDγ . PC denotes power control

Figure 5.6 presents the average e2e throughput for the considered UL schemes

at =1,SRγ 8 dB and =2,RDγ 15 dB as a function of 1,SDγ . Note in this scenario that when

≥1,SDγ 15 dB, that is when 2,1, RDSD γγ ≥ , the SSC scheme selects the S → D link

instead of the R → D link. Over this region, the ETSC scheme with power control, the

ETSC scheme without power control and the SSC scheme select the direct S → D

link. Even though the three schemes select the S → D link, the ETSC scheme with

power control significantly outperforms the ETSC scheme without power control and

the SSC scheme. This is due to the use of the power control at the RS that

significantly reduces the interference during the second phase, which is not the case

0 5 10 15 20 25 300.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

Average SINR in the S → D link, γSD,1 (dB)

Ave

rage

e2e

thro

ughp

ut (

bps/

Hz)

Proposed ETSC with PCETSC without PCConventional SSCJoint MRC-SC

_

162

for the other UL schemes. This reduction in the interference allows the MS to transmit

during the second phase using an MCS with spectral efficiency that is near or similar

to that of the first phase.

Figure 5.6 Average e2e throughput of the ETSC scheme with power control as

compared to the other considered UL schemes at =1,SRγ 8 dB and =2,RDγ 15 dB as a function of 1,SDγ

Figure 5.7 shows the impact of the difference between the average SINRs of

the S → D links during the first phase and the second phase on the average e2e

throughput of the proposed ETSC scheme with power control at =1,SRγ 11 dB,

=2,RDγ 30 dB and =1,SDγ 20 dB. Note that the difference between 1,SDγ and 2,SDγ is

because of the difference in the interference levels experienced by the MS during the

first phase and the second phase. When the difference between 1,SDγ and 2,SDγ is zero,

this means that the power control at the RS reduces the interference during the second

phase to a level equals to the interference level during the first phase. It is clear from

the results of Figure 5.7 that the proposed scheme significantly achieves higher e2e

0 5 10 15 20 25 300

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

Average SINR in the S → D link, γSD,1 (dB)

Ave

rage

e2e

thro

ughp

ut (

bps/

Hz)

Proposed ETSC with PCETSC without PCConventional SSCJoint MRC-SC

_

163

throughput compared to the other schemes when the difference between 1,SDγ and

2,SDγ is small. The e2e throughput of the proposed scheme decreases as the difference

between 1,SDγ and 2,SDγ is increased. However, when 2,SDγ is less than 3 dB, the

minimum SINR required for the receiver to obtain services as described previously in

Table 3.4, the proposed scheme shows similar performance with the ETSC scheme

without power control.

Figure 5.7 Average e2e throughput of the ETSC scheme with power control as compared to the other considered UL schemes at =1,SRγ 11 dB, =2,RDγ30 dB and =1,SDγ 20 dB as a function of the difference between 1,SDγ and 2,SDγ

Figure 5.8 shows the impact of the average SINR of the S → R link on the

average e2e throughput of the considered UL schemes with =1,SDγ 20 dB and

=2,RDγ 30 dB. Note in this scenario that the proposed ETSC scheme with power

control achieves the best average e2e throughput even when 1,SRγ and 2,RDγ are higher

than 1,SDγ . It is also clear from the results of Figure 5.8 that the average e2e

0 5 10 15 20 25 300.5

1

1.5

2

2.5

3

3.5

4

Difference between γSD,1 and γSD,2 (dB)

Ave

rage

e2e

thro

ughp

ut (

bps/

Hz)

Proposed ETSC with PCETSC without PCConventional SSCJoint MRC-SC

_ _

164

throughput performance of the joint MRC-SC scheme increases as 1,SRγ increases

according to Equation (2.5). However, at higher value of 1,SRγ , the joint MRC-SC

scheme obtains average e2e throughput similar to the ETSC scheme without power

control and the SCC scheme.

Figure 5.8 Average e2e throughput of the ETSC scheme with power control as

compared to the other considered UL schemes at =1,SDγ 20 dB and =2,RDγ 30 dB as a function of 1,SRγ

Figure 5.9 shows the average e2e throughput of the considered UL schemes as

a function of 1,SRγ for the same scenario as in Figure 5.8 but with =1,SDγ 11 dB. In

Figure 5.9, it can be seen that the proposed ETSC scheme with power control selects

the relay based transmission instead of the direct link when ≥1,SRγ 11.5 dB, that is

where the S → R link can support 16-QAM with 1/2 code rate. Note also in this

scenario that the joint MRC-SC schemes achieves similar performance to the other

0 5 10 15 20 25 300

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Average SINR in the S → R link, γSR,1 (dB)

Ave

rage

e2e

thro

ughp

ut (

bps/

Hz)

Proposed ETSC with PCETSC without PCConventional SSCJoint MRC-SC

_

165

considered schemes only at ≥1,SRγ 28 dB where the S → R link can support the

highest spectrally efficient MCS mode (64-QAM).

Figure 5.9 Average e2e throughput of the ETSC scheme with power control as

compared to the other considered UL schemes at =1,SDγ 11 dB and =2,RDγ 30 dB as a function of 1,SRγ

(b) Simulation Results

Figures 5.10 shows the interference experienced by the MS during the first phase and

the second phase for the ETSC scheme when power control at the RS is not used and

is used. It is clear from Figure 5.10 that the interference during the second phase is

significantly reduced when the power control is employed at the RS. In fact, when the

power control is employed at the RS, there is about 30 dB reduction in the second

phase interference compared to the scenarios where the power control is not

employed. Note also that the results of Figure 5.10 show the difference in the

interference level during the first phase and the second phase and hence the difference

between 1,SDγ and 2,SDγ . This proves that the assumptions made in the beginning of

0 5 10 15 20 25 300

0.5

1

1.5

2

2.5

3

Average SINR in the S → R link, γSR,1 (dB)

Ave

rage

e2e

thro

ughp

ut (

bps/

Hz)

Proposed ETSC with PCETSC without PCConventional SSCJoint MRC-SC

_

166

Subsection 5.3.2 regarding the value of 2,SDγ used in the analytical results are logical

and accurate.

Figure 5.10 Interference experienced by the MS during the first and second phases when the power control at the RS is not used and used

Figure 5.11 plots the CDF of the average e2e throughput of the various

considered UL schemes. The median e2e throughput of the proposed ETSC scheme

with power control, the ETSC scheme without power control, the SSC scheme and the

joint MRC-SC scheme are 1.22, 0.87, 0.86 and 0.78 bps/Hz, respectively. It is clear

that the proposed scheme achieves the highest e2e throughput among the considered

schemes. The average e2e throughput gains of the proposed ETSC scheme with power

control compared with the ETSC scheme without power control, the SSC scheme and

the joint MRC-SC scheme are 40%, 42% and 56%, respectively. It is interesting to

note that the ETSC scheme without power control and the conventional SSC scheme

have almost the same average e2e throughput performance. In addition, the joint

MRC-SC scheme has the lowest average e2e throughput compared to the other

considered schemes.

1 2-120

-100

-80

-60

-40

-20

0

Phase index

Inte

rfere

nce

expe

ienc

ed b

y th

e M

S (

dBm

)

Without PCWith PC

167

Figure 5.11 CDF of the average e2e throughput of the proposed ETSC scheme with power control as compared to the other considered UL schemes

Figure 5.12 illustrates the average e2e throughput of the various considered UL

schemes as a function of the target SINR of the R → D link, tγ . The considered

values for tγ correspond to the required SINRs to support the various MCS modes

shown previously in Table 3.4. In other words, this graph also implicitly shows the

impact of 2,RDγ and, as a result, the spectral efficiency of the MCS mode used in the

R → D link on the average e2e throughput of the considered UL schemes. As tγ

increases, on one hand the spectral efficiency of the R → D link increases, on the

other hand the RS transmitted power and hence the interference comes from the RS

during the second phase also increase. It can be seen from the results of Figure 5.12

that the proposed ETSC scheme with power control achieves the best performance

among the considered UL schemes for all tγ values. For the proposed ETSC scheme

with power control, as tγ increases from 3 dB to 23 dB, the average e2e throughput is

increased since the increment in the spectral efficiency dominates over the increment

0.2 0.4 0.6 0.8 1 1.2 1.4 1.60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Average e2e throughput (bps/Hz)

Pro

babi

lity

(Ave

rage

e2e

thro

ughp

ut <

Abs

ciss

a)

ETSC with PCETSC without PCConventional SSCJoint MRC-SC

168

in the second phase interference in this case. On the contrary, as tγ increases from

23 dB to 28 dB, the average e2e throughput is slightly decreased. This is because the

interference caused by the RSs to the MSs during the second phase becomes

significant in this case, thereby eliminating the enhancement in the spectral efficiency.

