prr.hec.gov.pkprr.hec.gov.pk/jspui/bitstream/123456789/7886/1/PhD... · Department of Computer...

Energy Efficiency in Wireless Sensor Networks UsingCollaborative Communication in Wideband Channels

Ph.D Thesis

By

Anwar Ghani70-FBAS/PHDCS/F11

Supervisor

Prof. Dr. Muahmmad SherDean, FBAS, IIU

Co-Supervisor

Dr. Syed Husnain Abbas NaqviChairman, DCS & SE, FBAS, IIU

Department of Computer Science & Software EngineeringFaculty of Basic & Applied Sciences

International Islamic University, Islamabad2016

A dissertation submitted to theDepartment of Computer Science & Software Engineering,

International Islamic University, Islamabadas a partial fulfillment of the requirements

for the award of the degree ofDoctor of Philosophy in Computer Science.

Anwar Ghani: 70-FBAS/PHDCS/F11 Page i of 153

Department of Computer Science & Software EngineeringInternational Islamic University Islamabad

Date: November 30, 2016

Final ApprovalIt is certified that we have examined the thesis report submitted by Mr. Anwar Ghani, RegistrationNo. 70-FBAS/PhD(CS)/F11, and it is our judgment that this thesis is of sufficient standard to war-rant its acceptance by the International Islamic University, Islamabad for the Doctor of Philosophyin Computer Science.

Committee:

External Examiners

Dr. Mujahid AlamPrincipal ScientificPakistan Atomic Energy Commission (PAEC)

Dr. Nadeem JavaidAssociate ProfessorDepartment of Computer ScienceCOMSATS Institute of Information Technology, Islamabad

Internal Examiner

Dr. Shehzad Ashraf Ch.Assistant ProfessorDepartment of Computer Science & Software EngineeringInternational Islamic University Islamabad

Supervisor

Dr. Muhammad SherProfessor/DeanDepartment of Computer Science & Software EngineeringFaculty of Basic & Applied SciencesInternational Islamic University Islamabad

Co-Supervisor

Dr. Syed Husnain Abbas NaqviChairmanDepartment of Computer Science & Software EngineeringInternational Islamic University Islamabad

Anwar Ghani: 70-FBAS/PHDCS/F11 Page ii of 153

Declaration

I hereby declare that this thesis, neither as a whole nor as a part thereof has been copied out fromany source. It is further declared that no portion of the work presented in this report has beensubmitted in support of any application for any other degree or qualification of this or any otheruniversity or institute of learning.

Anwar Ghani

Anwar Ghani: 70-FBAS/PHDCS/F11 Page iii of 153

Dedication

Dedicated to My family, especially to my parents, my wife and my sons Kashif, Yasir, Atif & Abu

Bakar.

Anwar Ghani

Anwar Ghani: 70-FBAS/PHDCS/F11 Page iv of 153

AcknowledgmentsI am very grateful to ALLAH the ALMIGHTY for without His grace and blessing this study wouldnot have been possible.

Foremost, I would like to express my sincere gratitude to my supervisor Prof. Dr. Muhammad Sher

for the continuous support of my Ph.D study and research, for his patience, motivation, enthusiasm,and immense knowledge. His guidance helped me in all the time of research and writing of thisthesis. I could not have imagined having a better advisor and mentor for my Ph.D study.

I would like to extend immeasurable appreciation and deepest gratitude for the help and support ofmy Co-Supervisor Dr. Husnain Abbas Naqvi, who gave all his knowledge, guidance and support toboost my confidence and learning. His mentoring and encouragement have been specially valuable,and his early insights launched the greater part of this dissertation.

I would also like to acknowledge my friends, students and colleagues especially Mr. Imran Khan,

Mr. Syed Muhammad Saqlain, Mr. Shehzad Ashraf Ch., Mr. Tawab Khan, Mr. Zahid Mahmood,

Syed Muhammad Bilal, Mr. Kamranullah, Mr. Atif Shabbir and Mr. Khalid Mehmood. All of themencouraged and provided logistic and technical help during this research.

I would also like to thank my wife who has supported me patiently and firmly during completionof this task.

I would like to admit that I owe all my achievements to my truly, sincere and most loving parentsand friends who mean the most to me, and whose prayers have always been a source of determina-tion for me.

Anwar Ghani: 70-FBAS/PHDCS/F11 Page v of 153

List of Publications

1. Published Article(s) (from thesis) in SCI/SCIE Journals (IF = 6.311)

(a) Anwar Ghani, Husnain Naqvi, Muhammad Sher, Imran Khan, Muazzam A. Khan,Azeem Irshad, “Spread Spectrum Based Energy Efficient Collaborative Communi-cation in Wireless Sensor Networksn” PLoS ONE, Public Library of Science, 2016,11(7), 1-24. (impact factor 3.057)

(b) Anwar Ghani, Husnain Naqvi, Muhammad Sher, Zeeshan Shafi Khan, Imran Khan,& Muhammad Saqlain, “Energy efficient communication in body area networks usingcollaborative communication in Rayleigh fading channel”, Telecommunication Sys-tems, Springer US, 2015, 63(3), 357-370.(impact factor 0.705)

2. Journal papers (from thesis) under review

(a) Anwar Ghani, Husnain Naqvi, Shehzad Ashraf Ch., Muhammad Sher, Imran Khan, M.Khurram Khan, “Energy efficient communication in multipath Rayleigh faded wirelesssensor networks using collaborative communication”, Wireless Networks (Springer).(impact factor 1.006).

(b) Anwar Ghani, Husnain Naqvi, Muhammad Sher, Imran Khan, Muhammad Saqlain,“Capacity Gain in Spread Spectrum Based Collaborative Communication with unsyn-chronized Phase in Sensor Networks”, Int. J. of Ad Hoc and Ubiquitous Computing.(impact factor 0.489)

3. Journal papers (others)

(a) E. Rahman, M. Sher, H. Naqvi, A. Ghani, “Enhancing Energy Efficiency and QoSusing Buffer Queue Management Algorithms in Stable Clustering for Wireless MobileSensor Network”, International Journal of Computer Science and Information Security,2016, 14(11), 714-722.

(b) K. Mansoor, A. Ghani, A. F. Baig, H. Naqvi, I. Khan, M. Saqlain, “Two Level Authen-tication Scheme for Securing Session Passwords”, International Journal of ComputerScience and Information Security, 2016, 14(11), 374-378.

(c) I. Khan, J. I. Khan, M. Sher, S. M. Saqlain, A. Ghani, M. U. Ashraf, “Clinical Doc-ument Construction using HL7 with Medical Drop Box for Exchange of Electronic

Anwar Ghani: 70-FBAS/PHDCS/F11 Page vi of 153

Health Records under Country Medical Law”, International Journal of Computer Sci-ence and Information Security, 2016, 14(10), 559-576.

(d) Khan I, Sher M, Aslam S, Saqlain SM, Ashraf MU, Khan JI, Rauf M, Ghani A. Medicaldrop box (MDB); A national health information exchange and management system formedical industry. Professional Medical Journal, 2016, 23(4), 489-498.

(e) Khan, I., Sher, M., Khan, J. I., Saqlain, S. M., Ghani, A., Naqvi, H. A., & Ashraf,M. U. “Conversion of Legal Text to a Logical Rules Set from Medical Law Using theMedical Relational Model and the World Rule Model for a Medical Decision SupportSystem”. In Informatics (Vol. 3, No. 1, p. 2), 2016.

(f) Ch, S., uddin, N., Sher, M., Ghani, A., Naqvi, H. & Irshad, A. “An efficient signcryp-tion scheme with forward secrecy and public verifiability based on hyper elliptic curvecryptography” Multimedia Tools and Applications, Springer US, 2014, 1-13. (impact

factor 1.058)

(g) Irshad, A.; Sher, M.; Faisal, M. S.; Ghani, A.; Ul Hassan, M. & Ashraf Ch, S. “A secureauthentication scheme for session initiation protocol by using ECC on the basis of theTang and Liu scheme”, Security and Communication Networks, 2014, 7, 1210-1218(impact factor 0.433)

(h) A. Irshad, M. Sher, E. Rehman, S. A. Ch, M. U. Hassan, and A. Ghani, “A singleroundtrip sip authentication scheme for voice over internet protocol using smart card”,Multimedia Tools and Applications, pp. 1-18, 2013. (impact factor 1.058)

4. Journal papers (others) under review

(a) Anwar Ghani, Saeedullah Jan, Khwaja M. Hassan, Imran Khan, Husnain Naqvi, EidRahman, Muhammad Saqlain, M. Usman Ashraf and M. Tawab Khan, “Cloud Com-puting Storage Architecture: Issues and Challenges”, IEEE Transaction on parallel

and Distributed Systems. (impact factor 2.661)

5. Conference Publication(s)

(a) Z. Mehmood, N. Nizamuddin, S. Ch, W. Nasar, and A. Ghani, “An efficient key agree-ment with rekeying for secured body sensor networks”, in Second International Con-ference Digital Information Processing and Communications (ICDIPC), IEEE, 2012,pp. 164-167

Anwar Ghani: 70-FBAS/PHDCS/F11 Page vii of 153

Abstract

Wireless Sensor Network (WSN) consists of a large number of tiny sensor nodes capable of beingdeployed in a range of environments in random or regular fashion. These networks are gettingattention of the research community due their broad application domain. They can be applied inmany fields like health care, homes, military, environment monitoring or in any commercial en-vironment. Due to resource constrained nature of the nodes, these networks need energy efficientsolutions. Since a network operation is many times more expensive than local operation, there-fore very efficient and highly adaptable communication systems need to be developed for thesenetworks in order to increase a node’s life time. To avoid resource limitation and achieve energyefficiency collaborative communication combines the power of multiple sensor nodes to transmitthe same data at the same time to a base station.

This research focuses on energy efficiency in Body Area Networks, sensor networks with multipathand wideband channels. All these models are investigated in the presence of Rayleigh FadingAdditive White Gaussian Noise (AWGN). The received signals from collaborative nodes at the basestation are considered to be out-of-phase (imperfect phase synchronization). The ultimate goal is toinvestigate the effect of collaborative communication on energy efficiency, channel capacity gain,received power, BER in Wireless Sensor Networks using narrow-band, multipath and widebandchannels.

A synchronization process is designed to reduce the phase and frequency synchronization errorsamong the transmitter (Collaborative) nodes and the receiver (base station). A theoretical model forunsynchronized phase using collaborative communication in the presence of Additive White Gaus-sian Noise (AWGN) and Rayleigh fading is proposed, analyzed and simulated. The performance ofcollaborative communication system is evaluated by investigating several figures of merits includ-ing “received power”, “BER”, “energy efficiency” and “channel capacity”. The theoretical findingsof collaborative communication system are verified using Monte Carlo simulation by consideringthe parameters of “off-the-shelf” products i.e. “CC2420” and “AT86RF212”.

Theoretical analysis of the derived models for received power gain, BER, energy efficiency andcapacity gain show that an increase in the number of collaborative nodes also increases the gain inreceived power and capacity of the system whereas inversely effect BER.

Simulation results for phase error intervals {−0.1 ∼ 0.1}, {−0.2 ∼ 0.2}, {−0.3 ∼ 0.3} and{−0.4 ∼ 0.4} in case of BAN show a 0.475N2 to 0.8N2 gain in received power whereas to achievea BER of 10−3, the required transmited power decreases from 12.5dB to 10dB with an increase

Anwar Ghani: 70-FBAS/PHDCS/F11 Page viii of 153

in the number of nodes from 5 to 11 over the phase error interval {−0.2 ∼ 0.2}. This requiredpower to achieve the desired BER raises 15dB to 12.5dB in case of phase error {−0.4 ∼ 0.4}. Inmultipath communication the gain in received power improves from 0.49N2 to 0.83N2 whereasrequired power for BER of 10−3 in case of {−0.1 ∼ 0.1} decreases from 10dB to 7.5dB andfor {−0.3 ∼ 0.3} the decrease is from 12dB to 9dB. In case of wideband communication thegain in received power ranges from 0.51N2 to 0.93N2 and the required power for BER of 10−3

for single node is 7.5dB and for nodes from 5 to 11 it is 3dB to 2dB. For trade-off-analysisof energy saving and transmission distances performed for off-the-shelf devices “CC2420” and“AT86RF212”, shows in all scenarios “CC2420” stabilizes before “AT86RF212”. On the basis ofthese results it can be concluded that collaborative communication is energy efficient and suitablefor resource limited networks like WSN.

Anwar Ghani: 70-FBAS/PHDCS/F11 Page ix of 153

Contents

List of Figures xiv

List of Tables xvii

1 Introduction 11.1 Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Wireless Sensor Networks Applications . . . . . . . . . . . . . . . . . . . . . . . 3

1.2.1 Military Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2.2 Healthcare Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2.3 Environmental Applications . . . . . . . . . . . . . . . . . . . . . . . . . 41.2.4 Home Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2.5 Commercial Applications . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3.1 Sensor Node Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3.2 Sensor Network Architecture . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.4 Sensor Networks and Digital Communication Systems . . . . . . . . . . . . . . . 81.5 Challenges and Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.6 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101.7 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Wireless Communication Preliminaries 142.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.2 Wireless Communication Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 152.3 Antennas and Antenna-Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.3.1 Multiple input multiple output (MIMO) Systems . . . . . . . . . . . . . . 182.4 Signal Power Loss in Wireless Channels . . . . . . . . . . . . . . . . . . . . . . . 19

x

Contents

2.4.1 Free space signal path loss model . . . . . . . . . . . . . . . . . . . . . . 202.4.2 Two-ray ground signal path loss model . . . . . . . . . . . . . . . . . . . 202.4.3 Log-normal shadowing signal power loss model . . . . . . . . . . . . . . 202.4.4 Simplified signal path loss model . . . . . . . . . . . . . . . . . . . . . . 21

2.5 Multipath and Small-scale Fading . . . . . . . . . . . . . . . . . . . . . . . . . . 222.5.1 Physical Factors Influencing Fading . . . . . . . . . . . . . . . . . . . . . 222.5.2 Types of Small Scale Fading . . . . . . . . . . . . . . . . . . . . . . . . . 232.5.3 Fade Margin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.6 Multiple Access Methods for Multi-user Communication Systems . . . . . . . . . 252.6.1 Time Division Multiple Access (TDMA) . . . . . . . . . . . . . . . . . . . 262.6.2 Frequency Division Multiple Access (FDMA) . . . . . . . . . . . . . . . . 262.6.3 Frequency Time Division Multiple Access (F/TDMA) . . . . . . . . . . . 262.6.4 Code Division Multiple Access (CDMA) . . . . . . . . . . . . . . . . . . 272.6.5 Space Division Multiple Access (SDMA) . . . . . . . . . . . . . . . . . . 28

2.7 Wideband Channels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282.7.1 Rayleigh Fading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282.7.2 Channel Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.8 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3 Literature Review 303.1 Multihop Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.2 Cooperative Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2.1 Amplify-&-Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.2.2 Decode-&-Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.2.3 Compress-&-Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.2.4 Cooperative Communication in Wireless Sensor Network . . . . . . . . . . 343.2.5 Limitations of Cooperative Communication . . . . . . . . . . . . . . . . . 36

3.3 Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.3.1 Beamforming in Wireless Sensor Networks . . . . . . . . . . . . . . . . . 373.3.2 Limitations of Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.4 Collaborative Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.4.1 Why Collaborative Communication . . . . . . . . . . . . . . . . . . . . . 413.4.2 Synchronization Process in Collaborative Communication . . . . . . . . . 42

3.5 Received Power Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Anwar Ghani: 70-FBAS/PHDCS/F11 Page xi of 153

Contents

3.6 Bit Error Rate(BER) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.7 Capacity Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483.8 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

4 Proposed Methodology 514.1 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514.2 Performance Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.2.1 Received Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534.2.2 Bit Error Rate(BER) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544.2.3 Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.2.4 Capacity Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4.3 Expected Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5 Collaborative Communication in Body Area Networks (BAN) 575.1 Proposed System Model for BAN . . . . . . . . . . . . . . . . . . . . . . . . . . 58

5.1.1 Theoretical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595.1.2 Probability of Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5.2 Collaborative Communication: Energy Consumption in BAN . . . . . . . . . . . . 645.2.1 BAN: SISO Energy Consumption Model . . . . . . . . . . . . . . . . . . . 655.2.2 BAN: Collaborative Communication Energy Consumption Model . . . . . 66

5.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685.4 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

6 Collaborative Communication in Multipath Fading 776.1 Proposed System Model for Multipath Collaborative Communication . . . . . . . 78

6.1.1 Theoretical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796.1.2 Average Probability of Error . . . . . . . . . . . . . . . . . . . . . . . . . 82

6.2 Collaborative Communication: Energy Efficiency . . . . . . . . . . . . . . . . . . 856.2.1 SISO Systems: Energy Consumption Model . . . . . . . . . . . . . . . . . 856.2.2 Collaborative Communication: Energy Consumption Model . . . . . . . . 86

6.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 886.4 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

7 Collaborative Communication in Spread Spectrum(wideband channels) 967.1 System Model for Collaborative Communication in Wideband Channel . . . . . . 98

Anwar Ghani: 70-FBAS/PHDCS/F11 Page xii of 153

Contents

7.1.1 Theoretical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 987.1.2 Average Probability of Error . . . . . . . . . . . . . . . . . . . . . . . . . 102

7.2 Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1057.2.1 SISO energy consumption model . . . . . . . . . . . . . . . . . . . . . . . 1057.2.2 Collaborative communication energy consumption model . . . . . . . . . . 106

7.3 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1087.4 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

8 Collaborative Communication in Spread Spectrum: Capacity Gain 1178.1 System Model for Capacity Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

8.1.1 Theoretical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1198.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1218.3 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

9 Comparison with other Approaches 1269.1 Comparison with Multihop communication . . . . . . . . . . . . . . . . . . . . . 1269.2 Comparison with Cooperative communication . . . . . . . . . . . . . . . . . . . . 1289.3 Comparison with Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

10 Conclusion and Future Work 13110.1 Summary of Key Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

10.1.1 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13410.2 Suggestions for Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

Appendix A Derivation of Trigonometric Functions 136A.1 Mean Of Cosine Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

A.1.1 Mean value of cos(θf ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136A.1.2 Mean value of cos2(θf ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

A.2 Variance Of Cosine Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137A.2.1 Variance of cos(θf ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137A.2.2 Variance of αX cos(θf ) . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

Bibliography 139

Anwar Ghani: 70-FBAS/PHDCS/F11 Page xiii of 153

List of Figures

1.1 A typical wireless network based on infrastructure . . . . . . . . . . . . . . . . . . 11.2 Hardware components of a sensor node . . . . . . . . . . . . . . . . . . . . . . . 61.3 Software components of a sensor node . . . . . . . . . . . . . . . . . . . . . . . 71.4 Wireless sensor network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.5 Chapter wise flow of the study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.1 Components of wireless digital communication system . . . . . . . . . . . . . . . 162.2 Fade margins in case of cellular application . . . . . . . . . . . . . . . . . . . . . 252.3 Time and Frequency domain representation of FDMA, TDMA, and CDMA . . . . 27

3.1 The cooperative relay network, also known as the relay channel. . . . . . . . . . . 323.2 Beamforming scenario between the nodes and the far-away base station . . . . . . 373.3 Top level of master slave based model for information transfer . . . . . . . . . . . 383.4 Collaborative Communication Model . . . . . . . . . . . . . . . . . . . . . . . . 413.5 Architecture of the phase synchronization process . . . . . . . . . . . . . . . . . . 433.6 System Model for Collaborative Communication with Random Distribution of

Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453.7 System model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.1 Geometry of sensor network used in collaborative communication based modelspresented in later chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4.2 Proposed theoretical model based on collaborative communication . . . . . . . . . 53

5.1 General architecture of BAN for healthcare . . . . . . . . . . . . . . . . . . . . . 585.2 A communication system with fixed antenna array . . . . . . . . . . . . . . . . . . 595.3 Collaboration based system model for BAN . . . . . . . . . . . . . . . . . . . . . 60

xiv

List of Figures

5.4 Normalized average received power vs. number of collaborative nodes with Rayleighfading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5.5 Average received power/N vs. number of collaborative nodes with Rayleigh fading. 705.6 Bit Error Rate over interval {−0.2π to 0.2π} for different number of nodes with

fading and total transmitted energy . . . . . . . . . . . . . . . . . . . . . . . . . 715.7 Bit Error Rate over interval {−0.4π to 0.4π} for different number of nodes with

fading and Eb/N total transmitted energy . . . . . . . . . . . . . . . . . . . . . . 725.8 Break-even distances and energy preservation in the presence of phase error dis-

tributed over {−0.3π ∼ 0.3π} using different varying number of transmitter nodesN for AT86RF212 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.9 Break-even distances and energy preservation in the presence of phase error dis-tributed over {−0.4π ∼ 0.4π} using different varying number of transmitter nodesN for AT86RF212 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.10 Break-even distances and energy preservation in the presence of phase error dis-tributed over {−0.3π ∼ 0.3π} using different varying number of transmitter nodesN for CC2420 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.11 Break-even distances and energy preservation in the presence of phase error dis-tributed over {−0.4π ∼ 0.4π} using different varying number of transmitter nodesN for CC2420 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

6.1 Multipath system geometry where the dotted lines represent multiple scatter com-ponents of a signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

6.2 Proposed Theoretical System Model for Multipath Collaborative Communication . 796.3 Normalized received signal power vs. number of transmitter nodes in the presence

of Rayleigh fading and phase error . . . . . . . . . . . . . . . . . . . . . . . . . . 896.4 Average received signal power “power/N” vs. number of transmitter nodes in the

presence of Rayleigh fading and phase error . . . . . . . . . . . . . . . . . . . . . 906.5 Bit error rate plot against number of collaborative transmitters for phase error in-

terval {−0.1π ∼ 0.1π} for m = 3 . . . . . . . . . . . . . . . . . . . . . . . . . . 916.6 Bit error rate plot against number of collaborative transmitters for phase error in-

terval {−0.3π ∼ 0.3π} for m = 3 . . . . . . . . . . . . . . . . . . . . . . . . . . 926.7 Energy efficiency and break-even distance for varying number of nodes for AT86RF212 936.8 Energy efficiency and break-even distance for varying number of nodes for CC2420 93

7.1 Geometry of sensor nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

Anwar Ghani: 70-FBAS/PHDCS/F11 Page xv of 153

List of Figures

7.2 System model for collaborative communication in wideband channel . . . . . . . . 1007.3 Normalized received signal power vs. number of transmitter nodes with Rayleigh

fading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1097.4 Average total received signal power “power/N” vs. number of transmitter nodes

with Rayleigh fading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1107.5 BER over interval {−0.1π ∼ 0.1π} for varying number of nodes with fading and

total transmitted energy Eb/N0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1117.6 BER over interval {−0.2π ∼ 0.2π} for varying number of nodes with fading and

total transmitted energy Eb/N0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1127.7 percentage energy savings and break-even distances for AT86RF212, for different

number of collaborative nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1147.8 Percentage energy savings and break-even distances for CC2420, for different

number of collaborative nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

8.1 Universal frequency reused by spread spectrum . . . . . . . . . . . . . . . . . . . 1188.2 System model for capacity gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1198.3 Capacity gain in case of perfect phase synchronization in the received signals i.e,

φ = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1218.4 Capacity for phase error over the interval “{−0.1π ∼ 0.1π}” and each node trans-

mits power P, so total transmitted power by the network is mP . . . . . . . . . . . 1228.5 Capacity for phase error distributed over “{−0.4π ∼ 0.4π}′′ and each node trans-

mits power P, so total transmitted power by the network is mP . . . . . . . . . . . 1238.6 Capacity for phase error over the interval “{−0.1π ∼ 0.1π}” and each node trans-

mits power P/m, so total transmitted power by the network is P . . . . . . . . . . . 1248.7 Capacity for phase error distributed over {−0.4π ∼ 0.4π} and each node transmits

power P/m, so total transmitted power by the network is P . . . . . . . . . . . . . . 125

9.1 Comparison of collaborative communication with Multihop system . . . . . . . . . 1279.2 Multi relay cooperative communication system . . . . . . . . . . . . . . . . . . . 128

Anwar Ghani: 70-FBAS/PHDCS/F11 Page xvi of 153

List of Tables

3.1 Maximum Energy Savings of Multi-Node Cooperative Systems. . . . . . . . . . . 35

5.1 Product parameters, data and description used in the simulation experiments . . . . 685.2 Break-even distance based on the parameters of the devices used in experiments

i.e. CC2420 and AT86RF212. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 755.3 Percentage energy preservation based on the parameters of CC2420. . . . . . . . . 755.4 Percentage energy preservation based on the parameters of AT86RF212. . . . . . . 76

6.1 Break-even distance based on the parameter of CC2420 and AT86RF212. . . . . . 946.2 Percentage energy preservation based on the parameters of CC2420. . . . . . . . . 946.3 Percentage energy preservation based on the parameters of AT86RF212. . . . . . . 94

7.1 Break-even distance based on the parameters of CC2420 and AT86RF212. . . . . . 1157.2 Percentage energy preservation based on the parameters of CC2420. . . . . . . . . 1157.3 Percentage energy preservation based on the parameters of AT86RF212. . . . . . . 115

9.1 Comparison of power gain & transmitted power . . . . . . . . . . . . . . . . . . . 129

xvii

Acronyms

AF Amplify-and-Forward.

AM Amplitude Modulation.

AWGN Additive White Gaussian Noise.

BAN Body Area Network.

BER Bit Error Rate.

BPSK Binary Phase Shift Keying.

BS Base Station.

CDMA Code Division Multiple Access.

DF Decode-and-Forward.

DSSS Direct Sequence Spread Spectrum.

DoA Direction of Arrival.

F/TDMA Frequency/Time Division Multiple Access.

FDD Frequency Division Duplex.

FDMA Frequency Division Multiple Access.

FHHS Frequency Hopping Spread Spectrum.

FM Frequency Modulation.

GSM Global System for Mobile Communication.

IEEE Institute of Electrical and Electronic Engineers.

xviii

Acronyms

IMT-2000 International Mobile Telephony 2000.

ISI Inter Symbol Interference.

ITU International Telecommunication Union.

LoS Line-of-Sight.

MIMO Multi-Input-Multi-Output.

MISO Multiple input single output.

QoS Quality of Service.

RMS Root Mean Square.

SDMA Space Division Multiple Access.

SER Symbol error rate.

SINR Signal to Interference and Noise Ratio.

SISO Single Input Single Output.

SNR Signal-to-Noise Ratio.

TD-SCDMA Time Division Synchronous Code Division Multiple Access.

TDD Time Division Duplex.

TDMA Time Division Multiple Access.

UMTS Universal Mobile Telecommunications Standard.

VCR Video Cassette Recorder.

WCDMA Wideband Code Division Multiple Access.

WLAN Wireless Local Area Network.

WPAN Wireless Personal Area Networks.

WSN Wireless Sensor Network.

WiMAX Worldwide Interoperability for Microwave Access.

Anwar Ghani: 70-FBAS/PHDCS/F11 Page xix of 153

Chapter 1

Introduction

With the fast growth of information technology wireless communication is getting more and moreattention every day. Computer users like to go beyond the restrictions of guided media to avoidSpaghetti of cables connected to their computers. Mobile industry is moving from simple phoneto Smartphones. Ubiquitous communication is becoming the need of the day, as it is the onlyanswer to a very important question: “How to access the computing and communication serviceson the move?”. In a typical wireless network a land area that should be covered with a radioservice is divided into cells. Each cell is covered by a Base Station (BS) having wired connectionto the backbone network for reception and transmission of data from a specific cell. However, thecommunication between a BS and a Mobile node is over a wireless channel. The BS and backbonenetwork work as a relay for the traffic among Mobile nodes. Switching from one base station toanother is allowed for mobile nodes in order to support mobility [1] as shown in Figure 1.1, e.g.,GSM, UMTS, WLAN.

WSN can be described as a wireless network of thousands of inexpensive miniature devices capable

Figure 1.1: A typical wireless network based on infrastructure

1

Chapter 1. Introduction

of computation, communication and sensing their environment. These devices provide a bridgebetween the real, physical and virtual worlds introduced as the “many-tiny” principle in the U.SNational Research Council Report of 2001 [2].

1.1 Wireless Sensor Networks

In traditional networks the devices are close to human users for interaction. However if we wantto interact with the environment instead of human users, what alternative do we have? The answeris, WSN which can be embedded in any environment in a regular or random fashion, in order tomeasure/influence the environment [3].

WSN is a network of a large number of tiny, battery powered, low cost sensor nodes [4–6]. A sensornode is capable of sensing or reading the physical data from its environment using on-board sensingdevices, and communicating this data wirelessly through intermediate nodes to the network users[6, 7]. It is not mandatory to know position of a sensor in advance providing the freedom to deploysensor nodes randomly to facilitate relief operations in disaster or monitor inaccessible terrains.However, this signifies that protocols and algorithms designed for such networks must have theabilities of self organization. Another important feature of sensor nodes is the cooperation forperforming various tasks like information exchange, monitoring etc. A sensor node with the helpof an on-board processor, locally performs basic computations in order to avoid the exchange ofunnecessary data and save its resources [3, 8].

The rapid growth of WSN is characterized by different features of sensor nodes like, small size,low cost, and high processing power. This growth shows the fact that sensor networks gained theirmomentum in 1990s, not only in research community but also in industrial standardization andthe global business market. Many companies in Europe and America successfully commercial-ized WSNs in 1995 [9, 10]. WSN got ranked among the 21 most important technologies of the21st century by the Business Week Magazine in September 1999 [9]. Later in July 2002, the setup of WSNs was recommended and sponsored by the Commission of the European Communityto be deployed across Europe [11]. Different wireless standards have been introduced due to thesignificant rise in the popularity graph of wireless networks. In 2003 Institute of Electrical andElectronic Engineers (IEEE) standardized Wireless Personal Area Networks (WPAN) for the firsttime, and is known as IEEE 802.15.4 standard [12]. In the IEEE 802.15.4 standard the parame-ters considered for the radio frequency devices are low-data-rate, low-power, and low-complexity

Anwar Ghani: 70-FBAS/PHDCS/F11 Page 2 of 153


short-range. Original Equipment Manufacturers later adopted and recognized the IEEE 802.15.4

standard for WSN transceivers [13]. To address real world problems using sensor networks, theChinese Government in 2006 sponsored and introduced application driven methodologies [14]. By2010, the market value of the WSN in North America was expected to reach USD $5.9 billion [15].

Pakistan is a developing country and agriculture sector is a main contributor having a GDP of25.1%. In addition this country has been hit by many natural disasters, like flood, earthquack etc,destroying almost its major infrastructure. WSN has a great application potential in the above men-tioned areas. Intelligent and precise farming and monitoring may help in promoting sustainabilityin the country by improving the quality of its product and recovering from natural disasters.

1.2 Wireless Sensor Networks Applications

Sensors in a WSN may be of various types for example seismic, magnetic with low sampling rate,visual, thermal, acoustic, infrared and radar. They have the ability to monitor a vast range ofambient conditions. These conditions described in [16], are: “temperature, humidity, vehicularmovement, lightning condition, pressure, soil makeup, noise levels, presence or absence of certainkinds of objects, mechanical stress levels on attached objects, and current characteristics such asspeed, direction, and size of an object”.

WSN has a broad application domain and on the basis of these applications they can be broadlydivided into military, healthcare, environmental, home and other business/commercial areas [3].These application have been discussed thoroughly in the literature.

1.2.1 Military Applications

WSN are very suitable for hostile environments specially in military applications. These networkscan be used in different areas in military like command & control, battle field monitoring, targeting,

battle damage assessment, nuclear, biological and chemical attack detection and reconnaissance

and many more [4].



1.2.2 Healthcare Applications

Various applications of WSN in healthcare may include but are not limited to “telemonitoring ofhuman physiological data, tracking and monitoring doctors and patients inside a hospital, drugadministration in hospitals, monitoring the movements and internal processes of insects or othersmall animals” etc. [17–21].

1.2.3 Environmental Applications

In these application effect of the environmental changes on the living may be explored. May suchexample can be found like; birds and animal movement form one area to another with a changein the weather as well as insects monitoring, crops and livestock monitoring with a change isthe environmental conditions. Some other may be “irrigation; macro instruments for large-scaleearth monitoring and planetary exploration; chemical/biological detection; precision agriculture;biological, earth, and environmental monitoring in marine, soil, and atmospheric contexts, forestmonitoring, meteorological or geophysical research; flood detection, bio-complexity mapping ofthe environment, and pollution study” [17–20, 22–27].

1.2.4 Home Applications

Home applications of WSNs include home automation for example automated control of homeappliances like “vacuum cleaners, micro-wave ovens, refrigerators”, and VCRs [28], and smartenvironment which can either be human-centered or technology-centered [29, 30].

1.2.5 Commercial Applications

Some of the commercial applications are; “monitoring material fatigue, building virtual keyboards,managing inventory, monitoring product quality, constructing smart office spaces, environmentalcontrol in office buildings, robot control and guidance in automatic manufacturing environments,interactive toys, interactive museums, factory process control and automation, monitoring disas-ter area, smart structures with sensor nodes embedded inside, machine diagnosis, transportation,factory instrumentation, local control of actuators, detecting and monitoring car thefts, vehicle



tracking and detection, and instrumentation of semiconductor processing chambers, rotating ma-chinery, wind tunnels, and anechoic chambers” [17–21, 29, 31–33].

1.3 Architecture

The architecture of WSN may be explored from two perspective: architecture of a single sensornode and architecture of the overall sensor network. A short description of both the types is pre-sented here in the following subsections.

1.3.1 Sensor Node Architecture

There are mainly two parts of a sensor node; sensor hardware and sensor software.

1.3.1.1 Sensor Hardware

The hardware of a sensor can further be divided into various modules [34], like processor, memory,transceivers, sensors and battery module as shown in Figure 1.2

a. Processor & Memory:

A sensor is usually designed with a general purpose processor mainly optimized for embed-ded applications with low power consumption. Different examples include; Texas Instru-ment “MPS430”, “16-bit RISC core, up to 4 MHz, versions with 2-10 KBytes RAM, severalDACs, RT clock, prices start at 0.49 US$, Atmel AT Mega 128 micro-controller, 16 MHz,128 kByte flash [1].

b. Radio Transceiver:

Main functionality of a radio transceiver is to transmit or receive radio signals producedusing a bit or stream of bytes. On receiver side the received signal must be decoded back intothe transmitted data. Some well known example are; “Chipcon CC1000 (315/433/868/915MHz), CC2400 (2.4 GHz)”.

c. Battery:

Battery is the most critical resource in WSN. It should be used with high precision either by



Sensor 1

Sensor n

Micro Controller

Battery Source

Memory

Storage

Radio Transceiver

Figure 1.2: Hardware components of a sensor node

developing algorithms for reducing radio activities of a node or by designing energy efficientcommunication systems. The later one is the topic of discussion of this thesis.

d. Sensors:

A sensor node is equipped with n sensors for different tasks like sensing light, temperatureand motion etc. These sensors sense different events occur in its environment which arehandled using specific event handler by passing it through the system function unit. Afterprocessing these events are transmitted/forwarded by the transceiver.

1.3.1.2 Sensor Software

Sensor nodes are resource constrained specially having scarce power resource, that determinesthe lifetime of a node as well as life time of the network. As sensor nodes may be deployedin different environments like remote and hostile regions where ad hoc communications are a keycomponent. The operating system of a sensor node is typically less complex than a general purposeoperating system. Tiny (OS) is specifically designed for WSNs. It works on the principle of event-driven programming instead of multi-threading [34, 35]. Figure 1.3, taken from [1] shows softwarecomponents of a sensor node.



System Components

Event handler

System function

New event(Data packet)

Event(Sensor)

Event(Timer)

Event(Transceiver)

System function

New event(Timer)

Figure 1.3: Software components of a sensor node

1.3.2 Sensor Network Architecture

A WSN is composed of a large number of low-cost small sized nodes, where the location of a nodeis either fixed or randomly deployed to keep an eye on the environment [3]. On one hand wheresmall size and low cost of a sensor node makes it easy to be deployed anywhere, on the other handits small size makes it vulnerable to different problems and limitations. For communication witheach other, nodes in a WSN, follow the multihop approach. All nodes forward the sensed data to acommon receiver known as the base station(also referred to as sink). A typical WSN is shown inFigure 1.4. A BS acts like a bridge (gateway) between a WSN and any other network to exchangeinformation (sensed data) for further processing. Base station is a powerful node with no resourceconstraints like a sensor node, as it has to process large amount of complex data. Due to the highprocessing and storage requirements these are powerful machine with laptop/workstation levellarge memory processor where there is no limitation on energy, computational power or storage forperforming different operations smoothly. For base station to base station communication initiallylarge bandwidth channels are used [1].



