A COGNITIVE PHY-MAC COOPERATIVE PROTOCOL FOR LOW-POWER SHORT-RANGE WIRELESS AD-HOC NETWORKS USING UWB PPM RADIOS
By JOSE M. ALMODOVAR-FARIA
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014 c 2014 Jose M. Almodovar-Faria
2 To my parents, Mabel and Joe, for their support, encouragement, and inspiration.
3 ACKNOWLEDGMENTS Throughout my years as a graduate student I have received help, support, and encouragement from many individuals. They all have contributed in different ways to this dissertation and for that I will be forever thankful. First, I want to thank my advisor, Dr. Janise McNair, for all her help and guidance through the entire doctorate program. She was always available and very helpful. I could not have had a better advisor and mentor. The support and encouragement from my family kept me always motivated and has been one of the main reasons I have come this far. A special feeling of gratitude goes to all of them, in particular to my parents, Mabel and Joe. Without their effort and inspiration, I would not be where I am today. I thank my sister, Amarilys, and brothers, Arturo and Jose Angel, for their unconditional support and for always being there for me. They have been and will always be a great motivation for me. Likewise I thank all my friends for being supportive and believing in me uncondition- ally. A special acknowledgement goes to Pablo Rivera, my housemate and an old-time dear friend who followed my progress as a graduate student and was always very en- couraging, and to Edward Latorre, my ECE partner and a very good friend throughout all these years at UF. I want to thank also all the students in the WAMS Laboratory. They were coop- erative and supportive throughout the entire doctorate program. That will be always appreciated. For agreeing to serve in my PhD supervisory committee, I want to acknowledge Dr. Xiaolin "Andy" Li, Dr. Richard Newman, and Dr. Haniph Latchman. I really appreciate their availability and time. Finally, I express my gratitude to Dr. David Wentzloff who was my first advisor as a graduate student and awakened my interest in UWB communications. For that I am deeply thankful.
4 Contents page ACKNOWLEDGMENTS...... 4 LIST OF TABLES...... 9 LIST OF FIGURES...... 10 LIST OF ABBREVIATIONS AND VARIABLES...... 13 ABSTRACT...... 18
CHAPTER 1 INTRODUCTION...... 20 2 BACKGROUND AND MOTIVATION...... 23 2.1 Chapter Contributions...... 23 2.2 History of the Development of UWB Communications...... 23 2.3 Key Concepts in Wireless Communications...... 24 2.3.1 Multipath and Small-Scale Fading...... 24 2.3.2 Path Loss and Large-Scale Fading...... 25 2.3.3 Noise...... 26 2.3.4 Interference...... 28 2.4 UWB Definitions...... 30 2.5 UWB Benefits...... 32 2.5.1 HDR and Low SNR Operation...... 32 2.5.2 Low Interference...... 33 2.5.3 Multipath Robustness...... 33 2.5.4 High Interference Rejection...... 34 2.5.5 Low-Cost, Low-Complexity, and Low-Energy Architectures..... 35 2.6 Common Digital Modulation Schemes For UWB Radios...... 36 2.6.1 Coherent Modulation Schemes...... 37 2.6.2 Non-Coherent Modulation Schemes...... 38 2.6.3 Other Coherent and Non-Coherent Modulation Schemes...... 39 2.7 UWB Applications...... 41 2.8 UWB Channel Modeling...... 43 2.8.1 Large-Scale Path Loss for Indoor UWB Channels...... 43 2.8.2 Small-Scale Fading Model for Indoor UWB Multipath Channels.. 44 3 OPTIMIZATION OF ENERGY-DETECTION PPM RECEIVERS...... 49 3.1 Chapter Contributions...... 50 3.2 Previous work...... 50 3.3 Energy-Detection Demodulation for PPM...... 51 3.3.1 Probability of Bit-Error...... 53
5 3.4 Optimal Receiver Bandwidth...... 53 3.4.1 Effect of Receiver Bandwidth Reduction...... 54 3.4.2 Modified Probability of Bit-Error and Optimal Receiver Bandwidth. 55 3.4.2.1 Probability of Bit-Error and Receiver Bandwidth...... 56 3.4.2.2 Optimal Receiver Bandwidth...... 57 3.4.3 Adjacent-Channel Interference...... 58 3.4.3.1 Effect of ACIon the Receiver Performance...... 59 3.4.3.2 An Approximation for the Optimal Receiver Bandwidth in the Presence of ACI...... 61 3.4.4 Simulation Setup and Validation...... 62 3.4.4.1 Setup...... 62 3.4.4.2 Validation...... 63 3.4.5 Analysis...... 64 3.4.5.1 Theory Corroboration...... 64 3.4.5.2 Numerical Results...... 65 3.5 Optimal Integration Time...... 66 3.5.1 Effect of Integration Time due to Multipath Fading...... 66 3.5.2 Modified Probability of Bit-Error and Optimal Integration Time... 68 3.5.2.1 Probability of Bit-Error and Integration Time...... 68 3.5.2.2 Optimal Integration Time...... 69 3.5.3 Inter-Symbol and Inter-Frame Interference...... 71 3.5.3.1 Effect of ISI and IFI on the Receiver Performance.... 