At =tγ 23 dB, the proposed ETSC scheme with power control obtains average e2e

throughput gains of 46% (0.85 to 1.24 bps/Hz), 49% (0.83 to 1.24 bps/Hz) and 61%

(0.77 to 1.24 bps/Hz) compared to the ETSC scheme without power control, the

conventional SSC scheme and the joint MRC-SC scheme, respectively. On the other

hand, at =tγ 3 dB, the proposed ETSC scheme with power control achieves average

e2e throughput gains of 83% (0.53 to 0.97 bps/Hz), 87% (0.52 to 0.97 bps/Hz) and

94% (0.5 to 0.97 bps/Hz) over the ETSC scheme without power control, the

conventional SSC scheme and the joint MRC-SC scheme, respectively.

Figure 5.12 Average e2e throughput of the proposed ETSC scheme with power

control as compared to the other considered UL schemes as a function of tγ

0 5 10 15 20 25 300.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

Target SINR in the R → D link, γt (dB)

Ave

rage

e2e

thro

ughp

ut (

bps/

Hz)

Proposed ETSC with PCETSC without PCConventional SSCJoint MRC-SC

_

169

5.3.3 Performance Evaluation of the Proposed EBSC Scheme

In this subsection, the instantaneous e2e BER performance of the proposed EBSC

scheme, and the conventional BSC and SSC schemes are evaluated and compared

using Equation (5.19)-(5.23) given in Subsection 5.2.4. The performance evaluation is

carried out for different values of 1,SRγ , 2,RDγ and 1,SDγ . The modulation schemes to

be used in the performance evaluation are restricted to 16-QAM and 64-QAM, and the

channel coding is not employed. The modulation levels used by the source in the first

phase and used by the relay in the second phase are denoted by 1M and 2M ,

respectively. It is to be noted here that the instantaneous e2e BER is simply referred to

in this subsection as the e2e BER.

Figure 5.13 illustrates the e2e BER for the proposed EBSC scheme, the BSC

scheme and the SSC scheme with =1,SDγ 13 dB, =2,RDγ 25 dB, =1M 16-QAM and

=2M 64-QAM at different values of 1,SRγ . The results in Figure 5.13 show that at

low values of 1,SRγ , the e2e BER of the proposed EBSC scheme is significantly lower

than that of the conventional BSC and SSC schemes. For instance, at =1,SRγ 5 dB, the

BSC and SSC schemes achieve an e2e BER of 2102.4 −× , whereas the proposed EBSC

scheme obtains an e2e BER of 5104.2 −× . However, as 1,SRγ increases, the e2e BER

performance difference between the proposed EBSC scheme and the conventional

BSC and SSC schemes is decreased. When ≥1,SRγ 13 dB, that is when 1,1, SDSR γγ ≥ , the

proposed EBSC scheme achieves the same e2e BER compared to the considered

conventional BSC and SSC schemes.

170

Figure 5.13 e2e BER for the proposed EBSC scheme, the BSC scheme and the SSC scheme with =1,SDγ 13 dB, =2,RDγ 25 dB, =1M 16-QAM and =2M 64-QAM at different values of 1,SRγ

Figure 5.14 plots the e2e BER of the proposed EBSC and the conventional

BSC and SSC schemes for the same scenario as in Figure 5.13 but with =2,RDγ 17 dB.

It is interesting to note that both the proposed EBSC scheme and the conventional

BSC scheme achieve the same performance in this scenario. This is because both

schemes select the S → D link wherein the BER and the e2e BER are equivalent.

Even though 2,RDγ is higher than 1,SDγ , the BER of the R → D link is higher than that

of the S → D link. This is because the R → D link uses less robust modulation

scheme, namely 64-QAM, whereas the S → D link uses more robust modulation

scheme, namely 16-QAM. Thus, the BSC scheme selects the S → D link rather than

the R → D link.

0 5 10 1510-7

10-6

10-5

10-4

10-3

10-2

10-1

100

SINR in the S → R link, γSR,1 (dB)

BE

Re2

e

Proposed EBSCConventional BSCConventional SSC

171

Figure 5.14 e2e BER for the proposed EBSC scheme, the BSC scheme and the SSC scheme with =1,SDγ 13 dB, =2,RDγ 17 dB, =1M 16-QAM and =2M 64-QAM at different values of 1,SRγ

Figure 5.15 shows the effect of 2,RDγ on the e2e BER for the proposed EBSC

scheme, the BSC scheme and the SSC scheme with =1,SRγ 5 dB, =1,SDγ 12 dB,

=1M 16-QAM and =2M 64-QAM. Note that all schemes show identical BER

performance at ≤2,RDγ 12 dB because all schemes select the S → D link in this case.

However, when >2,RDγ 12 dB, that is to say when 1,2, SDRD γγ > , the SSC scheme

chooses the R → D link, whereas the EBSC scheme keeps selecting the S → D link.

Due to the different error-resistance capabilities of the modulation schemes used in the

S → D link and the R → D link, the BSC scheme does not switch to the relay based

transmission directly after 2,RDγ is greater than 1,SDγ . In fact, it switches to the relay

based transmission when the BER in the R → D link is lower than the BER in the

S → D link that occurs when >2,RDγ 16 dB. Over this region, even though 2,RDγ is

0 5 10 1510-5

10-4

10-3

10-2

10-1

100

SINR in the S → R link, γSR,1 (dB)

BE

Re2

e

Proposed EBSCConventional BSCConventional SSC

172

higher than 1,SDγ , the e2e BER of the proposed EBSC scheme is lower than that for

the BSC scheme and the SSC scheme. This is due to the fact that the performance of

the relay based transmission in this case is limited by the high BER in the S → R link

which dominates over the BER in the R → D link.

Figure 5.15 e2e BER for the proposed EBSC scheme as compared to the BSC and SSC schemes at =1,SRγ 5 dB, =1,SDγ 12 dB, =1M 16-QAM and =2M

64-QAM as a function of 2,RDγ

Figure 5.16 plots the e2e BER for the proposed EBSC scheme, the BSC

scheme and the SSC scheme as a function of 2,RDγ for the same scenario as in Figure

5.15 but with =2M 16-QAM. Note in this scenario that because of the similar error-

resistance capabilities of the modulation schemes used in the S → D link and the

R → D link, the BSC scheme switches to the relay-based transmission directly after

2,RDγ is greater than 1,SDγ . In this scenario, however, the BSC scheme shows identical

e2e BER performance with the SSC scheme at the different values of 2,RDγ .

0 5 10 15 20 25 3010-4

10-3

10-2

10-1

SINR in the R → D link, γRD,2 (dB)

BE

Re2

e

Proposed EBSCConventional BSCConventional SSC

173

Figure 5.16 e2e BER for the proposed EBSC scheme as compared to the BSC and SCC schemes at =1,SRγ 5 dB, =1,SDγ 12 dB, =1M 16-QAM and =2M

16-QAM as a function of 2,RDγ

Figure 5.17 presents the impact of 1,SDγ on the e2e BER for the proposed

EBSC scheme, the BSC scheme and the SSC scheme with =1,SRγ 5 dB, =2,RDγ 25 dB,

=1M 64-QAM and =2M 64-QAM. When <1,SDγ 5 dB, namely when 1,1, SRSD γγ < , the

proposed and conventional schemes select the relay based transmission. However,

when ≥1,SDγ 5 dB, the proposed EBSC scheme changes selection to the direct S → D

link, whereas the conventional BSC and SSC schemes keep selecting the relay based

transmission. It is also clear from Figure 5.17 that as 1,SDγ increases, the e2e BER

performance difference between the proposed EBSC scheme and the conventional

BSC and SSC schemes is increased. At =1,SDγ 15 dB, for example, the proposed

EBSC scheme achieves an e2e BER of 4107.7 −× , whereas the BSC and SSC schemes

obtain an e2e BER of 110− .