Sink

Task managernode

User

Sensor FieldSensor nodes

E

B

AD C

Figure 1.4: Wireless sensor network

1.4 Sensor Networks and Digital Communication Systems

To investigate the role of digital communications in WSNs first let us have an introduction to digitalcommunication systems for wireless networks in general.

The broad application domain of WSNs make these networks more and more popular every day.There are limitless uses of these networks including but not limited to, AM and FM radio, televi-sion, cellular phones, WiMAX, Wi-Fi and wireless internet [36]. Wireless communication systemsoriginated from the wonderful invention of wireless telegraph system almost a century ago [37].There has been a rapid development in the methods of wireless communication that has not only re-sulted in significant improvements in size of a wireless radio device but also the processing cyclesper unit time. The fast growth can also be explained with logical reasons such as rate of com-petition and lower market prices of these devices. The broad band technologies like “WorldwideInteroperability for Microwave Access (WiMAX)”, mobile broadband and cellular mobile servicesi.e., Global System for Mobile Communication (GSM), Code Division Multiple Access (CDMA)and “Time Division Synchronous Code Division Multiple Access (TD-SCDMA)” have laid foun-dation for this growth. The projected statistics from the International Telecommunication Union(ITU) reported that in US and Europe every other person, in Asia every third person and in Africaevery fourth person would have a mobile phone by the end of 2001 [38]. Global Analyst House,CCS Insight, recently reported in its issue that the growth rate in Europe was expected to grow fast



thereby not only doubling the number of subscribers but also the revenue by the end of 2011 [39].

Microelectronics and its emergence in 1990s led to substantial development in communicationtechnology, not only in terms of size reduction but improvements in the processing speed. In thosetimes, many factors, like the invention of microelectronic circuits, improvements in sensing mate-rials and the development of high speed wireless and networking equipments made the inventionof Wireless Sensor Network (WSN) possible. The WSN is ubiquitous computing system enablingefficient communication [40–43].

1.5 Challenges and Requirements

Despite the wide range applications of WSNs, these networks are facing many challenges; nodesin WSNs usually suffer from resource limitation problem. Some important challenge ones are thedesign of WSN protocols that are energy efficient. Other challenges arise from hardware constraintslike dense deployment of nodes and infrastructural-free network [4].

Different challenges have been identified in literature [4, 44] which are as follow:

1. Resource constraints

2. Dynamic topologies and harsh environmental conditions

3. Quality of Service (QoS) requirements

4. Data redundancy

5. Packet errors and variable-link capacity

6. Security

7. Large-scale deployment and ad hoc architecture

8. Integration with Internet and other networks

For a network with large number of tiny sensor nodes with limited resources, important require-ment is to increase lifetime of the network. There is a need of designing low-complexity and easy-to-implement protocols for communication to run on sensor nodes with such processors. Moreover,in many situations like natural disasters, it is usually impossible that an infrastructure be set up be-forehand. In such situations WSN’s protocols must have provisions for organizing sensor nodeswithout human interventions in self organizing fashion (free of infrastructure) in order to meet the



scalability requirements. A sensor network is required to provide coverage reliability, coveragerobustness, throughput, transmission reliability and energy efficiency.

1.6 Objectives

As discussed earlier life of a WSN is mainly affected by the batteries of individual sensor nodes.Therefore, the main objective/goal of this work is to improve battery life of a sensor node by devel-oping an energy efficient communication system to provide satisfactory sensing and transmissionservices. One of the major objective of this research is to identify the factors that affect energy effi-ciency of sensor networks in Rayleigh fading (narrowband and wideband) channel in the presenceof noise i.e. AWGN. Some of the objectives of this research are given as follows:

1. Investigating/gathering various factors and parameters of the required energy consumptionfor communication in WSNs.

2. Development of communication system that is energy efficient using narrowband and wide-band channels in the presence of Rayleigh fading.

3. Determine the crucial factors that affects communication efficiency of the proposed commu-nication systems developed for WSNs.

4. Using “Bit Error Rate (BER)”, received signal power and channel capacity gain to evaluateperformance of the proposed communication systems developed for WSNs using collabora-tive communication in this research works.

5. Perform trade-off-analysis between energy consumption and transmission distances to verifywhether incorporating multiple transmitters in collaborative communication has positive ornegative impact on the overall energy consumption of the system.

For achieving the above listed objectives theoretical models are developed. To evaluate perfor-mance its performance the proposed models are simulated in MATLAB. Results from the sim-ulations (existing formulas) and theoretical expressions (derived formulas) are compared usingparameters of off-the-shelf devices.



1.7 Thesis Organization

This thesis is organized into nine chapters. Route map of the thesis is shown in Figure 1.5.

Chapter 2 presents the basic concepts of wireless communication system, specially related to phys-ical layer. Analysis of various models for calculating signal power loss is presented and theirmathematical expressions are analyzed. Different factors are identified which are crucial to energyefficiency in WSNs.

Chapter 3 presents an extensive review of the state of the art literature available on energy efficiencytechniques and approaches in sensor networks. The main focus of the study is the effect of fadingover power loss and multi-user access to channel and the roles it plays in the energy consumptionmodel. These models have been used in analyzing energy consumption in sensor networks.

Chapter 4 describes the methodology followed in this thesis. Factors affecting performance ofthe collaborative communication are described. Figures of merit to evaluate the theoretical andsimulated performance are explained.

Chapter 5 starts with description of the first algorithm based on collaborative communication forBody Area Network (BAN). In this chapter received power, probability of error BER and energyconsumption of collaborative communication are explored and analyzed. The chapter ends with acomparison of different energy efficiency techniques with collaborative communication.

Chapter 6 elaborates the same three algorithms as in chapter 4, but under different circumstances.In this chapter not only the effect of noise and unsynchronized phase is included but the fading isassumed to be multipath. Signal from each collaborative node undergoes scattering before arrivingat the receiver. This chapter analyzes how multipath effect can be explored as space diversity toachieve significant gains in received power, BER and energy consumption. Energy consumptionof multipath faded collaborative communication is analyzed by comparing its energy consumptionwith “SISO” system using parameters of off-the-shelf products, for example “AT86RF212” [45]and “CC2420” [46].

Chapter 7 turns its attention towards using collaborative communication in combination withspread spectrum(wideband channels) based channels. Since collaborative communication is verysuccessful in mitigating fading and unsynchronized phase effects, this combined the benefits ofspread spectrum technique like anti jamming, universal frequency reuse and specially noise sup-pression to produce high Signal-to-Noise Ratio (SNR) values for calculating power, BER and low



power transmission to improve energy efficiency.

Results obtained using MATLAB in all the proposed models in the above mentioned chapters basedon collaborative communication models are also analyzed and presented in each of the chapters.

Chapter 8 explores the effect of spread spectrum based collaborative communication on capacitygain. A closed form mathematical model based on wideband channels is developed and analyzed.Graphical demonstration and analysis of MATLAB based results in the presence of fading andAWGN is presented in this chapter.

Chapter 9 presents a brief theoretical comparison of collaborative communication with popular en-ergy efficiency techniques used in WSNs. The included techniques are; multihop communication,cooperative communication and beamforming. The comparison is based on energy consumptionin case of node failure, communication range and retransmission.

Chapter 10 presents the research findings and a brief summary of this research and its contribu-tions. This chapter also concludes this research work and future directions are given for exploringcollaborative communication further.

Chapter-wise flow of the study is shown in Figure 1.5



Chapter 1

Introduction

Chapter 3

Literature Review of Wireless Communication,

Fading, Fixed array antenna and Collaborative

Communication, Existing Problems

Chapter 4

Methodology Performance analysis

figures of merit for Evaluating Sensor

Networks performance

Chapter 10 Conclusions and future work

Chapter 5 Collaborative Communication in BAN with

Narrowband Channel

Received power gain with imperfect

Phase synchronization

Analysis and Results

Bit Error Rate with imperfect Phase

synchronization

Energy Efficiency with imperfect Phase

synchronization

Chapter 7 Collaborative Communication in Sensor Networks

with Wideband Channels





synchronization


synchronization

Chapter 6 Collaborative Communication with Multipath

Fading in Narrowband Channels





synchronization


synchronization

Chapter 8 Capacity calculation of Collaborative

Communication in Wideband Channels

Collaborative Communication with

imperfect Phase synchronization

Received Power in presence

Expression for Capacity Gain

Comparison with other

Techniques


Chapter 2

Description of wireless communication

preliminaries and the basic concept and

terminology related to physical layer

Chapter 9 Comparison with other Approaches

Figure 1.5: Chapter wise flow of the study


Chapter 2

Wireless Communication Preliminaries

2.1 Overview

Wireless communication though started in 1960’s, but it is today the most vibrant field of com-munication. This chapter briefly discusses the basics of wireless communication, its componentsand the factors which effect wireless communication. It also elaborates on antenna array, differentfading models and access techniques that are used in wireless communication. The knowledgeof these topics is mandatory because of the impassiveness of upper layers wireless sensor net-work protocols by the physical layer [47]. This chapter mainly focuses on the factors that are notonly responsible for the loss of signal power in transmission but also play a vital role in energyconsumption of sensor nodes.

Section 2.2 of this chapter is dedicated to the wireless communication system and its major compo-nents. A description of one of the crucial components of a wireless system i.e; antenna and antennaarray, is presented in 2.3. The benefits of antenna array over a single antenna based system are alsoelaborated. “Multi-Input-Multi-Output (MIMO)” systems are elaborated in section 2.3.1 which isby far the most important application of antenna array in wireless systems. MIMO system pro-vides the advantage of space diversity which is the basic motivation for developing collaborativecommunication based systems.

An elaboration of path loss models for wireless signals is presented in section 2.4. A descriptionof the factors which contribute to the path loss of a signal power in transmission is also presented.The description is also supported by mathematical models. Three main avenues are identified as

14

Chapter 2. Wireless Communication Preliminaries

the key source of path loss are i) channel path loss, ii) shadow fading and iii) small-scale fading[48]. These models are later used in derivation of energy efficiency models based on collaborativecommunication in sensor networks.

Channel fading is explored in section 2.4, the concept of multipath and its impact on signal powerhas been described. It includes shadowing and path loss of signal power, collectively known aslarge-scale fading whereas Rayleigh fading is referred to as small-scale fading [48, 49]. The effectsof shadowing and Rayleigh fading in case of indoor and outdoor environment are confirmed byresults from field tests reported in Fanimokun and Frolik [50] and Hara et al. [51].

A description of the fade margin referred to as the power required to compensate for the powerloss due to fading which is confirmed by different studies [48, 48, 52], is presented. Many com-munication algorithms in sensor networks are found which consider fad margin as conventionalapproach to be adapted to compensate for loss of power caused by channel fading [53, 54]. This isthe actual motivation for investigating Rayleigh fading and to explore methods and techniques formitigating its effect, for example collaborative communication which has been used in all of theproposed algorithms presented later in this thesis.

A discussion about multi-user systems in communication is presented in section 2.6, where acommunication path/channel is shared by multiple users. Different multiple access approacheslike “Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), Fre-quency/Time Division Multiple Access (F/TDMA), Code Division Multiple Access (CDMA) andSpace Division Multiple Access (SDMA)” have been presented and described.

2.2 Wireless Communication Systems

The term “communication” has been defined by the Encyclopedia Britannica dictionary [55] as“interchange of thoughts or opinions through shared symbols” and “system” as “an organizedintegrated whole made up of diverse but interrelated and interdependent parts”. Hence, a “com-munication system” can be described as transfer or exchange of data/symbols among various en-tities through components which are interrelated as well as independent. “Wireless” is defined as“medium without any physical connection”. Therefore, a wireless communication systems can bedescribed as a transfer or exchange of information among various entities through a non wired ornon physical medium for example air.



19

components are represented in simple white boxes. The description and functions of each

component shown in Figure 2.1 are explained below.

Format: Format is the necessary component of wireless digital communication system.

Data to be transferred is formatted into binary bits. Then a set of bits may be grouped.

The grouped bits are called message symbols.

Source encoder/decoder: Source encoder/decoder is optional component of wireless

digital communication system. This component is used for when the transmitted data is in

analog form. Analog data is converted to digital data using the encoder by the transmitter.

On the receiver the received digital data is converted to analog form by the decoder.

Channel encode/decode: Channel encode/decode is optional component of wireless

digital communication system. Transmitted message symbols are converted into channel

symbols known as channel code by the channel encoder. Channel coding is useful to

minimize the effect of noise and other channel impairments. On the receiver side channel

codes are converted to message symbols by channel decoder.

Pulse shaping: Pulse shaping is optional component of wireless digital communication

system. The pulse shaping filter adjusts the bandwidth of the transmitting message

symbols within the required channel bandwidth.

DECODER

Format Sourceencode

Channelencode

Pulseshaping

Bandpassmodulation

Frequencyspread

Multipleaccess

Transmitter Antenna

Receiver Antenna

Channel

Multipleaccess

Frequencydespread

De-modulation

DetectionChanneldecode

Sourcedecode

Format

Synchronization

Symbol sequence inthe digital form

Symbol sequence inthe signal waveform

Source/Transmitter

Destination/Receiver

Channel

Figure 2.1 Major components of wireless digital communication systemFigure 2.1: Components of wireless digital communication system

A typical digital wireless communication system as shown in Figure 2.1 (taken from [56]) mainlyconsists of three major components i.e. a transmitter for transmitting information, a receiver forreceiving the information and a channel through which the transmission actually takes place [57].In Figure 2.1 the mandatory parts are represented by boxes in Grey while the optional componentsare represented by boxes in white color. A description of each of these components is presented asfollow.

Format: The data to be transmitted must be converted into a specific format in order to be correctlyidentified by the receiver. Usually the bits are combined into groups called message symbols. Thisis a mandatory component of a wireless communication systems.

Source encoder/decoder: This is an optional component and is used only when analog data istransmitted. In this case the data first needs to be converted into digital form at the transmitterwhile at the receiver the original data is recovered using a decoder.

Channel encode/decode: Optional components responsible for converting message symbols intochannel symbols are also referred to as channel codes. This is a good way not only to avoid severenoise effects but also the effects of other impairments. The reverse happens at the receiver side.

Pulse shaping: An optional component responsible for filtering of message symbols of the trans-mitted signal to be adjusted in the available bandwidth.



Bandpass modulation/demodulation: A mandatory component responsible for modulating the in-formation signal over a high frequency carrier signal to make it suitable for broadcast through thewireless channel. This modulation causes the signal to travel over long distances which can bedemodulated at the receiver using the same carrier.

Frequency spread/de-spread: An optional component is responsible for converting low frequencysignal to high frequency to travel over long distances. It causes the signal bandwidth to becomewider and the signal becomes almost untouchable to noise and other induced impairments in thechannel. Such signals are referred to as wideband signals which are de-spread at the receiver toconvert them back to narrowband signals.

Multiple access: An optional component to enable multiple users share the same channel with-out effecting each other’s signal. Various techniques for multiple access and their features aredescribed in section 2.6

Transmitter antenna: A mandatory and last component of the transmitter side responsible forsending the modulated signal through the air.

Receiver antenna: A mandatory and first component on the receiver side responsible for capturingthe modulated signal sent by the transmitter antenna.

Detection: A mandatory component of the system responsible for deciding whether the receivedbit is zero or one while converting the modulated signal to binary data, based on some specific rule.

Synchronization: A mandatory component of the system responsible for synchronizing the sourceand destination in time and frequency to ensure efficient communication.

Wireless channel: A mandatory component of the system created through the air where the trans-mitted signal may experience different hazards due to reflection, diffraction and scattering.

As wireless sensor networks are also dependent on wireless channels for transmission of data,therefore, these networks also inherit most of the benefits and issues mentioned in this chapter.

2.3 Antennas and Antenna-Arrays

Efficient design of antennas which are used to emit electromagnetic waves that can carry infor-mation signal, is a challenging task in wireless systems. Any device made of metal can work asantenna using a suitable voltage [58]. One of the main problems is energy consumption caused



by internal resistance in antenna and the second issue is large reactive impedance which can bemitigated by producing large amount of radiation (large Q-factor also known as antenna’s band-width). Specially in single input single output systems to make the receive major portion of thetransmitted power, high directivity of the radiated power is required. Antenna array is one of thepossible solution to these problems [59].

Antenna array consists of a set of identical antennas where each one is independent and equidistantfrom the other. This set is organized in a way where a varied amplitude current is flowing througheach antenna. This arrangement of antennas can be used to produce highly directional beams withless power in comparison to single antenna system [58], which results in high gain in receivedpower at the receiver. Collaborative communication systems actually exploit the idea of antennaarray where the current distribution can be represented by discrete number of currents as antennasin an array are identical. Therefore linear equations can be used to represent the electromagneticfield equation [60] and the calculations can be performed either with a calculator or a computer.

A major application of antenna is radar systems. Other applications include but are not limitedto, radio-astronomy [61], “Direction of Arrival (DoA)” estimation, localization using acoustic sig-nals [62] and Multi-Input-Multi-Output (MIMO) systems. Antenna arrays also known as “smartantennas” or “adaptive antenna arrays” [63].

2.3.1 Multiple input multiple output (MIMO) Systems

“MIMO” system is an important application of the antenna array where antennas are organized inarrays both on transmitter and receiver side. Therefore, multiple streams data between transmitterand receiver can be transferred at the same time over the same frequency band. Hence largeamount of data can be transmitted over the same bandwidth in comparison to “SISO” systems [61].In addition, the idea of space diversity originated from this phenomenon which is very popular formitigating the effect of fading in wireless communication systems. Using space-time orthogonalcodes, diversity gain of MIMO can be analyzed.

Rayleigh fading channel model is used to analyze performance of MIMO systems and the effectof fading on these systems, over each antenna element. Motivation for using this channel modelis its simplicity which allowed the authors to focus more towards designing algorithm for bothtransmitter and receiver sides [64–68]. Results from fundamental information theoretic on MIMO

channel’s capacity [69] and antenna diversity approaches [70] used the same channel model.



A combination of multiple transmitters with a single antenna based on space-time coding approachcan also result in space diversity. This produces a phenomenon referred to as virtual MIMO wheremultiple nodes can transmit the same data cooperatively [68, 71]. Multi-user information theoryis used to analyze the system in [68, 71] using a baseband based model. However, analysis ofthe recent literature reveals that unsynchronized phase and frequency in virtual MIMO systems,significantly reduces performance [72]. To reduce the synchronization error and improve perfor-mance of sensor networks different algorithms and methods are developed and analyzed in the laterchapters.

Collaborative communication combines the transmission power of multiple transmitter to send thesame information towards a common single receiver (base station), making it a Multiple-Input-Single-Output (MISO) system.

2.4 Signal Power Loss in Wireless Channels

When a signal propagates through a medium whether guided or unguided, it loses power due todifferent factors like impairments, noise, cross talk, fading, path loss, etc. Various path loss modelsand their features in wireless channels are discussed in this section. If the signal “S” is transmittedover a distance “d” then the power received at the receiver “Pr”, represented as follow:

Pr(d) =Ptg(d)

(2.1)

Where Pr(d) represents power received at the receiver which is apart from the transmitter bya distance d, Pt represents the power transmitted and g(d) represents the “loss of power” as afunction of transmission distance d.

“Signal-to-Noise Ratio (SNR)” and Signal to Interference and Noise Ratio (SINR) are commonlyused to evaluate/characterize a particular communication system. The values of these ratios shouldbe bound by certain threshold which may vary from one wireless channel to another [73]. Thisthreshold actually is dependent on acceptable probability of error for receiver. Usually the ratio ofreceived signal power Pr and noise power is termed as SNR.

The loss of power g(d) is usually influenced by the channel, for example one such effect is fadingthat degrades the performance of a system [48]. Various models and their effects on g(d) are



explored in [47, 48, 57, 74], whereas this section presents an analysis of the power loss model forsignal which is very crucial to developing energy efficiency model for collaborative communicationpresented later in this thesis.

2.4.1 Free space signal path loss model

It is the most basic and ideal path loss wireless communication model which is adopted widelydue to its simple form. This model is suitable in situations where there is no obstacle betweentransmitter and receiver i.e. clear Line-of-Sight (LoS). Mathematical form of received power isexpressed as [47, 48, 74] follows:

pr(d) =PtGtGrλ

2w

(4π)2d2L(2.2)

Gr and Gt represents gain of the receiver and transmitter antennas respectively, λw representswavelength which is also given by c/fc, where c represents speed of light, fc represents carrierfrequency, L represents system loss factor and distance by which transmitter and receiver are apartis represented by d.

2.4.2 Two-ray ground signal path loss model

An extension to free space path loss model that includes the effect of ground-reflection and lineof sight (LOS) is referred to as “Two-ray” ground signal path loss model. Its accuracy is high andused in calculating the signal received power [47, 48, 74] and it’s mathematical form is as follows:

pr(d) =PtGtGrh

2th

2r

d4L(2.3)

Here ht represents transmitter’s while hr is the receiver’s antenna heights

2.4.3 Log-normal shadowing signal power loss model

Where there is shadow fading in the channel, then the path loss model for signal power usedis log-normal shadowing. Causes of shadow fading may be reflection,diffraction, scattering and



absorption [47]. From the experimental and analytical basis of this model it can be analyzed thatreceived signal power behaves as a logarithmic function of the distance d between a transmitter andreceiver and it follows log-normal distribution. Power loss of signal using this model is calculatedas [47, 48, 74] follows:

pr(d) = Pt + 10 log10

(λw4πd0

)2

− 10k log10

(d

d0

)2

+Xσ (2.4)

Here d0 represents far-field region distance of an antenna with loss exponent k, a zero-mean Gaus-sian random variable Xσ in dB, representing shadow fading with standard deviation of σ, in dB.k is dependent on wireless environment; in case of indoor environment it may be equal to a valueof 5 or larger while in case of outdoor environment the value may be as low as 2 or even more. Onthe other hand values of σ range from 6dB to 10dB or even more [48]

2.4.4 Simplified signal path loss model

The most widely used and a hybrid of free space and two-ray ground path loss model, excluding thefading effect, to calculate signal power is referred to as “simplified” path loss model. This modelhas resulted from the free space and two-ray ground path loss models and the approximation oftheir analytical results. Its accuracy in calculating signal power loss is claimed to be very high andis represented by the following mathematical form [47]:

pr(d) = Pt

(λw4πd0

)2(d

d0

)κ

(2.5)

Here the environment of propagation actually effects the value of κ and it may lie in range from2 to 8. Equation 2.5 is a generalized form of path loss model represented by equation 2.2 and2.3 with no fading effect. This allows for analyzing the energy consumption and saving due tomitigation of fading which leads to collaborative communication analyzed in different scenarios inlater chapters.



2.5 Multipath and Small-scale Fading

The term fading is defined as “the time-variation of the channel strengths due to the small-scale ef-fect of multipath scattering, as well as larger scale effects such as path loss via distance attenuationand shadowing by obstacles”[48]. In other words instantaneous variations in phases, amplitudesof a transmission signal due multipath effect for a short time period or over small distances. Thismay be caused by different obstacles present in any real environment like building, trees, hills etcpresent in the path of the transmission signal [75]. The signal when reflects from these obstaclesmay divide into multiple components resulting in multiple paths between the sender and receiverwhich may cause multipath interference referred to as multipath fading. In case of mobile objectsthis effect may even be more adverse and can degrade system performance even more. Effects ofsuch multipath fading are as follows:

• Loss of signal power abruptly in short time period and short transmission distance.

• Time dispersion and echoes result from delay in multipath propagation.

• Random frequency modulation resulted from Doppler shift produced due to multipath.

2.5.1 Physical Factors Influencing Fading

Various factors that effect fading during propagation through a radio channel are given here toimprove further understanding of the techniques discussed in the later chapters.

2.5.1.1 Multipath Effect

Obstacles in an environment may cause a signal to scatter into multiple components thereby cre-ating multiple paths between sender and receiver. It results either in constructive or destructiveinterference or causes phase variations in the received signals.

2.5.1.2 Mobility of Transmitter and Receiver

A sensor node is called as mobile node if it has the ability to move from one location to anotherin a sensor network and the phenomenon is known as mobility of a sensor node. Doppler effect isresulted due to the mobility of transmitter/receiver during signal propagation over multiple paths.



A change in frequency of a signal due to the movement of source and sink towards or away fromeach other resulting in a sudden variation in pitch is known as Doppler effect/shift. Doppler shiftalso causes random frequency modulation.

2.5.1.3 Mobility of Surrounding Objects

A time varying Doppler effect results in multipath components due to the motion/mobility of ob-jects in the transmission path. In case where the mobility of sender and receiver is lower than themobility of surrounding objects, the resultant fading becomes a dominant factor.

2.5.1.4 Transmission Bandwidth of the Signal

A significant power loss is experienced by the signal when bandwidth of the transmitted signal islarger than that of the multipath channel which is also known as the “coherence” bandwidth.

2.5.2 Types of Small Scale Fading

Signal, channel parameters and their relationship actually defines the fading. In the case of chan-nel, these parameters include delay spread and Doppler shift whereas in case of signal, these arebandwidth and time period. Various fading types are explored in this section.

2.5.2.1 Flat Fading

If coherence bandwidth of a channel exceeds the transmitted signal’s bandwidth or put it anotherway, if Root Mean Square (RMS) value of delay spread is smaller than the signal’s time period, thenit results in flat fading [48, 75]. Movement of transmitter and receiver effects Rayleigh fading andso does the coherence bandwidth. If the mobility of transmitter and receiver is low, then Rayleighfading becomes flat fading. In WSN mobility of transmitter and receiver is not usually high causingthe Rayleigh fading to be flat. Therefore collaborative communication systems discussed in laterchapters are developed considering Rayleigh fading channel. In order to moderate the effect ofRayleigh fading, many techniques based on advanced indication dispensation have been presented.The most conservative engineering method to moderate Rayleigh fading is to add sufficient fadeedges as presented in section 2.5.3.



2.5.2.2 Frequency Selective Fading

If a channel’s “coherence bandwidth” is smaller than the transmitted signal’s bandwidth or “RootMean Square (RMS)” value of delay spread is greater than the time period of the signal, then itresults in a fading called frequency selective fading [47, 48].

2.5.2.3 Fast Fading

Doppler effect is caused by the mobility of the receiver, transmitter or any surrounding object,which results in random frequency modulation which in turn produce fast fading. This fadingcauses instantaneous variation within transmitted signal’s time period [47, 48].

2.5.2.4 Slow Fading

If impulse response of a channel does not change within the transmitted signal’s time period, but astate change takes place after the time period, then the fading is termed as slow fading[47, 48].

2.5.3 Fade Margin

Usually channel fading causes loss of power and the required amount of power to compensate forthis loss is known as fade margin. A substantial fade margin is needed in case of various wirelessapplications to compensate for the loss of power caused by fading of the channel. Figure 2.2presents fade margin in the case of cellular system, where path loss is measured using equations2.2, 2.3, 2.4 or 2.5 with fade boundary of 20 to 30dB [47].

Figure 2.2 reveals that shadow fading has lower fade margins than small-scale fading. The largefade margin signifies the reception of significant transmit power to achieve the desired SNR atthe receiver. In WSNs not only efficient communication system is a primary requirement but alsosaving energy is the primary design goal [53, 54]. Therefore, techniques and methods to overcomeand mitigate the effect of fading to improve energy savings, are need of the day.

Various techniques proposed in literature to mitigate fading are; rake receiver, channel equalization,assortment of receiver and transmitter in time, space and frequency [47]. The implementationof these techniques require extra hardware as well as high signal processing where the later is



28

When channel does not changes impulse response within the time period of

transmitted signal, but changes its state after the time period, this type of fading is called

slow fading [2].

2.6.3 Fade Margin

Fade margin is defined as the power required to compensate the power loss due to the

channel fading. It is found in [2, 29] that for many wireless applications a substantial

amount of fade margin is required to compensate the power loss due to the channel

fading. Fade margins for the cellular system are shown in Figure 2.2. In Figure 2.2, path

loss may be calculated using equations (2.2), (2.3), (2.4) or (2.5), depending on the signal

power loss model. The typical value of the fade margin for shadow fading is often set to

be 6-10 dB [2] and the typical value of the fade boundary for Rayleigh fading is often set

to be 20 dB to 30 dB [2].

Figure 2.2 Fade margins in the cellular application [2]

From Figure 2.2 it is observed that small-scale fading having higher fade margin than

shadow fading. The large amount of necessary fade margin suggests that significant

amount of transmit power is received at receiver to achieve required SNR. Energy

efficient communication is the primary requirement in wireless sensor networks and

energy saving is the one of the primary design requirement in wireless sensor networks

[6, 7]. Therefore, there is need of appropriate signal processing technique or other method

to effectively mitigate the channel fading.

Transmitter Distance

Powertransmitted

Mean pathloss

Large-scalefade margin

Small-scalefade margin

6-10dB

20-30dB

Receiver

Received power thresholdPower

received

Figure 2.2: Fade margins in case of cellular application

assumed to require less energy, however, results from [76–78] show that in case of complex signalprocessing the energy expense is significantly high.

The proposed collaborative communication based system overcomes the effect of fading by ex-ploiting space or antenna diversity. It has been seen that collaborative communication based pro-posed approach is very successful in mitigating the fading effect at the expense of power requiredfor circuit operations. Hence, a trade off analysis is included in each chapter to compare the re-quired power and gain achieved in the received power using collaborative communication.

2.6 Multiple Access Methods for Multi-user CommunicationSystems

High capacity communication is a basic requirement of multi-user systems. Due to shared natureof wireless channel many users want to have access to the channel concurrently. In such systemsthe need for methods to share the available bandwidth with other users and keeping interference toa minimum is more than ever. Therefore, multi-user systems are more challenging in comparisonto the “transmitter-to-receiver” systems discussed in the previous section. In multi-user systemthe received signal is actually a collection of signals from the desired source as well as unwanted



signals from other transmitters. Unwanted or interference signals are considered as a noise to thedesired signal which may significantly affect the “SINR” hence, increasing the probability of error.

Various methods for access the publicly available wireless channel are proposed and are in wideuse in wireless communication. A few well known multiple access methods are:

• TDMA

• FDMA

• F/TDMA

• CDMA

• SDMA

2.6.1 Time Division Multiple Access (TDMA)

In this type of multiple access the available band is shared among multiple users. Each user receivesthe whole available band for a specific period of time referred to as time-slot as shown in Figure2.3b. In TDMA the transmission of data is in burst form rather than being continuous for any user.The transmission of information also involves synchronizing the sender and the receiver.

2.6.2 Frequency Division Multiple Access (FDMA)

The bandwidth is shared among users in frequency i.e. through channelization of the availableband into narrow band channels. Each user has its own channel and all users can transmit at thesame time. FDMA is shown in Figure 2.3a.

2.6.3 Frequency Time Division Multiple Access (F/TDMA)

It is clear from discussion on TDMA and FDMA that in each case user gets a dedicated portionof the band which in TDMA is a time slot whereas in FDMA is a narrowband channel in the totalband. At times some users in both these cases may not want to transmit data and since theseallocations are dedicated so their portion in the whole band will remain unused. Therefore, a



hybrid of TDMA and FDMA is proposed to improve capacity utilization of the system and is usedin various applications of wireless communication systems like GSM etc.

31

(a) FDMA (b) TDMA

(c) CDMA

Figure 2.3 FDMA, TDMA, and CDMA in the time domain and the frequency domain.

2.7.4 Code Division Multiple Access (CDMA)

In CDMA multiple users access the entire bandwidth at the same time. Users are

distinguished from each other using a pseudo sequence known as spreading codes. These

spread codes are orthogonal to each other. Therefore, multiple users can transmit and

receive the data using entire bandwidth at the same time at the expense of multi-user

interference induced in the receiver as shown in Figure 2.3 (c). CDMA systems can be

divided into two categories i.e., frequency hopping CDMA (FH-CDMA) and direct

sequence CDMA (DS-CDMA). In frequency hopping CDMA rapid frequency changes

are achieved using the spread codes. While in direct sequence CDMA transmitted data

bits are multiplied with a spreading code i.e., pseudo noise sequence also known as PN

Frequency

Time

Code

user 1

user 2

user 3

user N

Frequency

Time

Code

Frequency

Time

Code

(a) FDMA

31

(a) FDMA (b) TDMA

(c) CDMA












Frequency

Time

Code

user 1

user 2

user 3

user N

Frequency

Time

Code

Frequency

Time

Code

(b) TDMA

31

(a) FDMA (b) TDMA

(c) CDMA












Frequency

Time

Code

user 1

user 2

user 3

user N

Frequency

Time

Code

Frequency

Time

Code

(c) CDMA

Figure 2.3: Time and Frequency domain representation of FDMA, TDMA, and CDMA


Each user in this method, has its own pseudo random code (spreading code) which enables auser to share the same band at the same time with other users. Since the spreading codes areorthogonal, therefore it avoids signals from different transmitter to interfere with each other asshown in Figure 2.3c. It can further be classified as “frequency hopping CDMA (FH-CDMA),direct sequence CDMA (DS-CDMA)”. In Frequency Hopping Spread Spectrum (FHHS) thereare sudden changes from frequency to frequency based on the PN sequence whereas in DirectSequence Spread Spectrum (DSSS) the information signal is multiplied by the code. Since thecodes are orthogonal, at the receiving the receiver correlates the received signal with own PNsequence to extract (de-spread) signal meant for him.



2.6.5 Space Division Multiple Access (SDMA)

SDMA implemented through sectorised antenna is based on the idea of sectors [34], which isbroadly used in cellular networks. The division of wireless channel is carried out on the basis ofradio waves direction of motion i.e direction of arrival at the receiver and direction of departure atthe transmitter.

2.7 Wideband Channels

Wideband channels belong to the family of spread-spectrum techniques [79, 80]. Spread spec-trum deliberately spread a signal (e.g. an electrical, electromagnetic, or acoustic signal) generatedwith a particular bandwidth in the frequency domain resulting in wider bandwidth. Motivationsfor using wideband channels are; 1) universal frequency reuse, 2) jamming suppression, 3) reduc-tion in natural interference and 4) reduction in noise due to multi-path diversity. Collaborativecommunication in sensor networks using wideband channels has been investigated in chapter 7

A channel is termed as wideband channel, if coherence bandwidth of a channel is less than themessage bandwidth. The “coherence bandwidth” is termed as the statistical measure of the fre-quency range over which frequencies of a signal face correlated amplitude fading. If coherencebandwidth is Bc in Hz then it is given by the formula:

Bc ≈1

D(2.6)

where D is the time delay in seconds.

Examples: Different examples of wideband channels are; International Mobile Telephony 2000(IMT-2000), CDMA2000, WCDMA(FDD, TDD) etc.

2.7.1 Rayleigh Fading

Fading is the effect of environment on the signal transmitted through it, whereas Rayleigh fadingis a statical model to measure that effect [48, 79]. Rayleigh model is preferred in environmentwith many objects which deflect the transmitted signal dividing it into many “scatter components”before it arrives at the receiver. Rayleigh fading is a model that can be used to describe the form of



fading that occurs when multi-path propagation exists. In any terrestrial environment a radio signalwill travel via a number of different paths from the transmitter to the receiver. The most obviouspath is the direct or line of sight path.

2.7.2 Channel Noise

Every practical channel if able to transmit data may introduce some noise into the data. A channelwithout noise may be an ideal channel. In any communication system a transmitted signal is oftendistorted by noise. The most popular type is “Additive White Gaussian Noise (AWGN)”. It is calledadditive because it is added to a signal during propagation. AWGN actually mimics the behaviorof a random process and is very efficient in comparison to real noise from the point of view ofsimulating amplifier or background noise.

For example if we represent a signal in a digital system as

Y = X +N (2.7)

where Y represents the received signal composed of the actual transmitted signal X and the addednoise N is zero mean AWGN .

2.8 Chapter Summary

Some of the basic concepts of wireless communication system, specially for physical layer arepresented in this chapter. Physical layer plays an important role in the design and exploration ofdifferent communication systems for WSNs. Various model for calculating signal power loss areelaborated and their mathematical expressions are analyzed. In the later chapters these models areused to analyze energy consumption model based on collaborative communication for WSNs. Ithas been seen that to compensate for signal power loss due to channel fading, large fade margin isrequired, which motivates the investigation of techniques to mitigate channel fading. Another stepin achieving energy efficient communication is to reduce the number of collisions, therefore thereis a need for methods to allow sharing of a band among multiple users.