71 3.5.3.2 Optimal Integration Time...... 73 3.5.4 Simulation Setup and Validation...... 73 3.5.4.1 Setup...... 73 3.5.4.2 Validation...... 75 3.5.5 Analysis...... 76 3.5.5.1 Theory Corroboration...... 76 3.5.5.2 Numerical Results...... 76 3.6 Summary...... 79 4 ENERGY-INTEGRATION DETECTION FOR PPM RECEIVERS...... 81 4.1 Chapter Contributions...... 81 4.2 Previous Work...... 82 4.3 Energy-Integration Detection...... 82 4.3.1 Motivation...... 82 4.3.2 Bit Decision...... 83 4.3.3 Example...... 86 4.4 Probability of Bit-Error for EID...... 88 4.4.1 Bit Decision...... 88 4.4.2 Probability of Bit-Error...... 89 4.4.3 Modified Probability of Bit-Error...... 91 4.4.4 Energy Scaling Factors...... 92 4.5 Simulation...... 94
6 4.6 Analysis...... 94 4.6.1 Theory Corroboration...... 94 4.6.2 Bit-Error Rate...... 95 4.6.3 Integration Time...... 95 4.6.4 Signal Bandwidth...... 97 4.7 Summary...... 98 5 COGNITIVE PHY-MAC COOPERATIVE PROTOCOL...... 100 5.1 Chapter Contributions...... 101 5.2 Previous Work...... 101 5.3 System Model...... 102 5.3.1 Network and Signal Model...... 102 5.3.2 Modulation and Demodulation Schemes...... 103 5.3.3 Optimal Integration Time...... 104 5.3.4 Carrier Sense Multiple Access with Collision Avoidance...... 105 5.4 Channel Estimation...... 107 5.4.1 Signal and Energy Model...... 107 5.4.2 Energy Difference...... 109 5.4.3 Estimation of the Energy Scaling Factor...... 109 5.4.4 Achieving an Optimal Transmission Data Rate...... 111 5.5 UWB Cooperative PHY-MAC Protocol...... 113 5.5.1 Receiver Architecture for Channel Estimation...... 113 5.5.2 Cooperative PHY-MAC Protocol...... 114 5.5.3 PHYand MACFrame Formats...... 116 5.6 Simulation Setup...... 117 5.6.1 Network Simulator...... 117 5.6.1.1 Node class...... 118 5.6.1.2 Channel class...... 119 5.6.1.3 Other important classes and functions...... 119 5.6.2 Simulation Parameters and Setup...... 120 5.7 Analysis...... 121 5.7.1 Message Delivery Ratio...... 122 5.7.2 Average Transmission Time...... 123 5.7.3 Throughput...... 125 6 CONCLUSIONS AND FUTURE WORK...... 126
APPENDIX A DERIVATION OF THE PROBABILITY OF BIT-ERROR FOR PPM-ED RE- CEIVERS...... 130 B DERIVATION OF THE PROBABILITY OF BIT-ERROR FOR PPM-EID RE- CEIVERS...... 132
P P 2 2 C MEAN AND VARIANCE OF i j Xj FOR Xj ∼ N µj , σ ...... 134
7 REFERENCES...... 137 BIOGRAPHICAL SKETCH...... 146
8 LIST OF TABLES Table page 2-1 EIRP Limits for Indoor and Outdoor UWB Systems...... 31 2-2 Comparison between coherent and non-coherent modulation schemes.... 37 2-3 Wireless applications and their potential benefits from UWB technology.... 43 2-4 Path loss parameters for UWB channels in residential and commercial build- ings...... 44 2-5 Model parameters for UWB multipath channels...... 48 3-1 Constant values for the exponential fit given by Equation 3–15...... 58
3-2 Constant values for βopt ...... 58 3-3 Constant values for the exponential fit given by Equation 3–35...... 70
3-4 Constant values for( Tw )opt ...... 71
3-5 Constant values for( Tw )opt when ISI is considered...... 74 3-6 Optimal integration times for different values of signal bandwidth and BER... 75
4-1 Constant values for γb (t) ...... 92
0 00 4-2 Constant values for γb (t) and γb (t) ...... 93 5-1 Simulation parameters...... 121
9 LIST OF FIGURES Figure page 1-1 Timeline of popular commercial short-range wireless systems...... 21 2-1 Example of multipath propagation...... 25 2-2 Example of small-scale and large-scale fading...... 26 2-3 Example of additive white Gaussian noise...... 27 2-4 Example of co-channel and adjacent-channel interference...... 29 2-5 Example of inter-symbol interference...... 29 2-6 FCC spectral mask for indoor and outdoor UWB systems...... 31 2-7 Comparison of the theoretical channel capacities between UWB and Wi-Fi systems...... 32 2-8 Example of UWB low interference with narrowband and wideband signals... 33 2-9 Example of the UWB robustness against multipath propagation...... 34 2-10 Examples of coherent modulation schemes for UWB communications..... 38 2-11 Examples of non-coherent modulation schemes for UWB communications.. 40 2-12 Some UWB wireless applications...... 42 2-13 Path loss for UWB channels in residential and commercial buildings...... 45 3-1 General architecture forED receivers...... 49 3-2 Signal processing for aED-PPM receiver...... 52 3-3 Power spectral densities of a square pulse, Gaussian pulse, and AWGN.... 54 3-4 Signal and noise energy profile as a function of the receiver bandwidth..... 55 3-5 PSD of the transmitted signal and ACI signals...... 59 3-6 Simulator block diagram...... 62 3-7 Comparison between simulations and Equation 3–3 to validate the simulator. 64 3-8 Comparison between simulations, Equation 3–12, and Equation 3–22..... 65