0 5 10 15 20 25 3010

-4

10-3

10-2

10-1

SINR in the R → D link, γRD,2 (dB)

BE

Re2

e

Proposed EBSCConventional BSCConventional SSC

174

Figure 5.17 e2e BER for the proposed EBSC scheme as compared to the BSC and SSC schemes at =1,SRγ 5 dB, =2,RDγ 25 dB, =1M 64-QAM and =2M

64-QAM as a function of 1,SDγ

Figure 5.18 plots the e2e BER for the proposed EBSC scheme, the BSC

scheme and the SSC scheme as a function of 1,SDγ with =1,SRγ 5 dB, =2,RDγ 25 dB,

=1M 16-QAM and =2M 64-QAM. This scenario is similar to the one considered in

Figure 5.17 except for that in this scenario, more robust modulation scheme is used for

the S → R and S → D links. Consequently, at high values of 1,SDγ , the e2e BER

performance gain of the proposed scheme compared to the conventional schemes is

higher than that obtained in Figure 5.17. For instance, at =1,SDγ 15 dB, the proposed

EBSC achieves an e2e BER of 7108.1 −× , whereas the BSC and SSC schemes obtain

an e2e BER of 2102.4 −× .

0 5 10 1510

-4

10-3

10-2

10-1

100

SINR in the S → D link, γSD,1 (dB)

BE

Re2

e

Proposed EBSCConventional BSCConventional SSC

175

Figure 5.18 e2e BER for the proposed EBSC scheme as compared to the BSC and SSC schemes at =1,SRγ 5 dB, =2,RDγ 25 dB, =1M 16-QAM and =2M

64-QAM as a function of 1,SDγ

5.4 SUMMARY

This chapter proposed three efficient UL schemes for the MDHO technique of TDD-

OFDMA-based interference-limited multihop cellular networks. The first proposed

UL scheme used the MRC to combine the diversity branches signals in case of the

intra-cell MDHO scenarios in order to increase the spatial diversity gain, whereas it

used the conventional SSC scheme to select the appropriate diversity branch signal in

case of the inter-cell MDHO scenarios. This scheme was called the joint MRC-SC

scheme. The second proposed UL MDHO scheme combined the advantages of the e2e

throughput-based selection with the benefits of using the UL power control at the RS.

This scheme was referred to as the ETSC scheme with power control at the RS. The

third proposed UL MDHO scheme used the e2e BER as a selection metric to decide

0 5 10 1510

-7

10-6

10-5

10-4

10-3

10-2

10-1

SINR in the S → D link, γSD,1 (dB)

BE

Re2

e

Proposed EBSCConventional BSCConventional SSC

176

on the appropriate diversity branch. This scheme was called the EBSC scheme. The

selection criterion, the average SINR, the average e2e throughput and the e2e BER

were described for the considered UL schemes.

Evaluation results showed that the performance of the UL MDHO is limited by

the performance of the S → R link. The proposed joint MRC-SC scheme achieved an

average SINR gain of 1.33 dB compared to the conventional SSC scheme, whereas it

achieved the lowest average e2e throughput among the considered UL schemes. The

average e2e throughput of the joint MRC-SC scheme increases as 1,SRγ and/or 2,RDγ

increases. Furthermore, using UL power control at the RS significantly improved the

average e2e throughput of the ETSC scheme compared to the ETSC scheme when the

power control is not used, the conventional SSC scheme and the joint MRC-SC

scheme. The average e2e throughput of the proposed ETSC scheme with power

control at the RS increases as 2,RDγ increases. In fact, the proposed ETSC scheme with

power control obtained the best performance when the target SINR of the R → D link

=tγ 23 dB where it achieved average e2e throughput gains of 46% (0.85 to 1.24

bps/Hz), 49% (0.83 to 1.24 bps/Hz) and 61% (0.77 to 1.24 bps/Hz) compared to the

ETSC scheme without power control, the conventional SSC scheme and the joint

MRC-SC scheme, respectively. The average e2e throughput gain of the proposed

ETSC scheme with power control increases as the average SINRs in the S → D links

increase and when the difference between 1,SDγ and 2,SDγ decreases. For instance,

when =1,SRγ 11 dB, =2,RDγ 30 dB and at <1,SDγ 8.5 dB, the proposed ESTC scheme

with power control at the RS shows similar performance to the ESTC scheme without

power control and the SSC scheme. On the other hand, at =1,SDγ 16.5 dB, the

proposed ETSC scheme achieved e2e throughput gains of 100% (1.5 to 3 bps/Hz),

150% (1.2 to 3 bps/Hz) and 300% (0.75 to 3 bps/Hz) compared to the ETSC scheme

without power control, the SSC scheme and the joint MRC-SC scheme, respectively.

When >1,SDγ 16.5 dB, the proposed ETSC scheme with power control achieved much

higher gain than previous and this gain increases as 1,SDγ increases.

177

Finally, the proposed EBSC scheme significantly outperformed the

conventional BSC and SSC schemes in terms of the e2e BER. For instance, when

=1,SRγ 5 dB, =2,RDγ 25 dB, =1,SDγ 15 dB, =1M 16-QAM and =2M 64-QAM, the

proposed EBSC achieved an e2e BER of 7108.1 −× , whereas the BSC and SSC

schemes obtained similar e2e BER of 2102.4 −× . The BER performance gain of the

proposed EBSC scheme increases as 1,SDγ increases. In addition, the performance

gain also increases as 2,RDγ increases when 1,SRγ is low. However, the performance

enhancement of the EBSC scheme over the conventional BSC and SSC schemes

comes at the cost of increasing the complexity since the instantaneous CSI of the

S → R link is needed to be fed back to the BS. Next chapter will give the conclusions

for the work carried out in this research.

CHAPTER VI

CONCLUSIONS AND FUTURE WORK

6.1 CONCLUSIONS AND RESEARCH FINDINGS

Recently, there has been increasing interest in both academia and industry on the

multihop relaying as a promising cost-effective approach to extend the coverage and

significantly enhance the throughput and capacity of the future wireless networks.

Towards that end, the IEEE 802.16j relay task group specified OFDMA physical layer

and MAC layer enhancements to the IEEE 802.16e standard for licensed bands to

enable the operation of RSs. Handover is an essential component of mobile cellular

communication systems to allow full user mobility in the coverage areas. There are

three handovers supported within the IEEE 802.16j multihop cellular networks,

namely HHO, FASS and MDHO. MDHO is the process by which the MS

communicates with two or more access stations called a diversity set. Due to the

introduction of RS in the cellular network infrastructure, different intra-cell and inter-

cell MDHO scenarios occur and the number of handovers increases. Therefore, it is

essential to develop efficient MDHO techniques that get the full benefits from the new

features introduced into the systems as a result from the deployed RSs. It is first

essential to develop an efficient DL MDHO technique that receives all the data signals

transmitted by the diversity set members so that the topology of the diversity set

members is always fully exploited. It is secondly essential to develop an efficient UL

MDHO schemes that outperform the conventional SC schemes. These two issues were

tackled by this thesis. This chapter summarizes the work carried out and suggests

some possible future work to extend and improve the results presented in this thesis.

179

6.1.1 DL MDHO Technique

In Chapter 3 of this study, a new efficient topology-aware DL MDHO technique for

TDD-OFDMA-based interference-limited multihop cellular networks was proposed.

As opposed to the conventional MDHO technique, the proposed MDHO technique

receives all the data signals transmitted by the diversity set members. It ensures that

the topology of the diversity set members is always fully exploited. In the proposed

DL MDHO technique and whenever the diversity set members are two different-

topology access stations, that is a BS and an RS, the MS receives the signal

transmitted by the BS during the first phase; and at the same time it also receives the

simultaneous transmissions of the BS and the RS during the second phase. On the

other hand, whenever the diversity set members are two similar-topology access

stations, namely two RSs or two BSs, the proposed DL MDHO technique performs

similarly to the conventional MDHO where only the simultaneous transmissions of

the diversity set members are received by the MS. The mathematical model and the

detailed simulation model were developed. The average DL SINRs were formulated

and derived and the transmission sequences and MCS selection criterion were

described for the various considered handover techniques. The average DL e2e BERs

for the proposed MDHO and the conventional MDHO were also formulated and

derived. The implementation aspects for the proposed DL MDHO technique in the

IEEE 802.16j multihop cellular networks were also described.