Chapter 3

Literature Review

Recently a considerable amount of efforts and time has been invested by the research commu-nity in investigating diversity gains in cooperative communication [71] and beamforming [81, 82].Looking far behind in the past, reveals that multi-way channels had extensively been investigatedin [83] elaborating channels introduced till 1976 with an extensive analysis of relay channels. Theauthors in Pottie and Kaiser [31] argue that energy consumed by transmission is far more than thatof local execution. In an example, the authors considered the distance between source and sinknode to be 100m, transmission of one bit consumes energy equivalent to the energy required forexecuting 3000 instructions. Authors in [84, 85] compares the consumption of energy for commu-nicating one bit with its local execution stating the range to be 1, 000 − 10, 000. This shows theexpensiveness of transmission over the networks in terms of energy consumption. Therefore totalenergy consumption is mainly dominated by the energy consumed during transmission, and thisdominance may grow exponentially, with increase in transmission distances between source anddestination.

To make communication energy efficient in WSNs, recently various approaches have been pro-posed in the literature. Some popular among them include but not limited to; multihop commu-nication/routing [86], cooperative communication [87], beamforming [88, 89] and collaborativecommunication [56] . Rest of this chapter is dedicated to elaborate each of these approaches, theiradvantages and limitations.

30

Chapter 3. Literature Review

3.1 Multihop Communication

In energy constrained networks like WSNs communication over short distance is encouraged toavoid fast depletion of the power source. In WSNs the deployment of sensors is usually dense,which makes it suitable for a node to communicate with its neighbors using low transmissionpower. For such scenarios multihop communication is a viable option [90].

In multihop communication signal received from a node is forwarded to the next node and thenext to a next node until it reaches the destination. The use of intermediate hops between sourceand destination actually slows the battery depletion down by reducing the transmitted power [91].This is an excellent approach for saving a node’s energy by preventing it from sending high powersignals over long distances. Each node only has to forward the received signal to its next neigh-bor(next hop) in the direction of base station using a low power signal thereby, saving its powersource from being depleted fast.

Reliable and efcient multihop communication not only saves the power source of a sensor, but alsoenables it to transmit over long distances even to destinations outside its transmission range. Mul-tihop communication works in two phases: suitable path discovery from source to the destinationnode, followed by a data exchange over the discovered path established in the first phase [92].

Multihop network architecture may be flat or hierarchical [90]. In flat architecture, sensor nodeswork in peers, where each sensor uses its neighbor as a relay to forward data to the destinationnode. Queries from base station to sensors are usually flooded, because of the large number ofsensors, it is not feasible to maintain global identifiers for each sensor. Only the sensor having datamatching the query will respond to the query. In hierarchical architecture nodes are grouped intoclusters where a cluster head is responsible for collecting data from sensor nodes and forwarding itto the base station. This allows low energy nodes to carry out the sensing task whereas node withhigh energy can act as the cluster head. This reduces energy consumption as well the overall loadof the network.

However, in multihop communication, failure of a node in the multihop path breaks the communi-cation and retransmission of the data may be required. So the failure of single node not only causesthe sender node to transmit twice but also all the nodes in multihop path will have to re-forwardthe data, resulting in extra energy consumption. Secondly, multihop routing for increasing trans-mission range will have to include more nodes in the path. This increases the overall overhead andcomplexity of implementing routing process in these networks [86]. Sensor nodes in many cases



turned their radio off, when it is not required, to conserve its energy. However, multihop communi-cation requires a node to keep its radio turn on, so that it can be available for forwarding data [93].Moreover, clustering itself has major issues of cluster head selection as well cluster organization.

Some of these issues are addressed through the use of cooperative communication in WSNs [87]elaborated in the next section.

3.2 Cooperative Communication

The concept of cooperative communication was first introduced with the title “three-terminal com-munication channel” by van Der Meulen in 1971 [94] as shown in Figure 3.1. Later on the conceptof cooperation is further investigated in Cover and Gamal [95]. In cooperative communicationa signal transmitted by a source node S towards a destination D is improved by an intermediatenode known as the relay node R. The relay node R is a node which can hear transmission of thesource and it can transmit directly to the destination node D. A communication using intermediaterelay nodes to achieve cooperation is also known as relay protocol/communication. However, theidea of cooperative communication in WSNs is some how different from the basic relay channel.Firstly, the relay channel is investigated only in Additive White Gaussian Noise AWGN channelwithout fading. Secondly, the sole purpose of the relay channel in [94] is to support the main chan-nel whereas in WSNs the cooperative node not only behaves as a relay but also as an independentnode.

R

S D

Figure 3.1: The cooperative relay network, also known as the relay channel.

To achieve higher gains cooperative communication uses the idea of diversity which generally re-quires more than one transmitters at the sending side. Then signals from these multiple transmitters



are added at the receiver to achieve diversity gains. Diversity means transmitting different versionsof the same signal from different locations creates statistically independent channels effected bythe channel fading differently which may actually help in combating this effect [96]. However,mounting multiple antennas on a single node in WSNs may not be feasible for economic as well aspractical reasons [97]. Therefore, individual nodes in a WSN can be combined at the sending sideto form a virtual multi antenna transmission possible.

The process of cooperation is decomposed into two phases where in the first phase the source trans-mits the data which is received by both destination as well as the relay node. In the second phasethe relay can help forward or retransmit the information in the direction of the destination. Fromthis point of view, cooperative communication can be divided into different categories depend-ing upon processing performed at the relay node on the received signal from source [98]. Thesecategories are as follow:

3.2.1 Amplify-&-Forward

In the amplify-&-forward communication a signal received by the relay node is amplified and thenforwarded towards the destination. This amplification is intended to combat or equalize the effectof channel fading experienced by the signal from source to relay node. This is achieved by scalingthe received power by a factor inversely proportional to the received power. Both signals; the onereceived directly from the source and the one from relay are added at the destination using theconcept of Maximum Ratio Combiner (MCR) to improve gain in the received signal-to-noise ratio(SNR).

3.2.2 Decode-&-Forward

This type of cooperation is also known as “Detect-&-Forward”, where the signal received fromthe source is decoded and then re-encoded for hard decision. In such communication the relaynode does not add any information regarding the reliability of the link between source and relay.If the signal is received incorrectly, forwarding such signal to the destination will be meaningless.Therefore, performance of such system is limited by the worst link i.e. source-relay or source-destination link. Although, this approach is more efficient in mitigating the noise effect, but itmay propagate error to decoding incorrect signal, thereby degrading system performance. It is



reported in Liu [98] that the decode-&-forward and amplify-&-forward terminologies are actuallyintroduced in [72, 99].

3.2.3 Compress-&-Forward

In this approach the signal received at the relay node is transformed to obtain an estimate known assoft information which is then forwarded to the destination. The destination combines the messagedirectly received from the source with its compressed/quantized version received from the relaynode.

3.2.4 Cooperative Communication in Wireless Sensor Network

Cooperative communication uses the concept of diversity (spatial or antenna) which require morethan one transmitter antenna to achieve gains in power, bit error rate and capacity. As mentionedearlier, sensor networks usually consist of hundreds or even thousands of miniature devices havingvery small size, processing and battery power. Therefore, mounting more than one antenna on asingle sensor node is problematic both for economic and practical reasons [98].

To resolve this issue antennas of multiple individual nodes can be used to create a virtual coop-erative communication mechanism and still achieve the desired benefits. To achieve this goal theconcept of cooperative MIMO (C-MIMO) was introduced in WSNs in 2003 by Cui et al. [100]for the first time. The idea is further improved by Jayaweera [101] including channel estimationcapabilities with data gathering which is further enhanced by Gai et al. [102] and Islam et al. [103].The C-MIMO approach is based on multiple inputs and outputs which creates a cooperation in anetwork of sensor nodes with single antenna. This cooperation leads to the formation of a MIMOstructure which in turn leads to improved energy efficiency and reduced end-to-end delay.

Energy consumption model for MIMO proposed in[100, 101], presented a comparison betweenSISO and MIMO systems. It has been concluded that SISO performs well when the distance be-tween a transmitter node and the BS is small, however, in case the distance between the transmitternode and the BSis large, it is difficult for SISO to keep up with MIMO systems. A similar compari-son in terms of energy efficiency between virtual MISO and decode-and-forward based cooperativecommunication is presented in [104]. The authors concluded that decode-and-forward approachshows strong energy efficient behavior compared to virtual MISO. The maximum energy saving



Table 3.1: Maximum Energy Savings of Multi-Node Cooperative Systems.

Nt 2 3 4 5 6

Maximum Energy Saving 89.54% 94.65% 95.99% 96.53% 96.84%

reported in Simic et al. [104] for different values of Nt (number of transmitters) is shown in Table3.1

The feasibility of applying cooperative MIMO system to WSN has been investigated in[105]. Theauthors argued that the cooperative MIMO systems are promising in terms of low error perfor-mance, low power performance, tolerance to jitter and it outperforms other cooperative schemesin cases of imperfect synchronization. But it does not provide the best or optimal scheme becausethey relates diversity gain only with link reliability without considering circuit power in the overallenergy consumption of the system.

The authors in Laneman et al. [106] argued that maximum diversity gains in “amplify-and-forward”and “decode-and-forward” relays has a direct relation with the number of “degrees of freedom” inthe channel i.e. an increase in one leads to an increase in the other. This scheme has been extendedby Laneman and Wornell [72], to develop a more generalized expression for the case of space-timecoded system and prove that diversity gain can be achieved. In Zheng and Tse [107], a tradeoffbetween diversity and multiplexing for traditional MIMO in case of centralized array antenna hasbeen presented, where an optimal trade-off curve is given to compare the gain that can be achievedby increasing the number of degrees of freedom.

To calculate capacity of a cooperative communication system, one of the pioneer work can befound in Cover and Gamal [108]. In this proposal various channel are considered including feed-back relay channel, reversely and Gaussian degraded channel. Contribution of this work is thedevelopment of theorems for calculating system capacity for these channels. To calculate the up-per and lower capacity bounds for forward relay and cooperative diversity has been explored in[87, 109]. Here the authors have used a non-identical Rayleigh channels to develop a closed formsolution for system capacity bounds. It is argued in Kramer et al. [87] that cooperative MIMO isunable to produce multiplexing gain, however, it can be used to produce high SNR additive gain.



3.2.5 Limitations of Cooperative Communication

Apart from other issues discussed, synchronization of the received signal at the receiver is one ofthe main challenges for cooperative communication. To address synchronization related issuesvarious other techniques like beamforming and collaborative communication (discussed in thefollowing sections) have been proposed.

Furthermore, it can be concluded that the concept of spatial diversity used in cooperative commu-nication to resist against fading and achieve high throughput, lower delay and reduce interferencethrough wireless broadcast advantage(WBA). It must also be noted that cooperative communica-tion produces better results only if the convergence communication is synchronized. However, incase of imperfect synchronization cooperative communication fails to achieve the said benefits. Inaddition cooperative communication also suffers from several other disadvantages, for example,complex scheduling, increased overhead, increased interference, extra traffic etc with increase inthe number of relays.

3.3 Beamforming

Beamforming is a signal processing approach used in wireless communication to direct signal ina particular direction in order to achieve spatial selectivity. In beamforming the the transmittedenergy from multiple transmitters is concentrated in a specific direction as shown in Figure 3.2taken from [89].

Transmitted power of multiple antennas is combined in a certain way to produce constructive in-terference using specific angles in a specific direction. At the same time, some of the signals com-bining in different angles in other directions may also produce destructive interference, therebyreducing power of the transmitted signal in that direction. The improvement in signal recep-tion/transmission, in comparison to omnidirectional communication is known as receive/transmitbeamforming gain. Beamforming, some times, also termed as spatial filtering, was initially in-troduced for cellular networks and wireless LANs in order to focus the transmitted energy in aspecific direction [102].

From the look of it, beamforming clearly incorporates various overheads including the overhead interms of power and time for sharing data among collaborating sensors, precise location estimationand resolving synchronization issues. Various studies show that despite the overhead incorporated,


Chapter 3. Literature ReviewBÉJAR HARO et al.: ENERGY EFFICIENT COLLABORATIVE BEAMFORMING 497

replacement. There has been little attention to this issue in thecontext of beamforming applications. In [7] they consider en-ergy-efficiency when collaborative beamforming is used. How-ever, the work in [7] is oriented towards routing optimizationinstead of energy efficient beamforming. A routing schedulingpolicy has been recently proposed in [8] that helps to extendthe network’s lifetime. Again, in [7], [8], collaborative beam-forming is considered with no weight optimization; therefore,the network may be using more energy than strictly required. Itbecomes clear that the development of distributed optimizationtechniques that take into account energy efficiency are of para-mount importance in the context of sensor networks. Severaldistributed beamforming approaches have been proposed for thenetwork relay-beamforming scenario [9]–[14]. However, theseare based on centralized optimization and reduced feedback be-tween the relay network and the transmitter/receiver pair. In ourcontext of WSNs, such approaches may have limited applica-bility since they require constant feedback between sensors andbase station (e.g., battery level and channel state information).It is therefore preferable to devise distributed algorithms that donot require any communication with the base station. Anotherdifferentiating aspect to the relay-beamforming scenario is that,in the context of sensor networks, nodes are battery-powereddevices with finite energy resources.In this paper, we consider the distributed beamforming

problem with QoS constraints where the metric to be optimizedis the network’s lifetime (e.g., the time that the network canguarantee the specified QoS requirement). This work repre-sents a major extension of [15] where only an idealized sensingscenario was considered. We derive closed-form expressionsfor the optimal beamformer and provide iterative algorithmsfor its numerical computation. Using only local informationabout battery status and channel conditions, we use consensusalgorithms [16] to propose fully distributed solutions to theproblem that only require local communication among nodes.In the last part of the paper we consider the case where the

energy consumption at the nodes is not deterministic. It has beenshown that the energy consumption in a wireless sensor networkcan be modeled as random quantity [17] that depends on severalparameters related to data processing and sensing characteristics,node to node communication, transmission rate, duty cycle,MAC layer protocol, etc. In order to account for other sources ofbatterydepletiondifferent fromfar-away transmission to thebasestation we consider an additional random energy consumptionterm in our formulation. The problem then switches to a proba-bilistic design that generalizes the original problem. Conditionsunder which the general problem is convex are also provided. Insome specific scenarios, the more general problem is amenablefor its solution in a distributed fashion. However, this is not thecase ingeneral and furtherworkneeds tobedone in thatdirection.The paper is organized as follows: We first provide some

general definitions and describe the problem in Section II. InSection III we formulate the energy-efficient beamformingproblem and derive simple expressions for its computation.Section IV provides both centralized and distributed algorithmsfor the computation of the optimal beamvector. Section Vextends the results for the case of random energy consumption.Numerical simulations are provided in Section VI and conclu-sions are drawn in Section VII.

Fig. 1. Beamforming scenario between the nodes and the far-away base station.

Notation: Vector-valued quantities are denoted using boldlower-case letters. The optimal value of a variable in an opti-mization problem is denoted by . The symbols , , anddenote the set of complex, real, and non-negative real numbers,respectively. For a scalar , its complex-conjugate is denotedby . For vectors, denotes transposition whiledenotes complex-conjugate transposition.

II. SYSTEM MODEL

Consider a WSN composed of nodes scattered over a cer-tain area. Nodes are battery-powered elements equipped with asingle-antenna whose purpose is to sense and retrieve informa-tion from the environment. The information sensed is to be sentto a far-away base station where the data is further processedand analyzed. In order to reach the base station, nodes need tocooperatively form a virtual beamformer for transmitting the ac-quired data to the base station, see Fig. 1. At the base station, aminimumQoS requirement must be fulfilled that allows reliabledecoding of the received signal. We aim to maximize the timethat the network remains operative without human intervention(i.e., battery replacement). The problem is then to design a vir-tual beamformer that meets the required QoS at the base stationwhile maximizing the network’s lifetime. It is further assumedthat all elements (nodes and base station) lie on the same plane.Consider a scenario where all nodes of the network have ac-

cess to a (possibly) noisy version of the signal to be transmitted(e.g., they may measure independently the same quantity orthey might have obtained it through a joint estimation process).We assume that the network has two distinct modes of opera-tion, namely sensing (data harvesting) and data reporting. Onceenough measurements have been acquired, it is necessary to re-port those to the base station through a noisy wireless communi-cation channel. Let denote the discrete-time signal avail-able at the th node that needs to be transmitted at time

(1)

where is the true signal to be transmitted and corre-sponds to the observation noise at node , which is assumedto be independent and identically distributed (i.i.d.) with zero-mean and variance .

Figure 3.2: Beamforming scenario between the nodes and the far-away base station

beamforming can still achieve significant gain in energy savings. However, design of transmitter,data rate, size of the network and design of receiver circuit really effect the amount of energysavings. Therefore, for specific values of these parameters, beamforming is reported to have 90%

savings in energy as compared to single sensor transmission [110].

3.3.1 Beamforming in Wireless Sensor Networks

As mentioned earlier, mounting more than one antenna on a resource constrained sensor nodeis not appropriate for economic as well as practical reasons. Therefore, beamforming also usesthe concept of antenna diversity to combine the transmitted power of multiple nodes, and focusit toward a particular direction to achieve transmit and receive beamforming gain. From variousstudies found in the literature, it is clear that beamforming not only reduces the transmission powerper node, but also mitigates the loss of power due to propagation. Beamforming in sensor networkswas introduced in 2004 [81].

Beamforming actually works in two phases as shown in Figure 3.3. The phases are: i) Informationsource first shares the information it want to transmit, with the collaborating nodes (transmitternode, slave nodes, collaborative nodes). ii) Once all the collaborating nodes are synchronizedwith the information source, then all send the pre-shared data towards the receiver. Therefore, to



calculate total energy consumption, energy consumed in both phases must be considered. Numer-ous techniques for evaluating energy efficiency of beamforming can be found in literature. Somepopular of them are discussed here.

Barriac et al. [81] proposed a master slave architecture for collaborative beamforming in the con-text of sensor networks for sharing information between the source and collaborating nodes. Com-municating timing and carrier signal to slave nodes is the responsibility of the master node in thisarchitecture. They have achieved time and frequency synchronization between the master and slavenodes to estimate channel coefficient using one receiver. A top level view of the model used forcollaborative beamforming in Barriac et al. [81] is shown in Figure-3.3.

Figure 6: SIMULINK model of beamforming sensors

Information source

Sensor 1

e-jθ1 e-jθ2e-jθ3

Receiver

Sensor 2 Sensor 3

Local sensor communication

Figure 8: Top level view of sensor field

7. REFERENCES[1] G. Barriac and U. Madhow. Space-time communication

for OFDM with implicit channel feedback. In IEEEGlobal Telecommunications Conference, volume 3,pages 1321–1325, 2003.

[2] J. Chen, L. Yip, J. Elson, H. Wang, D. Maniezzo,R. Hudson, K. Yao, and D. Estrin. Coherent acousticarray processing and localization on wireless sensornetworks. In Proceedings of the IEEE, volume 91, pages1154 – 1162, August 2003.

[3] C. Intanagonwiwat, R. Govindan, D. Estrin,J. Heidemann, and F. Silva. Directed diffusion forwireless sensor networking. IEEE/ACM Transactionson Networking, 11:2–16, February 2003.

[4] J. D. Kraus. Antennas, Second Edition. Mc-Graw Hill,1988.

[5] D. Petrovic, R. Shah, K. Ramchandran, and J. Rabaey.Data funneling: routing with aggregation andcompression for wireless sensor networks. In IEEEInternational Workshop on Sensor Network Protocolsand Applications, pages 156 – 162, May 2003.

[6] A. Scaglione and S. Servetto. On the interdependenceof routing and data compression in multi-hop sensornetworks. In Proc. ACM Mobicom 2002.

[7] K. Yao, R. Hudson, C. Reed, D. Chen, T. Tung, andF. Lorenzelli. Array signal processing for a wirelessmem sensor network. In IEEE Workshop on SignalProcessing Systems, pages 11–20, Oct 1998.

[8] M. Yipeng Tang; Valenti. Coded transmitmacrodiversity: block space-time codes over distributedantennas. In Vehicular Technology Conference,volume 2, pages 1435–1438, May 2001.

Figure 3.3: Top level of master slave based model for information transfer

The key limitation of this architecture is that it considers the master node to be at equal distancefrom all the slave nodes, which is very difficult to achieve in practical scenario. Another problemis that the received power is analyzed without considering noise and fading in the channel.

In beamforming directivity or signal steering is a crucial factor effecting performance. variousstudies have been carried out, some claiming that even a partial synchronization in phase canproduce a significant increase in the network performance [71]. The idea of collaborative beam-forming is investigated by Ochiai et al. [82] where average beam pattern is evaluated against theuncertainty of sensor location as well as imperfection in phase. It has been shown that, collabora-tive beamforming offers the possibility of achieving power efficiency in wireless ad-hoc networks.



The authors concluded that a directivity of N is achievable where N is the number of transmitternodes. A similar study by Mudumbai et al. [111], has also explored directivity patterns achievedthrough random array. It has been shown that a 70% gain is achievable, even if there is a phaseerror of the order of 60o. Therefore, it is concluded that a directivity of linear scale with num-ber of nodes randomly distributed, can be achieved through physical propagation path model. Incase of “average beam pattern”, however, carrier phase synchronization across sensor nodes is notaddressed.

The effect of collaborative beamforming and cooperative transmission over the life time of sensornetworks by reducing the load or excluding nodes with critical battery power from packet for-warding request, has been studied in [112]. The proposed technique is analyzed for a special twodimensional disc case. It has been concluded that, extensive gain in received power is achievable atthe receiver using Cooperative transmission and collaborative Beamforming. A reduction of 90%in the payload of energy depleting nodes is observed, whereas a 10% improvement in the lifetimeof the general networks is recorded in comparison to existing approaches.

Distributed beamforming has been explored by Mudumbai et al. [113] for a randomly deployedsensor network. A master-slave approach is considered for sharing information between the trans-mitter and collaborating nodes. In case of random deployment, beamforming requires phase, fre-quency and time synchronization between the sensor nodes and the base station, to be achieved. Itis argued that, if synchronized in phase and frequency an N nodes distributed beamforming mayresult in N2 gain in the received power. However, it is considered more crucial to synchronize thecollaborating nodes in comparison to their distribution diversity. This is argued to be a a limitingfactor for many classes of diversity distribution which may be a restricting factor [114].

Koyuncu et al. [115] combined beamforming with amplify-&-forward relay networks and pro-posed an approach based on Quantized feedback. The proposed technique is claimed to achievesynchronization as well as significant gain in the performance. The use of quantized is meant foroptimization of system’s bit error rate. Furthermore, bound for the acceptable SNR values andapproximate bound for BER are derived. The authors concluded that using a few bits for feedback,their proposal can successfully achieve high array gain as well as full diversity.



3.3.2 Limitations of Beamforming

The practical implementation of beamforming involving multiple transmitters and single receiver(MISO beamforming) suffers from various issues. For example, in such scenario knowledge ofthe channel must be provided to each transmitter. Such a technique could be implemented in theform of multi-channel approach dedicating different channels for data and feedback transmission.However, in the context of WSN a multi-channel approach incorporate high complexity in hard-ware as well as processing at the sender and receiver side. Furthermore, frequency synchronizationrequirement is very tight while using and maintaining dual channel. This ultimately results in highenergy consumption of the overall network.

Collaborative beamforming may relatively seem simple, due to its successful application in wire-less networks, but in comparison to single user MIMO product, building a collaborative beamform-ing products are significantly complex. A serious concern in using multi transmitter beamformingis the interference experienced by a transmitter from another transmitter which is “too close” to it.Therefore, handling the interference caused by two nearby transmitters to each other, is a serioushurdle in the practicality of this approach.

Similarly, beamforming works for focusing energy transmitted by multiple transmitter at a singlereceiver. This means that most of the power of the transmitter is diverted to a specific direction.However, nodes outside the area actually intended for reception will experience significant degra-dation in the received signal.

3.4 Collaborative Communication

The concept of collaborative communication in WSNs has recently been introduced. In collabora-tive communication a transmitter node request its closer neighbors to help send some data to thebase station. These nodes are called collaborative nodes, collaborative transmitters or slave nodes.Once the source node shares its data and synchronizes with the collaborative nodes, all of themtransmit this same data at the same time towards the BS[116]. A pictorial diagram of collaborativecommunication is shown in Figure 3.4.

When signals from different collaborative nodes arrive at the base station, it combines these signalscoherently. This coherent combination of the signal at the receiver shows that a large amount ofpower is combined and transmitted. This phenomenon of combined power transmission is referred



CollaborativeNode 3

CollaborativeNode 1

CollaborativeNode 2

CollaborativeNode N

Figure 3.4: Collaborative Communication Model

to as constructive interference. The constructive interference not only results in high power gainin collaborative communications, but also reduces the bit error rate considerably and achieves asignificant gain in capacity.

It is the responsibility of the source node to send timing and carrier signal as well as the data tobe sent, to the collaborative nodes. The real question is “how is this sharing and synchronizationprocess carried out?”. To answer this question, a modified master-slave architecture has beenproposed by Naqvi et al. [117], based on the master-slave architecture presented in Mudumbaiet al. [113]. In this modified architecture any node can become a master node which is responsiblefor sending time and carrier signal to all the slave nodes. The distance between the master nodeand each collaborative (slave) nodes may vary. This same model has been adopted in this thesisfor analyzing energy savings of different proposed models in later chapters.

3.4.1 Why Collaborative Communication

Collaborative communication uses the concept of antenna diversity to achieve performance gain.However, the question is, how is the concept of collaborative communication different from othercooperative/diversity approaches?. To explain this, let us take an example where three transmitterssend information to a receiver. At the receiver, in normal communication, power of each receivedsignal is calculated and then the signals are added. If the received strength of signal from each



transmitter is 1volt, then the total received signal power gain will be 12 + 12 + 12 = 3volt. For thesame scenario in collaborative communication, the received signal gain is calculated as (1 + 1 +

1)2 = 9volt. This shows the significant improvement achievable due to the use of collaborativecommunication. However, it must be noted that, performance gain of collaborative communicationmay vary for different types of channels as well as the propagation environment.

In sensor networks the distribution of nodes may be fixed or random [118]. Achieving phase andfrequency synchronization in fixed array system is easy as the position of sensor node is known.Problem arises when the distribution of sensors in WSN is random where the accurate positionof a sensor node is hard to determine. This can lead to different problems like mismatching inphase and frequency. To overcome this problem and achieve high performance even when thereis a mismatch in phase and frequency, collaborative communication is used. Many architecturesregarding the issues related to collaborative communication can be found in literature.

It is argued in many studies that collaborative communication can tolerate error in phase and fre-quency synchronization i.e. good performance gain is achievable even if the signals received at thereceiver are not synchronized perfectly in phase and frequency [116]. This is achieved through asynchronization process, incorporated in collaborative communication, discussed in the followingsub-section.

3.4.2 Synchronization Process in Collaborative Communication

Collaborative communication considers collaborative nodes to be at different distances from theBSbecause of the difference in their locations. Due to this variation in distances of collaborat-ing nodes and the base station, signals from these collaborative nodes may arrive out-of-phase.To achieve high performance gain, collaborative communication needs a way to synchronize thereceived signal at the receiver.

A synchronization process has been presented in [56] which is adapted in this research work,illustrated in Figure 3.5. It is a multi step process, where a BStransmits a reference signal to all thecollaborative nodes. Each collaborative node passes the received signal through low noise amplifier(LNA) to perform the phase offset. Then, phase lock loop (PLL) locks on to the frequency of theincoming signal. This process is repeated after small interval if the channel is fast variant.

It is also shown in Naqvi [56] that the phase error θf can be calculate as θf = 2πfodf/c, wherethe carrier frequency is represented by fo, displacement error by df and speed of light by c. If we



61

in frequency and phase errors. Phase synchronization is multi-step and the periodic

process and its architecture are shown in Figure 4.7. For the synchronization process, the

base station sends a known signal to all collaborative nodes. The received signal from the

base station is passed through a Low noise amplifier (LNA). The phase lock loop (PLL)

locks itself on to the incoming signal frequency and performs the phase offset. This

process is repeated after a time interval depending upon the nature of the channel. If the

channel is fast variant, the time interval to repeat the synchronization process is small.

Base Station

Tref(t)

LNA

DECODE

R

PLL

Phase offset Correction

LNA

PLL


LNA

PLL


Collaborative Node 1 Collaborative Node 2 Collaborative Node N

...

Figure 4.7 Synchronization Model

The base station transmits a known signal to all sensor nodes in the network; this is

used as reference signal for the network and is given by

)2cos()( 000)2(

000

tfAeAtT

tfjref

(4.7)

Sensor nodes receive the signal; it then locks itself with the carrier frequency [29]

after achieving its steady state. The signal at the ith

sensor node can be written as

)2cos()( 000 ii tfAtT (4.8)

where Θi is the phase error between the transmitted and received signal and is due to the

displacement error and PLL error. But at steady state PLL error is approximately zero and

the Θi is due to a displacement error. The output of the PLL is passed through phase

offset correction to minimize phase error. Base station feedback is presented in order to

reduce phase error. The base station sends feedback to the sensor nodes. The feedback

from the base station is used for phase offset correction [13] and results in a significant

reduction in phase error.

Figure 3.5: Architecture of the phase synchronization process

consider fo = 915MHz, df = 5cm and c = 3× 108, a phase error of 60o is produced. However ifthis phase error becomes 180o then it produces destructive interference.

A thorough literature survey revealed that, synchronization in time, phase and frequency can beachieved in WSNs among signals from different nodes or scatter components of a signal, yieldsignificant gain in the received power . However, it is also argued that a significant gain in receivedpower is achievable in WSN using collaborative communication even if the received signal orscatter components of a signal are not completely synchronized in phase and frequency [88, 117–124]. Here it is important to note that, in collaborative communication there are two types ofcommunication scenario: communication among collaborative nodes and communication betweencollaborative nodes and the base station. As mentioned earlier, various mechanisms, like Master-Slave approach for communication among collaborative nodes(not between collaborative node &base station) in WSN is used to achieve frequency synchronization

The effect of time and frequency synchronization on collaborative communication has been inves-tigated to show what range(bounds) of frequency could be used. Two venues; efficient communi-cation system and task scheduling through smart algorithms for reducing energy consumption inWSNs has been explored in [125–129]. In both venues, reduction in energy utilization has beenclaimed. Synchronization among collaborative nodes is more crucial than their distribution diver-sity. But it must be noted that, synchronization among collaborative nodes can degrade system’sperformance in some groups of diversity distribution as mentioned earlier.

A spread spectrum based communication system for range extension in WSN to study the effectof collaboration has been investigated in[8]. The authors argued that using the natural diversity incollaboration, power from multiple sensor nodes can be combined to achieve robustness against



the channel impairments and produce significant extension in the coverage range. Our approachhowever is to improve the energy consumption through collaboration in combination with spreadspectrum by considering one receiver and multiple transmitter nodes from a sensor field as shownin figure 7.1.

To minimize the energy consumption in Body Area Networks, an energy efficiency techniquebased on energy efficient communication and battery friendly task scheduling has been presentedin [130]. There are two avenues where the energy consumption is claimed to be reduced. First,through using efficient communication system and secondly through an intelligent algorithm fortask scheduling. In collaborative communication it is more important that the collaborative nodesbe synchronized rather than diversity in their distribution.

In this thesis synchronization is considered as a degenerating factor as well as an averaging effect.In the first case it affects performance because of inter-symbol interference and co-channel inter-ference, whereas in the later case it improves predictability and reliability of the channel. Howeverthe requirement is more relaxed in comparison to distributed beamforming [81], where timing andphase synchronization of the collaborative nodes, is a strict requirement.

Naqvi et al. [118] pointed out that, in distributed deployment of sensors in an environment, it ishard to determine exact position of a sensor node. It may create problems in estimating positionof a node that ultimately results in displacement error. In collaborative communication shown inFigure 3.4, each node transmits data using its own oscillator producing a mismatch in frequencyamong the signals at the receiver. It has been shown in [118] that a significant gain in receivedpower is achievable even with error in phase and frequency synchronization using collaborativecommunication including the effect of noise AWGN(Additive White Gaussian Noise) and fading.

Naqvi et al. [123] has presented a theoretical model for both the received power and Bit ErrorRate (BER). Here it has been proved that received power is a function of the number of nodes incollaboration whereas BER is a function of signal to noise ratio. It must be noted that the proposedapproach also considered the presence of noise and fading. An energy consumption model isproposed to gauge the energy efficiency in terms of energy consumed for local communicationamong collaborative nodes as well as collaborative communication with the base station.

The proposed model is developed for collaborative communication by considering the parametersof off-the-shelf products, for example CC2420 and AT86RF212. The abstract system model of[123] is shown in Figure 3.7. The impact of unsynchronized frequency and time have not beendiscussed.



Collaborative Node 1

Integration (T interval )

Rr(t)

2cos(2πf0t+Θ0)

LPF

r(t)

R

Power Calculation

n(t)

Base Station

Collaborative Node i

h1

hi

Channel

Transmitter

Collaborative Node m

hm

Fig. 2 Mathematical model of a collaborative communication system.

NhhEjiESE

hEiESEPE

ji

m

i

m

jij

ff

i

m

ifR

1 1

2

2

1

22

))(cos())(cos(

))((cos][

, (5)

As E[S2]=P, all Θf(i) are i.i.d random variables, therefore E[Θf(i)] ≈E[Θf] and all hi are i.i.d. random variables, therefore, E[hi]=E[h]. The equation (5) can now be written as

NhEhEEPEmm

hEmPEPE

ff

ifR

)cos()cos()1(

][)]([cos][ 22

. (6)

Using the values of equations (A.1), (A.2), mean value of h and E[h2]=1 equation (6) becomes

NPbmm

mPPE R

22 )sin(

2

)1(

4

)2sin(

2

1][

. (7)

where φ is distribution limit of phase error and b is the mode of Rayleigh random variable h. The power of the received signal is the sum of the signal part and noise part. Thus the capacity of the collaborative communication system can be found as

N

Pbmmm

N

SC

22

2

2

)sin(

2

)1(

4

)2sin(

2

11log

2

1

)1(log2

1

(8) In the absence of phase error i.e., φ = 0, the channel capacity of collaborative communication system may be found as

N

Pbmmm

N

SC

2

)1(1log

2

1)1(log

2

1 2

22

(9)

Special Case: If the transmitted power of each collaborative node is P/m, equation (8) can be written as,

N

Pbm

N

SC

22

2

2

)sin(

2

)1(

4

)2sin(

2

11log

2

1

)1(log2

1

(10) This is the case when we reduce the power of each collaborative node by factor m, i.e., the total transmitted power is P. We will use this reduction to investigate the capacity with reduced power consumption of each collaborative node. This case can be used to investigate the effect of total transmitted power on capacity. The results obtained

Recent Researches in Communications and IT

ISBN: 978-1-61804-018-3 279

Figure 3.6: System Model for Collaborative Communication with Random Distribution of Sensors

H. Naqvi et al. / Physical Communication 3 (2010) 119–128 121

Fig. 1. Geometry of sensor nodes.

Θ0 = 2π f0d0/c , where c is the speed of light. Let df (i)be the displacement error between base station andthe ith collaborative node, then the phase error due todisplacement error is given byΘf (i) = 2π f0df (i)/c .Let cos(2π f0t) be the carrier/reference signal used by all

collaborative nodes. We assume that signal delay is verysmall with respect to the signal bit interval T , so thereis significant guard interval and inter symbol interference(ISI) can be ignored.The cumulative received signal at the Base station is

given by:

rm(t) = <

(N∑i=1

his(t)ej(2π f0t+Θ0+Θf (i)))+ n(t), (1)

where n(t) is AWGN and hi is the Rayleigh fading.At the base station the demodulated signal is integrated

and its output is given by

R =N∑i=1

hiS cos(Θf (i))+ n, (2)

where S = ±√Eb is the signal amplitude and n is the noise

amplitude at sampling time T .Power of R is given by

PR =

[N∑i=1

hiS cos(Θf (i))+ n

]2. (3)

AsΘf (i), hi and n are the independent random variablesand n is zero-mean, we have to calculate themean value ofreceived power as

E [PR] = E

[ N∑i=1

hiS cos(Θf (i))

]2+ σn2, (4)

where σ 2n is the variance of noise. Eq. (4) can also bedeveloped as follows

E[PR] =N∑i=1

S2E[cos2(Θf (i))

]E[h2i]

+

N∑i=1

N∑j=1i6=j

S2E[cos(Θf (i)) cos(Θf (j))

]E[hihj

]+ σn

2.