−3 3-9 Required SNRbit to achieve a BER = 10 ...... 66 3-10 Normalized optimal receiver bandwidth versus the signal’s 10 dB-bandwidth −3 for Tw = 30 ns, α = 1, and BER = 10 ...... 67
10 3-11 Signal and noise energy profile as a function of integration time...... 68
3-12 Energy scaling factor γ(Tw ) for each UWBCM reported in [28]...... 70 3-13 Illustration of ISI and IFI...... 72 3-14 Simulator block diagram...... 74 3-15 Comparison of simulations and Equation 3–3 to validate the simulator..... 76 3-16 Comparison between simulations and the modified BER equations with B = 2 GHz and Tw = 25, 30, 80, 100 ns forCM 1 through 4, respectively...... 77
−5 3-17 Required SNRbit to achieve a BER= 10 for B = 2 GHz...... 78
−5 3-18 Optimal integration time( Tw )opt to achieve BER= 10 ...... 78
−5 3-19 Required SNRbit to achieve BER = 10 ...... 79
4-1 Example of the actual and optimal probabilities of bit-error (Pe) for radios op- erating inCM 1 andCM2...... 84 4-2 General block diagram for an EID receiver...... 86 4-3 Example of a binary logic 1 demodulated usingED and EID...... 87 4-4 Energy scaling factors for UWB channels...... 93 4-5 Simulator Block Diagram...... 94 4-6 Comparison between simulation results and the derived BER equation for EID receivers...... 95 4-7 Probability of bit-error forED and EID forCM 1 through 4 and B = 2 GHz .. 96
−5 4-8 Required SNRbit forED and EID to achieve a BER = 10 for B = 2 GHz andCM 1 through 4...... 97
−3 4-9 Required SNRbit forED and EID to achieve BER = 10 ...... 98 5-1 Example of a wireless ad-hoc network...... 103 5-2 Illustration of the CSMA-CA protocol...... 106 5-3 Signal processing of the proposed channel estimation...... 111 5-4 Accuracy of the energy scaling factor estimation as more symbols are used.. 112 5-5 PPM ED receiver architecture with the proposed channel estimation...... 114 5-6 Cognitive PHY-MAC protocol summary...... 115 5-7 Frame format for each MPDU...... 117
11 5-8 Frame format of the PLDU...... 117 5-9 Diagrams of the two main classes used by the network simulator...... 118 5-10 Message delivery ratio as a function of (a) the message arrival rate and (b) the number of nodes in the network...... 122 5-11 Average transmission time as a function of (a) the message arrival rate and (b) the number of nodes in the network...... 124 5-12 Throughput as a function of (a) the message arrival rate and (b) the number of nodes in the network...... 125
12 LIST OF ABBREVIATIONS AND VARIABLES
Abbreviations ACI Adjacent-channel interference, page 28. ACK Acknowledgement packet, page 107. ADC Analog-to-Digital converter, page 54. AIC Akaike Information Criterion, page 45. ASK Amplitude shift keying, page 39. AWGN Additive white Gaussian noise, page 27. BCH Bose-Chaudhuri-Hocquenghem coding algorithm, page 118. BER Bit-error rate, page 36. BOK Bi-orthogonal keying, page 41. BPF Band-pass filter, page 29. BPSK Binary phase shift keying, page 37. CA-MAC Cognitive autonomous MAC protocol, page 102. CCI Co-channel interference, page 28. CCT Channel coding theorem, page 27. CIR Channel impulse response, page 46. CLT Central limit theorem, page 28. CM Channel model, page 47. CPLNC-MAC Cooperative PHY layer network coding MAC protocol, page 101. CRC cyclic redundancy check, page 116. CSMA Carrier sense multiple access, page 105. CSMA-CA CSMA with collision avoidance, page 105. CTS Clear-to-send packet, page 105. DATA Data packet, page 107. DCF Distributed coordination function, page 105. DCM Dual-carrier modulation, page 41.
13 DIFS DCF inter-frame spacing, page 106. DPSK Differential PSK, page 41. ED Energy detection, page 21. EID Energy-integration detection, page 81. EIRP Equivalent isotropically radiated power, page 30. ESD Energy spectral density, page 56. FCC Federal Communications Commission, page 21. FSK Frequency shift keying, page 39. HDR High data rate, page 32. HLDU Higher layer data unit, page 119. i.i.d Independent and identically distributed, page 47. IEEE Institute of Electrical and Electronics Engineers, page 20. IFI Inter-frame interference, page 51. IFS Inter-frame spacing, page 106. ISI Inter-symbol interference, page 28. JR Jam resistance, page 35. LNA Low-noise amplifier, page 54. LOS Line of sight, page 25. MA Multiple access, page 35. MAC Medium access control sublayer, page 21. MPDU MAC protocol data unit, page 116. MUI Multi-user interference, page 102. NAV Network allocation vector, page 106. NLOS Non-LOS, page 33. OOK On-off keying, page 39. OPSM Orthogonal pulse-shape modulation, page 38. PG Processing gain, page 34.