In Chapter 4, the superiority of the proposed DL MDHO was validated using

extensive analytical and simulation results developed using MATLAB software. The

derived equations for the average post-processing DL SINR were used to investigate

the effects of the average SINRs of the S → D links, the average SINR of the R → D

link and the interference ratio on the average post-processing DL SINR of the

proposed MDHO and the conventional MDHO. In addition, the derived equations for

the average DL e2e BER were used to investigate the effects of the average SINR of

the S → R link, the average SINRs of the S → D links, the average SINR of the

R → D link and the interference ratio on the BER performance of the proposed

MDHO and the conventional MDHO. In order to verify the accuracy of the BER

analysis, the BER performance curves predicted by the developed model were

180

compared to those predicted by the Monte Carlo simulation. The developed simulation

model was used to investigate the effects of some parameters, such as MS mobility

speed, RS transmitted power and relative RS location, on the performance of the

proposed MDHO, conventional MDHO, FASS and HHO techniques. The

performance evaluation using the developed simulation model was carried out in

multi-cell interference-limited environments and for users with high mobility speeds.

The performance metrics are the average DL SINR, the average DL spectral efficiency

and the outage probability. The MDHO probability and the MCS selection probability

were also presented.

The DL analytical results showed that the proposed MDHO significantly

outperforms the conventional MDHO in terms of the average SINR and the average

e2e BER. It was also found that the performance gain of the proposed MDHO over the

conventional MDHO decreases as the average SINR of the R → D link or as the

interference ratio increases. On the other hand, the performance gain of the

proposed MDHO compared with the conventional MDHO increases as the average

SINRs of the S → D links increase or the interference ratio decreases. The BER

performance curves predicted by the derived equations showed a good match with

those predicted by the Monte Carlo simulation. This proved that the analytical BER

expressions are almost exact. However, it was also found that if the RS has a decoding

error, this has a strong impact on the average e2e BER of the proposed MDHO and

the conventional MDHO. In other words, the average e2e BER performance of the

proposed MDHO and the conventional MDHO in which the RS always transmits is

limited by the error propagation.

In the different simulation environments, the evaluation results showed that the

proposed MDHO significantly outperforms the conventional MDHO, FASS and HHO

in terms of the average DL SINR, the average DL spectral efficiency and the outage

probability. For instance, over the MDHO regions in which the diversity set members

are BSs and RSs, the proposed MDHO achieved SINR gains of as much as 5.32, 7.2

and 9.67 dB compared to the conventional MDHO, FASS and HHO, respectively.

Over the same regions, the proposed MDHO obtained spectral efficiency gains up to

79% (1.07 to 1.92 bps/Hz), 116% (0.89 to 1.92 bps/Hz) and 191% (0.66 to 1.92

181

bps/Hz) compared to the conventional MDHO, FASS and HHO, respectively. It was

also found by the simulation results that the MS speed, the RS transmitted power and

the RS location had impacts on the MDHO probability. In the proposed MDHO, it

was found that the requirement on the RS transmitted power was largely reduced.

Deploying RS at a relative location of 2/3 between the BS and the cell boundary or

around this location resulted in the lowest outage probability among the considered

RS locations. This was a good choice for all the considered handover techniques. Due

to the various channel impairments experienced by the MS in the interference-limited

environments, the lower spectral efficiency MCSs were used by the MSs more often

than the higher spectral efficiency MCSs. It should be noted that the superiority of the

proposed MDHO comes at the expense of increased complexity since the MS needs to

buffer the signal transmitted by the BS during the first phase in order to be diversity

combined with the simultaneous transmissions of the BS and the RS occur during the

second phase. The proposed MDHO technique does not require any modification for

the MS configurations and hence the IEEE 802.16e compliant user can handover

seamlessly using the proposed MDHO without noticing that it is connected to the

IEEE 802.16j multihop cellular networks.

6.1.2 UL MDHO Technique

Chapter 5 was devoted to the UL of the MDHO technique. Three new efficient UL

schemes for the MDHO technique of TDD-OFDMA-based interference-limited

multihop cellular networks were proposed. The first proposed UL scheme used the

MRC to combine the diversity branches signals in case of the intra-cell MDHO

scenarios, whereas it used the conventional SSC scheme to select the appropriate

diversity branch signal in case of the inter-cell MDHO scenarios. This scheme was

called the joint MRC-SC scheme. The second proposed UL MDHO scheme combined

the advantages of the e2e throughput-based selection with the benefits of using the UL

power control at the RS. This scheme was referred to as the ETSC scheme with power

control at the RS. The third proposed UL MDHO scheme used the e2e BER as a

selection metric to decide on the appropriate diversity branch. This scheme was called

the EBSC scheme. The selection criterion, the average SINR, the average e2e

throughput and/or the e2e BER were provided for each of the proposed UL scheme.

182

To show the superiority of the proposed UL MDHO schemes, their performance were

compared to their conventional counterparts’ schemes in terms of the average SINR,

the average e2e throughput and/or the e2e BER.

The performance evaluation results showed that the UL performance of the

MDHO technique is limited by the performance of the S → R link, namely the link

between the MS and RS. The proposed joint MRC-SC scheme achieved better average

SINR compared to the conventional SSC scheme, whereas it achieved the lowest

average e2e throughput among the considered UL schemes. The average e2e

throughput of the joint MRC-SC scheme increases as 1,SRγ or 2,RDγ increases. It was

also found that using UL power control at the RS significantly reduces the

interference during the second phase, thereby allowing the MS to transmit during the

second phase using an MCS with spectral efficiency that is near or similar to that of

the first phase. Consequently, evaluation results showed that combining the ETSC

scheme with the power control at the RS significantly improved the e2e throughput

performance compared to the ETSC scheme when the power control at the RS is not

used, the conventional SSC scheme and the joint MRC-SC scheme. The e2e

throughput of the proposed ETSC scheme with power control at the RS increases as

the average SINR of R → D link, 2,RDγ , increases. In fact, the proposed ETSC scheme

with power control at the RS obtained the best performance when the target SINR of

the R → D link =tγ 23 dB where it achieved average e2e throughput gains of 46%

(0.85 to 1.24 bps/Hz), 49% (0.83 to 1.24 bps/Hz) and 61% (0.77 to 1.24 bps/Hz)

compared to the ETSC scheme without power control at the RS, the conventional SSC

scheme and the joint MRC-SC scheme, respectively. The e2e throughput gain of the

proposed ETSC scheme with power control at the RS increases as the average SINRs

in the S → D links increase and when the difference between 1,SDγ and 2,SDγ

decreases. Finally, the proposed EBSC scheme significantly outperformed the

conventional BSC and SSC schemes in terms of the e2e BER. It was also found that

the BER performance gain of the proposed EBSC scheme increases as 1,SDγ increases.

In addition, the performance gain also increases as 2,RDγ increases when 1,SRγ is low

compared to 1,SDγ . However, the performance enhancement of the EBSC scheme over

183

the conventional BSC and SSC schemes comes at the cost of increasing the

complexity since the instantaneous CSI of the S → R link needs to be fed back to the

BS.

In conclusion, all the objectives of this thesis defined in Chapter 1 have been

met and achieved. Although the proposed DL and UL schemes for the MDHO

technique significantly outperform the conventional DL and DL schemes, there are

still several open problems need to be addressed in order to further extend and

improve the results presented in this thesis.

6.2 FUTURE WORK

In this section, some recommendations are suggested to further extend and improve

the results presented in this thesis. These recommendations can be summarized as

follows:

(i) It is assumed in this research that the RS uses DF forwarding scheme. Thus,

the work presented in this research can be extended the scenario in which the

RS uses other forwarding schemes, such as AF.

(ii) The focus in this research was on the spatial diversity gain in which the same

data signals are transmitted during the two phases. The spectral efficiency of

the proposed DL MDHO could be further improved by considering the spatial

multiplexing gain wherein the BS may transmit new data signal during the

second phase. Thus, the spatial multiplexing gain of the proposed DL MDHO

should be studied and evaluated. The investigation from the channel capacity

point of view should be also carried out.

(iii) In this research, perfect or optimal MRC was assumed wherein the weight of

each diversity branch is the conjugate of the branch channel coefficient

normalized to the noise-plus-interference variance of that branch. However, if

the noise-plus-interference variance of each diversity branch is not the same

and the weight of the MRC does not take this fact into account, the MRC does

184

not yield the optimum performance in this case and it is called imperfect MRC.

Therefore, the analysis presented in this research needs to be extended to

include the effect of the imperfect MRC on the performance of the proposed

DL MDHO.