(5)

As all Θf (i) are i.i.d. random, therefore E[Θf (i)] ≈E[Θf ] and all hi are i.i.d. random variables, therefore,E[hi] = E[h]. The Eq. (5) can now be written as

E[PR] = NS2E[cos2(Θf )]E[h2i ] + N(N − 1)S2E[cos(Θf )

]× E

[cos(Θf )

]E [h] E [h]+ σn2. (6)

Using the values of Eqs. (A.1), (A.2), mean value of h andE[h2] = 1 Eq. (6) becomes

E[PR] = NS2[12+sin(2ϕ)4ϕ

]+N(N − 1)b2S2

2

[sin(ϕ)ϕ

]2+N02, (7)

where ϕ is distribution limit of phase error and b is themode of Rayleigh random variable h.The results obtained from Eq. (7) and Monte Carlo

simulation is discussed in Section 3.

2.2. Probability of error

In this section, we are calculating the probability oferror in the presence of Rayleigh fading in the channel.Suppose the data is sent using BPSK and the receivedsignal R depends upon independent random variables hi

Fig. 2. System model.Figure 3.7: System model



In collaborative communication if BSreceives all the sent data at the same time (time synchroniza-tion), all sending nodes use the same frequency (frequency synchronization) and the combinationof data from all the nodes results in a coherent signal (Phase synchronization), then a significantgain is achievable in terms of all the three parameters (time, frequency and phase) [82, 111, 131].To gauge the performance of a WSN the following key parameters need to be considered:

1. Power Gain

2. Capacity Gain

3. Bit Error Rate (BER)

In the following sub-sections each of these parameters are explained in the light of previous studies.

3.5 Received Power Gain

Received power gain is the effectiveness of receiving antenna in converting the incoming signalfrom a certain direction into an electrical power. With no specific direction the peak value isconsidered as the gain. Energy efficiency plays a crucial role and key requirement in WSN whiletransmitting information from source to a sink.

To remedy the problem, collaborative communication can be a viable solution. In collaborativecommunication, multiple transmitters transmit the same data towards the same receiver at the sametime.

If synchronization in time, phase and frequency is achieved, it can result in high power gain incollaborative communication. With no phase and frequency errors, and N nodes, collaborativetransmission is able to produce N2 gain in the received power. Channel fading is another factorthat significantly degrades the data transmission and increases the power requirement. In recentwork related to collaborative communication [81] and [19, 132], it is shown that high gain inreceived power is achievable, even if frequency is not fully synchronized.

In [133] a collaborative transmission based theoretical model has been presented. The performanceanalysis in presence of noise (AWGN) and fading for collaborative transmission of informationin sensor networks with frequency offsets a large power gain and low Bit Error Rate (BER) areachievable even with a mismatch in frequency synchronization. It is concluded that the numberof sensor nodes in collaboration affects the gain in received power as well as BER. But the power



requirement increases with a raise in the number of transmitter nodes.

So the total energy saving depends upon the energy gain and circuit energy used by the network.The models presented in [124, 134, 135] investigate the energy consumed by collaborative com-munication, when frequency synchronization may have mismatching errors, in WSN. The trade-offanalysis between the required circuit power and achieved power gain using collaborative commu-nication is carried out.

3.6 Bit Error Rate(BER)

Bit Error Rate (BER) is the number of bits received in error in unit time. The ratio of bits in errorwith the total number of transferred bit for a specified interval of time gives the bit error ratio. It isused to measure performance of a communication system and has no unit [136]

The expression of bit error rate (BER) for cooperative communication with multiple relays by ex-ploiting the concepts of Alamouti space-time code is presented and analyzed in [137]. In [49]and [138], the “Symbol error rate (SER)” performance of “decode-&-forward (DF)” cooperationis analyzed under Rayleigh fading channel. In [139] and [140], the authors analyzed the SER per-formance over Nakagami fading channels both for “Amplify-and-Forward (AF) and Decode-and-Forward (DF)” protocols, respectively. However, all the investigations consider only the classicalthree node model which needs to be expanded to more generalized cases instead of only consider-ing three nodes.

Similarly several studies have been conducted investigating performance of cooperative communi-cation in terms of BER and SER.

To conclude, the procedure for developing closed-form average BER expressions in the presenceof noise and fading including the effect of unsynchronized phase and frequency stays as an openissue. Moreover, such expressions are found very limited in existing literature and need furtherinvestigation. The BER may be used as baseline in the study of collaborative communicationsystems enhanced by advanced signal processing techniques. The fading mitigation ability ofcollaborative communication can also be analyzed using BER analysis.



3.7 Capacity Gain

Capacity gain represents the tight upper bound on the rate of information exchanged by a commu-nication system. It must be noted that the information exchange at this rate must ensure reliabilityof the information. During World War II, Claude E. Shannon developed a mathematical model forcalculating the capacity of channel stating that capacity of a channel is the maximum of informa-tion transferred between a transmitter and receiver [141]. The mathematical formula developed byClaude E. Shannon is shown in Equation 3.1.

C =1

2log2

(1 +

S

N

)(3.1)

One of the basic design requirements in WSNs is to determine their required capacity. If the sensornetwork is of a star configuration, where a BSor destination communicates with the surroundingsensor nodes that are sources of information, capacity can be found in a similar way in which it isdone for wireless cellular networks. However, when the network uses multiple relays to achievediversity in signal transmission, the problem that arises is how to determine the network capacityand it has been the subject of recent extensive research work.

Van der Meulen [83] presents an extensive survey of multi-way channels developed until 1976.This review included the analysis of relay channels. Two years later, capacity theorems for Gaus-sian degraded, reversely degraded and feedback relay channels are proved and reported in [108].A general wireless network with n nodes randomly located in a region of 1m2 is investigated in[142] to calculate the capacity of the system. The author concluded that using channelization tocreate multiple channels for information transfer does not actually change any of the results. Inmulti-relay network various strategies in signal processing at relay nodes can be applied. Theanalysis of cooperative strategies is conducted and capacity theorems proved for relay networksare based on either compress-&-forward or decode-&-forward relay strategies. Upper and lowercapacity bounds for cooperative diversity are calculated assuming that the channel contains AWGN

and Rayleigh fading [87, 109].

The networks based on the adaptive decode-&-forward strategy are analyzed in [143–145]. Theexact formula for the probability of error and channel capacity are derived over independent andno identical Rayleigh fading channels for selecting best relay[145] and Nakagami channel [143].A network with a number of relay channels and a single source to destination channel is designedto minimize the error probability at the destination in [146]. A cooperative network that selects a



set of relays to achieve Communications between the source and destination and comply with theoutage constraints is presented in [147]. In existing literature, the multi-relay systems have beentheoretically analyzed under the assumption that the channels between the source node and therelays are characterized by the “Additive White Gaussian Noise (AWGN)” or a combination of theAWGN and Rayleigh fading.

However, the procedure for developing closed-form expressions for the channel capacity of col-laborative communication systems with imperfect phase and frequency synchronization remainsan open issue. Moreover, such expressions are absent in the existing literature.

3.8 Problem Statement

From the survey of literature, it is clear that energy consumption in WSNs is the most crucialproblem. Various approaches have been proposed in order to optimize energy utilization in thesenetworks. In this thesis, an investigation, of applying the concept of collaboration and its effecton various scenarios (BANs, multipath system, wideband channels) in WSNs, is performed. GivenN collaborative nodes sharing the same information, are ready to send this shared informationto a common receiver; the base station. This research strive to design energy efficient systemsbased on collaborative communication in the context of WSNs. The effect of this collaboration isinvestigated, both in narrowband as well as Wideband channels, assuming the presence of noiseand fading in the channel. The following four sub problems are addressed in this research work.

Prob-1: In the first sub problem collaborative communication and its effect is investigated in thecontext of BANs using narrow band channel. The channel is assumed to have the effect of noiseand fading. Most importantly, the signals arrived at the BS from collaborative communication areconsidered to have synchronization errors in phase. The effectiveness of collaboration in mitigatingnoise and fading is analyzed as well as the effect of transmission distances on energy consumptionis investigated.

Prob-2: In this sub problem, the performance gain of collaborative communication is investigatedin multipath environment. The effect of multipth/shadow fading on collaborative communicationhas been investigated in the presence of noise, where sensors are randomly deployed. The impactof scatter components of a signal in mitigating the fading effect is also explored. It is consideredthat, not only the signals from different nodes, but also scatter components may arrive out-of-phase.



Prob-3: This sub problem investigates the behavior of collaborative communication in WSNs withan entirely different channel as compared to the previous two sub problems. In this sub problemcollaborative communication is studied using spread spectrum based system for energy consump-tion. The effect of collaborative communication and wideband system on each other has beeninvestigated in the presence of noise and fading. Furthermore, the phase synchronization is con-sidered imperfect for the received signals.

Prob-4: This last sub problem investigates the impact of collaborative communication, in terms ofcapacity gain, using wideband systems. The capacity of the system is calculated in the presence ofnoise and fading using the Shannon capacity formula. The impact of collaborative communicationover the capacity gain of wideband systems is investigated, both, in case of perfect and imperfectsynchronization in phase of the received signal.


Chapter 4

Proposed Methodology

The proposed model is simulated and tested using MATLAB® environment with parameters ofoff-the-shelf devices. To evaluate the performance of the proposed collaborative communicationsystems, several figure of merits are considered i.e. received power, BER and gain in channelcapacity. Theoretical/mathematical models are derived and analyzed for each of the parameters.Trade-off analysis between energy consumption and transmission distances are derived.

MATLAB/SIMULINK is developed by MathWorks and is used for programming and analyzingmulti-domain dynamic system. SIMULINK is a tool for modeling, analyzing and simulating dif-ferent systems using data flow graphical programming language. There are customizable set ofblock and libraries provided in the interface. It provides integration features that tightly integrateit with other MATLAB®environments i.e. it can drive MATLAB®and it can be scripted in MAT-LAB®. SIMULINK has vast usage in control theory and digital signal processing for multi-domainsimulation and Model-Based Design.

4.1 System Model

This section presents a brief and general description of the model used in this research work indifferent forms. In this model for collaborative communication, N number of collaborative nodesi.e, Collaborative Node1, Collaborative Node2, up to Collaborative Node N as shown in the Fig-ure 4.1 adopted from [116]. Let, all these N collaborative nodes collectively transmit the sameinformation at the same time to a common receiver as shown in Figure 4.1. This model considers

51

Chapter 4. Proposed Methodology

random deployment of sensor nodes, therefore information transmitted by each node at the sametime may arrive at different times at the base station due to variation in each node’s distance fromthe base station, resulting in an imperfect synchronization.

CollaborativeNode 3

CollaborativeNode 1

CollaborativeNode 2

CollaborativeNode N

Figure 4.1: Geometry of sensor network used in collaborative communication based models pre-sented in later chapters

As there is no perfect phase and frequency synchronization between the collaborative nodes andthe BAN, there is a random phase and frequency offset between all transmitted signals. The modelis tested in different channels scenarios i.e narrowband as well as wideband (spread spectrum)including the effect of Rayleigh fading. Rayleigh fading for each transmitted signal has an inde-pendent effect.

Figure 4.2 is a detailed structure of the proposed model. On the receiver side, denoted as theBAN, the received signal is demodulated. The demodulator correlates the received signal with acosine signal, passes it through a low pass filter (LPF) and integrates it. After integration of allcomponents, power of the demodulated signal is calculated.

From study of different wireless network algorithms, it is clear that significant fade margin is oftenneeded to compensate for the effects of channel fading on signal power loss. The exploitation ofthe space diversity by collaborative communication is considered, which confirms its effectivenessin reducing channel fading. In this regard, the fading-mitigation capability of collaborative com-munication has been investigated which confirms that collaborative communication is a feasiblesolution to save energy during data exchange in fading channels.



Integration

LPF

Power Calculation

DECODE

R

Base Station

Channel Collaborative Nodes Transmitters

cosine wave generator with random

phase and frequency error 1

cosine wave generator with random

phase and frequency error N

cosine wave generator

Wideband Channel

Binary pulse

strain generator

Constant

Multiplier

Figure 4.2: Proposed theoretical model based on collaborative communication

4.2 Performance Matrix

In this research collaborative nodes compliant with “IEEE 802.15.4” standard are used. Energyefficiency for different collaborative communication systems is simulated using parameters of “off-the-shelf products i.e. CC2420 [46] and AT86RF212” [45]. An investigation of energy savings iscarried out using collaborative communication system. The results obtained are compared with thestandard values established in “IEEE 802.15.4” standard for WSNs to confirm that the proposedmodel extends the lifetime of the network.

4.2.1 Received Power

This is a very important figure of merit used to measure performance of wireless systems. Itactually represents the effectiveness of receiving antenna, in converting the incoming signal froma certain direction into an electrical power. In the absence of specific direction, gain is representedby the peak value. Energy efficiency plays a crucial role and key requirement in WSN whiletransmitting information from source to a sink.

It is observed that the gain produced by collaborative communication increases with an increasein the number of collaborative nodes. However, this increase in the number of collaborative nodesincur more overhead in terms of the required circuit energy for data transmission. Therefore,an investigation into power savings through collaborative communication at the expense of circuit



power is performed. In this regard, energy efficiency models have been developed for collaborativecommunication system and SISO systems.

However, as mentioned in the previous chapter, collaborative communication performs coherentaddition of the received signals to obtain high power gain. But in case of 180o phase error, thereis 100% error in the information received at the BAN due to destructive interference and stillresulting in high value of received power. Therefore, to analyze performance of collaborativecommunication, considering received power only as a figure of merit, will not be sufficient.

4.2.2 Bit Error Rate(BER)

Bit Error Rate (BER) is the number of bits received in error in unit time. The ratio of bits in error

with the total number of transferred bit for a specified interval of time gives the bit error ratio. It

used to measure performance of a communication system and has no units [136].

BER analysis is widely used and common figure of merit to characterize and analyze performanceof various wireless communication systems. To see how successful is the proposed system againstfading, an expression for BER is derived based on collaborative communication systems (chapter5, 6 and 7) in case of both narrowband and wideband channels in the presence of Rayleigh fading.Specifically, mathematical expressions for BER in collaborative transmission systems with unsyn-chronized phase are derived. The derived BER expressions is verified by implementing it usingparameters of “off-the-shelf” products i.e. “CC2420 [46] and AT86RF212[45]”. For a given BER

value, the signal-to-noise ratio value needed by collaborative communication systems in compari-son to SISO systems are determined through simulation.

It is clear from the BER results in later chapters that, if collaborative communication is used, thefade margin values for the required BER can be significantly reduced. This reduction in fade mar-gin attributes significant savings in transmitted power. However, increase in number of transmitternodes, incorporate more circuit power which degrades power efficiency of the overall network.Therefore, the proposed model must also be evaluated using the trade-offs between energy con-sumption and the number of transmitter nodes used.



4.2.3 Energy Consumption

This figure of merit is used to gauge energy efficiency of a system. It is used to evaluate the pro-posed models and investigate the trade-offs between the gain in received power and the requiredcircuit power. For this purposed break-even distances are calculated for the proposed collaborativecommunication based models. Break-even distances are those distances where the energy con-sumption of collaborative communication and single transmitter (SISO) systems become equal.This value is related with the transmission distances to analyze performance of the multi transmit-ter (collaborative communication) in comparison to SISO systems.

4.2.4 Capacity Gain

Capacity of communication system is basically the upper limit on its data rate provided the re-ceived data is reliable. Capacity gain is a common figure of merit to evaluate performance of acommunication system.

To explore the effect of collaborative communication on capacity gain of WSNs, a close formmathematical relation for channel capacity is developed and analyzed. In this relation the pro-posed collaborative communication system is considered with unsynchronized phase for the caseof wideband channels. To verify results of the developed expression, simulation is performed usingparameters of the devices like, CC2420 [46] and AT86RF212 [45].

4.3 Expected Scenarios

From the study of several synchronization models for wireless communication, it is observed that100% synchronization among the transmitters is very difficult to achieve. Therefore, this researchaims to propose, model and analyze several collaborative communication systems for sensor net-works using narrowband as well as Wideband channels in the presence of Rayleigh fading andnoise.

The following system models are designed and analyzed:

1. Collaborative communication based energy efficient system with unsyncrhonized phase, as-suming perfect time and frequency synchronization for “Body Area Network (BAN)”.



2. Collaborative communication based energy efficient system with unsyncrhonized phase, as-suming perfect time and frequency synchronization in case of “multipath” communicationfor sensor networks.

3. Collaborative communication based energy efficient system with unsyncrhonized phase, as-suming perfect time and frequency synchronization in case of “wideband” (spread spectrum)systems for sensor networks.

4. Collaborative communication based energy efficient system with unsyncrhonized phase, as-suming perfect time and frequency synchronization in case of “wideband” (spread spectrum)systems for capacity gain in sensor networks.


Chapter 5

Collaborative Communication in Body AreaNetworks (BAN)

A network with small sized sensor nodes around human body for sensing biological, physical andchemical changes to alarm the wearer is known as Body Area Network. It has the ability of automedication in case of emergency, and also the capability to transmit the sensed information overthe Internet. BAN have recently emerged having a broad application domain, such as healthcare,detection of chronic diseases [121, 148, 149], monitoring performance of athletes [8, 129], militaryand multimedia applications, and BAN with cloud computing capabilities [150]. Tele-medicine isby far an important application, where doctors remotely monitor patient’s condition and prescribemedicine and precautions.

Being powered by battery, the most critical design constraint in BAN is energy consumption [149].Small sensor size increases comfort level of human body implanted with sensors, but on the otherhand it reduces battery size. Battery size determines its capacity, which in turn affects its life time.Keeping small battery size makes it difficult for designer to design sophisticated security and otherprotocols for these networks. Therefore these devices need to be highly energy efficient.

A vital part of energy consumption is communication system. For an efficient communication sys-tem a possible solution could be multihop approach in situation where the base station is distantfrom the transmitter nodes. However implementing multihop routing incurs increased overheadwhen transmitter and receiver get further away from each other, and also complicates implemen-tation of the routing process. In case of retransmission, data has to pass through the multihop pathfrom sender to the receiver which may result in an even higher energy consumption by draining

57

Chapter 5. Collaborative Communication in Body Area Networks (BAN)

Patient with Body sensors

Base Station

Medical Expert

Medical Report

Diagnosis

B.P: 130/90

Sugar: Nil

Temp: 101

Urine test:

Normal

Prescription

Medicine 1

usage

Medicine 2

usage

Medicine 3

usage

Next Appointment

15-06-2014

Focus of this article

Figure 5.1: General architecture of BAN for healthcare

batteries of all nodes on the multihop path [86, 151]. To address these problems, an energy efficienttransmission model based on collaborative communication for BAN has been proposed as shownin Figure 5.1.

In collaborative communication “a group of collaborative nodes in a network transmit the same

data at the same time toward a common destination known as the Base Station (BS)”. The BaseStation (BS) is the most powerful node with no resource limitations or constraints [87, 118, 123,124, 143]. The collaborative nodes are synchronized [81, 110] with each other and each node is atone hop from the base station. It is unlike cooperative communication where the information to betransfered is relayed by multiple intermediate cooperative nodes.

5.1 Proposed System Model for BAN

In collaborative fixed array antenna where position of the transmitter is fixed, signal at the receiver“BS” is the total of transmitted signal by each antenna (cumulative signal), shown in Figure 5.2.This constructive addition of individual signals leads to a large gain in received power as well as asignificant improvement in bit error rate. To achieve synchronization in phase, frequency and timein such systems, a common network controller is used. Since the antennas are fixed, the commonnetwork controller knows how far each transmitter node is apart from the other transmitter node.Therefore, the node’s position is determined accurately. Such systems do not suffer from problemslike phase error or displacement error [152, 153].

However, unlike fixed array antenna, determining accurate position of a transmitter in a random



52

antenna array system there is a common network node that is used to achieve frequency,

phase and time synchronization. The distances between transmitter antennas are fixed and

are accurately known to the controller. As the accurate positions of the antennas are

known, there are no distance estimation errors or phase errors in this system [23-26].

However in a distributed sensor network, the accurate position of the sensor nodes is

generally not known and the theory of the fixed antenna array cannot be applied directly.

Namely, there is no network node that performs phase, time and frequency

synchronization. This results in a position estimation error (called the displacement error),

that is translated into the phase error of the transmitted signal. To achieve a high power

gain and to reduce the BER using collaborative transmission, the synchronization of

frequency, time and phase needs to be achieved.

Base Station

Network

Node

Antennas

DECODE

R

Figure 4.1 Fixed Antenna Array

In regard to the advantages stated above, the study in this chapter is dedicated to the

design of a collaborative communication system that can provide significant power gain

with imperfect phase and frequency synchronization in the presence of AWGN and

Rayleigh fading.

Of particular interest to this study is the achievement of the following:

To understand the development of collaborative communication model;

To investigate the effect of phase and frequency errors on power gain achieved

using collaborative communication in wireless sensor networks;

To design an energy efficient collaborative communication system;

To establish a WSN simulator to assess power gain using collaborative

communication with imperfect phase and frequency synchronization in the

presence of AWGN and Rayleigh fading using the parameters of off-the-shelf

products i.e. CC2420 [27] and AT86RF212 [28].

Figure 5.2: A communication system with fixed antenna array

distributed deployment is not always possible. Therefore, fixed antenna array scenario cannot beapplied directly to such systems, as there is no central controller to achieve synchronization. As aresult an error may occur in estimating position of a transmitter, known as displacement error. Thisdisplacement error may be caused by the difference in arrival time of the signals at the receiver.The high gain in received power and reduction in BER of collaborative transmission is possibledue to the achievement of phase, time and frequency synchronization.

Due to estimation error, perfect synchronization however cannot be achieved in these systems.Therefore, a general architecture has been proposed where a set of implanted nodes transmit datacollaboratively towards the base station using the same carrier frequency as shown in Figure 5.1.Before transmitting the data towards the base station, it is shared among the collaborative nodes.

5.1.1 Theoretical Model

For N number of collaborative nodes, the detailed system’s structure used for theoretical modelingis shown in Figure 5.3. There are two main communication components of the proposed system i.e.communication among the implanted collaborative nodes(on-body channel) and communicationbetween the implanted nodes and the base station (body-to-external channel) [154–156]. Theauthors in [156], suggested Rician fading channel for “on-body channel”, but to consider worstcase scenario here, the line-of-sight(LOS) component of the Rician fading has been ignored. AsRician fading with no LOS is equivalent to Rayleigh fading. Similarly, in the second case thepatient is considered to be in a place like a room or a Hall furnished with different obstacles likedesks, computer monitors, metal windows or other equipments As well as the patient may changehis/her position which prevent LOS communication of the implanted collaborative nodes with base



X

T Intervals Integration

LPF

Calculate Power

Collab_Node 1 x(t)cos(2πf0t+θ0 + θr (1))

2cos(2πf0t+ θ0)

Collab_Node 2 x(t)cos(2πf0t+θ0 + θr (2))

Collab_Node N

x(t)cos(2πf0t+θ0 + θr (N))

α1

α2

αN

n(t)

ym(t)

Error Rate Calculation

R

Collaborative Nodes Transmitters Base Station

Figure 5.3: Collaboration based system model for BAN

station, making it worst case scenario and suitable for Rayleigh fading.

Let us suppose that N collaborative nodes are implanted in a patient’s body transmitting informa-tion collaboratively to a Base Station as shown in Figure 5.1. Due to obstacles and movements ofpatient’s body, there are negligible chances of “line-of-sight” communication between the collab-orative nodes and the base station.

If we consider x(t) as the information or data signal transmitted by each collaborative node de-ployed randomly, then there may be an estimation error in calculating distances between the trans-mitter nodes and the receiver (BS), known as the displacement error [123]. Let the initial distancebe d0 between each collaborative node and base station with carrier frequency f0. So the phasedue to distance d0 is given by θ0 = 2πf0d0/c, where c is the speed of light. The sensor nodes inpatient’s body may trigger re-estimation of the distances between nodes and base station due bodymoment. Let dr(i) be the estimated distance, which when translated into phase error, produces aphase of θr(i) = 2πf0dr(i)/c. Most of the mathematical concept and relations in this chapter andrest of the chapters in this research work, are adopted from [157].

The carrier/reference signal cos(2πf0t) is used by collaborative nodes. The bit interval T, is as-sumed to be very large in comparison to signal delay, leading to a sufficient guard interval. Theguard interval provides immunity against “ISI” therefore, ISI can be ignored. The accumulatedsignal from the collaborative transmitter nodes at the receiver (BS) is given by the following equa-tion:



ym(t) = <

(N∑i=1

αix(t)ej(2πf0t+θ0+θr(i))

)+ n(t) (5.1)

where αi denotes fading and and n(t) denotes AWGN.

The received demodulated signal in Equation 5.1 at the base station, when integrated and simplifiedusing trigonometric identity “2 cos(a) cos(b) = cos(a + b) + cos(a − b)” produce the followingresult:

Y =N∑i=1

αiX cos(θr(i)) + n (5.2)

where X = ±√E denotes the signal amplitude and n denotes noise amplitude at sampling time T.

Power of the signal at the receiver can be achieved as follow:

PY =

[N∑i=1

αiX cos(θr(i)) + n

]2(5.3)

In Equation 5.3, θr(i), αi and n are “identical independent random” variables. As n is also zero-mean AWGN, therefore we need to calculate the expected value of received power:

E[PY ] = E

[ N∑i=1

αiX cos(θr(i))

]2+ σ2n (5.4)

As σ2n is the variance of noise, this equation can also be rewritten as

E[PY ] =N∑i=1

X2E[cos2(θr(i))

]+ E

[α2i

]+

N∑i=1

N∑j=1i 6=j

X2E [cos(θr(i)) cos(θr(j))]E [αiαj] + σ2n (5.5)



As all θr(i) and αi are “i.i.d” random variables therefore E[θr(i)] ≈ E[θr] and E[αi] ≈ E[α]. Alsoθr(i) is “i.i.d” distributed uniformly over the interval {−φ → φ} for all i, then θr(i) ≈ θr. Usingvalues from Equations A.1,A.2, Equation 5.5 can be reduced to the following expression assumingmean value of α and E[α2] = 1.

E[PY ] = NX2

[1

2+

(sin(2φ)

4φ

]+

N(N − 1)b2X2

2

[sin(φ)

φ

]+

N0

2(5.6)

In the above Equation 5.6, φ is the limit of distribution of phase error whereas b represents themode of α which is a Rayleigh random variable. Section 5.2 will discuss and analyze the resultsobtained using Equation 5.6, and Monte Carlo Simulation.

5.1.2 Probability of Error

This sections is dedicated to the calculation of “Bit Error Rate (BER)” of collaborative commu-nication. Let the information is transmitted using BPSK and the received signal Y depends uponαi and θr(i) which are independent random variables. Equation 5.2 shows the input Y given todecision circuit.

If two “i.i.d” random variables are added then by “central limit theorem”, the distribution of thesum tends to Gaussian for sufficiently large value of N. Received signal Y can also be shown assum of i.i.d as Y = s + n where, s =

∑Ni=1X cos(θr(i)). Since s and n are both i.i.d random,

therefore Y tends to Gaussian where the error rate of the communication system is given by

Pe = 0.5erfc

(µY√2σ2

Y

)(5.7)

“erfc” is a complementary error function. As θr(i) and αi are two different identical indepen-dent random variables and due to statistical independence of two random variables, therefore, theexpected or mean value µY of the received signal at the base station, we may have

µY = µs + µn (5.8)



By putting values in Equation 5.8, we get:

µY = E

[N∑i=1

αiX cos(θr(i))

]+ E[n]

= XN∑i=1

E [αi]E [cos(θr(i))] (5.9)

Now by putting values from Equation A.1(see Appendix for derivation) and values of E [αi], theabove relation in Equation 5.9 will become:

µY =

√πbNX sin(φ)√

2φ(5.10)

Here φ is the limit of distribution of phase error and b is the mode of αi, which is a Rayleigh fadedrandom variable.

Since adding two independent random variables resulted in Y , so its variance σ2Y is given by the

following formula

σ2Y = σ2

s + σ2n (5.11)

σ2Y = V ar

[N∑i=1

αiX cos(θr(i))

]+ V ar [n]

=N∑i=1

V ar [αiX cos(θr(i))] +N0

2(5.12)

Using values from Equation A.5(see Appendix for derivation), Equation 5.12 becomes

σ2Y = Nb2X2

[1− π

2

(sin(φ)

φ

)2

+sin(2φ)

2φ

]+N0

2(5.13)

Now by putting values from Equation 5.10 and 5.13, in Equation 5.7 we get the relation in



Pe = 0.5erfc

√πbNX sin(φ)√

2φ√2

(Nb2X2

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

]+ N0

2

) (5.14)

Equation(5.14)

Let X2 = Eb, we can simplify the probability of error given in Equation 5.14 into a form shown inEquation (5.15) as follows:

Pe = 0.5erfc

√πb sin(φ)√

2φ

√√√√√ N2(Eb/N0)(2Nb2

N

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

](EbN0

)+ 1

) (5.15)

Special Case: For each collaborative node, if the amplitude of each bit is given by X = ±√Eb/N

then Equation 5.15 can take a more simpler form shown in Equation 5.16:

Pe = 0.5erfc

√π sin(φ)√

2φ

√√√√√ (Eb/N0)(2b2

N

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

](EbN0

)+ 1

) (5.16)

In order to study the system behavior with reduced power consumption, the signal amplitudes ofall collaborative nodes need to be reduced by some factor. In this case a reduction of factor N canbe achieved using Eb/N in the total transmitted power. It is also used to analyze the impact on BitError Rate with the transmitted power.

5.2 Collaborative Communication: Energy Consumption in BAN

The proposed collaborative communication model is a multi transmitter system. Energy con-sumption in BAN is modelled in this section using collaborative communication and its naturalcounterpart single transmitter “SISO” system is presented. Due to differences in distance of eachcollaborative node with the base station, the received signal may arrive out of phase at the base



station. Required power for the network circuit operation, transmission distance between the basestation and collaborative nodes and total received power are parameters of this model. To measureand analyze energy efficiency in BAN a comparison of energy consumption in SISO and collabora-tive communication is presented where the phase synchronization is considered to have matchingerrors.

5.2.1 BAN: SISO Energy Consumption Model

In SISO systems total energy consumed by the system is obtained by adding the power consumedby transmitter (Ptr) and power consumed by receiver (Prv). Therefore, the Energy consumptionover a single bit can be obtained as follows:

ESISO = (Ptr + Prv)/Rs (5.17)

where Rs represents data transfer rate

It is shown in [48], for information transfer the required power calculation can be achieved usingsimplfied path loss model having a chance format also known as long-distance path loss, whichcapture the real essence of the propagated signal [47]. LetGt andGr be antenna gain for transmitterand receiver respectively and both Gt and Gr are equal to 1, then Ptr can be obtained as[47].

Ptr = Pcir +(4π)2Prd

β

dβ−2r λ2(5.18)

where Pcir represents the power requirement of transmitter circuit operation, Pr represents thesignal power, λ = c/f0, here c is “speed of light”, f0 represents “carrier frequency”, β is theexponent of path loss, receiver and transmitter are apart by a distance d, whereas dr represents thedistance used for “far-field region” as reference distance.

The minimum requirement of the received power Pr to get the desired bit error rate(BER) can becalculated as follows:

Pr = Ps + reber (5.19)

where Ps represents sensitivity of the receiver (in Watt) needed to get the desired BER in the



presence of AWGN only and reber isEb/N0 (in Watt) used to obtain the suitable BER in the presenceof noise and fading. The reber is calculated in [104], as

reber =((1− 2Pe)

2/1− (1− 2Pe)2)

(erfc−1(2Pe))2 (5.20)

where erfc−1 is the inverse function of the complimentary error function erfc(x) = 2√π

∫ +∞x

e−t2dt.

By putting values from Equations (5.18), (5.19) and (5.20), in Equation(5.17, the total energyconsumption in SISO systems can be achieved as

ESISO =

(Pcir +

(4π)2Psreberdβ

drβ−2λ2+ Prx

)/Rs (5.21)

5.2.2 BAN: Collaborative Communication Energy Consumption Model

Total energy consumption in the proposed developed model, can be represented as the sum ofenergy consumed during local communication by collaborative nodes El and the energy consumedby communication with the base station Et. Therefore, the overall energy consumption in case ofcollaborative communication can be represented as:

ECOL = El + Et (5.22)

Here both the channels among collaborative nodes and between base station and collaborativenodes are considered to be Rayleigh Faded. For distances among collaborative nodes, worst casehas been considered where the maximum distance is taken which gives maximum consumption incase of local communication, although the distance among collaborative nodes may be different.Energy consumption for local communication among collaborative nodes can be written as

El = (Ptr l +NPrv l)/Rs (5.23)

Here N represents the implanted collaborative nodes in BAN. Ptr l can be derived from Equation5.18 and will take the following form:



El =(Pcir +

(4π)2Psrrber ldβl

dβ−2l λ2+NPrv l

)/Rs (5.24)

The energy consumption for transmission between BAN and the base station can be calculated as:

Et = (Ptr t + Prv)/Rs (5.25)

where Ptr t represents the energy total which is consumed by all (N) collaborative nodes and canbe written as:

Ptr t = NPcir +(4π)2Pr td

β

Ndrβ−2λ2(5.26)

To achieve the required Bit Error Rate(BER), the minimum required received power Pr t can becalculated as follow:

Pr t = PS + rcol ber (5.27)

where rcol ber represents a “ratio”, for a system with phase error, between Eb/N0, Rayleigh fadingand AWGN. For a system with AWGN only, The ratio is between Eb/N0 and AWGN, to obtain thedesired BER.

The rcol ber may be represented as:

rcol ber =BER−1(Pe, N)

(erfc−1(2Pe))2(5.28)

in Equation 5.28, BER−1 denotes a function which is inverse of Equation (5.15).

Et =(NPcir +

(4π)2Pr tdβ

Ndβ−2r λ2+ Prv

)/Rs (5.29)

Combining values from Equations 5.24 and 5.29 in Equation 5.22, the total energy consumption incollaborative communication can be represented as follow:



ECOL =

((Pcir +

(4π)2Psrrber ldβl

dβ−2l λ2+NPrv l

)+(NPcir +

(4π)2Pr tdβ

Ndβ−2r λ2+Prv

))/Rs (5.30)

For collaborative communication energy saving is defined by the following equation

Esaving(%) =

(ESISO − ECOL

ESISO× 100

)%. (5.31)

Energy saving achieved by collaborative communication is dominated by the circuit energy con-sumption for small distances. Energy saving is 0 for the case whenESISO = ECOL and the distancein this case is known as break-even distance.

5.3 Results and Discussion

To analyze the proposed system, parameters of “off-the-shelf”( CC2420” [46] and “AT86RF21”[45])products are used. Various attributes and their values for each product are shown in Table 5.1.

Table 5.1: Product parameters, data and description used in the simulation experiments

Symbol Description AT86RF212 CC2420

- Modulation BPSK BPSKf0 Operating frequency 915MHz 2.45GHzRs Transmission data rate (BPSK) 40Kbps 250KbpsU Operating voltage (typical) 3V 3VIrx Currency for receiving states 9mA 17.4mAPrx Receiving power, Prx = UIrx 27mW 52.2mWIidle Currency for idle states 0.4mA 0.4mAPcir Electronic circuitry power, Pcir = UIidle 1.2mW 1.2mWPs Receiver sensitivity -110dBm 95dBm

To evaluate performance of the proposed system, Monte Carlo simulation has been performed.Results of analytical model (derived in this chapter) and simulation (existing formula) are com-pared for each figure of merit. The analysis is performed assuming the received signal are notperfect synchronized in phase. For the this purposed, four different uniform distributions of phase



124 H. Naqvi et al. / Physical Communication 3 (2010) 119–128

Fig. 3. Normalized average received power vs. no. of collaborative nodeswith Rayleigh fading.

our analysis four phase error distributions are consideredi.e. {−0.1π to 0.1π}, {−0.2π to 0.2π}, {−0.3π to 0.3π}and {−0.4π to 0.4π}. Fig. 3 shows the analytical and sim-ulation results of our model in the presence of Rayleighfading for normalized received power at the base sta-tion. Results show that analytical results match with thesimulation results. It is analyzed that received power de-creases by approximately 11% for phase error distributedover {−0.1π to 0.1π}, by 20% for phase error distributedover {−0.2π to 0.2π}, by 35% for phase error distributedover {−0.3π to 0.3π} and by 50% for phase error dis-tributed over {−0.4π to 0.4π} than the received powerwithout phase errors i.e. N2. Fig. 4 shows the analyti-cal and simulation results for total received power/N atthe base station in the presence of fading. The total re-ceived power/N increases as the number of collaborativenodes increases and total received power decreases as thephase error increases. It is observed that high power gaincan be achieved using collaborative communication in thepresence of phase errors and fading. The power gain of0.62N2 can be achieved if phase error is distributed over{−0.3π to 0.3π} and 0.48N2 can be achieved if phase er-ror is distributed over {−0.4π to 0.4π}.To perform BER analysis, we set the energy per bit of

each collaborative node to be Eb/N2, so the total energyused by all collaborative nodes is Eb/N . Figs. 5 and 6show analytical (calculated using Eq. (16) as given inBox II) and simulated results for BER with phase errorsdistributed over {−0.1π to 0.1π} and {−0.3π to 0.3π},respectively in the presence of AWGN and Rayleighfading for different number of collaborative nodes. It isconfirmed that analytical and simulated BER curves arevery close to each other for different phase error. Thesmall difference between analytical and simulation resultis due to approximation used for calculating BER. It isshown that BER decreases as the number of transmitterincreases, which is the confirmation of the fact thatcollaborative communication overcomes the fading effect.Form analytical and simulated results, it is shown that forN = 1, to achieve BER 10−3, 7 dB power is required in

Fig. 4. Average total received power/N vs. no. of collaborative nodeswithRayleigh fading.