14 PHY Physical layer, page 21. PLDU PHY layer data unit, page 116. PPM Pulse-position modulation, page 21. PSD Power spectral density, page 27. PSK Phase shift keying, page 37. PSM Pulse-shape modulation, page 38. QoS Quality of service, page 106. QPSK Quadrature phase shift keying, page 37. RF Radio frequency, page 30. RFID Radio-frequency identification, page 20. RTS Request-to-send packet, page 105. Rx Receiver, page 25. SIFS Short inter-frame spacing, page 106. SINR Signal-to-interference-and-noise ratio, page 72. SIR Signal-to-interference ratio, page 35. SNR Signal-to-noise ratio, page 32. std. dev. Standard deviation, page 44. T-R Transmitter-receiver, page 25. TR Transmitted reference, page 41. Tx Transmitter, page 25. UWB Ultra-Wideband, page 21. WBAN Wireless body area networks, page 20. WLAN Wireless local area networks, page 20. WPAN Wireless personal area networks, page 20. WSN Wireless sensor networks, page 20.
Variables
α Channel spacing normalized to the B10dB of the Tx signal, page 60.
15 B Receiver bandwidth, page 67.
B10dB Signal 10dB-bandwidth, page 28.
Bf Fractional bandwidth, page 30.
β Receiver bandwidth normalized to the B10dB of the Tx signal, page 55.
βopt Optimal β , page 57. C Shannon’s channel capacity, page 32.
d0 Reference distance, page 43.
E0 Noise energy, page 67.
Eb Energy per bit, page 53.
Ei Interference energy, page 59.
fc Center frequency, page 28.
fch Channel frequency separation, page 28.
fs Sampling frequency, page 27.
γPL Path loss exponent, page 43.
I0 Interference PSD constant, page 59.
N0 Noise PSD constant, page 27.
P0 Noise power, page 27.
Pb Average power per bit, page 67.
PED Probability of bit-error forED receivers using PPM, page 53.
PEID Probability of bit-error for EID receivers using PPM, page 90. PL Path loss, page 44.
PL0 Path loss at the reference distance d0 , page 43. PL Average path loss, page 43.
SNRbit SNR per bit, page 57.
Tc Chip duration (same as pulse width), page 34.
Tp Pulse time width, page 56.
Ts Symbol period, page 34.
16 ts Sampling period, page 27.
Tw Integration window length or integration time, page 52.
(Tw )opt Optimal Tw , page 69.
XPL Shadowing parameter for the log-normal path loss model, page 44.
17 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy A COGNITIVE PHY-MAC COOPERATIVE PROTOCOL FOR LOW-POWER SHORT-RANGE WIRELESS AD-HOC NETWORKS USING UWB PPM RADIOS By Jose M. Almodovar-Faria May 2014 Chair: Janise McNair Major: Electrical and Computer Engineering Nowadays low-power short-range wireless ad-hoc networks are becoming more popular as the demand for wireless applications such as sensor and personal area networks continue to grow. Recently, in particular since the Federal Communications Commission approval in 2002, ultra-wideband (UWB) communications have been proposed as a viable and efficient alternative to implement short-range wireless ap- plications. For the past decade, numerous investigations and research works have been done in order to employ UWB technology in wireless applications that have been traditionally implemented with conventional narrowband technologies. The vast range of benefits offered by UWB makes it, in many cases, an ideal solution when implementing wireless radios and networks. Low-power operation, low-complexity and low-cost radio architectures, and high data rates are among the many advantages of UWB. In most short-range wireless ad-hoc networks, low-power operation as well as multiple access control (MAC) is crucial in the network design. Pulse-position modulation (PPM) is a well-known digital modulation scheme that when used in UWB radios can achieve simple low-cost architectures and more importantly a very low-power operation while offering relatively good data rates and bit-error rate (BER) performance. The DCF function described by the IEEE 802.11 WLAN standard is used quite often as the MAC protocol when implementing wireless networks in general and has proven to be efficient for many applications. This doctoral
18 dissertation presents a new cognitive and cooperative protocol between the physical (PHY) layer and the MAC sublayer for wireless ad-hoc networks using PPM UWB radios. By a cognitive estimation of the wireless channel and the cooperation between the MAC and PHY layers, the cognitive protocol can dynamically adjust the transmission data rate between two nodes optimizing their communication. Simulations show that the protocol improves the overall network performance in terms of message delivery ratio and average transmission delay.