(iv) It was assumed in this research that the CSI is accurately estimated at the

destination terminals and fed back to the BS at the end of each uplink

subframe using the fast feedback CQICH. It was also assumed that no delay or

transmission errors can occur in the feedback channel. However, further

investigations on the effect of the channel estimation error and feedback delay

on the performance of the proposed MDHO should be carried out.

(v) An adaptive thresholds MDHO algorithm should be developed. Unlike the

conventional MDHO algorithm where the same threshold is used to add the RS

or the BS into the diversity set, different thresholds can be used for the RSs

and the BSs in the developed algorithm. This implies to transmit a different

threshold for each of the stations. This should be done to ensure that the RS is

added to the diversity set only when it enhances the end-to-end performance.

The developed algorithm should also be optimized to yield the best

performance. Unnecessary handover and control overhead can be reduced in

the developed algorithm.

(vi) In the proposed ETSC scheme with power control at the RS, the AMC is

performed based on the average SINR of the different links. In addition, the

link between the BS and the RS is a fixed wireless link. When the MCS for the

R → D link is adapted based on the average SINR, it is reasonable to assume

that the MCS used for that link is fixed. The power control in this case adapts

the UL transmitted power of the RS to satisfy the required SINR for that MCS

schemes. However, instead of the average SINR, the instantaneous SINR of

the R → D link can be used as CSI to select the appropriate MCS. For

throughput maximization in this case, however, a joint optimization of the

power control and the AMC should be studied.

185

(vii) Apart from using the power control at the RS, some OFDMA-based inter-cell

interference mitigation techniques may be used in order to manage the

interference and further improve the performance of the multihop cellular

networks. Therefore, the effect of adopting interference cancellation and

interference avoidance techniques on the system performance should be

studied and evaluated.

(viii) Finally, the EBSC scheme needs to be extended to the scenarios include high

mobility speed users. In these scenarios, the selection criterion is based on the

average e2e BER and the performance evaluation should be carried out based

on the derived equations for the average e2e BER.

REFERENCES

3GPP. 2002. Technical Specification Group Radio Access Network; Radio Resource Management Strategies (Release 1999). Technical Report, 3GPP TR 25.922, V3.7.0. http://www.3gpp.org/ftp/Specs/html-info/25922.htm [5 February 2011]

3GPP. 2010a. Universal Mobile Telecommunications System (UMTS); High Speed Downlink Packet Access (HSDPA); Overall Description (Stage 2). 3GPP TS 25.308 version 9.4.0 Release 9. http://www.3gpp.org/ftp/Specs/html-info/25308.htm [5 February 2011]

3GPP. 2010b. Universal Mobile Telecommunications System (UMTS); Enhanced Uplink; Overall Description (Stage 2). Technical Specification, 3GPP TS 25.319 version 9.4.0 Release 9. http://www.3gpp.org/ftp/Specs/html-info/25319.htm [5 February 2011]

3GPP. 2010c. Universal Mobile Telecommunications System (UMTS); LTE; Requirements for Evolved UTRA (E-UTRA) and Evolved UTRAN (E-UTRAN). Technical Report, 3GPP TR 25.913 version 9.0.0 Release 9. http://www.3gpp.org/ftp/Specs/html-info/25913.htm [5 February 2011]

3GPP2. 2001. cdma2000 High Rate Packet Data Air Interface Specification. 3GPP2 C.S0024-0 ver. 3.0. http://www.3gpp2.org/Public_html/specs/alltsgscfm.cfm [11 February 2011]

3GPP2. 2006. cdma2000 High Rate Packet Data Air Interface Specification. 3GPP2 C.S0024-A ver. 3.0. http://www.3gpp2.org/Public_html/specs/alltsgscfm.cfm [11 February 2011]

3GPP2. 2009. cdma2000 High Rate Packet Data Air Interface Specification. 3GPP2 C.S0024-B ver. 3.0. http://www.3gpp2.org/Public_html/specs/alltsgscfm.cfm [11 February 2011]

Adinoyi, A., Fan, Y., Yanikomeroglu, H., Poor, H.V. & Al-Shaalan, F. 2009. Performance of selection relaying and cooperative diversity. IEEE Transactions on Wireless Communications 8(12): 5790-5796.

Agilent Technology. 2009. 3GPP long term evolution: System overview, product development, and test challenges. Application Note. http://www.3gpp.org/Industry-White-Papers [8 February 2011]

Ahmed, M. H. & Yanikomeroglu, H. 2009. Throughput fairness and efficiency of link adaptation techniques in wireless networks. IET Communications Journal 3(7): 1227-1238.

187

Ahmadi, S. 2009. An Overview of next-generation mobile WiMAX technology. IEEE Communications Magazine 47(6): 84-98.

Andrews, J.G., Ghosh, A. & Muhamed, R. 2007. Fundamentals of WiMAX: Understanding broadband wireless networking. New Jersey: Prentice Hall.

Azarian, K., Gamal, H.E. & Schniter, P. 2005. On the achievable diversity-vs-multiplexing tradeoff in cooperative channels. IEEE Transactions on Information Theory 51(12): 4152–4172.

Bacioccola, A., Cicconetti, C., Eklund, C., Lenzini, L., Li, Z. & Mingozzi, E. 2010. IEEE 802.16: History, status and future trends. Elsevier Computer Communications 33(2): 113–123.

Baum, D.S., Hansen, J., Salo, J., Del Gardo, G., Milojevic, M. & Kyosti, P. 2005. An interim channel model for beyond-3G systems: Extending the 3GPP spatial channel model (SCM). Proceedings of 61st IEEE Vehicular Technology Conference (VTC 2005-Spring) 5: 3132-3136.

Becvar, Z. & Zelenka, J. 2007. Implementation of handover delay timer into WiMAX. Proceedings of 6th Conference on Telecommunications, Lisbon, pp. 401-404.

Becvar, Z. 2007. Impact of relay stations implementation on the handover in WiMAX. Proceedings of Personal Wireless Communications, Springer 245: 107-114.

Becvar, Z., Mach, P. & Bestak, R. 2008. Optimization of handover scanning procedure in WiMAX networks with relay stations. Proceedings of 3rd IEEE International Symposium on Wireless Pervasive Computing (ISWPC ’08), pp. 581-585.

Becvar, Z., Mach, P. & Bestak, R. 2009. Initialization of handover procedure in WiMAX networks. Proceedings of ICT-Mobile Summit 2009.

Becvar, Z. & Mach, P. 2010. Reduction of scanning reporting overhead in IEEE 802.16 networks with relays. Proceedings of 2010 ninth IEEE International Conference on Networks (ICN), pp. 109-114.

Biglieri, E. & Taricco, G. 2004. Transmission and reception with multiple antennas: theoretical foundations. Boston: Now Publishers Inc.

Bletsas, A., Khisti, A., Reed, D.P. & Lippman, A. 2006. A simple cooperative diversity method based on network path selection. IEEE Journal on Selected Areas in Communications 24(3): 659-672.

188

Boyer, J., Falconer, D.D. & Yanikomeroglu, H. 2004. Multihop diversity in wireless relaying channels. IEEE Transactions on Communications 52(10): 1820-1830.

Camp, T., Boleng, J. & Davies, V. 2002. A survey of mobility models for ad hoc network research. Wireless Communications and Mobile Computing Journal 2(5): 483-502.

Can, B., Yomo, H. & Carvalho, E.D. 2007. Link adaptation and selection method for OFDM based wireless relay networks. Journal of Communications and Networks 9(2): 118-127.

Chen, Q.-B., Liu, Y., Tang, L. & Chai, R. 2010. Handover algorithm based on movement state for cellular relaying networks. Proceedings of 2nd International Conference on Future Computer and Communication (ICFCC) 3: 81-85.

Cho, S., Kim, J. & Kim, J.-H. 2008. Relay assisted soft handover in multihop cellular networks. Proceedings of 2nd International Conference on Ubiquitous Information Management and Communication (ICUIMC’08), pp. 136-139.

Cho, S., Jang, E.W. & Cioffi, J.M. 2009. Handover in multihop cellular networks. IEEE Communications Magazine 47(7): 64-73.

Cudak, M. 2010. IEEE 802.16m system requirements. IEEE 802.16m-07/002r10. http://www.ieee802.org/16/tgm/docs/ [18 February 2011]

Das, S., Klein, T., Rajkumar, A., Rangarajan, S. & Turner, M. 2006. System aspects and handover management for IEEE 802.16e. Bell Labs Technical Journal 11(1): 123-142.