Fig. 5. BER for phase error distributed over {−0.1π to 0.1π} for differentN with fading and total transmitted energy Eb/N .

AWGN case without fading, 24 dB power is required infading and for N = 5 required power is 11 dB for phaseerror is distributed over {−0.1π to 0.1π} and 13 dB forphase error is distributed over {−0.3π to 0.3π}.From Figs. 3 and 4 it is analyzed that power gain de-

pends upon N , however as the number of nodes increasesmore circuit energy is required. To analyze the energy ef-ficiency of our collaborative communication system westart our analysis by calculating the break-even distancefor different number of collaborative nodes for differentfrequency and phase errors. Circuit parameters are con-sidered from the off-the-shelf RF products i.e., CC2420 andAT86RF21. Maximum local distance between collaborativenodes is considered1mand the requiredBER is 10−5. Valueof path loss exponent α is 4.0–6.0 [20]. Product data andthe parameters used for calculation of energy efficiency areshown in Table 1.Energy saving for different number of collaborative

nodes and break-even distance is shown in Figs. 7–10. Itis analyzed that the break-even distance increases as thenumber of collaborative nodes increases. AT86RF212 has

Figure 5.4: Normalized average received power vs. number of collaborative nodes with Rayleighfading.

error are considered: i) “{−0.1π ∼ 0.1π}, ii) {−0.2π ∼ 0.2π}, iii) {−0.3π ∼ 0.3π} and iv){−0.4π ∼ 0.4π}”. This distribution is used in order to avoid a phase error of 180o which mayresult in destructive interference, as explained in chapter 3. A Rayleigh fading channel having theeffect of noise is used.

First of all the system is analyzed using received power figure of merit. In this analysis, normalizedaverage power both for the derived mathematical (analytical) model and simulated models arepresented in Figure 5.4 in the presence of noise and fading. The figure clearly shows that bothsimulation and analytical results are a match, which confirms accuracy of the proposed system.Secondly, it is also evident from the figure that, an increase in the number of slave/collaborativenodes have a positive impact on the received power. However, an increase in phase error seems tohave negative impact on the received power.

Furthermore, we can quantify the impact of phase error by drawing a straight line from the rightmost end where the graphs for different phase error distribution terminates, back to the y-axisrepresenting normalized average power. Then subtract the value given by the line from 0.95 andcalculate its percentage, gives us the approximate decrease in the received power. Following thisapproach, an approximate decrease in case of {−0.1π ∼ 0.1π} is 13%, in case of {−0.2π ∼ 0.2π}the decrease is 21%, in case of {−0.3π ∼ 0.3π} the decrease is 34%, whereas in case of {−0.4π ∼0.4π}, the decrease in received power reaches a maximum of 51%. This decrease in the received




Fig. 3. Normalized average received power vs. no. of collaborative nodeswith Rayleigh fading.

our analysis four phase error distributions are consideredi.e. {−0.1π to 0.1π}, {−0.2π to 0.2π}, {−0.3π to 0.3π}and {−0.4π to 0.4π}. Fig. 3 shows the analytical and sim-ulation results of our model in the presence of Rayleighfading for normalized received power at the base sta-tion. Results show that analytical results match with thesimulation results. It is analyzed that received power de-creases by approximately 11% for phase error distributedover {−0.1π to 0.1π}, by 20% for phase error distributedover {−0.2π to 0.2π}, by 35% for phase error distributedover {−0.3π to 0.3π} and by 50% for phase error dis-tributed over {−0.4π to 0.4π} than the received powerwithout phase errors i.e. N2. Fig. 4 shows the analyti-cal and simulation results for total received power/N atthe base station in the presence of fading. The total re-ceived power/N increases as the number of collaborativenodes increases and total received power decreases as thephase error increases. It is observed that high power gaincan be achieved using collaborative communication in thepresence of phase errors and fading. The power gain of0.62N2 can be achieved if phase error is distributed over{−0.3π to 0.3π} and 0.48N2 can be achieved if phase er-ror is distributed over {−0.4π to 0.4π}.To perform BER analysis, we set the energy per bit of

each collaborative node to be Eb/N2, so the total energyused by all collaborative nodes is Eb/N . Figs. 5 and 6show analytical (calculated using Eq. (16) as given inBox II) and simulated results for BER with phase errorsdistributed over {−0.1π to 0.1π} and {−0.3π to 0.3π},respectively in the presence of AWGN and Rayleighfading for different number of collaborative nodes. It isconfirmed that analytical and simulated BER curves arevery close to each other for different phase error. Thesmall difference between analytical and simulation resultis due to approximation used for calculating BER. It isshown that BER decreases as the number of transmitterincreases, which is the confirmation of the fact thatcollaborative communication overcomes the fading effect.Form analytical and simulated results, it is shown that forN = 1, to achieve BER 10−3, 7 dB power is required in

Fig. 4. Average total received power/N vs. no. of collaborative nodeswithRayleigh fading.


AWGN case without fading, 24 dB power is required infading and for N = 5 required power is 11 dB for phaseerror is distributed over {−0.1π to 0.1π} and 13 dB forphase error is distributed over {−0.3π to 0.3π}.From Figs. 3 and 4 it is analyzed that power gain de-

pends upon N , however as the number of nodes increasesmore circuit energy is required. To analyze the energy ef-ficiency of our collaborative communication system westart our analysis by calculating the break-even distancefor different number of collaborative nodes for differentfrequency and phase errors. Circuit parameters are con-sidered from the off-the-shelf RF products i.e., CC2420 andAT86RF21. Maximum local distance between collaborativenodes is considered1mand the requiredBER is 10−5. Valueof path loss exponent α is 4.0–6.0 [20]. Product data andthe parameters used for calculation of energy efficiency areshown in Table 1.Energy saving for different number of collaborative

nodes and break-even distance is shown in Figs. 7–10. Itis analyzed that the break-even distance increases as thenumber of collaborative nodes increases. AT86RF212 has

Figure 5.5: Average received power/N vs. number of collaborative nodes with Rayleigh fading.

power is recorded in comparison to the signal power at the receiver without phase errors i.e. N2.

Another comparison between simulation results and analytical results has been shown in Figure5.5, for total received “power/N” (average power) at the receiver (at station). It is revealed that thetotal average received “power/N” at the receiver behaves as a function of the collaborative nodes,where the an increase in the number of collaborative transmitters have positive, while an increasein the phase error has negative impact on received power gain. It can be seen that a high gain inreceived power is achievable using collaborative communication in a Rayleigh faded channel evenin the presence of phase error in the received signals. To quantify, a gain of 0.62N2 in receivedpower observed for a phase error interval over {−0.3π ∼ 0.3π} and a gain of 0.48N2 in receivedpower is achievable for a phase error interval distributed over interval of {−0.4π to 0.4π}.

To analyze the proposed system in terms of Bit Error Rate (BER), a comparison between analyt-ical (computed using equation 5.15 and 5.16) and simulated results is performed. Let Eb/N2, isthe energy expense over transmission of single bit from source to destination. Therefore, energyconsumed by all collaborative nodes is Eb/N . Analytical results of values computed using theabove mentioned equation, are depicted in Figure 5.6 and Figure 5.7 against the simulated re-sults. To show the overall trend only two phase error interval distribution {−0.2π ∼ 0.2π} and{−0.4π ∼ 0.4π}, are considered. The results are computed, considering Rayleigh fading channelwith noise (AWGN) for different number of collaborative nodes.

It can be observed that the simulation and analytical curves are nearly a match. The small mismatch



−2 0 2 4 6 8 10 12 14 1610

−5

10−4

10−3

10−2

10−1

Eb/No, dB

Bit

Err

or R

ate

AWGN onlyAnalytical N=1Analytical N=5Simulation N=5Analytical N=7Simulation N=7Analytical N=9Simulation N=9Analytical N=11Simulation N=11

Figure 5.6: Bit Error Rate over interval {−0.2π to 0.2π} for different number of nodes with fadingand total transmitted energy

between analytical and simulated results could be the cause of approximation used in derivationof BER. Furthermore, it is also clear that BER behaves inversely to the number of transmitters. Itmeans an increase in collaborative transmitters will reduce the error rate, confirming that collabo-rative communication successful in mitigating the effect of fading in the channel. It can be seen,both from simulated and analytical results, for a single collaborative node (N = 1), achieving aBER of 103 requires a power of 7dB in case when no fading considered but only noise (AWGN)is taken into account. The power requirement raises to 24dB if fading effect is included. If thenumber of collaborative nodes is raised to five (N = 5), and the phase error distributed over“{−0.1π ∼ 0.1π}”, a power of 11dB is required whereas this power requirement raises to 13dB

for a phase error over {−0.3π ∼ 0.3π}.

An analysis from Figure 5.4 and Figure 5.5 clearly indicates the dependency of gain in power onnumber of collaborative nodes. However an increase in number of collaborative nodes incorporatemore energy consumption in form of circuit energy. Therefore, it is mandatory to explore the trade-offs between the number of transmitters incorporated for collaboration and overall energy of thesystem and well as transmission distances. For this purpose, break-even distances are calculatedin order to analyze energy consumption of the proposed collaborative communication system, asshown in Table 5.2.



−2 0 2 4 6 8 10 12 14 1610

−5

10−4

10−3

10−2

10−1

Eb/No, dB

Bit

Err

or R

ate

AWGN only Analytical N=1Simulation N=7Analytical N=7Simulation N=9Analytical N=9Simulation N=11Analytical N=11

Figure 5.7: Bit Error Rate over interval {−0.4π to 0.4π} for different number of nodes with fadingand Eb/N total transmitted energy

Break-even distance is a distance where energy consumption of a collaborative communicationbecomes equal to the single transmitter (SISO) system. These distances are calculated for dif-ferent phase error values using varying number of collaborative nodes. To analyze circuit energyconsumption, simulation is performed using parameters of “off-the-shelf” products i.e. “CC2420”[46] and “AT86RF21” [45]. The collaborative nodes are considered to be at most 1m apart fromeach other with accept BER value of 10−5 and “path loss exponent” β is 4.0–6.0 [158]. Informationabout the products and their parameters are shown in Table 5.1.

The next series of Figures 5.8–5.11 shows percentage energy preservation for varying number oftransmitter and “break-even” distances. Analysis reveals that break-even distance is “directly” re-lated to the number of transmitters nodes. It is further analyzed that, “AT86RF212” has greater“break-even” distance than “CC2420”, but “AT86RF212” is more energy efficient than “CC2420”at a distance of “100m and 200m”. An exponential energy saving growth for AT86RF212 hasbeen observed compared to CC2420 as increase in distance exceeds the break-even distance, re-flected from Figures 5.8–5.10. It has further been observed that AT86RF212 stabilizes earlier thanCC2420. AT86RF212 gets stable at a distance of 120m whereas CC2420 achieves stability at adistance approximately equal to 150m.




Table 1Product data and parameters.

Symbol Description AT86RF212 [21] CC2420 [22]– Modulation BPSK BPSKf0 Operating frequency 915 MHz 2.45 GHzRs Transmission data rate (BPSK) 40 Kbps 250 KbpsU Operating voltage (typical) 3 V 3 VIrx Currency for receiving states 9 mA 17.4 mAPrx Receiving power, Prx = UIrx 27 mW 52.2 mWIidle Currency for idle states 0.4 mA 0.4 mAPcir Electronic circuitry power, Pcir = UIidle 1.2 mW 1.2 mWPs Receiver sensitivity −110 dB m −95 dB m


Fig. 7. Energy saving and break-even distance with phase error{−0.1π to 0.1π} for different N for product AT86RF212.

larger break-even distance than CC2420. But the energysaving in AT86RF212 is greater than CC2420 at distance100 m and 200 m. From Figs. 7–10 it is observed thatenergy saving is growing very fast for AT86RF212 thanCC2420 as the distance increases then the break-evendistance. AT86RF212 achieves the steady state earlier thanthe CC2420. AT86RF212 achieves the steady statewhen thedistance is nearly 120 m and CC2420 achieves steady statewhen the distance is approximately equal to 150 m.


Fig. 9. Energy saving and break-even distance with phase error{−0.1π to 0.1π} for different N for product CC2420.

The break-even distance for CC2420 and AT86RF212is summarized in Table 2. It is analyzed that as thedistance increases the energy saving using collaborativecommunication also increases and that after a certaindistance it becomes constant steady. The summary ofenergy saving for different phase errors at distance 100 mand 200 m for CC2420 and AT86RF212 are shown inTables 3 and 4.

Figure 5.8: Break-even distances and energy preservation in the presence of phase error distributedover {−0.3π ∼ 0.3π} using different varying number of transmitter nodes N for AT86RF212










Figure 5.9: Break-even distances and energy preservation in the presence of phase error distributedover {−0.4π ∼ 0.4π} using different varying number of transmitter nodes N for AT86RF212












Figure 5.10: Break-even distances and energy preservation in the presence of phase error dis-tributed over {−0.3π ∼ 0.3π} using different varying number of transmitter nodes N for CC2420



Table 2Break-even distance for CC2420 and AT86RF212.

N Break-even distance CC2420 (m) Break-even distanceAT86RF212 (m)

2 39 43.53 43 46.54 46.2 495 49 516 51 52.57 53.5 53.59 57 5611 59.7 58

5. Conclusions

Wehavepresented collaborative communicationmodelfor sensor networks with imperfect phase synchronization

in the presence of AWGN and Rayleigh fading. The theo-retical analysis of the system is presented and the expres-sions for received power are derived and expressed as thefunction of the number of collaborative nodes. Then theprobability of error for both AWGN and for Rayleigh fadingchannel are derived and expressed as a function of signal tonoise ratio for the number of collaborative nodes as param-eter. To investigate the energy saving, energy consump-tionmodel is presented and expression for energy saving isderived.It is concluded that a significant power gain can be

achieved by increasing the number of collaborative nodeswithout increasing the total transmitted power. This is theconsequence of the collaborative communication that canbe considered as a space diversity system. By collabora-tion the fading in the channel is efficiently mitigated. It isanalyzed that the proposed collaborative communicationsystem outperforms the SISO systems in terms of energyefficiency at medium/long transmission distance (greaterthan break-even distance). Results revealed that usingcollaborative communication 99% energy can be savedwith imperfect frequency and phase synchronization. Thebreak-even distance increases as the number of collabo-rative nodes increases. It is also concluded that collabora-tive communication is useful when the distance betweentransmitters and base station is greater than break-evendistance.The analysis of the impact of imperfect frequency

synchronization and time synchronization can also be afuture research direction.

Appendix

AsΘf has uniform distribution from {−ϕ to ϕ}.

Table 3Energy saving (%) for CC2420.

N Phase error 0.1π Phase error 0.2π Phase error 0.3π Phase error 0.4π200 m 100 m 200 m 100 m 200 m 100 m 200 m 100 m

2 97.7 95 97.5 94.5 96.5 93.8 94.5 91.53 99.1 95.5 99 95 98.8 95 98.5 94.24 99.6 94.5 99.4 94 99.2 94 99 945 99.4 93 99.5 93 99.3 93 99.1 936 99.3 92 99.2 92 99.3 92 99.1 927 99.2 91 99.1 91 99.2 91 99 919 99.1 89 99 89 99 90 99 8911 98.9 86.5 98.5 86 98.1 89 98 86.5

Table 4Energy saving (%) for AT86RF212.


2 98 97 97.7 96.7 96.5 95.6 95 943 99.4 98 99.2 98 99.1 97.8 98.8 97.54 99.7 98 99.7 98.2 99.6 98 99.4 97.75 99.8 97.6 99.7 97.5 99.7 97.6 99.6 97.56 99.8 97.4 99.8 97.3 99.7 97.5 99.7 97.37 99.8 97.2 99.8 97.2 99.8 97.2 99.8 949 99.8 96.5 99.9 96.5 99.8 96.5 99.8 96.511 99.9 96 99.9 96 99.8 96 99.8 96

Figure 5.11: Break-even distances and energy preservation in the presence of phase error dis-tributed over {−0.4π ∼ 0.4π} using different varying number of transmitter nodes N for CC2420



Table 5.2: Break-even distance based on the parameters of the devices used in experiments i.e.CC2420 and AT86RF212.

N Break-even distance CC2420(m) Break-even distance AT86RF212(m)

2 39.5 43.53 42.5 45.54 46.3 49.15 48.9 50.96 51.2 52.77 53.4 53.49 56.9 55.910 59.9 58.2

Table 5.2 summarizes the break-even distances for both the products i.e. CC2420 and AT86RF212.It is analyzed that, prior to becoming constantly stable at some specific distance, the energy savingfor collaborative communication increases with increase in distance. Energy saving percentage forboth the products with different values of phase errors over distances of 100m and 200m has beenshown in Table 5.3 and Table 5.4.

Table 5.3: Percentage energy preservation based on the parameters of CC2420.

N phase error 0.1π phase error 0.2π phase error 0.3π phase error 0.4π

200m 100m 200m 100m 200m 100m 200m 100m

2 97.8 95.1 97.6 94.6 96.7 94 94.6 91.73 99 95.4 99.1 95.1 98.9 95.1 98.7 94.34 99.6 94.6 99.4 94 99.2 94 99 94.15 99..5 93.1 99.5 93 99.3 93 99.1 936 99.3 92 99.2 92.1 99.3 92 99.1 927 99.3 91.1 99.2 91 99.2 91 99 919 99.1 89 99 89 99 90 99.1 8911 98.8 86.4 98.7 86.2 98.3 89.2 98.1 86.6

A column wise (top down) analysis of the tables show that, percentage energy savings are increas-ing with increase in the number of transmitter nodes. A sudden but slight decrease in energy savingmay be noted while going down in a specific column. This may be the effect of incorporating moretransmitter, thereby the circuit power consumption contributes negatively to the overall energyconsumption of the system. Similarly row wise (left to right) analysis of the tables show that in-creasing phase error has adverse effect on the energy savings in case of both product. However,



Table 5.4: Percentage energy preservation based on the parameters of AT86RF212.


200m 100m 200m 100m 200m 100m 200m 100m

2 98.1 97.1 97.7 94.5 96.6 95.5 95.1 94.13 99.4 98 99.1 94.9 98.2 98 98.9 97.64 99.6 97.9 99.7 94 98.2 98 99.4 97.75 99.8 97.6 99.7 93 97.6 97.7 99.6 97.56 99.8 97.4 99.8 92 97.5 97.7 99.7 97.37 99.8 97.2 99.8 91 97.3 97.3 99.8 949 99.8 96.5 99.9 89 96.4 96.4 99.8 9611 99.9 96 98.9 86 96 96 98.8 96.5

the overall analysis reveals that, AT86RF212 is more energy efficient than CC2420.

5.4 Chapter Summary

A collaborative communication model for BAN has been presented in case of out-of-phase re-ceived signal in the presence of noise and Rayleigh fading. A mathematical relation has beenderived. Analysis shows that the cumulative power received at the BS increases as the number ofcollaborative nodes increase. A mathematical expression in case of AWGN and Rayleigh fading,for the probability of error has been derived which is expressed as a function of SNR (Eb/No), fornumber of collaborative nodes. An energy consumption model is presented to analyze the energyefficiency.

If number of collaborative nodes is increased, a large gain in received power is achieved withoutchanging the transmission power. It is proved that channel fading can be effectively mitigatedusing collaborative communication. A performance analysis at medium and long distances(greaterthan break-even distance), shows that collaborative communication system beats the SISO systemsin terms of energy efficiency. Results revealed that even with imperfect phase synchronizationcollaborative communication can save 99% energy. As the number of collaborative nodes increaseso as the break-even distance but collaborative communication is still useful even if the distancebetween BS and transmitter nodes is more than the break-even distance.


Chapter 6

Collaborative Communication in MultipathFading

This chapter analyzes collaborative communication in wireless sensor networks, using multipath[31, 159] environment in the presence of Rayleigh fading, to achieve energy efficient communica-tion. In collaborative communication, each collaborative node sends a copy of the same informa-tion to a common sink (base station) simultaneously [123]. Each collaborative node transmits thepreshared data to the base station. Unlike cooperative communication where the transmitted in-formation is relayed by intermediate cooperative nodes [160], in collaborative communication allnodes communicate directly with the base station. Study of the literature reveals that if the trans-mitted signals could be synchronized in time, phase and frequency, a significant gain in receivedpower may be achieved at the receiver [98, 115, 117, 131, 148, 152, 161–163]. But with collab-orative communication such high gain in received power is achievable, even if, phase, frequencyand time synchronization is imperfect [118].

In a multipath system, a signal transmitted by each node, may scatter into multiple componentswhen it stricks with various objects in the propagation path. The accumulation of all these com-ponents may result in even more power gain as compared to single-path systems. Mathematicalmodels for received power, BER and energy consumption are developed based on collaborativecommunication using multipath scenario. Each of these theoretical models are compared withsimulated results obtained using existing formula.

77

Chapter 6. Collaborative Communication in Multipath Fading

6.1 Proposed System Model for Multipath Collaborative Com-munication

In a fixed array antenna systems using collaborative communication, all components (direct plus

scatter) of a received signal at base station are added synchronously due to common networkcontroller; responsible for synchronization. The controller knows how far each transmitter node isapart from other, and also the antennas are fixed, so location of transmitter is determined accurately,leaving out any chances of error in phase of the added signal at the base station [152, 153]. Thiscoherent addition of incoming signals and its components result in a significant gain in receivedpower, as well as reduction in BER.

However, in distributed environment, antenna array is not fixed as well as the absence of centralnetwork controller, making it infeasible to apply the theory of fixed antenna array directly. As aconsequence, an error due to displacement may occur while estimating position of the transmitter.This displacement error results in out-of-phase reception of the received signal at the base station.Therefore in collaborative communication, received signal needs to be synchronized in time, phaseand frequency to yield gain in the received power.

N2

N3

N1

N

Base Station

Collaborative Nodes

Sensor Nodes

Sensor Field

Figure 6.1: Multipath system geometry where the dotted lines represent multiple scatter compo-nents of a signal

Location estimation gives most probable location rather than exact location of a transmitter, there-for it is very difficult to fully synchronize the system. Therefore, a general collaborative architec-ture where each node uses same carrier frequency, has been proposed in this chapter, as shown inFigure 6.1. The information intended to be transferred to Base Station(BS), is first shared amongthe collaborative nodes. Once all theses nodes get synchronized with each other, they collabora-



Node1

Node2

NodeN

ym(t)

2cos(2π fot)

LPF

Calculate Power

T Interval Integeration


n(t)

X(t1)

X(t2)

X(tm)

αm

X(t1)

X(t2)

X(tm)

X(t1)

X(t2)

X(tm) Base Station

Figure 6.2: Proposed Theoretical System Model for Multipath Collaborative Communication

tively start transmitting this information towards the BS.


For a system of “N” collaborative nodes, a detailed system structure used in theoretical modelingis shown in Figure 6.2. In a system of N randomly deployed collaborative nodes, transmittingtowards a Base Station where the distance varies between the base station and each node. Thesignal components received at base station are deflected from different objects while travelingfrom source to BS, which causes signal from a node to scatter(dotted lines in Figure 6.1) intomultiple components with no line-of-sight component.

An information signal x(t) transmitted by each node, splits into M scatter components due todeflection from different objects during the transit. Since each node is at different distance fromthe BS and each scatter component arrives over a different path. In addition, since the nodes aredeployed randomly, therefore, there is a chance of error in calculating distances between collabo-rative nodes and the BS [118]. If d0 is the “initial” displacement between each collaborative nodeand the BS, f0 is the carrier frequency, then phase produced due to d0 is given by θ0 = 2πf0d0/c,where c is speed of light. Similarly if dr(i) is the displacement error due to variations in distances



between collaborative nodes and BS as well as due to delay in arrival of different scatter compo-nents of the same signal at BS. This displacement error when transformed into phase error, resultsin a phase of θr(i) = 2πf0dr(i)/c. It should be noted that the mathematical concept and derivationis based on the mathematical modelling in John G. Proakis [157].

If we consider cos(2πf0t) the modulation signal and signal delay is negligible in comparison to bitlength, which diminishes the chances of inter symbol interference by producing enough guard in-terval. The cumulative received signal from collaborative nodes each with M scatter components,at the receiver is given by the following equation.

ym(t) = <

(N∑i=1

αix(t)ej(2πf0t+θ0+θr(i))

)+ n(t) (6.1)

Here αi is the “fading effect” some times also called channel response and n(t) is AWGN.

Signal in Equation(6.1) is demodulated, integrated at the base station and then passed through LPF,yields the following result.

Y = LOS +N∑i=1

M∑j=1

αijXj cos(θr(ij)) + n (6.2)

Here X = ±√E and n represent signal and noise amplitudes respectively.

Since LOS = 0 in case of Rayleigh fading therefore, received signal power at the receiver (basestation) can be achieved as follows:

PY =

[N∑i=1

M∑j=1

αijXj cos(θr(ij)) + n

]2(6.3)

Equation 6.3, consists of multiple identical independent random variables, like θr(ii), αij and zeromean noise n. Therefore, expected value of the received power in Equation 6.3 is computed andthe above equation is expanded as follows:



E[PY ] = E

[N∑i=1

M∑j=1

αijXj cos(θr(ij))

]2+

E

( N∑i=1

M∑j=j

αijXj cos(θr(ij))

) N∑k=1k 6=i

M∑l=1l 6=j

αklXl cos(θr(kl))

+ σ2

n (6.4)

σ2n in Equation 6.4, represents variance of noise. The sum of M scatter components must not be

greater than the original signal i.e,∑M

j=1Xj ≈ X , we consider the maximum value here i.e, X.Now expanding Equation 6.4, will take the following form:

E[PY ] = X2

N∑i=1

M∑j=1

E [αij cos(θr(ij))]2 +X2

(N∑i=1

M∑j=j

E [αij cos(θr(ij))]

) N∑k=1k 6=i

M∑l=1l 6=j

E [αkl cos(θr(kl))]

+ σ2n (6.5)

Since θr(i), θr(j), θr(k), θr(l), αi, αj , αk and αl, are “i.i.d” random variables, therefore “E[θr(i)] ≈E[θr], E[θr(j)] ≈ E[θr], E[θr(k)] ≈ E[θr], E[θr(l)] ≈ E[θr] and E[αi] ≈ E[α], E[αj] ≈ E[α],E[αk] ≈ E[α], E[αl] ≈ E[α]. Also θr(i), θr(j), θr(k) and θr(k)” are i.i.d distributed uniformlyover the interval “{−φ → φ}” for all i,j,k and l, then θr(i) ≈ θr, θr(j) ≈ θr, θr(j) ≈ θr andθr(j) ≈ θr.

E[PY ] = X2

[MNE

[α2]E[cos2(θr)

]+

MN (E [α]E [cos(θr)])MN(MN − 1)E [α]E [cos(θr)]

]+ σ2

n (6.6)

We know that variance V ar(α) = σ2α =

(2− π

2

)and mean E[α] = µα =

√π2. Using values from

Equation A.1 and Equation A.2; Equation 6.6 can be simplified as follows, considering mean valueof α2 i.e, E[α2] = 1.



E[PR] = N × MX2

[(1

2+

sin(2φ)

4φ

)+πMN(MN − 1)b2

2

(sin(φ)

φ

)2]

+N0

2(6.7)

Here, φ is the limit of distribution of phase error whereas b represents the mode of α which is aRayleigh random variable. It is clear form Equation 6.7, that not only the number of collaborativenodes “N”, but also the number of scatter components “M”, contributes to the gain in receivedpower.

6.1.2 Average Probability of Error

If BPSK is used to modulate the information signal transmitted over a Rayleigh faded channel. Thereceived signal Y , depends on two identical independent random variables α and θr. We wish tocalculate the error rate in bits as given in Equation 6.2. By central limit theorem, the sum of twoindependent random variables tends to Gaussian if value of N is sufficiently large. The receivedsignal Y = s + n, is the sum of two i.i.d’s where, Y =

∑Ni=1

∑Mj=1X cos(θr(ij)) (in internal

summation j 6= i). Therefore, Y tends to Gaussian and the error rate is given by the followingrelation.

Pe = 0.5erfc

(µY√2σ2

Y

)(6.8)

Since θr and α are two different i.i.d random variables, therefore, the average value of the receivedsignal is obtained by computing average values of the transmitted signal (represented by µs) andthe average value of the noise (represented µn) added to it, during propagation. So the averagereceived value µY is computed as follows:

µY = µs + µn (6.9)

Putting values of µs and µn in Equation 6.9, we get



µY = E

[N∑i=1

M∑j=1

αijXj cos(θr(ij))

]+ E[n]

= X

N∑i=1

E [αii]E [cos(θr(ii))] +X

N∑i=1

M∑j=1j 6=i

E [αij]E [cos(θr(ij))] (6.10)

The above equation produces different values depending upon the value of N and M. It producesone value in case when N =M another for N > M and a different value when N < M .

Taking values from Equation A.1 and putting value of E [α], the relation in Equation 6.10, canbecome,

µY =

√πbNMX sin(φ)√

2φ(6.11)

Here φ represents distribution of phase error limit and b is the mode of αi, which is a Gaussiandistributed zero mean random variable.

Y is identical independent random variable because it is the sum of two i.i.d random variables, soits variance can be calculated as follows:

σ2Y = σ2

s + σ2n (6.12)

σ2Y = V ar

N∑i=1

M∑j=1j 6=i

αijX cos(θr(ij))

+ V ar [n]

=N∑i=1

V ar [αiiX cos(θr(ii))] +N∑i=1

M∑j=1j 6=i

V ar [αijX cos(θr(ij))] +N0

2(6.13)

putting values from Equation A.5, Equation 6.13 will become



σ2Y = NMb2X2

[1− π

2

(sin(φ)

φ

)2

+sin(φ)

2φ

]+N0

2(6.14)

Now taking values from Equation 6.11 and Equation 6.14, in Equation 6.15 we get the followingrelation

Pe = 0.5erfc

√πbNMX sin(φ)√

2φ√2

(NMb2X2

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

]+ N0

2

) (6.15)

As X2 = Eb, the error rate relation in Equation 6.15, can be simplified into the following form

Pe = 0.5erfc

√πb sin(φ)√

2φ

√√√√√ N2M2(Eb/N0)(2NMb2

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

](EbN0

)+ 1

) (6.16)

Special Case:

If X = ±√Eb/N is the amplitude of the signal from each transmitter node, multiplying the 2 with

both the terms in denominator and dividing both numerator and denominator by No to representthe above equation in the form of SNR (Eb/No). Therefore, the relation in Equation 6.16, canfurther be simplified as follows:

Pe = 0.5erfc

√πb sin(φ)√

2φ

√√√√√ (Eb/N0)(2b2

NM

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

](EbN0

)+ 1

) (6.17)

The above equation is developed using transmitted power X by each transmitter node divided bythe number of total nodes. In order to analyze the system behavior under low power consumption,the transmitted power of each collaborative node is reduced by a factor of N . Then power trans-mitter by all collaborative nodes is constructive added at the receiver to compute total transmittedpower. The same approach is also used to analyze error rate in the transmitted power. It is clear



from Equation 6.17, that an increase in the number of collaborative nodes N as well as multi-path components M , reduces the required power transmitted by each node i.e,

√Eb/N , thereby

prolonging a node’s life.

6.2 Collaborative Communication: Energy Efficiency

Energy consumption models are presented for collaborative communication and “SISO” systemsin this section. Signals are received out of phase at the base station. For total energy consumption,circuit power required for the operation of the network, the distance by which transmitter andbase station are apart, and the total received power are used as parameters. To analyze energyefficiency of the proposed system, energy consumption of SISO and collaborative communication

are compared.

6.2.1 SISO Systems: Energy Consumption Model

Total energy consumption in SISO system is given by the sum of energy consumption of the trans-mitter (Ptr) and energy consumption of the receiver (Prv). Therefore, energy consumed per bit isgiven by the following relation:

ESISO = (Ptr + Prv)/Rs (6.18)

Rs represents the rate of information exchange.

It is shown in [48] that for information transmission the required power can be estimated usingsimplified path loss(long distance) model. If gain of transmitter antenna is Gt, and that of receiverisGr and gain for both receiverGr and transmitterGt are equal to 1, then total power consumptionPtr of the system is given by [47].


β

dβ−2r λ2(6.19)

Power required for transmitter circuit operation is represented by Pcir, power of received signal Pr,λ = c/f0, where c represents speed of light, f0 represents carrier frequency, β represents exponent



of path loss, d represents distance between receiver and transmitter, dr represents the referencepoint in terms of distance for the case of “far-field region”.

The minimum required received signal power Pr, for calculation of the desired bit error rate is asfollow:


Ps denotes sensitivity of the receiver (in Watt), required to obtain the desired BER in an AWGN

channel only, and “reber is Eb/N0 (in Watt)” used to obtain desired BER for a Rayleigh fadedAWGN system. reber in [104] is calculated as

reber =((1− 2Pe)

2/1− (1− 2Pe)2)

(erfc−1(2Pe))2 (6.21)

erfc−1 is complimentary inverse error function erfc(x) = 2√π

∫ +∞x

e−t2dt.

Taking values from Equation 6.19, Equation 6.20 and Equation 6.21, total consumption of energyof SISO systems can be obtained using the following equation

ESISO =

(Pcir +

(4π)2Psreberdβ

drβ−2λ2+ Prv

)/Rs (6.22)

6.2.2 Collaborative Communication: Energy Consumption Model

Energy consumption using SISO model is given by the sum of energy consumed in node-to-nodecommunication El (local communication) and energy consumed by communicating with BS Et.Energy consumption of both local communication and communication with the BS may be repre-sented as following:

ECOL = El + Et (6.23)

The channel between a each transmitter node and the BS are considered to be Rayleigh faded. Thedistance among collaborative nodes is considered to be maximum (local communication) leading



to maximum energy consumption, although distance of collaborative nodes from BS may vary.Energy consumption in case of local communication can be represented as following:

El = (Ptr l +NPrv l)/Rs (6.24)

Here “N” is the number of collaborative transmitter nodes in the sensor network. Ptr l can bederived from Equation 6.19, can be written in the following form

El =(Pcir +

(4π)2Psrrber ldβl

dβ−2l λ2+NPrv l

)/Rs (6.25)

Energy consumed during the communication between the BS and each collaborative transmitternode is given by

Et = (Ptr t + Prv)/Rs (6.26)

Ptr t is total energy, which is consumed by all (N) collaborative nodes and can be rewritten as


β

Ndrβ−2λ2(6.27)

To obtain the desired “Bit Error Rate (BER)”, the minimum required received power Pr t may bewritten as


In the above equation rcol ber represents ratio between Eb/N0 (for a system with imperfect phase),“AWGN” and “Rayleigh” fading. For systems with “AWGN” only, this ratio is equal to “Eb/N0 (inWatt)”, to achieve the desired BER. rcol ber can be re-written as


(erfc−1(2Pe))2(6.29)

Here BER−1 is function which is inverse of Equation 6.16. Therefor Et can be written as follow



Et =(NPcir +

(4π)2Pr tdβ

Ndβ−2r λ2+ Prv

)/Rs (6.30)

Taking values from Equation 6.25 and Equation 6.30, total energy consumption of collaborativecommunication can be represented as follows:

ECOL =

((Pcir +

(4π)2Psrrber ldβl

dβ−2l λ2

)+

(NPrv +NPcir +

(4π)2Pr tdβ

Ndβ−2r λ2+ Prv

))/Rs

(6.31)

The total energy saving for the proposed collaborative communication model can be achieved usingthe following equation.