19 CHAPTER 1 INTRODUCTION In the last three decades, as the wireless communications industry has advanced, we have seen a great increase in the development of not only long-range and medium- range wireless communications (e.g. radio and television broadcasts, satellite com- munications and cellular networks among others) but in short-range wireless systems as well. For the last 15 years, wireless systems involving radio-frequency identification (RFID), wireless sensor networks (WSN), wireless local area networks (WLAN), per- sonal area networks (WPAN) and body area networks (WBAN) have been increasingly developed to meet the demands of our technology-hungry society. These systems are used in many of today’s wireless applications such as mobile devices, wireless routers, wireless audio and video systems, advanced remote controlling, and much more. Figure 1-1 shows a timeline of some popular short-range wireless applications that have been commercialized over the last six decades. Until early 1990s, there were few commercial applications for short-range systems (e.g. remote controls (RCs), cordless phones and few other applications). At the time, most of the wireless applications commercially available focused on medium and long-range communications. However, thanks to the research done during the 1970s and early 1980s [61], the first WLAN products started to appear at the end of the 1980s. By 1997, the original version of the IEEE1 802.11 standard for WLAN was finalized and with it came a rapid development of this technology for residential and commercial use (e.g. Wi-Fi routers). Other short- range systems such as WPAN and WBAN started around the idea of WLAN and have extended in the last decade to numerous other systems such as WSN and RFID. Today, these systems have a very wide range of applications such as bluetooth in mobile
1 IEEE stands for Institute of Electrical and Electronics Engineers
20 Figure 1-1: Timeline of popular commercial short-range wireless systems devices and remote controllers, high-speed wireless routers, RFID tags, and body implantable sensors to mention just a few. Recently, in particular since the Federal Communications Commission (FCC) ap- proval in 2002, ultra-wideband (UWB) communications have been proposed as a viable and efficient alternative to implement short-range wireless applications. Thus, for the last decade, research and numerous investigations have been done in which the con- ventional narrowband communications are being substituted by UWB communications. This mainly due to the wide range of benefits offered by UWB communications including low-power operation, low-complexity and low-cost radio architectures, and high data rates among several others. In this doctoral dissertation, non-coherent UWB radios using pulse-position mod- ulation (PPM) and energy-detection (ED) demodulation are studied and optimized. In addition, a modification to theED technique is proposed and discussed in detail. Based on the extensive study and the theory developed, a new cognitive and cooperative pro- tocol involving the physical (PHY) layer and the medium access control (MAC) sublayer is introduced. The rest of this doctoral dissertation is divided in 5 additional chapters. Chapters 2,3, and4 correspond to the work done on UWB PPM radios. Chapter2, provides a literature overview of UWB communications along with the explanation of several
21 key concepts directly related to the discussion throughout this document. Chapter3 discusses in detailED and its optimal receiver bandwidth and optimal integration time in the presence noise and certain interference sources. The theory presented in this chapter is the base for the channel estimation that will be used for the cognitive and cooperative protocol. Chapter4 presents a modification made to theED demodulation technique presented in Chapter3 that improves the receiver performance. This modifi- cation along with the original demodulation method will be used in the implementation of the new protocol presented in Chapter5. In this chapter, the cognitive and cooperative PHY-MAC protocol is presented, discussed and analyzed. Finally, conclusions are presented in Chapter6. This chapter also provides a brief summary of the future work.
22 CHAPTER 2 BACKGROUND AND MOTIVATION The field of UWB wireless communications has been rapidly growing in the past decade since its FCC approval in 2002. The wide variety of advantages that UWB offers has motivated a significant interest toward its development and application to a vast range of wireless applications including medium and long range communications [30, 90]. However, at least 80% of UWB commercial applications are envisioned to be short-range wireless communications [38]. Although the concepts presented in this chapter focus on short-range wireless communications, they are still valid for potential applications in the medium-range and even in the long-range domain of UWB wireless communications.
2.1 Chapter Contributions
This chapter provides a literature overview of UWB communications. It presents the key concepts, benefits, and challenges of UWB signaling as well as its applications and technologies. In addition, a review of UWB channel modeling is offered including a characterization of the channel models proposed by the IEEE P802.15.3a task group.
2.2 History of the Development of UWB Communications
UWB communications employ narrow pulses in order to achieve large bandwidths and, therefore, its early name was impulse communications –the term UWB became popular during the 1990s. The first experiments with narrow pulses can be traced back to the late the 19th century when Heinrich Hertz experimented with spark discharges [42] to verify Maxwell’s equations on electromagnetic theory. The equipment Hertz used is probably the first impulse radio in history. A few years later, Guglielmo Marconi’s experiments using spark-gap transmissions expanded Hertz’s work and demonstrated its practical application: wireless communications [8]. Ironically Marconi, known today as the inventor of radio, was using UWB communi- cations for its radio applications by employing spark-gap transmissions. In fact, for about
23 20 years after Hertz’s first experiments, this was the dominant technology [37] for the early research in wireless communications. Later, mainly due to the lack of appropriate hardware for pulse-based modulation and demodulation as well as wideband inter- ference mitigation techniques, sinusoidal waves became the leading form of wireless communications. It was not until the late 1960s and early 1970s that pulse-based com- munications resurfaced with the pioneering contributions of researchers like Henning Harmuth, Paul Van Etten, and Gerald Ross [7]. Harmuth publications presented the ba- sic receiver and transmitter design for UWB while Van Etten’s experiments in UWB radar systems resulted in the development of the basic concepts for UWB antennas. In 1971, Ross filed a patent on the transmission and reception of pulse signals without distortion [68] and in 1973 it became the first US patent awarded for UWB communications. For the next two decades after Ross’ patent, UWB was mostly used by the military in communications, radar, sensing, and niche applications [18]. In the 1990s, a few startup companies –in particular, Time Domain Company (TDC) [80]– staged a movement toward the commercialization of UWB systems which, after years of much opposition, culminated with its approval by the FCC in April of 2002. A few months later, the PulsOn chipset from TDC became the first UWB communications product certified by the FCC. Since then, UWB has been a major research area in wireless communications as evidenced by the numerous articles and books published in the last 10 years.