De Sanctis, A., Rastell, M. & Tortora, D. 2009. A method for preventing ping-pong handover effect in mobile WiMAX networks. US Patent Application, US 2009/0274118 A1. http://www.freepatentsonline.com/20090274118.pdf [20 December 2010]

Ding Z., Ratnarajah, T. & Cowan, C.F.N. 2009. Spectrally efficient cooperative diversity protocols for uplink wireless transmissions. IET Signal Processing Journal 3(5): 368-380.

Erceg, V., Greenstein, L.J., Tjandra, S.Y., Parkoff, S.R., Gupta, A., Kulic, B., Julius, A.A. & Bianchi, R. 1999. An empirically based path loss model for wireless channels in suburban environments. IEEE Journal on Selected Areas in Communications 17(7): 1205-1211.

189

Erceg, V., Hari, K.V.S, Smith, M.S., Baum, D.S., Soma, P., Greenstein, L.J., Michelson, D.G., Ghassemzadeh, S., Rustako, A.J., Roman, R.S., Sheikh, K.P., Tappenden, C., Costa, J.M., Bushue, C., Sarajedini, A., Schwartz, R., Branlund, D., Kaitz, T. & Trinkwon, D. 2003. Channel models for fixed wireless applications. Technical Report, IEEE 802.16a-03/01. http://www.ieee802.org/16/tga/docs/ [19 January 2011]

Ergen, M. 2009. Mobile broadband including WiMAX and LTE. New York: Springer Science + Business Media.

Etemad, K. 2008. Overview of mobile WiMAX technology and evolution. IEEE Communications Magazine 46(10): 31-40.

Feng, M., Wang, H. & Chen, T. 2005. System level modeling and algorithms for a B3G system employing OFDMA. Proceedings of 2nd IEEE International Conference on Mobile Technology, Application and Systems, pp. -7.

Frenger, P., Orten, P. & Ottosson, T. 1999. Convolutional codes with optimum distance spectrum. IEEE Communications Letters 3(11): 317-319.

Gage, B., Martin, C., Sich, E. & Tong, W. 2005. WiMAX: Untethering the Internet user. Nortel Technical Journal 2: 31-38.

Garg, V.K. 2002. Wireless network evolution: 2G to 3G. New Jersey: Prentice Hall.

Ghassemian, M., Friderikos, V. & Aghvami, H. 2005. Hybrid handover in multihop radio access networks. Proceedings of IEEE Vehicular Technology Conference (VTC 2005-Spring) 4: 2207–2211.

Glavieux, A. 2007. Channel coding in communication networks: from theory to turbocodes. London: ISTE Ltd.

Glisic, S.G. 2006. Advanced wireless networks 4G technologies. England: John Wiley & Sons Ltd.

Goldsmith, A. 2005. Wireless communications. New York: Cambridge University Press.

Gradshteyn, I.S. & Ryzhik, I.M. 2007. Table of integrals, series, and products. 7th ed. California: Academic Press.

190

Hasna, M.O. & Alouini, M.-S. 2003. End-to-end performance of transmission systems with relays over Rayleigh-fading channels. IEEE Transactions on Wireless Communications 2(6): 1126-1131.

Herhold, P., Zimmermann, E. & Fettweis, G. 2004. A simple cooperative extension to wireless relaying. Proceedings of International Zurich Seminar on Communications, pp. 36-39.

Herhold, P., Zimmermann, E. & Fettweis, G. 2005. Cooperative multihop transmission in wireless networks. Elsevier Computer Networks Journals 49(3): 299-324.

Holma, H. & Toskala, H. 2004. WCDMA for UMTS, radio access for third generation mobile communications. 3rd ed. England: John Wiley & Sons Ltd.

Homnan, B., Kunsriruksakul, V. & Benjapolakul, W. 2000. A comparative performance evaluation of soft handoff between IS-95A and IS-95B/cdma2000. Proceedings of the 2000 IEEE Asia-Pacific Conference on Circuits and Systems (APCCAS 2000), pp. 34-37.

Hu, H. 2003. Performance Analysis of Cellular Networks with Digital Fixed Relays. MA Thesis, Department of Systems and Computer Engineering, Carleton University, Ottawa, Ontario, Canada.

Hu, H., Yanikomeroglu, H., Falconer, D. & Periyalwar, S. 2004. Range extension without capacity penalty in cellular networks with digital fixed relays. Proceedings IEEE GLOBECOM’04 5: 3053-3057.

Huo, D. 2005. Clarification on the wrap-around hexagon network structure. Technical Report, IEEE C802.20-05/15. http://ieee802.org/20/Contribs/ [17 October 2010]

Hytonen, V. 2009. Handover Performance in IEEE 802.16 Mobile Networks. MA Thesis, Department of Mathematical Information Technology, University of Jyvaskyla, Jyvaskyla, Finland.

IEEE. 2004. IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed Broadband Wireless Access Systems. IEEE 802.16-2004.

IEEE. 2006. IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems; Amendment 2: Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands and Corrigendum 1. IEEE 802.16e.

191

IEEE. 2008. IEEE Standard for Local and Metropolitan Area Networks- Part 21: Media Independent Handover Services. IEEE 802.21.

IEEE. 2009. IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Broadband Wireless Access Systems Amendment 1: Multihop Relay Specification. IEEE 802.16j.

Ikeda, T., Asa, M. & Saito, K. 2006. Comments on IEEE 802.16j path-loss models in IEEE802.16j-06/013. Technical Report, C802.16j-06/113.

http://www.ieee802.org/16/relay/contrib/ [21 January 2011]

Ikki, S. & Ahmed, M.H. 2007. Performance analysis of cooperative diversity wireless networks over Nakagami-m fading channel. IEEE Communications Letters 11(4): 334-336.

Ikki, S.S. & Ahmed, M.H. 2010. Performance analysis of adaptive decode-and-forward cooperative diversity networks with best-relay selection. IEEE Transactions on Communications 58(1): 68-72.

ITU-R. 2003. Recommendation ITU-R M.1645: Framework and Overall Objectives of the Future Development of IMT-2000 and Systems beyond IMT-2000.

Kawai, H., Morimoto, A., Higuchi, K. & Sawahashi, M. 2010. Investigation of inter-node B macro diversity for single-carrier based radio access in evolved UTRA uplink. IEICE Transactions on Communications E93–B(1): 125-134.

Khadivi, P., Todd, T.D., Samavi, S., Saidi, H. & Zhao, D. 2008. Mobile ad hoc relaying for upward vertical handoff in hybrid WLAN/cellular systems. Elsevier Ad hoc Networks 6(2): 307-324.

Kim, D., Wang, H., Lee, J. & Hong, D. 2008. Analysis of handoff algorithm for multihop cellular networks. Proceedings of 10th IEEE International Conference on Advanced Communication Technology (ICACT), pp. 29-32.

Kim, J.Y. & Cho, D.-H. 2009. Pre-buffering scheme for seamless relay handover in relay based cellular systems. Proceedings of 69th IEEE Vehicular Technology Conference (VTC 2009-Spring), pp. 1-5.

Ko, Y.C., Luo, T. & Jeong, G. 2003. Effect of noise variance in the channel estimation on dual-MRC over Rayleigh fading channels. Proceedings of 57th IEEE Vehicular Technology Conference (VTC 2003-Spring) 4: 2538-2542.

192

Kramer, G., Gastpar, M. & Gupta, P. 2006. Cooperative strategies and capacity theorems for relay networks. IEEE Transactions on Information Theory 51(9): 3037-3063.

Laiho-Steffens, J., Jasberg, M., Sipila, K., Wacker, A. & Kangas, A. 1999. Comparison of three diversity handover algorithms by using measured propagation data. Proceedings of 49th IEEE Vehicular Technology Conference (VTC’99) 2:1370-1374.

Laneman, J.N. & Wornell, G.W. 2003. Distributed space–time coded protocols for exploiting cooperative diversity in wireless networks. IEEE Transactions on Information Theory 49(10): 2415-2525.

Laneman, J.N., Tse, D.N.C. & Wornell, G.W. 2004. Cooperative diversity in wireless networks: Efficient protocols and outage behavior. IEEE Transactions on Information Theory 50(12): 3062-3080.