Esaving(%) =

(ESISO − ECOL

ESISO× 100

)%. (6.32)

Energy saving for small distances is dominated by circuit energy consumption. Saving is 0 in casewhere ESISO = ECOL, and the distance in this case is known as “break-even-distance”.

6.3 Results and discussion

For analysis of the proposed system, Monte Carlo simulation is used. Analysis is performed withthe same parameter values as in chapter 5. The distribution of phase error is uniform over theinterval “{−φ ∼ φ}”. To analyze the system with different phase error values, four different phases“{−0.1π ∼ 0.1π}, {−0.2π ∼ 0.2π}, {−0.3π ∼ 0.3π} and {−0.4π ∼ 0.4π}” are used. Oneadditional parameter has been added to the experiment; the number of scatter components. All theexperiments are performed, to analyze the effect of multipath scattering, on energy consumptionusing collaborative communication in WSN.

Results for the normalized received power for the proposed model are shown in Figure 6.3. It isclear from the results that both simulation and analytical results are a match. In comparison tothe normalized and average received power plots in chapter 5, a slight improvement in both can beseen. This improvement in received power is clearly the result of multipath components, mitigating



2 2.5 3 3.5 4 4.5 5 5.5 60.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95Received Power− Noise Power for Noise Power 0.1 Watt

No. of Nodes ( N )

(E[P

r]/N

) (W

atts

)

Phase Error = 0.1Phase Error = 0.2Phase Error = 0.3Phase Error = 0.4

Figure 6.3: Normalized received signal power vs. number of transmitter nodes in the presence ofRayleigh fading and phase error

the fading effect. It confirms that even if some of the components a signal are blocked due to fading,a few of its component may be successful in reaching the receiver, not only mitigating the fadingeffect but also contributing to the received signal strength.

Figure 6.3, shows an approximate decrease of 10% in case of phase error from {−0.1π ∼ 0.1π},19% in case of phase error from {−0.2π ∼ 0.2π}, 33% in case of phase error from {−0.3π ∼0.3π} and 49% when the phase error is from {−0.4π ∼ 0.4π} in comparison to the signal powerat receiver with no error in phase i.e. N2. Figure 6.4 shows another comparison in case of totalaverage received power power/N at the receiver between simulated and analytical results.

Figure 6.4 depicts that the average received power “power/N” increases with a raise in the numberof collaborative transmitter nodes and vice versa. It can be seen that the total received power andBER are directly related with the number of nodes and are also inversely proportional to the phaseerror. It is clear that using collaborative communication in Rayleigh fading produces significantgain in the received power. A power gain of 0.64N2 is recorded when the interval of phase erroris {−0.3π to 0.3π} and a power gain of 0.49N2 is recorded when the interval of phase error is“{−0.4π to 0.4π}”.

To analyze “Bit Error Rate (BER)”, Eb/N2 is set to be the energy expense per bit. So the energyconsumed by all collaborative nodes is Eb/N . Analytical results of values calculated in Equation



2 3 4 5 61

2

3

4

5

6

7

No. of Collaborative Nodes ( N )

Ave

rage

Pow

er (

E[P

r]/N

) (W

atts

)

Phase Error (0.1)Phase Error (0.2)Phase Error (0.3)Phase Error (0.4)

_____ Simulated−.−.−.−.−. Analytical

Figure 6.4: Average received signal power “power/N” vs. number of transmitter nodes in thepresence of Rayleigh fading and phase error

6.16 has been shown in Figure 6.5 and Figure 6.6 against the simulated results. Here the phaseerror distribution interval “{−0.1π to 0.1π} and {−0.3π to 0.3π}” for BER with Rayleigh fadingand AWGN for varied number of collaborative nodes.

Simulation and analytical curves shown in Figure 6.5 and Figure 6.6 are nearly a match. Thesmall variation between analytical and simulated results could be caused by approximation used inderiving BER. Furthermore, it can be seen that BER behaves inversely to the number of transmittersleading to the success of collaborative communication even in the presence of fading effect. It isevident both from simulated and analytical results, in case of a single node(N = 1), achievinga BER of 10−3 requires a power of 6dB in case where fading is not considered but only noise(AWGN) is taken into account, whereas the power requirement raises to 20.7dB if fading effect isincluded. In collaborative communication, if the number of nodes is raised to five (N = 5), anderror distribution of phase interlude over “{−0.1π to 0.1π}” a power of 9dB is required whereasthis power requirement raises to 11dB for a phase error over {−0.3π to 0.3π}. In case of N = 7,and phase error from {−0.1π to 0.1π} the required power is 9dB whereas in case of phase errorfrom {−0.3π to 0.3π} the required power raises to 10dB. The similar effect can be seen forN = 9

and N = 11.

By comparing results with previous chapter, it can be observed that in case of multipath communi-cation, bit error rate is significantly reduced thereby reducing the required transmission power per



0 5 10 15 20 2510

−5

10−4

10−3

10−2

10−1

Signal−to−Noise ratio (Eb/No), dB

Bit

Err

or R

ate

BER for Phase Error 0.1


Figure 6.5: Bit error rate plot against number of collaborative transmitters for phase error interval{−0.1π ∼ 0.1π} for m = 3

node to mitigating the effect of fading and noise. This is due to the reason that even though someof the scatter components of the transmitted signal may be blocked due to fading, however, othercomponents of the same signal may arrive at the destination following a different path, contributingthe SNR value at the receiver.

Figure 6.4 shows the direct relation of power gain with the number of nodes. But increasing thenumber of node incur more power consumption in terms of circuit operations. To analyze the en-ergy consumption of the proposed system, break-even distances are calculated for different phaseerror intervals and different number of nodes. For circuit energy consumption simulation, param-eters of “off-the-shelf” products i.e. “CC2420 and AT86RF21” are used. Nodes are considered tobe at most 1m apart from each other, the accepted BER value is 10 and path loss exponent β isfrom 4.0–6.0 [158]. The product information for CC2420 and AT86RF21 are summarized in Table5.1 in chapter 5.

The next series of Figures 6.7–6.8, show percentage energy preservation for varying number of col-laborative nodes and break-even distances. Analysis shows that the break-even distances increasewith a raise in the number of transmitter nodes. The “break-even distances” for “AT86RF212”are greater than the break-even distance for CC2420. However, AT86RF212 is more energy effi-cient than CC2420 at “100m and 200m” distances. A significant growth in energy saving in case



0 5 10 15 20 2510−5

10−4

10−3

10−2

10−1

Eb/No, dB

Bit

Err

or R

ate

BER for Phase Error 0.1


Figure 6.6: Bit error rate plot against number of collaborative transmitters for phase error interval{−0.3π ∼ 0.3π} for m = 3

of AT86RF212 has been observed in comparison to CC2420 as increase in distances exceeds thebreak-even distances as shown in Figures 6.7–6.8. It has been seen that AT86RF212 is more stablethan CC2420. AT86RF212 gets stable at a distance of 118m, whereas CC2420 achieves stability ata distance approximately equal to 148m.

“Break-even” distances for “CC2420 and AT86RF212” are shown in Table 6.1. It has been seenthat before becoming constantly stable at some specific distance, there is an increase in energysaving with increase in distances. Energy saving in percentage in case of both the products withdifferent phase errors over distances of 100m and 200m are summarized in Tables 6.2 and Table6.3.

A comparative analysis with chapter 5 reveals that there is a slighter improvement in break-evendistances, however, this improvement in break-even distance is not significant. This is due to thefact that the sum of all the scatter components result in signal with power equal to or less the powertransmitted by the transmitter node. This in some cases of multipath communication, may evenslightly reduce the break-even distances.












(a) Interval {−0.3π → 0.3π}










(b) Interval {−0.4π → 0.4π}

Figure 6.7: Energy efficiency and break-even distance for varying number of nodes for AT86RF212










(a) Interval {−0.3π → 0.3π}





2 39 43.53 43 46.54 46.2 495 49 516 51 52.57 53.5 53.59 57 5611 59.7 58

5. Conclusions





Appendix




2 97.7 95 97.5 94.5 96.5 93.8 94.5 91.53 99.1 95.5 99 95 98.8 95 98.5 94.24 99.6 94.5 99.4 94 99.2 94 99 945 99.4 93 99.5 93 99.3 93 99.1 936 99.3 92 99.2 92 99.3 92 99.1 927 99.2 91 99.1 91 99.2 91 99 919 99.1 89 99 89 99 90 99 8911 98.9 86.5 98.5 86 98.1 89 98 86.5



2 98 97 97.7 96.7 96.5 95.6 95 943 99.4 98 99.2 98 99.1 97.8 98.8 97.54 99.7 98 99.7 98.2 99.6 98 99.4 97.75 99.8 97.6 99.7 97.5 99.7 97.6 99.6 97.56 99.8 97.4 99.8 97.3 99.7 97.5 99.7 97.37 99.8 97.2 99.8 97.2 99.8 97.2 99.8 949 99.8 96.5 99.9 96.5 99.8 96.5 99.8 96.511 99.9 96 99.9 96 99.8 96 99.8 96

(b) Interval {−0.4π → 0.4π}

Figure 6.8: Energy efficiency and break-even distance for varying number of nodes for CC2420



Table 6.1: Break-even distance based on the parameter of CC2420 and AT86RF212.


2 39.9 43.93 43.2 46.14 47.4 49.75 49.3 51.36 51.6 53.57 54 549 57.1 56.410 60.3 59



200m 100m 200m 100m 200m 100m 200m 100m

2 97.5 95 97.4 94.4 96.5 93.8 94.5 91.53 98.9 95.2 99 95 98.8 95 98.5 94.14 99.3 94.3 99.2 93.9 99 93.8 98.9 945 99.4 93 99.3 92.8 99.1 92.8 99 936 99.1 91.9 99.1 92 99.1 91.8 99 91.97 99.1 91 99 90.9 99 90.8 99 919 99 88.9 99 88.9 99 90 99 88.911 98.7 86.3 98.5 86 98.1 89 98 86.5



200m 100m 200m 100m 200m 100m 200m 100m

2 97.9 96.9 97.5 94.5 96.5 95.4 95 943 99.3 97.9 99 94.8 98 97.8 98.6 97.34 99.3 97.6 99.4 93.8 98.1 97.9 99.1 97.45 99.5 97.3 99.5 92.8 97.4 97.5 99.4 97.36 99.5 97.1 99.6 91.8 97.3 97.4 99.47 977 99.6 97 99.6 90.8 97.1 97.1 99.7 93.99 99.5 96.2 99.6 88.7 96.2 96.2 99.5 95.711 99.7 95.8 98.8 85.9 95.8 95.8 98.6 96.3



6.4 Chapter Summary

A multipath collaborative communication model for WSN with Rayleigh faded noisy channel ispresented in this chapter.The signal received at the BS is considered to be unsynchronized in phase.The proposed mathematical model shows that the gain in received power is directly proportionalto the number of collaborative nodes and the BER is directly related to SNR (Eb/No) for a specificnumber of nodes. The multipath scatter components of transmitted signal contributes significantlyto the SNR values at the receiver thereby, mitigating the fading effect. To gauge energy con-sumption of proposed system, energy consumption models for SISO and multipath collaborativecommunication have been developed. Analysis of these models show that multipath collabora-tive communication performs better than SISO at long distances making it suitable for distributeddeployment in sensor networks. To conclude, increase in number of collaborative nodes producehigh power gain in the received power while keeping the transmission power constant. The mul-tipath effect in collaborative communication not only strengthen it to mitigate the fading effect,but it also performs better than SISO systems in case of medium and long distances(greater thebreak-even-distance) in terms of energy consumption. It is evident from the results that even withphase error, saving in energy consumption of collaborative communication are very close to 99%.The number of collaborative nodes also positively effect the break-even-distances. Further more,even if the distance between the nodes and BS is larger than the break-even-distance, collaborativecommunication can still be useful.


Chapter 7

Collaborative Communication in SpreadSpectrum(wideband channels)

The resource limited nature of WSN make battery life of a sensor node very crucial resource. Thepower source in these networks should be used intelligently, as in many applications it is almostimpossible to replace or recharge it [164].

Many approaches, like cooperative communication [87, 98, 160], multi-hop routing [86], collabo-rative communication [117, 118, 123, 124, 149, 165] and beamforming [115, 148, 161, 162], forsensible use of power source in sensor networks can be found in the literature. Each of these ap-proaches have their pros and cons for example in case of multi-hop system [86] over long distancesaffect power source of all the nodes from source to destination specially in case of retransmission.Similarly, in case of distributed beamforming schemes, source and cooperative relays node makesure that their signal is transmitted with such phases so that they can be constructively added at thedestination. The issue with distributed beamforming is that it needs modification to the existingfront end RF which can further increase complexity and cost of the system [87].

This chapter presents model based on collaborative communication for BSs using Rayleigh fad-ing and wideband channels, to achieve energy efficient communication. A set of sensor nodes incollaboration transmits the same information towards a common sink (base station) concurrently[82, 123]. The data is shared among collaborative nodes before the actual collaborative communi-cation starts. Unlike cooperative communication, where the transmitted information is relayed byintermediate cooperative nodes [160], in the proposed system there is one hop distance betweena transmitter and the base station. The study of the current literature shows that if sender and

96

Chapter 7. Collaborative Communication in Spread Spectrum(wideband channels)

receiver could be synchronized in time, frequency and phase, substantial gain in received powercan be achieved at the receiver [98, 115, 117, 131, 148, 152, 161–163]. However, in collaborativecommunication high gain in received power is achievable, even with imperfect phase, frequencyand time synchronization [118].

A simple motivation here is to combine the benefits of collaborative communication with benefitsof wideband channels. Wideband channels offer benefits like immunity against jamming, interfer-ence and noise as well as using single frequency for all nodes (universal frequency reuse). Thesefeatures are helpful in different applications of sensor networks, such as gaining secure and unin-terrupted information transmission from hostile environments like battle field or volcanic area. Achannel is wideband if the ratio of its bandwidth (W) and the information rate (R) is much greaterthan unity i.e. Be = W/R� 1, where Be is known as the expansion factor [166].

Contribution(s) of this chapter are; (a) derivation of mathematical model for collaborative commu-nication (power gain, BER and energy consumption) with imperfect phase synchronization basedon spread spectrum with noise and fading. (b) Theoretical model of received power and BER showa direct proportionality of power gain with the number of collaborative nodes N and BER withSNR. (c) Derivation of theoretical model to capture not only the energy consumed during com-munication between nodes and base station but also the energy consumption during nodes syn-chronization process to provide more realistic results. (d) To propose an energy efficiency modelbased on the parameters of “off-the-shelf” products like, “CC2420” and “AT86RF212” [45, 46]as shown in Table 5.1. Transmission distance, number of transmitters nodes and phase error aretaken into consideration while calculating the total consumed energy of the proposed system. At“break-even-distance” energy consumption of collaborative communication becomes equal to thatof SISO. It is used for “trade-off” analysis between the power consumed during circuit operationsand transmission power. (e) A technique to reduce the effect of fading using collaborative com-munication by achieving significant gain in the received power (improving SNR at the receiver)and reduce the required power to be transmitted by a factor N (N is the number of collaborativenodes). (f) The use of spread spectrum approach to mitigate the effect of noise and inter symbolinterference and transmit information at even low power and single frequency so that it is immuneto interference, jamming and eavesdropping.

The use of this model, ensures prolongation in network’s life time by prolonging the life of a singlenode and improves the coverage range as the number of transmitter nodes increases. Therefor, col-laborative communication is suitable for networks with limited resources like sensor networks. Thetheoretical expressions derived for received power and bit error rate are confirmed by simulation.



7.1 System Model for Collaborative Communication in Wide-band Channel

The assumption under which the proposed model is developed are listed as follow:

1. From a sensor field of randomly distributed sensor nodes, a set of sensors (collaborativenodes) transmit the same information concurrently to a common receiver (Base Station).

2. Only one signal component from each collaborative node using identical modulation andsingle frequency is received subject to synchronization error making the system behave likea virtual multipath fading system.

3. The noise used in this model is AWGN

Unlike fixed array antenna systems in distributed deployment, it is often very difficult to determinethe exact location of a node. Secondly, there is no central network controller, to synchronize thereceived signal, so the theory of fixed array antenna is not applicable[152, 153]. As a result theremay be an estimation error in determining the position of a node in distributed deployment, knownas the displacement error. Due to this displacement error the reception of the signals at the basestation may be out-of-phase. So collaborative communication needs to synchronize the signal intime, frequency and phase to achieve power gain. In addition the use of spread spectrum providesopportunity to exploit the transmitter diversity for achieving improved SNR values at the receiver[105].

But the position of transmitters is not fixed in the proposed scenario, and the application of estima-tion technique to determine positions of the transmitters gives most probable location rather thanthe exact location of a transmitter. Therefore it is very difficult to fully synchronize the system.However, a general collaborative architecture where each node uses same carrier frequency basedon spread spectrum approach has been proposed in this article as shown in Figure 7.1. The col-laborative nodes in the figure, are considered to be synchronized with each other before they begininformation exchange with the base station.


Let’s consider N collaborative nodes in a random deployment, for example in an urban or volcanicarea where there is no line of sight communication between the nodes and base station. Therefore,



N2

N3

N1

N

Collaborative Nodes

Sensor Nodes

Sensor Field

Figure 7.1: Geometry of sensor nodes

Rayleigh fading channel with AWGN is considered for the proposed system. A detail representationof the proposed system for N nodes is shown in Figure 7.2.

Each node transmits a signal x(t) correlated with the pseudo random number (PN) also called“chipping code” ci(t) of length `. Let’s assume d0 as the initial displacement of a node from thebase station and f0 the carrier frequency. Therefore, the phase is given by θ0 = 2πf0d0/c, where crepresents the speed of light. Let the distance due to displacement error be di, which is translatedinto phase error, produces a phase of θi = 2πf0di/c. If the carrier signal is cos(2πf0t), and notonly the negligible signal delay in comparison to bit length lessens the chances of inter symbolinterference but the use of spread spectrum approach fully mitigate this problem. The spreadspectrum approach also exploits the transmit diversity to produce significant power gain withoutbandwidth expansion but at lower spatial rate.

The transmitted data x(t) from each node, is modulated using cos(2πfct) and correlated with thechip code c(t), to produce the transmitted data signal s(t) as follows

s(t) =1

`x(t)c(t) cos(2πfct) (7.1)

where ` is the length of chip code and fc is the carrier frequency.



Node1x(t)cj(t)cosθ1

2cos(2π f ot)

LPF

Calculate Power



α1

Base Station

Node2x(t)cj(t)cosθ2

NodeNx(t)cj(t)cosθi

α2

αi

Y

X

n(t)

y(t)

ChannelCollaborative Transmitters

Figure 7.2: System model for collaborative communication in wideband channel

Signals from all N nodes are received at the base station with phase error as follows

y(t) =1

`

N∑i=1

∑j=1

αix(t)cj(t) cos(2πfct+ θi) + n(t) (7.2)

Here αi is the channel response/attenuation factor for ith channel and n(t) is AWGN.

Signal in Equation 7.2 is demodulated and decorrelated with chip sequence, yielding the followingresult;

Y =1

`

N∑i=1

∑j=1

αiXcj × cj cos(θi) + n× cj (7.3)

Here X = ±√E is amplitude of the received signal, whereas n represents noise amplitude.

Also we know from spread spectrum properties that 1/`∑`

j=1 cj × cj is equal to 1. Also let PYrepresents the signal power at the base station therefore the above equation can be rewritten as:

PY =

[N∑i=1

αiX cos(θi) +1

`

∑j=1

n× cj

]2(7.4)



Since αi, θi, n and cj are identically independent random (i.i.d), therefore the expected value ofthe above equation should be calculated in order to derive the power gain.

E[PY ] = E

[[N∑i=1

αiX cos(θi)

]]2+σ2n

`(7.5)

As summation is a linear operator, so we can take the expectation inside to evaluate the expression.

E[PY ] =N∑i=1

X2E[cos2(θi)]E[α2i ] +

N∑i=1

N∑j=1j 6=i

X2E [cos(θi) cos(θj)]E[αiαj] +σ2n

`(7.6)

Since all θi, θj hi and hj are i.i.d random therefore, E[θi] ≈ E[θj] ≈ E[θ], E[αi] ≈ E[αj] ≈ E[α]

and E[α2i ] ≈ E[α2]. Also we know from [123] that V ar(α) = σ2

α =(2− π

2

)b2 and E[α] = µα =(√

π2

)b. So equation (7.6) can be rewritten as follows

E[PY ] = NX2E[cos2(θ)]E[α2] +N(N − 1)X2E [cos(θ)]E[cos(θ)]E[α]E[α] +σ2n

`(7.7)

Now putting values from Equation A.1, Equation A.2 (for derivation see Appendix) and values ofE[α], whereas E[α2] = 1 in Equation 7.7, we get the following result.

E[PY ] = NX2

[1 +

sin(2φ)

2φ

]+πN(N − 1)b2X2

2

[sin(φ)

φ

]2+σ2n

`(7.8)

Here φ is the distribution limit over the phase error whereas b denotes the mode of Rayleigh fading.Notice in equation 7.8 that the noise effect is mitigated by the length of pseudo noise (PN) chipcode. As the length of chip code increases, the effect of noise decreases and that is the magic ofthe spread spectrum system. Detailed analysis of results obtained by implementing equation 7.8,is presented in section 7.2.



7.1.2 Average Probability of Error

To evaluate the bit error rate of the proposed system, let the transmitted signal be modulated usingBPSK. The signal received at the base station is a combination of two identically independentrandom variable i.e, Y = s + n. The normalized probability of error is given by the followingexpression

Pe = 0.5erfc

(µY√2σ2

Y

)(7.9)

Since θi and α are two different i.i.d random variables, therefore

µY = µs + µn (7.10)

Putting values of µs and µn in Equation 7.10 and decorrelate with pseudo noise chipping code cj ,we get

µY =1

`E

[N∑i=1

∑j=1

(αiXicj cos(θi))× cj

]+

1

`E

[∑j=1

n× cj

](7.11)

In the first part of Equation 7.11, 1/`∑`

j=1 cj × cj = 1, whereas in the second part since nrepresents AWGN and its expectation results in 0, therefore the whole second part becomes zeroand Equation 7.11 is reduce to the following form.

µY = XN∑i=1

E[αi]E[cos(θi)] (7.12)

In Equation 7.12, αi, αj, θi and θj are identically distributed random process, therefore their ex-pectation is approximately equal to their unsubscripted values. Putting values from Equation A.1(see appendix for derivation) and the expectation of E [α], Equation 7.12 can be reduced to thefollowing form.

µY =

(√π

2

)b(N ×X)

sin(φ)

φ(7.13)



Here φ is the limit on the distribution of phase error whereas b denotes the mode of Rayleigh fadingαi, a zero mean Gaussian distributed random process.

Since Y is the sum of two “i.i.d random” variables, so its variance may be calculated as follow

σ2Y = σ2

s + σ2n (7.14)

Putting values of the decorelated signal and noise variance in Equation 7.14. We know that noise nand chipping code cj are independent of each other and variance of AWGN noise correlated withPN sequence summed over M is σ2

n/M , where σ2n = N0/2 to distinguish it from Equation 7.14.

We also know that αi and θi are independent random processes so αi ≈ α and θi ≈ θ, so the aboveequation will take the following form.

σ2Y = V ar

[N∑i=1

αiX cos(θi)

]+ V ar

[∑j=1

n× cj

]

=N∑i=1

V ar [αiX cos(θi)] +N0

2`(7.15)

Equation 7.15 shows that the effect of noise is considerably mitigated by the PN sequence. Asthe length ` of PN sequence increases the effect of noise decreases which means that the noise isevenly distributed over all the communicating nodes instead of just one or two.

Now taking values from Equation A.5 (see appendix for derivation), Equation 7.15 will become

σ2Y = NX2b2

[1− π

2

(sin(φ)

φ

)2

+sin(2φ)

2φ

]+N0

2`(7.16)

Now by putting values from Equation 7.13 and Equation 7.16, in Equation 7.9 we get the followingrelation.



Pe = 0.5erfc

(√

π2

)b(N ×X) sin(φ)

φ√2

(NX2b2

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

]+ N0

2`

) (7.17)

Equation 7.17 can be represented in form of SNR with some further simple manipulation. AsX2 = Eb, the probability of error relation in Equation 7.17 can be rewritten in the following form

Pe = 0.5erfc

√π

2

(sin(φ)

φ

)b

√√√√√ N2(Eb/N0)(2Nb2

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

](EbN0

)+ 1

`

) (7.18)

Special Case: Reduction in transmit power

Spread spectrum technique allow nodes to transmit at a low power which extends the battery timeof each node. This extension in battery time is further improved by collaborative communicationby reducing the transmit power by a factor of N , where N is the number of collaborative nodes.To prove this let the amplitude of signal transmitted by each collaborative node is X = ±

√Eb/N ,

then the expression in Equation 7.18 can be written as follows.

Pe = 0.5erfc

√π

2

(sin(φ)

φ

)b

√√√√√ (Eb/N0)(2b2

N

[1− π

2

(sin(φ)φ

)2+ sin(2φ)

2φ

](EbN0

)+ 1

`

) (7.19)

It is clear from Equation 7.19 that the transmit power is reduced by a factor ofN in the collaborativecommunication as in 2b2/N . It is also interesting to see that the length ` of PN sequence (chipcode) also contributes to this reduction in power. The energy efficiency of the proposed system canbe analyzed against varying number of transmitter nodes as well as for different values of BER.



7.2 Energy Consumption

Energy consumption is an important figure of merit in evaluating performance of BSs. In orderto analyze how optimum collaborative communication can be in terms of consuming energy?,two models are developed in this section. Firstly a model for single transmitter SISO systemis developed followed an energy consumption model for collaborative communication. Both themodels are derived and analyzed for the sake of comparison and evaluating performance of thecollaborative communication. These theoretical models reflect the effect on energy consumptionwhen collaborative communication is used in combination with the spread spectrum approach.These models have been used to compute break-even-distances, and an analysis of both the modelshas been presented in the results and discussion section to analyze the effect of both systems oncoverage range.

7.2.1 SISO energy consumption model

Energy consumption of a communication system is typically the combined energy consumed bythe transmitter (Ptr) and receiver (Prv). Therefore, energy consumption over a single bit in SISO

can be represented by the following relation

ESISO =Ptr + Prvci ×Rs

(7.20)

Where Rs is the bit transmission rate and Ci is the PN sequence or chipping code.

In this case, a simplified path loss model be applied to calculate the desired transmission power asargued in [48]. Consider Gt and Gr to be the gain of transmitter and receiver antenna respectively,and if Gt = Gr = 1, then the total energy consumption of the transmitter Ptr is given by thefollowing relation [47]:


β

dβ−2r λ2(7.21)

Power required for transmitter circuit operation is represented by Pcir, power of received signal Pr,λ = c/f0, where c represents speed of light, f0 represents carrier frequency, β represents exponentof path loss, d represents distance between receiver and transmitter, dr represents reference distance



for “far-field region”.

The least required received signal power Pr, for achieving the required “BER” can be calculated asfollow:


Ps represents sensitivity of the receiver (in Watt) which a minimal requirement to obtain the desired“BER” in noisy (AWGN) channel only, and reber is Eb/N0 (in Watt) used to obtain the required“BER” in a Rayleigh faded AWGN system. reber in [104] is shown to be computed as:

reber =((1− 2Pe)

2/1− (1− 2Pe)2)

(erfc−1(2Pe))2 (7.23)

erfc−1 is a complimentary inverse error function erfc(x) = 2√π

∫ +∞x

e−t2dt.

Taking values from Equation 7.21, Equation 7.22 and Equation 7.23, total consumption of energyfor SISO systems can be achieved as;

ESISO =

(Pcir +

(4π)2Psreberdβ

drβ−2λ2+ Prv

)/ci ×Rs (7.24)

7.2.2 Collaborative communication energy consumption model

Energy consumption in SISO model is given by the sum of energy consumed in node-to-nodecommunication El (local communication) and energy consumed by communicating with BS Et.Energy consumption of both local communication and communication with the BS may be repre-sented as follows:

ECOL = El + Et (7.25)

The distance among collaborative nodes is considered to be maximum (local communication) lead-ing to maximum energy consumption, although distance of collaborative nodes from BS may vary.Energy consumption in case of local communication El can be represented as follow:



El =Ptr l +NPrv l

Rs

(7.26)

Here N is the number of collaborative nodes in the sensor network. Ptr l can be derived fromEquation 7.21, can be written in the following form

El =(Pcir +

(4π)2Psrrber ldβl

dβ−2l λ2+NPrv l

)/Rs (7.27)

Energy consumed during communication between collaborative nodes and BS is given by

Et =Ptr t + Prvci ×Rs

(7.28)

Where Ptr t is total energy, which is consumed by all (N) collaborative nodes and can be rewrittenas


β

Ndrβ−2λ2(7.29)

To obtain the desired BER, the minimum required received power Pr t may be written as


Where rcol ber represents ratio between Eb/N0 (for a system with erroneous phase), AWGN andRayleigh fading, and for systems with “AWGN” only, this ratio is equal to Eb/N0 (in Watt), toobtain the desired “BER”. rcol ber can be re-written as;


(erfc−1(2Pe))2(7.31)

Here the function BER−1 is inverse of the Equation 7.19. Therefor Et can be written as follows:

Et =(NPcir +

(4π)2Pr tdβ

Ndβ−2r λ2+ Prv

)/ci ×Rs (7.32)



Taking values from Equation 7.27 and Equation 7.32, to represent total energy consumption ofcollaborative communication as follows:

ECOL =1

Rs

(Pcir + (4π)2Psrrber ldβl

dβ−2l λ2+NPrv

)+

NPcir + (4π)2Pr tdβ

Ndβ−2r λ2

+ Prv

ci

(7.33)

The total energy saving for the proposed collaborative communication model can be achieved usingthe following equation.

Esaving(%) =

(ESISO − ECOL

ESISO× 100

)%. (7.34)

Energy saving for small distances is dominated by circuit energy consumption. Saving is 0 in casewhere ESISO = ECOL, and the distance in this case is known as “break-even-distance”.

7.3 Results and Discussions

This section is devoted to analyzing the behavior of the proposed system. For this purpose resultsobtained from analytical and simulated experiments are compared shown in the following seriesof figures. The comparison shows that the theoretical and simulation results are a perfect match.Both analytical and simulated results are obtained using Monte Carlo simulation.

For experimental purposes, the distribution of phase error is considered to be uniform over theinterval “{−φ to φ}”. Four different ranges of phase values has been considered for the sake ofexperimental analysis. These intervals include; “{−0.1π ∼ 0.1π}, {−0.2π ∼ 0.2π}, {−0.3π ∼0.3π} and {−0.4π ∼ 0.4π}”. 64-bit Hadamard codes [157], are used as PN sequences to spreadand de-spread the signal at the sender and receiver. It has been observed that the use of these codes(spread spectrum), reduces the effect of noise, thereby improving not only the received power butalso the BER.

First of all the proposed system is evaluated using received signal power. For this purpose, plotsof results of normalized received power are shown in Figure 7.3. In this plot the received signalpower is plotted against number of collaborative transmitters. This figure serves the purpose to



2 3 4 5 60.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

No. of collaborative Nodes ( N )

(E[P

r]/S

QR

(N))

(W

atts

)


Simulated

Analytical

Figure 7.3: Normalized received signal power vs. number of transmitter nodes with Rayleighfading.

check the effect of number of collaborative nodes on the decrease in received signal power.

Figure 7.3 clearly shows that an increase in phase error has inverse effect on the normalized re-ceived signal power. As the phase error increases, the normalized received signal power decreases.A closer look at the figure reveals that 9-10% approximate decrease occurred for the phase errorover the interval {−0.1π ∼ 0.1π}, 17-18% for the phase error over the interval {−0.2π ∼ 0.2π}.The error seems to increase with increase in phase error i.e, the interval over {−0.3π ∼ 0.3π} pro-duces an approximate decrease of 31-32%, whereas the interval over {−0.4π ∼ 0.4π} producesan approximate of 47-48% decrease in the normalized received power in comparison to the N2

received power without phase errors.

Another comparison between simulated and analytical results is shown in Figure (7.4), based onthe total average received power i.e, “power/N” in the presence of fading. The purpose of thisfigure is to check the trend in received signal power while increasing the number of collaborativetransmitters. This will confirm whether collaborative communication produces gain or loss in thereceived signal power.

It is clear from the figure linear increase occurs in the total received power with an increase inthe number of collaborative nodes. Therefore, it is safe to conclude that in collaborative com-munication, the number of collaborative nodes have a impact on the total received power. Evenin the presence of phase error and fading a significant gain in the received power is recorded. Itproves that collaborative communication in combination with spread spectrum not only can mit-



2 3 4 5 61

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6

No. of collaborative Nodes ( N )

E[P

r]/N

(Wat

ts)


Figure 7.4: Average total received signal power “power/N” vs. number of transmitter nodes withRayleigh fading.

igate the fading effect but also the noise to achieve such gain in the received power. Figure 7.4shows an approximate of 0.65 − 0.66N2 gain in the received power over the phase error intervalof {−0.3π ∼ 0.3π}. In case of interval over {−0.4π ∼ 0.4π} there seems to be a negative impacton the gain in received power, but still it resulted in a gain of 0.51− 0.52N2.

A comparative analysis with the received power plots in chapter 5 and chapter 6 shows a significantimprovement. This is due to the use of wideband channel as it naturally suppress the noise and in-terference as shown in Equation 7.8 by the last term. This shows that collaborative communicationis successful in obtaining more gain in received power using wideband channels in comparison tonarrowband as well as multipath communication.

Now it is clear that collaborative communication performs better in terms of received power usingwideband channels. However, gain in received signal power does not guarantee that the data isreceived correctly. A bit if flipped, may also produce high power signal at the receiver, but in factthe bit is received in error. To confirm that the proposed system achieve a good error rate as well,BER analysis is presented here.

In the following Figure 7.5, BER is plotted against the SNR values for different number of col-laborative transmitters which shows the effect of number of transmitter nodes for achieving thedesired BER. This figure also shows a plot with AWGN only where fading is not considered anda plot with N = 1 where noise is not considered. The plots reveal that increase in the number of



Figure 7.5: BER over interval {−0.1π ∼ 0.1π} for varying number of nodes with fading and totaltransmitted energy Eb/N0

collaborative nodes has a positive impact on the SNR values thereby decreasing the BER.

To analyze BER, let the energy spent per bit be represented by Eb = N2. Consequently, the energyconsumed by all collaborative nodes must be Eb = N . Analytical results based on the calculationfrom Equation 7.19 versus the simulated results are plotted in Figure 7.5 and Figure 7.6. Forthe sake of space limitation, two of the four results are included for phase error over the interval“{−0.1π to 0.1π} and {−0.2π to 0.2π}”, but they should be enough to reflect the trend of BER

with Rayleigh fading and AWGN for a varied number of collaborative node.

Results shown in Figures 7.5 and Figure 7.6 reflect an approximate match between analytical andsimulation results, of course with a slight gap between the two. The slight mismatch may be theresult of approximation used while deriving the mathematical model for BER. A close observationof the results reveal that an increase in the number of collaborative nodes improves the BER. Thatis the reason collaborative communication is successful in mitigating the fading effect. In additionto this, the use of spread spectrum further mitigates the effect of noise as shown in Equation 7.8 andEquation 7.19, thereby leading to a significant improvement in BER in comparison to narrowbandapproaches like Naqvi et al. [123].

Further analysis of the BER curves reveal the effect of number of collaborative nodes. From Figure7.5, with phase error interval over {−0.1π to 0.1π}, it is clear to achieve a BER of 10−4 in case



Figure 7.6: BER over interval {−0.2π ∼ 0.2π} for varying number of nodes with fading and totaltransmitted energy Eb/N0

of AWGN only, a power of approximately 5dB is required with no fading. Whereas in case ofa single node with fading only, a power of 17dB is required. By increasing the number of nodefrom one to five with fading, the required power to achieve a BER of 10−4, is approximately 8dB.Similarly, the required power reduces to approximately 8dB, 7.5dB and 7.2dB for a number of 7,9and 11 collaborative nodes respectively. Figure 7.5 in case of phase error over {−0.2π to 0.2π},the required power in case AWGN only is same but there is a slight raise in the remaining cases likefading only requires a power of 17.6dB, for N = 5 it is 9.2dB for N = 7 it is 8dB, for N = 9 it isapproximately 7.7dB and for N = 11 it is 7.5dB. Here, it can be seen that an increase in numberof collaborative nodes N, reduces the desired power requirement, whereas an increase in the phaseerror has inverse effect over the required power. Similar trends have been observed for phase errorinterval over “{−0.3π to 0.3π} and {−0.4π to 0.4π}”.