2.3 Key Concepts in Wireless Communications
This section briefly defines a few key concepts in wireless communications that are relevant to the discussion that will be developed in subsequent chapters. These key concepts are: small-scale fading, large-scale fading, noise, and interference.
2.3.1 Multipath and Small-Scale Fading
A radio signal that is transmitted through a wireless channel travels through multiple paths before reaching the receiver’s antenna. This phenomena is called multipath
24 (a) Multiple propagation paths (b) Transmitted and received signals Figure 2-1: Example of multipath propagation propagation, or simply multipath, and is the cause of the rapid amplitude fluctuations that the radio signal undergoes over a short period of time, i.e. short distance. Figure 2-1(a) shows an example of multiple signal paths. The direct path from transmitter (Tx ) to receiver (Rx) is called the line-of-sight (LOS) path and it is usually the dominant multipath component of the received radio signal. Small-scale fading describes the effects caused by multipath propagation and other factors such as the signal bandwidth and receiver motion relative to the transmitter. The most important effects described by small-scale fading are: time dispersion due to multipath propagation delays, rapid changes in signal strength and polarity, and random frequency modulation due to Doppler shifts.1 Figure 2-1(b) shows an example of these effects on a transmitted pulse. As can be seen, the received pulse has been dispersed in time with a decreasing average signal strength and changes in frequency and polarity.
2.3.2 Path Loss and Large-Scale Fading
Small-scale fading describes the rapid changes in signal strength over short transmitter-receiver (T-R) separation distances. Large-scale fading, in contrast, de- scribes the mean signal strength attenuation over longer T-R separation distances, i.e.
1 When a transmitter and a receiver are moving relative to each other, the frequency of the received signal changes based on their motion. This phenomenon is known as the Doppler effect (or Doppler shift).
25 (a) Linear Scale (b) Logarithmic Scale Figure 2-2: Example of small-scale and large-scale fading longer periods of times. Figure 2-2(a) shows an example of signal strength variations as a function of T-R distance illustrating the rapid small-scale fading and the slower large-scale fading. The figure also shows that the mean signal strength (i.e. large-scale fading) attenuates exponentially as the T-R distance increases. For this reason as well as to simplify related calculations, the signal strength attenuation is often described in logarithmic scales as shown in Figure 2-2(b). In wireless communications, the signal strength attenuation is commonly referred to as path loss. Since the signal attenuates more and more as it travels further along its wireless path, the path loss increases with larger T-R distances. Path loss is very useful when defining that radio coverage area of a transmitter.2
2.3.3 Noise
There are several types of noise (e.g. shot noise, burst noise, Brownian noise) but the most common when it comes to wireless communications is thermal noise. This
2 The radio coverage area refers to how far a receiver can be from the transmitter so that the received signal strength is large enough to be detected.
26 (a) Time Domain (b) Frequency Domain Figure 2-3: Example of additive white Gaussian noise
unavoidable phenomenon is a random process caused by thermal motions of electrons in any conducting material and has three main properties:
1. It is an additive process because a received signal can be represented by the sum of the transmitted signal and the noise signal. 2. It has a constant power spectral density (PSD) for all frequencies (i.e. white noise).3 3. It follows a zero-mean Gaussian distribution with finite variance σ2 equal to the average noise power P0. Hence, thermal noise is often called additive white Gaussian noise (AWGN). Figure 2-
3(a) shows an example of AWGN when using a sampling period ts = 1/fs, where fs is the sampling frequency. Figure 2-3(b), on the other hand, shows AWGN in the frequency
4 domain, that is, a flat spectrum with a PSD of N0/ 2. AWGN is fundamental in the understanding of wireless communications. A basic theorem of Information Theory is the Channel Coding Theorem (CCT) [14] from which
3 A signal with constant PSD for all frequencies is called white noise in analogy to white light which covers all wavelengths.
4 The PSD constant value for AWGN is N0 for all frequencies (positive and negative). However, real-world devices only use the positive half of the spectrum. Therefore, N0 / 2 is used instead.
27 can be concluded that the worst-case background noise in wireless channels is AWGN [50, 76] as it minimizes the channel information capacity. Furthermore, recently, it has been suggested that AWGN is also the worst-case additive noise in wireless networks in general [76, 77]. Intuitively, it makes sense the use of AWGN to model all noise since, in many cases, the combined noise sources should approach a Gaussian random distribution by the Central Limit Theorem (CLT).
2.3.4 Interference
Interference can be defined as any unwanted signal from an external source that alters or disrupts the intended signal. In wireless communications there is a wide range of interference signals, or just interferers, that are taken into account when designing wireless systems. Broadly speaking, interference in wireless communications can be divided in:
1. Co-channel interference (CCI): the frequency bands (channels) of the interferer and the intended signal overlap. 2. Adjacent-channel interference (ACI): the interferer is at a neighboring channel and part of its energy is leaked into the frequency band of the intended signal 3. Inter-symbol interference (ISI): a previous intended signal interferes with the current intended signal due to the time dispersion caused by multipath propagation (see section 2.3.1).