Lee, K. & Cho, D. 2007. Hierarchical constellation based adaptive relay scheme in multi-hop networks. IEEE Communications Letters 11(3): 225-227.

Li, J., Fan, C., Yang, D. & Gu, J. 2005. UMTS soft handover algorithm with adaptive thresholds for load balancing. Proceedings of 62nd IEEE Vehicular Technology Conference (VTC 2005-Fall) 4: 2508-2512.

Lin, Z., Erkip, E. & Ghosh, M. 2005. Adaptive modulation for coded cooperative systems. Proceedings of 6th IEEE Workshop on Signal Processing Advances in Wireless Communications, pp. 615-619.

Lin, S.-J, Sheen, W.H., Fu, I.-K. &. Huang, C.-C. 2007. Resource scheduling with directional antennas for multi-hop relay networks in Manhattan-like environment. Proceedings of IEEE Mobile WiMAX Symposium, pp: 108-113.

Lin, L. & Chen, K.-C. 2008. Introduction to mobile WiMAX. In Chen, K.-C. & de Marca, J.R.B. (ed.). Mobile WiMAX, pp. 1-11. England: John Wiley & Sons Ltd.

Liu, T., Rong, M., Shi, H., Yu, D., Xue, Y. & Schulz, E. 2006. Reuse partitioning in fixed two-hop cellular relaying networks. Proceedings of IEEE Wireless Communications and Networking Conference (WCNC’06), pp. 177-182.

Makelainen, A. 2007. Analysis of Handoff Performance in Mobile WiMAX Networks. MA Thesis, Department of Electrical and Communications Engineering, University of Helsinki, Finland.

193

Nabar, R.U., Bolcskei, H. & Kneubuhler, F.W. 2004. Fading relay channels: Performance limits and space-time signal design. IEEE Journal on Selected Areas in Communications 22(6): 1099-1109.

Nee, R.V. & Prasad, R. 2000. OFDM for wireless multimedia communications. Boston: Artech House.

Nourizadeh, H., Nourizadeh, S. & Tafazolli, R. 2006. Impact of the inter-relay handoff on the relaying system performance. Proceedings of 64th IEEE Vehicular Technology Conference (VTC 2006-Fall), pp. 1-5.

Oestges, C. & Clerckx, B. 2007. MIMO wireless communication: from real-world propagation to space-time code design. Oxford: Elsevier.

Onat, F.A., Adinoyi, A., Fan, Y., Yanikomeroglu, H., Thompson, J.S. & Marsland, I.D. 2008. Threshold selection for SNR-based selective digital relaying in cooperative wireless networks. IEEE Transactions on Wireless Communications 7(11): 4226-4237.

Oyman, O. 2007. End-to-end throughput and latency measures for multi-hop routing in relay-assisted broadband cellular OFDM systems. Proceedings of IEEE Radio and Wireless Symposium, pp. 51-54.

Oyman, O. & Laneman, J.N. 2007. Multihop relaying for broadband wireless mesh networks: from theory to practice. IEEE Communications Magazine 45(11): 116-122.

Pabast, R., Walke, B.H., Schultz, D.C., Herhold, P., Yanikomeroglu, H., Mukherjee, S., Viswanathan, H., Lott, M., Zirwas, W., Dohler, M., Aghvami, H., Falconer, D.D. & Fettweis, G.P. 2004. Relay-based deployment concepts for wireless and mobile broadband radio. IEEE Communications Magazine 42(9): 80-89.

Panajotovic, A.S., Stefanovic, M.C. & Draca, D.L. 2009. Effect of microdiversity and macrodiversity on average bit error probability in shadowed fading channels in the presence of interference. ETRI Journal 31(5): 500-505.

Park, J.H., Han, K.Y. & Cho, D.H. 2007. Reducing inter-cell handover events based on cell ID information in multi-hop relay systems. Proceedings of 65th IEEE Vehicular Technology Conference (VTC 2007-Spring), pp. 743-747.

Park, Y. & Adachi, F. 2007. Overview of mobile communication. In Park, Y. & Adachi, F (ed.). Enhanced radio access technologies for next generation mobile communication, pp. 1-37. New York: Springer Science + Business Media.

194

Park, K., Ryu, H.S., Kang, C.G., Chang, D., Song, S., Ahn, J. & Ihm, J. 2009. The performance of relay-enhanced cellular OFDMA-TDD network for mobile broadband wireless services. EURASIP Journal on Wireless Communications and Networking, Article ID 261815, 10 pages.

Paulraj, A.J., Nabar, R.U. & Gore, D.A. 2003. Introduction to space-time wireless communications. Cambridge: Cambridge University Press.

Perez-Romero, J., Sallent, O., Agusti, R. & Diaz-Guerra, M.A. 2005. Radio resource management strategies in UMTS. England: John Wiley & Sons Ltd.

Prasad, R. & Velez, F.J. 2010. WiMAX networks: Techno-economic vision and challenges. Dordrecht: Springer Science + Business Media.

Proakis, J.G. & Salehi, M. 2008. Digital communications. 5th ed. Singapore: The McGraw-Hill.

Puthenkulam, J. 2010. 4G broadband wireless standards based mobile Internet evolution. Mobile Wireless Group, Intel Corporation. http://ewh.ieee.org/r10/pune/WiMAX2010/4G-Wireless-Stds-Mobile-Internet-Ecosystem-Technology-Roadmap-01.pdf [8 March 2011].

Rappaport, T. 2002. Wireless communications: principles and practice. 2nd ed. New Jersey: Prentice Hall.

Ray, S.K., Pawlikowski, K. & Sirisena, H. 2010. Handover in mobile WiMAX networks: The state of art and research issues. IEEE Communications Surveys & Tutorials 12(3): 376-399.

Salem, M., Adinoyi, A., Rahman, M., Yanikomeroglu, H., Falconer, D., Kim, Y.-D., Kim, E. & Cheong, Y.-C. 2010. An Overview of radio resource management in relay-enhanced OFDMA-based networks. IEEE Communications Surveys & Tutorials 12(3): 422-438.

Sediq, A.-B. 2008. Diversity Combining of Signals with Different Modulation Levels and Constellation Rearrangement in Cooperative Relay Networks. MA Thesis, Department of Systems and Computer Engineering, Carleton University, Ottawa, Ontario, Canada.

Sediq, A.-B. & Yanikomeroglu, H. 2009. Performance analysis of SNR-based selection combining and BER-based selection combining of signals with different modulation levels in cooperative communications. Proceedings of 70th IEEE Vehicular Technology Conference (VTC 2009-Fall), pp. 1-5.

195

Senarath, G., Tong, W., Zhu, P., Zhang, H., Steer, D., Yu, D., Naden, M., Kitchener, D., Hart, M., Vadgama, S., Cai, S., Chen, D., Xu, H., Peterson, R., Fu, I.-K., Wong, W.C., Srinivasan, R., Lee, H.H., Johnsson, K., Sydir, J., Ahmadi, S., Hamzeh, B., Timiri, S., Oh, C., Wang, P., Sun, Y., Basgeet, D.R., Chindapol, A., Hu, T., Kong, P.-Y., Wang, H., Ahn, J.B., Kang, H., Cho, J., Lim, H., Molisch, A.F., Zhang, J. & Kuze, T. 2007. Multi-hop relay system evaluation methodology (Channel model and performance metrics). Technical Report, IEEE 802.16j-06/013r3. http://ieee802.org/16/relay/docs/80216j-06_013r3.pdf [23 January 2011]

Sendonaris, A., Erkip, E. & Aazhang, B. 1998. Increasing uplink capacity via user cooperation diversity. Proceedings of IEEE International Symposium on Information Theory (ISIT), pp. 156.

Sendonaris, A., Erkip, E. & Aazhang, B. 2003a. User cooperation diversity – Part I: System description. IEEE Transactions on Communications 51(11): 1927-1938.

Sendonaris, A., Erkip, E. & Aazhang, B. 2003b. User cooperation diversity – Part II: Implementation aspects and performance analysis. IEEE Transactions on Communications 51(11): 1939-1948.

Seyfi, M., Muhaidat, S. & Liang, J. 2011. Performance analysis of relay selection with feedback delay and channel estimation errors. IEEE Signal Processing Letters 18(1): 67-70.

Shah, A. & Haimovich, A.M. 2000. Performance analysis of maximal ratio combining and comparison with optimum combining for mobile radio communications with cochannel interference. IEEE Transactions on Vehicular Technology 49(4): 1454-1463.