A comparative analysis of BER with the plots in chapter 5 and chapter 6 show that collaborativecommunication using wideband channel is a significant improvement over the previous techniques.Since BER is mainly dependent on the SNR value which is improved by the increasing the numberof collaborative transmitter as well as the noise is suppressed by the used of pseudo ransom code.This contributes to increase in the SNR values which in turn improve BER.

As mentioned above, it is clear from Figure 7.3 and Figure 7.4 that the gain in received powerraises as the number of collaborative nodes N increases, but increasing the number of nodes also



mean that an increase in power consumption due to circuit operations. Here a trade-off analysis isrequired to check if circuit power consumption overwhelm the overall energy consumption of thesystem by incorporating more collaborative nodes.

To perform an analysis of the energy consumed by the proposed approach, break-even distancesare measured over different phase error intervals for different number of collaborative nodes (N).Parameters of the “off-the-shelf” products, like “CC2420” and “AT86RF21” are considered forthis analysis. If the maximum distance between any two nodes is considered to be 1m, acceptedvalue of BER for energy consumption analysis is 10−5, whereas the path loss exponent β rangesfrom 4.0–6.0 [158]. A detailed summary of product information about “off-the-shelf” products i.e.“CC2420 and AT86RF21” is given in Table 5.1.

The following Figures 7.7a,7.7b,7.8a,7.8b present the relation between percentage energy savingswith the number of collaborative nodes (N). It can be observed that an increase in the number ofcollaborative nodes result in a raise in break-even distances. For the sake of analysis break-evendistances for AT86RF212 and CC2420 are presented over different phase error intervals. CC2420

has long break-even distance but less energy efficiency than AT86RF212 both at 100m and 200m

distances. It can be seen that AT86RF212 has fast energy saving convergence than CC2420 afterreaching the break-even distances showing that AT86RF212 is more stable than CC2420. Acloser look shows that AT86RF212 stabilizes at approximately 117m, whereas CC2420 stabilizeat approximately 146m.

Table 7.1 shows a detailed summary of Break-even distances for both CC2420 and AT86RF212. Itis analyzed that an increase in the distances causes increase in the energy preservation of collabo-rative communication and becomes constantly steady after a specific distance. Percentage energypreservation for both the products over different phase error intervals for distances of 100m and200m are shown in Tables 7.2 and Table 7.3.

From the Tables it is also clear that an increase in the phase error has inverse effect on the energysavings. As the phase error increases, energy saving percentage decreases.












(a) Phase error interval {−0.1π ∼ 0.1π}










(b) Phase error interval {−0.3π ∼ 0.3π}

Figure 7.7: percentage energy savings and break-even distances for AT86RF212, for differentnumber of collaborative nodes










(a) Phase error interval {−0.1π ∼ 0.1π}





2 39 43.53 43 46.54 46.2 495 49 516 51 52.57 53.5 53.59 57 5611 59.7 58

5. Conclusions





Appendix




2 97.7 95 97.5 94.5 96.5 93.8 94.5 91.53 99.1 95.5 99 95 98.8 95 98.5 94.24 99.6 94.5 99.4 94 99.2 94 99 945 99.4 93 99.5 93 99.3 93 99.1 936 99.3 92 99.2 92 99.3 92 99.1 927 99.2 91 99.1 91 99.2 91 99 919 99.1 89 99 89 99 90 99 8911 98.9 86.5 98.5 86 98.1 89 98 86.5



2 98 97 97.7 96.7 96.5 95.6 95 943 99.4 98 99.2 98 99.1 97.8 98.8 97.54 99.7 98 99.7 98.2 99.6 98 99.4 97.75 99.8 97.6 99.7 97.5 99.7 97.6 99.6 97.56 99.8 97.4 99.8 97.3 99.7 97.5 99.7 97.37 99.8 97.2 99.8 97.2 99.8 97.2 99.8 949 99.8 96.5 99.9 96.5 99.8 96.5 99.8 96.511 99.9 96 99.9 96 99.8 96 99.8 96

(b) Phase error interval {−0.3π ∼ 0.3π}

Figure 7.8: Percentage energy savings and break-even distances for CC2420, for different numberof collaborative nodes



Table 7.1: Break-even distance based on the parameters of CC2420 and AT86RF212.


2 40.5 44.53 43.7 46.74 47.6 50.55 50.2 52.16 52.5 547 54.7 54.79 58.1 57.110 61.1 59.5



200m 100m 200m 100m 200m 100m 200m 100m

2 98 95.5 97.5 94.9 97 94.3 95 923 99.4 95.7 99.5 95.5 98.3 95.5 99 94.64 99.8 94.8 99.7 94.4 99.5 94.3 99.4 94.55 99.9 93.5 99.8 93.3 99.6 93.3 99.5 93.56 99.6 92.4 99.6 92.5 99.6 92.3 99.5 92.47 99.6 91.5 99.5 91.4 99.5 91.3 99.5 91.59 99.5 88.4 99.5 88.4 99.5 90.5 99.5 89.411 99.2 86.8 99 86.5 98.6 89.5 98.5 87



200m 100m 200m 100m 200m 100m 200m 100m

2 98.4 97.4 98 95 97 95.9 95.5 94.53 99.8 98.4 99.5 95.3 98.5 98.3 99.1 97.84 99.8 98.1 99.9 94.3 98.6 98.4 99.6 97.95 99.9 97.8 99.9 93.3 97.9 98 99.9 97.86 99.9 97.6 99.9 92.3 97.8 97.9 99.97 97.57 99.9 97.5 99.9 91.3 97.6 97.6 99.99 94.49 99.9 96.7 100 89.2 96.7 96.7 99.99 96.211 99.9 96.3 99.3 86.4 96.3 96.3 99.1 96.8



7.4 Chapter Summary

A spread spectrum based energy efficiency mechanism using collaborative has been derived andanalyzed in this chapter. The analysis is performed through a comparison between analytical andsimulated results using received power, BER and energy consumption as figures of merit.It hasbeen proved that even if the received signals are unsynchronized in phase, using collaborativecommunication a significant power gain and well as better BER rates can be achieved. It has beenobserved in the analysis that BER is a function of SNR whereas received power is a function ofthe number of collaborative nodes. For analysis of energy consumption the proposed system iscompared with its counterpart SISO system and it has been observed that the performance of SISO

is better at short distance whereas collaborative communication performs well at medium and largedistances. To conclude it can be argued that although an increase in the number of collaborativenode offers many benefits but the this number should be kept moderate so that the power consump-tion of circuit operation be kept from overwhelming the overall energy consumption of the system.The same approach can be extended to perform the same analysis in case of imperfect frequencyas well as frequency and phase synchronization for collaborative communication systems.


Chapter 8

Collaborative Communication in SpreadSpectrum: Capacity Gain

The scarce nature of energy source in WSNs makes energy efficient transmission a crucial re-quirement. It has been seen that high gain in received power is possible through collaboration inWireless networks [81, 110, 113, 124, 152, 153, 165, 167, 168], if frequency and phase synchro-nization is achieved. But a fundamental problem with such kind of networks is the measurementof their capacity.

It has been observed that a significant gain in received power as well as better error rate are achiev-able using collaborative communication even if synchronization in phase is imperfect [165]. Thischapter however, presents a hybrid approach based on collaborative communication and spreadspectrum to even further enhance the power gain and investigates its impact on the capacity gainin WSNs with fading and noise. Spread spectrum is famous for its success against (ISI) and miti-gation of noise effect [114, 169, 170]. In addition spread spectrum provides other benefits, like 1)immunity against crosstalk interference, 2) multipath fading, 3) jamming, 4) inherent security and5) greater coverage capability. Spread spectrum uses a much wider band than conventional nar-rowband systems (“order of 20 to 254 times the bandwidth of narrowband transmissions”) [170].The wastage of bandwidth due to spreading is recovered from, by allowing multiple users to usethe same frequency, Figure 8.1 shows the scenario.

The traditional method of measuring capacity of a system in cellular networks can be applied tosensor networks as well within a star configuration, where a sensor node is at one hop distance fromthe base station [171]. Use of collaboration in sensor networks aims at achieving spatial diversity

117

Chapter 8. Collaborative Communication in Spread Spectrum: Capacity Gain

USER1 + USER2 + USER3 + .........................USER N

DATA IN BB PROCESSING GAIN

Figure 8.1: Universal frequency reused by spread spectrum

to achieve benefits like gain in the received power and ultimately high capacity gain. Calculatingcapacity for such systems is a challenge and was thoroughly investigated but in the recent years itcaught attention of the research community [83, 108, 109].

The major contribution from this chapter is a mathematical model for calculating capacity gain forspread spectrum based collaborative communication WSNs. The received signals are consideredto be unsynchronized in phase and the channel to be Rayleigh faded with AWGN. Second contri-bution is the development of the theoretical expression which proves that a raise in the number ofcollaborative nodes significantly improves power and capacity gain of the system. Thirdly wide-band channel is used in combination with collaborative communication to reduce the transmissionpower per node.

It is evident from the results that simulated and theoretical analysis produces almost matchingresults and the use of wideband channels with collaborative communication successfully mitigatedthe noise and fading effects thereby producing significant gain in the capacity of the proposedsystem.

8.1 System Model for Capacity Gain

In random distribution of nodes in WSN, exact position of a sensor is generally difficult to deter-mine. secondly in the absence of central controller [152, 153], it is almost impossible to achievephase, frequency and time synchronization. This may lead to errors in estimating position of sen-sor nodes, referred to as displacement error. As a consequence the reception at the base stationis out-of-phase. To achieve power gain collaborative communication must synchronize the re-ceived signals in time, phase and frequency, along with exploiting features of spread spectrum to



Node1

x(t)c1(t) 2cos(2π f ot)

LPF

Calculate Power


Capacity Calculation

α1

Base Station

Nodei

x(t)ci(t)

Nodem

x(t)cm(t)

αi

αm

X

n(t)

yr(t)

Noise power

ChannelCollaborative Transmitters

Figure 8.2: System model for capacity gain

achieve space/antenna diversity to improve the “SNR” and reduce “BER” [105]. Only one signalcomponent is considered to be received at the receiver from each collaborative node i.e, multipathscenario is not considered. The same system model is adopted as in section 7.1.


Theoretical model for calculating capacity gain in wideband channels is shown in Figure 8.2. Itis clear from the figure that to calculate capacity gain, signal as well as noise power must becomputed.

For computing signal and noise power, the equation derived in in section 7.1.1 has been adoptedwith some minor modifications. Here the number of collaborative nodes is represented using “m”instead of N to distinguish between number of nodes and noise power “N”.

E[PY ] = mX2

[1

2+

sin(2φ)

4φ

]+πm(m− 1)b2X2

2

[sin(φ)

φ

]2+N

`(8.1)

φ and b represent Rayleigh fading distribution bound and mode of Rayleigh fading respectively.An observation of Equation 8.1 shows that an increase in number of transmitter nodes m raisesthe gain in received signal power. This gain is further improved by the use of spread spectrum,where the effect of noise of noise is mitigated by PN sequence. It is clear that the power of noise



is inversely effected by the length of the PN sequence. So the long the PN sequence, the less theeffect of noise.

Capacity of a communication system having the effect of noise in the medium can be calculatedusing the standard mathematical formula of Shannon (introduced in chapter 3) given as follows:

C =1

2log2

(1 +

S

N

)(8.2)

here C is system capacity, S is signal power whereas N is the noise power.

Since the values of signal and noise power are already given in Equation 8.1. If we represent thesignal power X2 by P and put the values from Equation 8.1 in Equation 8.2, the following relationcan be obtained after simplification.

C =1

2log2

(1 +

(m

[1

2+

sin(2φ)

4φ

]+πm(m− 1)b2

2

[sin(φ)

φ

]2)`P

N

)(8.3)

It can be seen that Equation 8.3 is combination of signal and noise components, therefore capacitygain of collaborative communication can be calculated as follows:

In case where signals from all collaborative nodes are perfectly synchronize which means φ = 0,then capacity gain of collaborative communication is given by

C =1

2log2

(1 +

[m+

πm(m− 1)b2

2

]`P

N

)(8.4)

Special Case:

Collaborative communication can also improve the transmitted power of each node. In case eachnode transmits a power of P/m, then Equation 8.3 can take the following form

C =1

2log2

(1 +

[(1

2+

sin(2φ)

4φ

)+π(m− 1)b2

2

(sin(φ)

φ

)2]`P

N

)(8.5)

Equation 8.5 shows a reduction in transmitted power of a factor m. It means that a raise in thenumber of collaborative transmitters can reduce transmit power per node, there by increasing thenetwork’s lifetime. Secondly, the effect of power reduction per node on capacity gain can be



investigated. The same case could also be extended to investigate the effect of total transmittedpower on capacity gain. An analysis of results produced in this section has been presented insection 8.2.

8.2 Results and Discussion

To gauge capacity gain of spread spectrum based collaborative communication, the proposed sys-tem is analyzed using Monte Carlo simulation. The analysis is performed for the cases wherereceived signal from different nodes are not in perfect synchronization. AWGN and Rayleigh fad-ing blocks of SIMULINK environment of MATLAB are used. The curves obtained during theexperiments show that the simulated and analytical results are a perfect match.

To confirm the accuracy of the mathematical model derived in the previous section, initially theproposed system is tested for a case of perfect phase synchronization. This means that there isno phase error among the received signals i.e. φ = 0. The results for this case are plotted inFigure 8.3.

Figure 8.3: Capacity gain in case of perfect phase synchronization in the received signals i.e, φ = 0

In Figure 8.3, capacity gain is plotted against SNR values for different number of nodes. It can



Figure 8.4: Capacity for phase error over the interval “{−0.1π ∼ 0.1π}” and each node transmitspower P, so total transmitted power by the network is mP

be clearly observed from the figure that the theoretical and simulated results overlap each other,confirming the correctness of the proposed mathematical model. Furthermore, it can also be seenthat increasing the number of nodes has positive impact on the capacity gain of the system.

In addition to the unsynchronized phase, the proposed approach is analyzed under the effect offading and noise. For the sake of analysis the phase error is considered to be uniformly distributedover different intervals. Results of each interval are plotted and considered in different cases. Toimprove understanding of the proposed approach and systematic analysis, two different cases arediscussed here. In the first case, the system is analyzed under normal transmitted power P by eachnode. In the second case the system is analyzed under reduced transmitted power by a factor of thenumber of nodes i.e. P/m.

Case-I:

In the first case, the phase error is considered be uniformly distributed over the intervals {−0.1π ∼0.1π} and {−0.4π ∼ 0.4π}. If the power transmitted by each node is P . Then the total transmittedpower by all m collaborative transmitters is m×P . Results for this case are plotted in Figures 8.4using the phase error distributed over “{−0.1π ∼ 0.1π}”and Figure 8.5 for phase error intervaldistributed over “{−0.4π ∼ 0.4π}”.

A closer observation of both Figure 8.4 and Figure 8.5, confirms that a raise in in the number of



collaborative nodes has positive impact on the capacity gain. Therefore, increasing the numberof transmitter nodes will increase the gain in capacity of the system. Furthermore, a slight gapebetween analytical and simulated results can be seen. The reason for the presence of this gapbetween the simulated and analytical results may be the approximation used during the derivationof the proposed approach. However, it must also be noted that the this gape increases with increasein the phase error.

Figure 8.5: Capacity for phase error distributed over “{−0.4π ∼ 0.4π}′′ and each node transmitspower P, so total transmitted power by the network is mP

Case-II:

In the second case the same phase error intervals {−0.1π ∼ 0.1π} and {−0.4π ∼ 0.4π} are con-sidered. The objective of this case is to test the proposed system under reduced power. Therefore,the power transmitted by each node is considered to be P/m. In other words it can be said thateach node transmits an average amount of power, leading to a total network transmitted power of P .Results produced in this case are plotted in Figure 8.6 for phase error interval “{−0.1π ∼ 0.1π}”and Figure 8.7 for phase error interval {−0.4π ∼ 0.4π}. It can be seen that the analytical andsimulated results are a match.

A closer look at Figure 8.6 and Figure 8.7, reveals that the power transmitted by the network intotal is reduced by a factor m, where m represents the number of collaborative nodes. It meansreducing the transmit power by this factor and still produces significant capacity gain, is the suc-cess of collaborative communication. An analysis of the analytical results from Equation 8.3 and



Figure 8.6: Capacity for phase error over the interval “{−0.1π ∼ 0.1π}” and each node transmitspower P/m, so total transmitted power by the network is P

Equation 8.5, and simulated results show that if the transmit power is reduced by a factor m, iteffects the capacity gain only by a factor of 0.5 log(m), even if there is noise and fading in thechannel and the received signals are unsynchronized in phase.

Overall analysis from Figure 8.4, Figure 8.5, Figure 8.6 and Figure 8.7 confirms that capacity gainhas a direct relation with the number of transmitter nodes. It is also evident that phase error hasinverse effect on the capacity gain. A decrease of 0.3 bits/sec/Hz is recorded in capacity gain overthe phase error distribution from {−0.1π ∼ 0.1π} to {−0.4π ∼ 0.4π}

To conclude capacity calculation in the proposed model of collaboration can be very helpful as thedesired capacity is maintained even with a reduced power transmitted by the network which in turncan lead to longer network life. If the network configuration is complex the proposed method canbe used to reduce the overall power consumed by the network.

8.3 Chapter Summary

A theoretical model for capacity gain in WSNs using spread spectrum based collaborative com-munication has been derived and analyzed. The model is developed with unsynchronized received



Figure 8.7: Capacity for phase error distributed over {−0.4π ∼ 0.4π} and each node transmitspower P/m, so total transmitted power by the network is P

signal with phase error and the effect of noise and fading. It has been observed that capacity gainbehaves as a function of SNR for varied number of collaborative nodes. It has also been noted that“phase error degrades capacity gain”, but the use of spread spectrum approach mitigates the effectof noise thereby further improving capacity gain. It has been seen that collaborative communica-tion exploits the spatial diversity of to reduce the transmitted power by a factor of the number ofnode m. As a conclusion it can be argued that a decrease of a factor m times the transmit power,results in a decrease of 0.5 log2(m) in the capacity gain. It is clear that maintaining capacity highin this case is possible therefore, significant saving in power of the system can be recorded.


Chapter 9

Comparison with other Approaches

Many approaches for improving energy efficiency, power gain and capacity gain in WSNs havebeen introduced as well as various figure of merits are identified to gauge performance of these ap-proaches. Improvement in some of the factors like power gain and capacity are related as capacityis calculated from the received power. Therefore, if an approach produces gain in received power,it ultimately improve capacity gain. Apart from collaborative communication there are three veryimportant techniques for improving energy consumption in sensor networks. They are being stud-ied in the literature and in the spot light for quite some time. These techniques include; multihopcommunication, cooperative communication and beamforming[86, 115, 151, 162]. This chapterpresents a brief theoretical comparison of the proposed collaborative communication based sys-tems with these techniques. This comparison is based on coverage range and energy consumption.

9.1 Comparison with Multihop communication

In multihop communication the transmission from a sender node passes through different interme-diate nodes before it reaches the base station. These intermediate nodes are termed as relay nodes[35, 86]. In this technique a relay node receives the transmission from the sender and relay it to thenext node thereby reducing the distance over which the transmitter node has to transmit in orderto reach the base station. Since transmitting over long distances requires a node to consume morepower. The next node either forwards the data to the base station if in range or forwards it in thesame fashion to another neighbor that is closer to the base station as shown in Figure 9.1a

126

Chapter 9. Comparison with other Approaches

This approach can greatly improve signal reception since the signal is reinforced by each interme-diate node. However, complexity in the routing process also increases with the number of hopsin the multihop path leading to higher energy consumption in the routing process [86]. This con-sumption may get even more higher in case of retransmission of certain data blocks when receivedin error. This retransmission effects not only the energy consumption of the source node but alsoall the nodes in the multihop path.

For range wise comparison of multihop and collaborative communication let’s consider a runningexample for a case of four nodes (N = 4) each having 4m range. In multihop setup these fournodes are able to transmit information up to a maximum of 16 meters, so the range is (N× trans-mission power of a node). If the required range exceeds this limit, another hop will have to beadded to the multihop path which will not only increase the routing overhead but also energy con-sumption of the system. But in case of collaborative communication the range is “64” meters ascollaborative communication produces square of the input power from each node. The multipatheffect in collaborative communication further improves the range and energy consumption of thesystem.

N2 N3N1

N

Base Station

Multihop Nodes

Sensor Nodes

Sensor Field

(a) Multihop system in case of node death

N2

N3

N1

N

Base Station

Collaborative Nodes

Sensor Nodes

Sensor Field

Node dies

(b) collaborative communication in case of node death

Figure 9.1: Comparison of collaborative communication with Multihop system

Secondly, multihop communication unlike collaborative communication faces communication breakupissues when a node in the multihop path fails (dies). In this case the system is unable to transmitdata as well as BER has exponential growth at least till the system finds a replacement for the fail(dead) node. Once the failed node is replaced the data lost during the failure has to be retrans-mitted causing even more energy consumption of the nodes in the multihop path upto the newly



introduced node. This issue becomes more serious in distributed deployment where it is difficult toreplace a node if not impossible, as shown in Figure 9.1a. On the other hand in collaborative com-munication the failure (death) of a node will not effect continuity of the communication. However,BER may slightly increase as well as the range of the system may be reduce by a factor equal tothe square of the power of a single node, as shown in Figure 9.1b.

Collaborative communication is almost immune to a single node failure as shown in Figure 9.1b,keeping the capacity gain of the system high enough for smooth transmission of the information.However, in case of multihop communication the failure (death) of node may bring capacity gainof the system even to zero and BER to its highest value until the system is restored by replacingthe failed node with new node in the multihop path, as shown in Figure 9.1a.

9.2 Comparison with Cooperative communication

In cooperative communication a node in range of a communication, if receives any data from itsneighbor will forward it to its next hop neighbor which closer towards the destination (BS) [87].There may be “multiple nodes” in range at the same level of a node which can forward its dataunlike multihop communication where at “each level there is one hop”. This creates a scenarioof multipath multihop communication and it can remedy the problem of node failure (dead) byredundancy, as shown in Figure 9.2.

2 Mathematical Model of a Multi-Relay System 2.1 General Considerations A general structure of a multi-relay system, which is composed of a source node, a destination node and n relay nodes R1 to Rn, is presented in Fig. 1. The multi-relay channel of this system consists of a direct channel from the source to the destination and n relay channels. Each channel, which includes a channel between the source node and a relay node (relay) and the channel between the relay and the destination, we will call the relay channel. The message signal sent by the source is always received by the destination and all relays. The message signal at each relay is retransmitted and after some delay received at the destination. Our aim is to calculate the capacity of this multi-relay channel.

Source Destination

Rn

Rn-1

R1

R2

D1 Direct Channel

Relay Channel

……

……

Multi-relay Channel

Fig.1 Multi-relay communication system.

The mathematical model of the multi-relay channel for the system in Fig. 1 is developed using the multi-relay system model presented in Fig. 2. Each relay channel is decomposed into two sub-channels. The first sub-channel is between the source and any relay and the second sub-channel is between the relay and the destination. These two channels are cascaded to form a relay channel. The first sub-channel is represented by the additive Gaussian noise samples and the flat fading coefficients. We assume that the signals from all relays are received synchronously at the destination. Thus, second sub-channel is represented by a new set of flat fading coefficients and a common additive Gaussian noise that is added to all channels. 2.2 Capacity Calculation

We will analys the system and calculate the capacity havind in mind these assumptions: 1. The source node communicates directly to the

receiver;

2. There are (n-1) nodes that can be used as relays; 3. Channel is characterized by AWGN and

Raleigh fading; and 4. All relay nodes can receive the source signal. 5. All relay nodes may or may not send the

received signal to the destination depending on the assumed signal processing at the nodes.

Signal received at the kth relay at time t is

)()()( 1 tWtXtY kkk , (1)

where αk is fading coefficient and Wk (t) is AWGN. Suppose all αk and Wk(t) have the same probability density functions. Then the expected values of the random function Yk and its squared value at time instant t are )()()( 1 tWtXtY kkk . (2)

22

1222 )()()(2)()( tWtWtXtXtY kkkkkk .

(3) Let us express the powers in this form: Yk PtY )(2 Xk PtX )(2 , Wk PtW )(2 , Ptk )(2 . Then we

may have WXY PPPP , at t = t + 1. We can find the signal that is resent at the relay in this form

kkkkkk WXtWtXtYtX 11 )()()()1( , (4)

where ς is a constant. The signal affected by the fading in the second sub-channel

kkkkkkkkk WXtYtXtY 1)()1()1( , (5)

where βk is constant and the total signal at antenna may be written as

1

11

1

1)1(

n

kkkkk

n

kk WXtYY . (6)

The signals at the output of all relay channels may be written as

nr

n

kkkkknrn WWXWYtZ

∑

1

111 )1(

(7)

The signals at the output of the direct channel may be written as

Recent Researches in Communications and IT

ISBN: 978-1-61804-018-3 272

Figure 9.2: Multi relay cooperative communication system

However, involving multiple nodes at each level result in high energy dissemination in the form ofcircuit operations. It may also be noted that even if forwarding by a certain node does not make



Table 9.1: Comparison of power gain & transmitted power

S/No. Schemes Power Gain Transmission Power

1 Proposed Approach (narrowband) N2 1N×Txp

2 Proposed Approach (multipath) M ×N2 1M×N ×Txp

3 Proposed Approach (wideband) `×N2 1`×N ×Txp

4 Multihop Approach N N × Txp5 Cooperative Approach N 1

L× Txp

6 Beamforming Approach N2 1N× Txp

any difference in the overall communication, it still has to forward the data in order to fulfill theduty of cooperation. So cooperative communication effect energy consumption of more nodes byinvolving them in forwarding data which multihop or collaborative communication can do withcomparatively less number of nodes. For the same amount of data to be transmitted, cooperativecommunication may use too many nodes draining, their energy source as compared to collaborativecommunication. The number of effected nodes depends on how dense the deployment of thenetwork is?

The situation becomes worst when a data is to be retransmitted. This can happen in two cases; oneif the data received has errors or due to the failure of nodes at a particular level. Retransmissionis even more costly in cooperative communication than in multihop communication. However, therange of multihop communication is improved in cooperative communication by a factor of the“number of levels”. The more the number of levels the larger the range. But at the same time itsevers the problem of routing complexity as well as retransmission.

For N number of cooperative nodes cooperative communication is able to achieve N times gain inthe received power, thereby improving capacity gainN times as capacity is directly calculated fromthe power of the received signal. But it can reduce the transmission power L times per node whereL is the number of relay level. Therefore, instead of sending information over long distances, inthis type of communication a node just forwards its data to a nearby relay node, thereby savingprecious power which results in prolonged network life as shown in Table 9.1. But collaboration incomparison not only produces N2 gain in received power but also reduces transmitted power Txpby a factor of N × ` leading to significant improvement in energy consumption as well as capacitygain of the system.



9.3 Comparison with Beamforming

Beamforming [131, 148] and collaborative communication work alike as beamforming focuses asignal beam at a single point. So technically there is no difference but the direction of the sig-nal. We consider collaborative communication independent of the antenna design i.e. no speciallydesigned antennas are required to apply collaborative communication in WSNs. however collabo-rative communication can be combined with beamforming to achieve various goals like extendingthe range of a node, improving coverage in a specific direction using low transmission power. Infact it can work as an extension to collaborative communication. Since collaborative communi-cation considers transmission over a single hop therefore, combining it with beamforming mayextend its range as well as improve the power and capacity gain. Further combining it with col-laborative communication in wideband channel may further improve the system as in widebandcommunication the exchange of data can take place at even lower transmission power.


Chapter 10

Conclusion and Future Work

The problems tackled in this research work are concerned with the resource limitation of WSNsand focuses on the energy source and its efficient usage during information communication. Theultimate objective is to prolong the network’s life while providing services up to the expectation.The theory of independent transmitters using virtual array along with Rayleigh fading and AWGN

to develop communication systems which improve energy efficiency of WSNs. The basic require-ment in developing collaborative communication systems using virtual array, as well as the keyfactors that worsen its performance are also identified.

An algorithm providing energy efficiency in WSNs is developed based on collaborative communi-cation where the channel is considered to have the effect of Rayleigh fading as well as AWGN andoffers the following advantages.

• Produces significant gain in power at the receiver.

• Mitigation of fading effects considerably.

• Significant gain in channel capacity.

• For medium and long distances Collaborative communication overtakes SISO systems.

Key findings of this research in developing energy efficient algorithms and their performance eval-uation are presented in this chapter. In remainder of the chapter suggestion for future work arepresented.

131

Chapter 10. Conclusion and Future Work

10.1 Summary of Key Findings

From the feasibility study of collaborative communication it is found that determining exact lo-cation of a sensor node in randomly distributed sensor networks is almost impossible that is whyapplication of centralized antenna theory in this case is not possible directly. The assumption ofphase and frequency synchronization in randomly distributed sensor networks are even more chal-lenging in comparison to centralized antenna array. It is found that frequency, time and phasesynchronization between base station and collaborative nodes, are the key requirements while de-veloping collaborative communication based systems for WSNs.

Survey of various synchronization algorithms show that fully synchronized communication inwireless networks is very difficult if not impossible. Therefore in this thesis various collabora-tive communication based systems are modeled, analyzed theoretically and simulated in case ofRayleigh fading and AWGN, for WSNs.

Systems developed in this thesis for WSNs are:

1. Collaborative communication based energy efficient system with unsynchronized phase andassuming perfect time and frequency synchronization in “body area networks”.

2. Collaborative communication based energy efficient system with unsynchronized phase andassuming perfect time and frequency synchronization in “multipath Rayleigh faded sensornetworks”.

3. Collaborative communication based energy efficient system with unsynchronized phase andassuming perfect time and frequency synchronization “spread spectrum based systems”.

4. Collaborative communication based energy efficient system with unsynchronized phase andassuming perfect time and frequency synchronization for “capacity gain” in spread spectrumsystems.

Several figure of merits like, BER, Received Power and Channel Capacity are used to evaluate theproposed collaborative communication systems and their performance.

Review of various algorithm for wireless networks reveals the need of a significant fade margin asa compensation for the loss of signal power due to channel fading. It has been found and confirmedthat the space diversity provided by collaborative communication is very effective in reducing thefading effect, making it an efficient physical layer algorithm in terms of energy. Therefore collab-orative communication was then employed for sensor nodes transmitting data in fading channel to



save energy.

Closed form mathematical expression for calculating BER are developed with the effect of Rayleighfading and AWGN based on collaborative communication in order to observe the fading mitigationcapabilities of the proposed system. Parameters of devices like AT86RF212 [45] and CC2420 [46]are used to verify the mathematical results through simulation showing a decreased SNR require-ments in case of collaborative communication in comparison to SISO for a specific BER value. Forexample a decrease of 13dB is observed by collaborative communication system for a case of 7collaborative nodes when SNR is set to 10−3, in comparison to SISO system. This shows that therequired fade margin for collaborative communication is much less than that of SISO for channelfading compensation.

It is clear from the results that number of collaborative node has a direct impact over the gain ofcollaborative communication but an increase in the number of collaborative node may incur moreenergy consumption in the form of circuit operations. An investigation of the matter is carriedout in order to calculate power savings at the expense of th incorporated circuit power while usingcollaborative communication. Models for energy consumption are developed both for collaborativecommunication and Single Input Single Output (SISO) systems to perform comparisons of theresults obtained using both systems.

Parameter of devices used in experimentation are compliant with the IEEE standard 802.15.4, tosimulate energy consumption of collaborative communication. It is revealed that collaborativecommunication is successful in producing significant improvements in energy savings. It is alsoclear that these savings in energy are effected by the path loss exponent and the distance over whichthe information is transmitted. Based on the theoretical analysis and results from simulations, anincrease in the transmission distance and the value of path loss exponent result in the prolongationof network’s lifetime in sensor networks.

For analyzing capacity gain of collaborative communication in wideband channels a closed formmathematical expression with the effect of unsynchronized phase is derived and simulated withparameter of the “off-the-shelf” devices i.e, AT86RF212 [45] and CC2420 [46]. Its clear fromthe results that collaborative communication successfully produce high gain in the capacity of thecommunication system.

This concludes our discussion on energy efficiency and its improvements resulted by utilizingcollaborative communication.



10.1.1 Contributions

A list of the contributions of this thesis is as follows:

1. A collaborative communication based ideal model for WSN is developed with the determi-nation of the key factors that affect performance of this system. Further contribution is thedevelopment of a synchronization process to achieve phase synchronization at base station.A mathematical model for gain in received power is developed in the presence of fading andnoise. The received signal is considered to be unsynchronized in phase and expressed asfunction of the nodes.

2. For confirming the theoretical results of gain in received power, a simulation-based verifica-tion is performed. This simulation is also performed for collaborative communication withAWGN and Rayleigh fading as well for sensor networks.

3. To mitigate fading effect of the channel, energy efficiency of collaborative communicationfor sensor networks is explored which results in a mathematical expression for BER. AWGN

and Rayleigh fading is considered for developing expression for BER with unsynchronizedphase.

4. To verify theoretical BER results, a simulation-based comparison is performed and the resultsof collaborative communication are compared with SISO system which reflect that collabo-rative communication performs better than the SISO system both in case of narrow band aswell as wideband channels.

5. An increase in the number of collaborative nodes incur more power for circuit operationswhich affects the overall energy expense of the system. To verify the performance of col-laborative communication in terms of energy efficiency in comparison to SISO system andcalculate energy savings of the system. The comparison in terms of energy is performedusing parameters of off-the-shelf products for example; “CC2420” [46] and “AT86RF212”[45].

6. An investigation of the capacity gain of collaborative communication is performed by de-veloping a closed form mathematical expression for the capacity gain in sensor networks.Specifically mathematical model with unsynchronized phase in case of wideband channelsis derived.

7. An overall comparison of collaborative communication is performed for each technique with



other energy efficiency mechanism like multihop, cooperative communication and beam-forming.

10.2 Suggestions for Future Work

The broad application domain of Wireless Sensor Network (WSN) caused a rapid growth of thistechnology specially the emergence of smart home, smart office and smart city based applications.This may create new venues for research both in military and non-military applications. Thisemerging nature of WSN and the ever growing demand for better communication among sensornodes, this research work may always have a room for further improvement. To provide a fewdirections forward, some suggestions are in order.

First of all, there is need for developing algorithm to improve not only the synchronization issuesbut also some improved estimation techniques are also required in order to accurately estimate theposition of a sensor in random deployment.

Secondly, more complicated types of channels may be used to evaluate performance of collab-orative communication, for example channels with different fading than Rayleigh fading, suchas fast fading and frequency selective fading channels. Furthermore, acquiring a channel and itssynchronization in time, using collaborative communication is very crucial.

The third point is related to investigation of channel capacity gain in other communication ap-proaches like, MIMO technique for signal processing and cooperative communication.

Last but not the least, the proposed collaborative communication technique may also be evaluatedin different circumstances. For example in this work an unsynchronized phase is considered, butin future it may tested with errors in frequency synchronization or both phase and frequency.


Appendix A

Derivation of Trigonometric Functions

A.1 Mean Of Cosine Function

A.1.1 Mean value of cos(θf)

E [cos(θf )] =

∞∫−∞

cos(θf )P (θf )d(θf )

=

φ∫−φ

cos(θf )1

2φd(θf )

=1

2φ

φ∫−φ

cos(θf )d(θf )

=sin(φ)

φ(A.1)

136

Appendix A. Derivation of Trigonometric Functions

A.1.2 Mean value of cos2(θf)

E[cos2(θf )

]=

∞∫−∞

cos2(θf )P (θf )d(θf )

=

φ∫−φ

cos2(θf )1

2φd(θf )

=1

2φ

φ∫−φ

cos2(θf )d(θf )

=1

2φ

(φ+

sin(2φ)

2

)=

1

2+

sin(2φ)

4φ(A.2)

A.2 Variance Of Cosine Function

A.2.1 Variance of cos(θf)

V ar (cos(θf )) = E [cos2(θf )]− (E [cos(θf )])2

Using equations A.1 and A.2 in the above equation we get

V ar (cos(θf )) =

(1

2+

sin(2φ)

4φ

)−(sin(φ)

φ

)2

(A.3)

A.2.2 Variance of αX cos(θf)

since all of alpha, X and cos(θf ) are independent random variables, therefore the multiplicationcan be calculated as


Appendix A. Derivation of Trigonometric Functions

V ar [αX cos(θf )] = X2

[V ar[α] (E [cos(θf )])

2 + (E [α])2V ar [cos(θf )] +

V ar [α]V ar [cos(θf )]

](A.4)

Since we know from Equations (A.1) and (A.3) that, V ar(α) = σ2α = (2− π

2)b2 and E [α] = µα =(√

π2

)b

by putting these values in EquationA.4, we get

V ar [αX cos(θf )] = b2X2

[1− π

2

(sin(φf )

φ

)2

+sin(φf )

2φ

](A.5)


Bibliography

[1] V. Garg, Wireless Communications & Networking. Morgan Kaufmann, 2010.