Figure 2-4(a) shows an example of CCI for a signal with center frequency fc. In this example, the frequency bands of the interference signals overlap with the frequency
5 band of the intended signal, i.e. the channel frequency separation (fch) is less than
6 the 10dB-bandwidth of the signal (B10dB). Similarly, Figure 2-4(b) shows an example of ACI which, in contrast to CCI, fch is greater or equal than B10dB . In both figures, the
5 The channel frequency separation is the frequency spacing between the center fre- quencies of two channels. 6 In wireless communications, the frequency band of a signal is usually determined by its 10dB-bandwidth B10dB , i.e. the frequency band in which the signal’s PSD falls 10dB from its highest point.
28 (a) Co-channel interference (CCI) (b) Adjacent-channel interference (ACI) Figure 2-4: Example of co-channel and adjacent-channel interference dashed yellow lines represent a non-ideal band-pass filter (BPF). On the other hand, an example of ISI is illustrated in Figure 2-5. This example shows the wireless transmission of two symbols.7 As seen in the figure, the first received symbol interferes with the second since its time dispersion due to multipath propagation is larger than the symbol period.
(a) Transmitted Signal (b) Received Signal Figure 2-5: Example of inter-symbol interference
7 In digital communications, a symbol is a signal modulated to represent one or more logic bits.
29 2.4 UWB Definitions
The FCC final rule on UWB Transmission systems defines a UWB transmitter as “an intentional radiator 8 that at any point in time, has a fractional bandwidth equal to or greater than 0.20 or has a UWB bandwidth equal to or greater than 500 MHz, regardless of the fractional bandwidth”[26]. In this definition, the fractional bandwidth Bf is
Bf = (fH − fL) /fc (2–1) where fc = (fH + fL) /2 is the center frequency and fH and fL are, respectively, the upper and lower boundaries of the 3dB-bandwidth9 of the transmitted signal. In addition to the bandwidth requirement, the FCC also defines a maximum trans- mission power. The power limits for the equivalent isotropically radiated power (EIRP )10 of indoor and outdoor UWB systems when measured at a distance of 3 m with a bandwidth resolution of 1 MHz are tabulated in Table 2-1. The FCC also sets a power limit for unintentional radiators.11 For frequencies above 960 MHz, this type of radiators cannot exceed an electric field strength of 500 µV /m measured at a T-R separation distance of 3 m over a 1 MHz frequency band [25]. To compare this limit to the UWB EIRP limit, the conversion equation given by [12]
8 Intentional radiators are devices that generate radio frequency (RF) energy on pur- pose such as wireless transmitters, imaging sensors, and ground penetrating radars among many others. 9 The 3dB-bandwidth of a signal is the frequency band in which its PSD falls 3dB from the highest point. 10 EIRP, as defined by the FCC, refers to the highest signal power strength measured at 3 m from the source at any frequency and in any direction. 11 Unintentional radiators are devices not designed to emitRF energy on purpose such as digital electronics, electric chargers, and audio amplifiers among many others.
30 Table 2-1: EIRP Limits for Indoor and Outdoor UWB Systems
Frequency Range (MHz) Indoor EIRP( dBm/MHz) Outdoor EIRP( dBm/MHz) 960 − 1610 −75.3 −75.3 1610 − 1990 −53.3 −63.3 1990 − 3100 −51.3 −61.3 3100 − 10600 −41.3 −41.3 Above 10600 −51.3 −61.3
Figure 2-6: FCC spectral mask for indoor and outdoor UWB systems.
2 2 Prad = 4 · π · drad · Erad /Z0 (2–2) can be used. In this case, the distance from the measurement location to the radiator is drad = 3 m, the electric field strength from the radiator is Erad = 500 µV /m and the characteristic impedance of free space is Z0 = 120 · π Ω. With these values, the radiated power limit for unintentional radiators is Prad ≈ 75nW ≈ −41.3dBm per 1MHz. This limit is often called the noise floor and is shown in Figure 2-6 along with the FCC spectral mask for indoor and outdoor UWB transmitters. From the figure, it is evident that the best frequencies to operate UWB systems range from 3.1 GHz to 10.6 GHz where the EIRP limit is the highest.
31 Figure 2-7: Comparison of the theoretical channel capacities between UWB and Wi-Fi systems.
2.5 UWB Benefits
UWB technology offers several advantages over the traditional narrowband tech- nologies. Among these, the key benefits can be summarized as: high data rate (HDR), low signal-to-noise ratio (SNR) operation, low interference, multipath robustness, high interference rejection, and low-cost, low-complexity, and low-energy architectures.
2.5.1 HDR and Low SNR Operation
The CCT states that information can be transmitted at any data rate R that does not exceed the channel capacity C [14], i.e. R ≤ C. This capacity is given by the well-known formula derived by Shannon in 1948 [75]
C = B · log2 (1 + SNR) (2–3) where B is the transmission bandwidth and SNR is the signal-to-noise ratio. From Equation 2–3, it is clear that UWB systems have the potential to achieve HDR due to their large transmission bandwidth. This is illustrated in Figure 2-7 which shows the theoretical channel capacity as a function of SNR for an UWB system with B = 500 MHz and current Wi-Fi systems. From the figure, it is easy to see that even at low values of SNR, UWB systems can still offer relatively large data rates as a result of the their large
32 Figure 2-8: Example of UWB low interference with narrowband and wideband signals. bandwidth. For instance, in theory, the UWB system (B = 500 MHz) with SNR= 0.5 dB has the same channel capacity than Wi-Fi 802.11ac (B = 80 MHz) at SNR= 20 dB.