Shen, G., Ni, W., Zou, W., Zhang, K., Liu, J., Jin, S., Fahldieck, T. & Muenzner, R. 2006. Handover schemes in IEEE 802.16j. Technical Report, IEEE C802.16j-06_005r1. http://www.ieee802.org/16/relay/contrib/ [5 January 2011]

Simon, M.K. & Alouini, M.-S. 2000. Digital communication over fading channels – A unified approach to performance analysis. New York: John Wiley & Sons, Inc.

Sizun, H. 2005. Radio wave propagation for telecommunication applications. Heidelberg: Springer Science + Business Media.

Sklar, B. 2001. Digital communications: Fundamentals and applications. 2nd ed. New Jersey: Prentice Hall.

196

Soldani, D. & Dixit, S. 2008. Wireless relays for broadband access. IEEE Communications Magazine 46(3): 58-66.

Sun, Y., Basgeet, D.R. & Fan, Z. 2008. Fast relay station handover. US Patent Application, US 2008/0316968 A1. http://www.freepatentsonline.com/20080316968.pdf [14 December 2010]

Sun, L., Zhang, T., Lu, L. & Niu, H. 2010. On the combination of cooperative diversity and multiuser diversity in multi-source multi-relay wireless networks. IEEE Signal Processing Letters 17(6): 535-538.

Sydir, J. 2006. Harmonized contribution on 802.16j (mobile multihop relay) usage models. Technical Report, IEEE 802.16j-06/015. http://ieee802.org/16/relay/docs/80216j-06_015.pdf [26 February 2011]

Ulvan, A., Becvar, Z., Zelenka, J. & Bestak, R. 2007. Using the relative thresholds in handover procedure. Technical Report, IEEE C802.16j-07/086. http://www.ieee802.org/16/relay/contrib/ [5 January 2011]

Upase, B. & Hunukumbure, M. 2008. Dimensioning and cost analysis of multihop relay-enabled WiMAX networks. Fujitsu Scientific & Technical Journal 44(3): 303-317.

Vanghan, R. & Andersen, J.B. 2003. Channels, propagation, and antennas for mobile communications. London: The Institution of Electrical Engineer.

Wang, T., Cano, A., Giannakis, G.B. & Laneman, J.N. 2007. High-performance cooperative demodulation with decode-and-forward relays. IEEE Transactions on Communications 55(7): 1427-1438.

Wang, F., Ghosh, A., Sankaran, C., Fleming, P. J., Hsieh, F. & Benes, S. J. 2008. Mobile WiMAX systems: performance and evolution. IEEE Communications Magazine 46(10): 41-49.

Wang, H., Yin, Q. & Xia, X.-G. 2011. Full diversity space-frequency codes for frequency asynchronous cooperative relay networks with linear receivers. IEEE Transactions on Communications 59(1): 236-247.

WiMAX Forum. 2006a. Mobile WiMAX – Part II: A comparative analysis. White Paper.

WiMAX Forum. 2006b. Mobile WiMAX – Part I: A technical overview and performance evaluation. White Paper.

197

WiMAX Forum. 2006c. WiMAX Forum mobile system profile (Release 1.0 Approved Specification; Revision 1.2.2). White Paper.

WiMAX Forum. 2008a. WiMAX system evaluation methodology version 2.1.

WiMAX Forum. 2008b. The WiMAX Forum certifiedTM program. White paper.

WiMAX Forum. 2010. Network architecture – architecture tenets, reference model and reference points base specification. White Paper, WMF-T32-001-R016v01.

Wornell, G.W. 1998. Linear diversity techniques for fading channels. In Poor, H.V. & Wornell, G.W. (ed.). Wireless communications, signal processing perspectives, pp. 1-63. New Jersey: Prentice Hall.

Yagoobi, H. 2004. Scalable OFDMA physical layer in IEEE 802.16 wireless MAN. Intel Technology Journal 8(3): 201-212.

Yang, H, Lee, H. & Lee, M. 2008. A mobility management protocol for multi-hop relay networks. Proceedings of 10th IEEE International Conference on Advanced Communication Technology (ICACT ’08) 1: 37-42.

Yang, W.-H. & Tseng, Y.-C. 2008. Mobile WiMAX networks and their integration with WiFi networks. In Xiao, Y. (ed.). WiMAX/MobileFi advanced research and technology, pp. 31-48. New York: Taylor & Francis Group.

Yanikomeroglu, H. 2002. Fixed and mobile relaying technologies for cellular networks. Proceedings of Second Workshop on Applications and Services in Wireless Networks (ASWN’02), pp. 75-81.

Yanikomeroglu, H. 2006. Tutorial: Infrastructure-based wireless multihop/relay/mesh networks. IEEE Wireless Communication and Networking Conference (WCNC’06).

Yap, J.H., Yang, X., Ghaheri-Niri, S. & Tafazolli, R. 2002. Position assisted relaying and handover in hybrid ad hoc WCDMA cellular system. Proceedings of the 13th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, (PIMRC ‘02) 5: 2194-98.

Yoo, H.-I., Kim, Y.-J., Woo, K.-S., Kim, J., Yun, S. & Cho, Y.S. 2009. A hierarchical preamble design technique for efficient handovers in OFDM-based multi-hop relay systems. IEICE Transactions on Communications E92.B(12): 3907-3910.

198

Zhao, S., Teo, K.H., Tao, Z. & Zhang, J. 2008. Macro diversity handover and fast access station switching in wireless multi-user multi-hop relay networks. US Patent Application, US 2008/0165736 A1.

http://www.freepatentsonline.com/20080165736.pdf [14 December 2010]

199

LIST OF PUBLICATIONS

Journal Papers 1. Sultan, J., Ismail, M., Misran, N. & Jumari, K. 2008. Spectral Efficiency

Evaluation of Downlink Mobile Multi-hop Relay Systems Employing Macro Diversity Handover Technique. International Journal of Computer Science and Network Security (IJCSNS) 8(5): 122-129.

2. Sultan, J., Misran, N., Ismail, M. & Islam, M.T. 2010. An enhanced Macro Diversity Handover Technique for IEEE 802.16j. IEICE Electronics Express Journal (ELEX) 7(10): 732-737.

3. Sultan, J., Misran, N., Ismail, M. & Islam, M.T. 2011. Topology-Aware Macro Diversity Handover Technique for IEEE 802.16j Multi-hop Cellular Networks. IET Communications Journal 5(5):700-708.

4. Sultan, J., Misran, N., Ismail, M. & Islam, M.T. 2011. A spectrally Efficient Macro Diversity Handover Technique for Interference-Limited IEEE 802.16j Multihop Wireless Relay Networks. ETRI Journal 33(4): 558-568.

Proceeding Papers 5. Sultan, J., Ismail, M. & Misran, N. 2008. Downlink Performance of Handover

Techniques for IEEE 802.16j Multi-hop Relay Networks. Proceedings of 4th IEEE International Conference on Internet (ICI2008), Uzbekistan, pp. 1-4.

6. Sultan, J., Misran, N., Ismail, M. & Islam, M.T. 2008. Handover Techniques for Relay-Enhanced Mobile Wireless Broadband Networks. Proceedings of Engineering Postgraduate Conference (EPC2008), UKM, pp. 1-8.

7. Sultan, J., Misran, N., Ismail, M. & Islam, M.T. 2008. Performance Evaluation of Handover Techniques for Mobile Multi-hop Relay Systems. Proceedings of 3rd Brunei International Conference on Engineering and Technology (BICET2008), Brunei.

8. Yacoob, N., Ismail, M., Sultan, J., Ibrahim, M.Y., Mohamad, H. & Misran, N. 2010. Performance Evaluation of IEEE 802.16j with Fixed and Mobile Relay Stations. Proceedings of 4th International Symposium on Broadband Communication (ISBC2010), Melaka, Malaysia.

9. Sultan, J., Misran, N., Ismail, M., Islam, M.T. & Mohamad, H. 2010. Performance Evaluation of Macro Diversity Handover Technique for Multi-hop Relay Cellular Networks. Proceedings of 16th IEEE Asia-Pacific Conference on Communications (APCC2010), New Zealand, pp. 414-418.

200

Patent 10. Ismail, M., Sultan, J., Jumari, K., Misran, N. & Mohamad, H. 2010. Improved

Macro Diversity Handover in Wireless Multi-hop Relay Networks. Patent Application Number: D722. Under filing.