[2] e. a. D. Estrin, “Embedded everywhere, a research agenda for networked systems of embed-ded computers,” 2001.

[3] I. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “Wireless sensor networks: asurvey,” Computer Networks, vol. 38, no. 4, pp. 393 – 422, 2002.

[4] C.-Y. Chong and S. Kumar, “Sensor networks: evolution, opportunities, and challenges,”Proceedings of the IEEE, vol. 91, no. 8, pp. 1247–1256, 2003.

[5] I. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “A survey on sensor networks,”IEEE Communications Magazine, vol. 40, no. 8, pp. 102–114, Aug 2002.

[6] M. Tubaishat and S. Madria, “Sensor networks: an overview,” IEE Potentials, vol. 22, no. 2,pp. 20–23, 2003.

[7] “Ideas for the 21st century,” Business Week, August 1999.

[8] E. H. Callaway Jr, Wireless sensor networks: architectures and protocols. CRC press,2004.

[9] “CrossBow Technology,” September 2013. [Online]. Available: http://www.test.org/doe/

[10] “Arch Rock Technology,” September 2013. [Online]. Available: http://www.ecsirt.net

[11] “European csirt network,” September 2013. [Online]. Available: http://www.ecsirt.net/

[12] IEEE Standards Association, “IEEE 802.15.4 version 2006,” 2013. [Online]. Available:http://standards.ieee.org/getieee802/download/802.15.4-2003.pdf

[13] “Zigbee Alliance,” 2013. [Online]. Available: http://www.zigbee.org/

139

Bibliography

[14] L. M. Ni, Y. Liu, and Y. Zhu, “China’s national research project on wireless sensor net-works,” IEEE Wireless Communications, vol. 14, no. 6, pp. 78–83, 2007.

[15] M. Horton, “Crossbow wireless sensing solutions conference, chicago,” 2004.

[16] D. Estrin, R. Govindan, J. Heidemann, and S. Kumar, “Next century challenges: Scalablecoordination in sensor networks,” in 5th annual ACM/IEEE International conference on

Mobile computing and networking. ACM, 1999, pp. 263–270.

[17] J. M. Rabaey, M. J. Ammer, J. L. da Silva Jr, D. Patel, and S. Roundy, “Picoradio supportsad hoc ultra-low power wireless networking,” Computer, vol. 33, pp. 42–48, 2000.

[18] N. Noury, T. Herve, V. Rialle, G. Virone, E. Mercier, G. Morey, A. Moro, and T. Porcheron,“Monitoring behavior in home using a smart fall sensor and position sensors,” in 1st Annual

International, Conference On Microtechnologies in Medicine and Biology. IEEE, 2000,pp. 607–610.

[19] J. M. Kahn, R. H. Katz, and K. S. Pister, “Next century challenges: mobile networking forsmart dust,” in 5th annual ACM/IEEE international conference on Mobile computing and

networking. ACM, 1999, pp. 271–278.

[20] N. Bulusu, D. Estrin, L. Girod, and J. Heidemann, “Scalable coordination for wireless sensornetworks: self-configuring localization systems,” in International Symposium on Communi-

cation Theory and Applications (ISCTA 2001), Ambleside, UK, 2001.

[21] B. Warneke, M. Last, B. Liebowitz, and K. S. Pister, “Smart dust: Communicating with acubic-millimeter computer,” Computer, vol. 34, no. 1, pp. 44–51, 2001.

[22] J. Agre and L. Clare, “An integrated architecture for cooperative sensing networks,” Com-

puter, vol. 33, no. 5, pp. 106–108, 2000.

[23] M. Bhardwaj, T. Garnett, and A. P. Chandrakasan, “Upper bounds on the lifetime of sensornetworks,” in International Conference on Communications, ICC, vol. 3. IEEE, 2001, pp.785–790.

[24] P. Bonnet, J. Gehrke, and P. Seshadri, “Querying the physical world,” IEEE Personal Com-

munications, vol. 7, no. 5, pp. 10–15, 2000.

[25] A. Cerpa and D. Estrin, “Ascent: Adaptive self-configuring sensor networks topologies,”IEEE transactions on mobile computing, vol. 3, no. 3, pp. 272–285, 2004.


Bibliography

[26] A. Cerpa, J. Elson, D. Estrin, L. Girod, M. Hamilton, and J. Zhao, “Habitat monitoring:Application driver for wireless communications technology,” ACM SIGCOMM Computer

Communication Review, vol. 31, no. 2 supplement, pp. 20–41, 2001.

[27] S. Cho and A. P. Chandrakasan, “Energy efficient protocols for low duty cycle wirelessmicrosensor networks,” in International Conference on Acoustics, Speech, and Signal Pro-

cessing, (ICASSP’01)., vol. 4. IEEE, 2001, pp. 2041–2044.

[28] E. M. Petriu, N. D. Georganas, D. C. Petriu, D. Makrakis, and V. Z. Groza, “Sensor-basedinformation appliances,” IEEE Instrumentation & Measurement Magazine, vol. 3, no. 4, pp.31–35, 2000.

[29] C. Herring and S. Kaplan, “Component-based software systems for smart environments,”IEEE Personal Communications, vol. 7, no. 5, pp. 60–61, 2000.

[30] G. D. Abowd and J. P. Sterbenz, “Final report on the inter-agency workshop on researchissues for smart environments,” IEEE Personal Communications, vol. 7, no. 5, pp. 36–40,2000.

[31] G. J. Pottie and W. J. Kaiser, “Wireless integrated network sensors,” Communications of the

ACM, vol. 43, no. 5, pp. 51–58, 2000.

[32] J. Rabaey, J. Ammer, J. da Silva Jr, and D. Patel, “Picoradio: Ad-hoc wireless networkingof ubiquitous low-energy sensor/monitor nodes,” in IEEE Computer Society Workshop on

VLSI. IEEE, 2000, pp. 9–12.

[33] E. Shih, S.-H. Cho, N. Ickes, R. Min, A. Sinha, A. Wang, and A. Chandrakasan, “Physicallayer driven protocol and algorithm design for energy-efficient wireless sensor networks,” in7th Annual international Conference on Mobile Computing and Networking. ACM, 2001,pp. 272–287.

[34] H. Karl and A. Willig, Protocols and architectures for wireless sensor networks. Wiley.com, 2007.

[35] D. Medhi and K. Ramasamy, Network routing: algorithms, protocols, and architectures.Morgan Kaufmann, 2010.

[36] P. J. Nahin, The Science of Radio: With Matlab and Electronics Workbench Demonstration.Springer, 2001.

[37] A. Odlyzko, “The history of communications and its implications for the internet,” 2000.


Bibliography

[38] “Global ICT developments,” 2007. [Online]. Available: http://www.itu.int/ITU-D/ict/statistics/ict/index.html

[39] C. I. by Global Analyst House, “Analysts at CCS Insight,” September 2013. [On-line]. Available: http://www.contextworld.com/c/document library/get file?p l id=47472&folderId=52910&name=DLFE-2968.pdf

[40] J. G. Proakis, E. M. Sozer, J. A. Rice, and M. Stojanovic, “Shallow water acoustic net-works,” IEEE Communications Magazine, vol. 39, no. 11, pp. 114–119, 2001.

[41] A. Swami, Q. Zhao, Y.-W. Hong, and L. Tong, Wireless Sensor Networks: Signal Processing

and Communications. John Wiley & Sons, 2007.

[42] H. Saito, O. Kagami, M. Umehira, and Y. Kado, “Wide area ubiquitous network: thenetwork operator’s view of a sensor network,” IEEE Communications Magazine, vol. 46,no. 12, pp. 112–120, 2008.

[43] “Ubiquitous sensor networks,” February 2008. [Online]. Available: $http://www.itu.int/dms\ pub/itu-t/oth/23/01/T23010000040001PDFE.pdf$

[44] V. C. Gungor and G. P. Hancke, “Industrial wireless sensor networks: Challenges, designprinciples, and technical approaches,” IEEE Transactions on Industrial Electronics, vol. 56,no. 10, pp. 4258–4265, 2009.

[45] “AT86RF212, ATMEL products,” 2012. [Online]. Available: http://www.atmel.com/devices/at86rf212.aspx

[46] “CC2420, Texas Instruments Chipcon products,” 2013. [Online]. Available: http://www.ti.com/product/cc2420

[47] A. Goldsmith, Wireless communications. Cambridge university press, 2005.

[48] B. Sklar, “Rayleigh fading channels in mobile digital communication systems .i. character-ization,” IEEE Communications Magazine, vol. 35, no. 7, pp. 90–100, 1997.

[49] W. Su, A. K. Sadek, and K. R. Liu, “Ser performance analysis and optimum power al-location for decode-and-forward cooperation protocol in wireless networks,” in Wireless

Communications and Networking Conference, vol. 2. IEEE, 2005, pp. 984–989.

[50] A. Fanimokun and J. Frolik, “Effects of natural propagation environments on wireless sensor


Bibliography

network coverage area,” in 35th Southeastern Symposium on System Theory. IEEE, 2003,pp. 16–20.

[51] S. Hara, D. Zhao, K. Yanagihara, J. Taketsugu, K. Fukui, S. Fukunaga, and K.-i. Kitayama,“Propagation characteristics of ieee 802.15. 4 radio signal and their application for locationestimation,” in 61st Vehicular Technology Conference, VTC 2005-Spring, vol. 1. IEEE,2005, pp. 97–101.

[52] K. Iniewski, Wireless technologies: circuits, systems, and devices. CRC press, 2007.

[53] W. B. Heinzelman, A. P. Chandrakasan, and H. Balakrishnan, “An application-specific pro-tocol architecture for wireless microsensor networks,” IEEE Transactions on Wireless Com-

munications, vol. 1, no. 4, pp. 660–670, 2002.

[54] O. Younis and S. Fahmy, “Heed: a hybrid, energy-efficient, distributed clustering approachfor ad hoc sensor networks,” IEEE Transactions on Mobile Computing, vol. 3, no. 4, pp.366–379, 2004.

[55] M. El-Sayed and J. Jaffe, “A view of telecommunications network evolution,” Communica-

tions Magazine, vol. 40, no. 12, pp. 74–81, 2002.

[56] S. H. A. Naqvi, “Energy-efficient collaborative communications for wireless sensor net-works,” Ph.D. dissertation, ResearchSpace@ Auckland, 2012.

[57] B. Sklar, Digital Communications: Fundamentals and Applications. BEIJING: PublishingHouse of Electronics Industry, 2002.

[58] J. D. Kraus, Antennas. {McGraw-Hill Education}, 1988.

[59] M. F. Iskander, Electromagnetic fields and waves. Prentice Hall, 1992.

[60] S. Uda and Y. Mushiake, Yagi-Uda Antenna. Research Institute of Electrical Communica-tion, Tohoku University, 1954.

[61] A. R. Thompson, B. Clark, C. Wade, and P. J. Napier, “The very large array,” The Astro-

physical Journal Supplement Series, vol. 44, pp. 151–167, 1980.

[62] K. Yao, R. E. Hudson, C. W. Reed, D. Chen, and F. Lorenzelli, “Blind beamforming ona randomly distributed sensor array system,” Selected Areas in Communications, vol. 16,no. 8, pp. 1555–1567, 1998.


Bibliography

[63] J. H. Winters, “Smart antennas for wireless systems,” Personal Communications, vol. 5,no. 1, pp. 23–27, 1998.

[64] V. Tarokh, N. Seshadri, and A. R. Calderbank, “Space-time codes for high data rate wire-less communication: Performance criterion and code construction,” IEEE Transactions on

Information Theory, vol. 44, no. 2, pp. 744–765, 1998.

[65] G. J. Foschini, “Layered space-time architecture for wireless communication in a fadingenvironment when using multi-element antennas,” Bell Labs Technical Journal, vol. 1, no. 2,pp. 41–59, 1996.

[66] V. Tarokh, H. Jafarkhani, and A. R. Calderbank, “Space-time block coding for wirelesscommunications: performance results,” Selected Areas in Communications, vol. 17, no. 3,pp. 451–460, 1999.

[67] V. Tarokh, A. Naguib, N. Seshadri, and A. R. Calderbank, “Combined array processing andspace-time coding,” IEEE Transactions on Information Theory, vol. 45, no. 4, pp. 1121–1128, 1999.

[68] S. M. Alamouti, “A simple transmit diversity technique for wireless communications,” Se-

lected Areas in Communications, vol. 16, no. 8, pp. 1451–1458, 1998.

[69] G. J. Foschini and M. J. Gans, “On limits of wireless communications in a fading environ-ment when using multiple antennas,” Wireless personal communications, vol. 6, no. 3, pp.311–335, 1998.

[70] A. Narula, M. D. Trott, and G. W. Wornell, “Performance limits of coded diversity methodsfor transmitter antenna arrays,” IEEE Transactions on Information Theory, vol. 45, no. 7,pp. 2418–2433, 1999.

[71] A. Sendonaris, E. Erkip, and B. Aazhang, “User cooperation diversity. part i. system de-scription,” IEEE Transactions on Communications, vol. 51, no. 11, pp. 1927–1938, 2003.

[72] J. N. Laneman and G. W. Wornell, “Distributed space-time-coded protocols for exploit-ing cooperative diversity in wireless networks,” IEEE Transactions on Information Theory,vol. 49, no. 10, pp. 2415–2425, 2003.

[73] C. de Morais Cordeiro and D. P. Agrawal, Ad hoc and sensor networks: theory and appli-

cations. World Scientific, 2011.


Bibliography

[74] T. S. Rappaport et al., Wireless communications: principles and practice. Prentice HallPTR New Jersey, 1996, vol. 2.

[75] D. Tse and P. Viswanath, Fundamentals of wireless communication. Cambridge universitypress, 2005.

[76] R. Olexa, Implementing 802.11, 802.16, and 802.20 Wireless Networks: Planning, Trou-

bleshooting, and Operations. Newnes, 2004.

[77] P. Agrawal, J.-C. Chen, and K. M. Sivalingam, “Energy efficient protocols for wirelesssystems,” in Wireless Multimedia Network Technologies. Springer, 2002, pp. 187–204.

[78] W. R. Heinzelman, A. Sinha, A. Wang, and A. P. Chandrakasan, “Energy-scalable algo-rithms and protocols for wireless microsensor networks,” in International Conference on

Acoustics, Speech, and Signal Processing, ICASSP’00, vol. 6. IEEE, 2000, pp. 3722–3725.

[79] L. M. Correia, Mobile broadband multimedia networks: techniques, models and tools for

4G. Elsevier, 2010.

[80] H. Holma, A. Toskala et al., WCDMA for UMTS. Citeseer, 2000, vol. 4.

[81] G. Barriac, R. Mudumbai, and U. Madhow, “Distributed beamforming for information trans-fer in sensor networks,” in 3rd International Symposium on Information Processing in Sen-

sor Networks. ACM, 2004, pp. 81–88.

[82] H. Ochiai, P. Mitran, H. V. Poor, and V. Tarokh, “Collaborative beamforming for distributedwireless ad hoc sensor networks,” IEEE Transactions on Signal Processing, vol. 53, no. 11,pp. 4110–4124, 2005.

[83] E. C. Van der Meulen, “A survey of multi-way channels in information theory: 1961-1976,”IEEE Transactions on Information Theory, vol. 23, no. 1, pp. 1–37, 1977.

[84] J. Hill, R. Szewczyk, A. Woo, S. Hollar, D. Culler, and K. Pister, “System architecturedirections for networked sensors,” ACM SIGOPS operating systems review, vol. 34, no. 5,pp. 93–104, 2000.

[85] W. Merrill, K. Sohrabi, L. Girod, J. Elson, F. Newberg, and W. J. Kaiser, “Open standard de-velopment platforms for distributed sensor networks,” in AeroSense. International Societyfor Optics and Photonics, 2002, pp. 327–337.


Bibliography

[86] B. Karp and H.-T. Kung, “Gpsr: Greedy perimeter stateless routing for wireless networks,”in 6th annual international conference on Mobile computing and networking. ACM, 2000,pp. 243–254.

[87] G. Kramer, M. Gastpar, and P. Gupta, “Cooperative strategies and capacity theorems forrelay networks,” IEEE Transactions on Information Theory, vol. 51, no. 9, pp. 3037–3063,2005.

[88] J. Feng, Y.-H. Lu, B. Jung, D. Peroulis, and Y. C. Hu, “Energy-efficient data disseminationusing beamforming in wireless sensor networks,” Transactions on Sensor Networks (TOSN),vol. 9, no. 3, p. 31, 2013.

[89] B. B. Haro, S. Zazo, and D. P. Palomar, “Energy efficient collaborative beamforming inwireless sensor networks,” IEEE Transactions on Signal Processing, vol. 62, no. 2, pp.496–510, 2014.

[90] J. Zheng and A. Jamalipour, Wireless sensor networks: a networking perspective. JohnWiley & Sons, 2009.

[91] E. Erkip, A. Sendonaris, A. Stefanov, and B. Aazhang, “Cooperative communication inwireless systems,” DIMACS Series in Discrete Mathematics and Theoretical Computer Sci-

ence, vol. 66, pp. 303–320, 2004.

[92] Z. Zou, T. Nagayama, and Y. Fujino, “Efficient multihop communication for static wirelesssensor networks in the application to civil infrastructure monitoring,” Structural Control and

Health Monitoring, vol. 21, no. 4, pp. 603–619, 2014.

[93] D. Ganesan, A. Cerpa, W. Ye, Y. Yu, J. Zhao, and D. Estrin, “Networking issues in wirelesssensor networks,” Journal of Parallel and Distributed Computing, vol. 64, no. 7, pp. 799–814, 2004.

[94] E. C. Van Der Meulen, “Three-terminal communication channels,” Advances in applied

Probability, pp. 120–154, 1971.

[95] T. Cover and A. E. Gamal, “Capacity theorems for the relay channel,” IEEE Transactions

on Information Theory, vol. 25, no. 5, pp. 572–584, Sep 1979.

[96] A. Nosratinia, T. E. Hunter, and A. Hedayat, “Cooperative communication in wireless net-works,” IEEE Communications Magazine, vol. 42, no. 10, pp. 74–80, Oct 2004.

[97] H. Xu, L. Huang, C. Qiao, W. Dai, and Y.-e. Sun, “Joint virtual mimo and data gathering for


Bibliography

wireless sensor networks,” IEEE Transactions on Parallel and Distributed Systems, vol. 26,no. 4, pp. 1034–1048, 2015.

[98] K. R. Liu, Cooperative communications and networking. Cambridge University Press,2009.

[99] J. N. Laneman, D. N. Tse, and G. W. Wornell, “Cooperative diversity in wireless networks:Efficient protocols and outage behavior,” IEEE Transactions on Information theory, vol. 50,no. 12, pp. 3062–3080, 2004.

[100] S. Cui, A. J. Goldsmith, and A. Bahai, “Energy-efficiency of mimo and cooperative mimotechniques in sensor networks,” Selected Areas in Communications, vol. 22, no. 6, pp. 1089–1098, 2004.

[101] S. K. Jayaweera, “Virtual mimo-based cooperative communication for energy-constrainedwireless sensor networks,” IEEE Transactions on Wireless Communications, vol. 5, no. 5,pp. 984–989, 2006.

[102] Y. Gai, L. Zhang, and X. Shan, “Energy efficiency of cooperative mimo with data aggre-gation in wireless sensor networks,” in IEEE Wireless Communications and Networking

Conference, March 2007, pp. 791–796.

[103] M. R. Islam, T. A. Huynh, and J. Kim, “Energy efficient cooperative technique for ieee1451 based wireless sensor network,” in IEEE Wireless Communications and Networking

Conference, March 2008, pp. 751–756.

[104] L. Simic, S. M. Berber, and K. W. Sowerby, “Energy-efficiency of cooperative diversitytechniques in wireless sensor networks,” in 18th International Symposium on Personal, In-

door and Mobile Radio Communications, PIMRC. IEEE, 2007, pp. 1–5.

[105] M. R. Ahmad, E. Dutkiewicz, X. Huang, M. K. Suaidi et al., “Cooperative mimo systemsin wireless sensor networks,” Radio Communications, 2010.

[106] J. N. Laneman, G. W. Wornell, and D. N. C. Tse, “An efficient protocol for realizing coop-erative diversity in wireless networks,” in International Symposium on Information Theory.IEEE, 2001, p. 294.

[107] L. Zheng and D. N. Tse, “Diversity and multiplexing: a fundamental tradeoff in multiple-antenna channels,” IEEE Transactions on Information Theory, vol. 49, no. 5, pp. 1073–1096, 2003.


Bibliography

[108] T. M. Cover and A. E. Gamal, “Capacity theorems for the relay channel,” IEEE Transactions

on Information Theory, vol. 25, no. 5, pp. 572–584, 1979.

[109] A. Høst-Madsen, “Capacity bounds for cooperative diversity,” IEEE Transactions on Infor-

mation Theory, vol. 52, no. 4, pp. 1522–1544, 2006.

[110] R. Mudumbai, D. R. Brown, U. Madhow, and H. V. Poor, “Distributed transmit beamform-ing: challenges and recent progress,” Communications Magazine, vol. 47, no. 2, pp. 102–110, 2009.

[111] R. Mudumbai, B. Wild, U. Madhow, and K. Ramchandran, “Distributed beamforming using1 bit feedback: from concept to realization,” in 44th Allerton conference on communication,

control and computation, 2006, pp. 1020–1027.

[112] Z. Han and H. V. Poor, “Lifetime improvement in wireless sensor networks via collabora-tive beamforming and cooperative transmission,” IET microwaves, antennas & propagation,vol. 1, no. 6, pp. 1103–1110, 2007.

[113] R. Mudumbai, G. Barriac, and U. Madhow, “On the feasibility of distributed beamformingin wireless networks,” IEEE Transactions on Wireless Communications, vol. 6, no. 5, pp.1754–1763, 2007.

[114] R. Mudumbai, G. Barriac, and U. Madhow, “Spread-spectrum techniques for distributedspace-time communication in sensor networks,” in Thirty-Eighth Asilomar Conference on

Signals, Systems and Computers, vol. 1. IEEE, 2004, pp. 908–912.

[115] E. Koyuncu, Y. Jing, and H. Jafarkhani, “Distributed beamforming in wireless relay net-works with quantized feedback,” IEEE Journal on Selected Areas in Communications,vol. 26, no. 8, pp. 1429–1439, 2008.

[116] H. Naqvi, S. Berber, and Z. Salcic, “Energy efficient collaborative communications in awgnand rayleigh fading channel in wireless sensor networks,” Wireless Communications and

Mobile Computing, vol. 14, no. 14, pp. 1382–1396, 2014.

[117] H. Naqvi, S. M. Berber, and Z. Salcic, “Performance analysis of collaborative communica-tion in the presence of phase errors and awgn in wireless sensor networks,” in International

Conference on Wireless Communications and Mobile Computing: Connecting the World

Wirelessly. ACM, 2009, pp. 394–398.

[118] H. Naqvi, S. Berber, and Z. Salcic, “Collaborative communication with imperfect phase


Bibliography

and frequency synchronization in AWGN channel in wireless sensor networks,” in Sensors

Applications Symposium (SAS). IEEE, 2010, pp. 241–244.

[119] M. Chen, S. Gonzalez, A. Vasilakos, H. Cao, and V. C. Leung, “Body area networks: Asurvey,” Mobile Networks and Applications, vol. 16, no. 2, pp. 171–193, 2011.

[120] N. Malik, M. Esa, S. Yusof, M. Ismail, and S. Hamzah, “Intelligent circular collaborativebeamforming array in wireless sensor network for efficient radiation,” in Asia-Pacific Mi-

crowave Conference (APMC). IEEE, 2013, pp. 1009–1011.

[121] U. Varshney, “Pervasive healthcare and wireless health monitoring,” Mobile Networks and

Applications, vol. 12, no. 2-3, pp. 113–127, 2007.

[122] K.-K. Wu, M.-K. Chang, S.-Y. Lee, and Y.-W. Chan, “Novel protocols of modulation levelselection in decode-and-forward multinode cooperative communication systems,” in Multi-

media and Ubiquitous Engineering. Springer, 2014, pp. 133–138.

[123] H. Naqvi, S. Berber, and Z. Salcic, “Energy efficient collaborative communication withimperfect phase synchronization and rayleigh fading in wireless sensor networks,” Physical

Communication, vol. 3, no. 2, pp. 119–128, 2010.

[124] H. Naqvi, S. Berber, and Z. Salcic, “Energy efficient collaborative communications inAWGN and rayleigh fading channel in wireless sensor networks,” Wireless Communica-

tions and Mobile Computing, vol. 14, no. 14, pp. 1382–1396, 2012.

[125] P. A. Parker, P. Mitran, D. W. Bliss, and V. Tarokh, “On bounds and algorithms for frequencysynchronization for collaborative communication systems,” IEEE Transactions on Signal

Processing, vol. 56, no. 8, pp. 3742–3752, 2008.

[126] H. Mehrpouyan and S. D. Blostein, “Bounds and algorithms for multiple frequency off-set estimation in cooperative networks,” IEEE Transactions on Wireless Communications,vol. 10, no. 4, pp. 1300–1311, 2011.

[127] D. R. Brown III and H. V. Poor, “Time-slotted round-trip carrier synchronization for dis-tributed beamforming,” IEEE Transactions on Signal Processing, vol. 56, no. 11, pp. 5630–5643, 2008.

[128] U. Madhow, D. Brown, S. Dasgupta, and R. Mudumbai, “Distributed massive mimo: algo-rithms, architectures and concept systems,” in Information Theory and Applications Work-

shop (ITA). IEEE, 2014, pp. 1–7.


Bibliography

[129] M. Bachlin and G. Troster, “Swimming performance and technique evaluation with wear-able acceleration sensors,” Pervasive and Mobile Computing, vol. 8, no. 1, pp. 68 – 81,2012.

[130] D. Qiao, Y. Li, and H. Li, “Energy efficient techniques for implanted nodes in body sensornetworks,” in International Conference on Health Informatics, ser. IFMBE, Y.-T. Zhang,Ed. Springer, 2014, vol. 42, pp. 167–170.

[131] R. Mudumbai, J. Hespanha, U. Madhow, and G. Barriac, “Distributed transmit beamformingusing feedback control,” IEEE Transactions on Information Theory,, vol. 56, no. 1, pp. 411–426, Jan 2010.

[132] S. M. B. H. Naqvi and Z. Salcic, “Collaborative communication in sensor networks,” Uni-versity of Auckland, Tech. Rep. N.762, 2010.

[133] S. B. Husnain Naqvi and Z. Salcic, “Performance analysis of collaborative communicationwith imperfect frequency synchronization and awgn in wireless sensor networks,” in Com-

munication and Networking. Springer, 2009, pp. 114–121.

[134] H. Naqvi, S. Berber, and Z. Salcic, “Capacity calculation for collaborative communicationwith imperfect phase synchronization, awgn and fading in sensor networks,” in 5th Interna-

tional Conference on Communications and Information Technology. World Scientific andEngineering Academy and Society (WSEAS), 2011, pp. 277–282.

[135] S. Berber, H. Naqvi, M. Temerinac, and V. Kovacevic, “Capacity calculation for multi-relaychannel in the presence of noise and fading,” in 5th International Conference on Communi-

cations and Information Technology, ser. CIT’11. Stevens Point, Wisconsin, USA: WorldScientific and Engineering Academy and Society (WSEAS), 2011, pp. 271–276.

[136] G. Breed, “Bit error rate: Fundamental concepts and measurement issues,” High Frequency

Electronics, vol. 2, no. 1, pp. 46–47, 2003.

[137] P. Duangkird, S. Kraisingsomnuek, and C. Kotchasarn, “Ber of alamouti stc with multiplerelays using amplify and forward cooperative diversity over rayleigh fading channel,” in 2nd

International Conference on Intelligent Systems, Modelling and Simulation (ISMS). IEEE,2011, pp. 324–328.

[138] A. K. Sadek, W. Su, and K. R. Liu, “Performance analysis for multi-node decode-and-


Bibliography

forward relaying in cooperative wireless networks,” in International Conference on Acous-

tics, Speech, and Signal Processing (ICASSP’05), vol. 3. IEEE, 2005, pp. iii–521.

[139] S. Atapattu and N. Rajatheva, “Exact ser of alamouti code transmission through amplify-forward cooperative relay over nakagami-m fading channels,” in International Symposium

on Communications and Information Technologies ISCIT’07. IEEE, 2007, pp. 1429–1433.

[140] Y. Lee and M.-H. Tsai, “Performance of decode-and-forward cooperative communicationsover nakagami-m fading channels,” IEEE Transactions on Vehicular Technology, vol. 58,no. 3, pp. 1218–1228, 2009.

[141] T. M. Cover and J. A. Thomas, Elements of information theory. John Wiley & Sons, 2012.

[142] P. Gupta and P. R. Kumar, “The capacity of wireless networks,” IEEE Transactions on In-

formation Theory, vol. 46, no. 2, pp. 388–404, 2000.

[143] S. S. Ikki and M. H. Ahmed, “Performance analysis of adaptive decode-and-forward cooper-ative diversity networks with best-relay selection,” IEEE Transactions on Communications,vol. 58, no. 1, pp. 68–72, 2010.

[144] T. Wang, A. Cano, G. B. Giannakis, and J. N. Laneman, “High-performance cooperativedemodulation with decode-and-forward relays,” IEEE Transactions on Communications,vol. 55, no. 7, pp. 1427–1438, 2007.

[145] S. S. Ikki and M. H. Ahmed, “Exact error probability and channel capacity of the best-relay cooperative-diversity networks,” IEEE Signal Processing Letters, vol. 16, no. 12, pp.1051–1054, 2009.

[146] B. Liu, B. Chen, and R. S. Blum, “Minimum error probability cooperative relay design,”IEEE Transactions on Signal Processing, vol. 55, no. 2, pp. 656–664, 2007.

[147] R. Madan, N. Mehta, A. Molisch, and J. Zhang, “Energy-efficient cooperative relaying overfading channels with simple relay selection,” IEEE Transactions on Wireless Communica-

tions, vol. 7, no. 8, pp. 3013–3025, 2008.

[148] S. Junnila, H. Kailanto, J. Merilahti, A.-M. Vainio, A. Vehkaoja, M. Zakrzewski, and J. Hyt-tinen, “Wireless, multipurpose in-home health monitoring platform: Two case trials,” IEEE

Transactions on Information Technology in Biomedicine, vol. 14, no. 2, pp. 447–455, March2010.


Bibliography

[149] A. Milenkovi, C. Otto, and E. Jovanov, “Wireless sensor networks for personal health mon-itoring: Issues and an implementation,” Computer Communications, vol. 29, no. 1314, pp.2521 – 2533, 2006.

[150] X. Xu and M. Zhong, “Wireless body sensor networks with cloud computing capability forpervasive healthcare: Research directions and possible solutions,” in Frontier and Future

Development of Information Technology in Medicine and Education, ser. Lecture Notes inElectrical Engineering. Springer Netherlands, 2014, vol. 269, pp. 979–988.

[151] J. Elias, “Optimal design of energy-efficient and cost-effective wireless body area networks,”Ad Hoc Networks, vol. 13, Part B, no. 0, pp. 560 – 574, 2014.

[152] S. A. Astaneh and S. Gazor, “Collaborative communications: Joint relay and protocol selec-tion,” in 11th Canadian Workshop on Information Theory, CWIT. IEEE, 2009, pp. 25–28.

[153] K. Zarifi, S. Affes, and A. Ghrayeb, “Distributed beamforming for wireless sensor networkswith random node location,” in International Conference on Acoustics, Speech and Signal

Processing, ICASSP. IEEE, 2009, pp. 2261–2264.

[154] K. Sayrafian-Pour, W.-B. Yang, J. Hagedorn, J. Terrill, K. Y. Yazdandoost, and K. Ham-aguchi, “Channel models for medical implant communication,” International Journal of

Wireless Information Networks, vol. 17, no. 3-4, pp. 105–112, 2010.

[155] S. L. Cotton and W. G. Scanlon, “Characterization of the on-body channel in an outdoor en-vironment at 2.45 ghz,” in 3rd European Conference on Antennas and Propagation, EuCAP.IEEE, 2009, pp. 722–725.

[156] H. Viittala, M. Hamalainen, J. Iinatti, and A. Taparugssanagorn, “Different experimentalwban channel models and ieee802. 15.6 models: comparison and effects,” in 2nd Inter-

national Symposium on Applied Sciences in Biomedical and Communication Technologies,

ISABEL. IEEE, 2009, pp. 1–5.

[157] M. S. John G. Proakis, Digital Communications, 5th ed. McGraw-Hill Higher Education,2008.

[158] D. Blumenfeld, Operations research calculations handbook. CRC Press, 2010.

[159] Y. Zhang, N. Ansari, and W. Su, “Optimal decision fusion based automatic modulationclassification by using wireless sensor networks in multipath fading channel,” in Global

Telecommunications Conference (GLOBECOM 2011). IEEE, Dec 2011, pp. 1–5.


Bibliography

[160] R. Urgaonkar and M. J. Neely, “Delay-limited cooperative communication with reliabilityconstraints in wireless networks,” in INFOCOM. IEEE, 2009, pp. 2561–2565.

[161] F. Quitin, U. Madhow, M. Rahman, and R. Mudumbai, “Demonstrating distributed transmitbeamforming with software-defined radios,” in IEEE International Symposium on a World

of Wireless, Mobile and Multimedia Networks (WoWMoM), June 2012, pp. 1–3.

[162] Z. Ding, W. H. Chin, and K. K. Leung, “Distributed beamforming and power allocation forcooperative networks,” IEEE Transactions on Wireless Communications, vol. 7, no. 5, pp.1817–1822, 2008.

[163] Z. H.-l. Q. Wei-ping, “Collaborative beamforming for distributed wireless ad hoc sensornetworks,” Shanxi Electronic Technology, vol. 5, p. 032, 2007.

[164] B. Bejar Haro, S. Zazo, and D. Palomar, “Energy efficient collaborative beamforming inwireless sensor networks,” IEEE Transactions on Signal Processing,, vol. 62, no. 2, pp.496–510, Jan 2014.

[165] A. Ghani, H. A. Naqvi, M. Sher, Z. Khan, I. Khan, and M. Saqlain, “Energy efficient com-munication in body area networks using collaborative communication in rayleigh fadingchannel,” Telecommunication Systems, pp. 1–14, 2015.

[166] J. G. Proakis, Intersymbol Interference in Digital Communication Systems. Wiley OnlineLibrary, 2001.

[167] S. Savazzi and U. Spagnolini, “Cooperative fading regions for decode and forward relaying,”IEEE Transactions on Information Theory, vol. 54, no. 11, pp. 4908–4924, 2008.

[168] A. El-Keyi and B. Champagne, “Collaborative uplink transmit beamforming with robustnessagainst channel estimation errors,” IEEE Transactions on Vehicular Technology, vol. 58,no. 1, pp. 126–139, 2009.

[169] A. Perrig, J. Stankovic, and D. Wagner, “Security in wireless sensor networks,” Communi-

cations of the ACM, vol. 47, no. 6, pp. 53–57, 2004.

[170] D. Torrieri, Principles of spread-spectrum communication systems. Springer, 2015.

[171] P. Gupta and P. R. Kumar, “The capacity of wireless networks,” IEEE Transactions on In-

formation Theory, vol. 46, no. 2, pp. 388–404, 2000.


prr.hec.gov.pkprr.hec.gov.pk/jspui/bitstream/123456789/7886/1/PhD... · Department of Computer...

Documents

Transcript of prr.hec.gov.pkprr.hec.gov.pk/jspui/bitstream/123456789/7886/1/PhD... · Department of Computer...