2.5.2 Low Interference
As explained in section 2.4, the EIRP limit for UWB radios is −41.3 dBm/MHz which is the noise floor. Due to this power limit and large bandwidth, UWB signals appear as regular channel noise to traditional narrowband and wideband radios operating in the same frequency band, that is, UWB signals produce very low interference to in-band radios. This is illustrated in Figure 2-8.
2.5.3 Multipath Robustness
UWB pulses have a very short duration. This makes UWB systems less sensitive to multipath propagation than narrowband systems that use wider pulses. The reason is that the pulse propagating through a non-LOS(NLOS) path has a very small window of opportunity to collide with the pulse propagating through the LOS path which causes signal degradation [58]. To illustrate this concept, an example is shown in Figure 2-9. Let us assume that a transmitted pulse propagates only through a LOS path and a NLOS path as shown in Figure 2-9(a) with travel distances of 10 m and 11 m, respectively. Then, assuming a propagation speed equal to the speed of light c = 3 × 108 m/s, the signal propagating
33 (a) Propagation paths (b) Arrival time profile Figure 2-9: Example of the UWB robustness against multipath propagation
through the LOS path arrives at the receiver after 33.3 ns while the NLOS signal arrives 3.3 ns later. The wide pulse representing a narrowband signal in Figure 2-9(b) has a time duration larger than the difference in the arrival times of the multipath signals which causes a collision. In contrast, the narrow pulse in the same figure has a smaller duration than 3.3 ns and thus no collision occurs.
2.5.4 High Interference Rejection
An approximate measure for the capability of a system to reject interference is the processing gain (PG. HigherPG results in greater ability to suppress in-band interference [66]. A common way to definePG is
R T PG = c = s (2–4) Rs Tc
where Rc = 1/Tc is the chip rate (Tc is the chip duration or pulse width) and Rs = 1/Ts is the symbol rate (Ts is the symbol period). In UWB systems, very narrow pulses are used in order to generate large signal bandwidths and, hence, Tc is very small in comparison to Ts. Consequently, the Ts/Tc ratio, sometimes called the spreading factor, is usually quite large.
34 In multiple access (MA) applications, in particular, highPG is desired since the dominant interference comes from in-band signals. The jam resistance (JR)12 margin offers a measure of how capable a system is when rejecting in-band interference and can be defined as [95]
JR = PG − SIRmin [dB] (2–5) where SIRmin is the minimum signal-to-interference ratio (SIR) required to meet a de- sired system performance. Clearly, UWB is an attractive technology forMA applications since it can provide highPG resulting in a high resistance to narrowband interference signals.
2.5.5 Low-Cost, Low-Complexity, and Low-Energy Architectures
The low transmission power and very large signal bandwidth of UWB radios bring advantages in the hardware implementation such as small antennas and other passive elements, relatively simple architectures, and low-energy operation. The high frequency band (3.1 − 10.6 GHz) allocated for UWB results in signals with small wavelengths. This, in turn, helps reducing the size of antennas since it is typically proportional to the signal wavelength. In addition, passive elements such as inductors and capacitors used mainly for impedance matching and resonance are also reduced in size due to the UWB high frequency band.13 Having smaller antennas and passive elements significantly reduce the size of UWB integrated circuits resulting in a considerable reduction in the cost of manufacturing.
12 Although the term “jamming” nowadays is usually used to refer to an intentional attempt of disrupting a communication, in the past, it was often used as a synonym to in-band interference. Thus,JR can be interpreted as interference resistance.
13 Circuits designed for higher frequencies need smaller values√ of inductance (L) and capacitance (C) as the resonance frequency is proportional to 1/ L·C.
35 The pulse-based transmission of UWB systems allows for low-complexity archi- tectures. For instance, pulses can be generated directly in the UWB frequency band without requiring frequency translation [84]. This eliminates the need of an oscillator for frequency up-conversion reducing then the complexity of the transmitter and its energy consumption. Similarly, in the receiver end, oscillators for down-conversion can be omitted by employing non-coherent modulation schemes such as energy-detection (discussed in detail in Chapter3). In addition, the FCC limit of −41.3 dBm/MHz on the EIRP of UWB radios implies a very low power transmission which reduces the need of power amplifiers in the transmitter architectures. In general, the low transmission power and the potential low complexity of UWB radios result in low-energy and low-cost systems. These add to the list of benefits that make UWB a very attractive technology for short-range wireless communications.
2.6 Common Digital Modulation Schemes For UWB Radios
This section briefly discusses several basic modulation schemes used in UWB digital systems. These can be divided in two main groups: coherent and non-coherent modulation. Coherent modulation exploits the phase and shape of the carrier signal in order to transmit information. Non-coherent modulation, in contrast, uses only the instantaneous power of the signal eliminating the need for coherent carrier recovery[86]. Both coherent and non-coherent modulation techniques have advantages over each other (see Table 2-2) and choosing one over the other will strongly depend on the target application. For instance, in terms of data rate and bit-error rate (BER), coherent modulation schemes will typically provide a better system performance [67]. However, non-coherent modulation schemes require less energy to operate and can be realized with relatively simpler architectures mainly due to the fact that no coherent carrier recovery is needed.
36 Table 2-2: Comparison between coherent and non-coherent modulation schemes
Parameter Coherent Non-coherent Carrier Recovery Yes No Energy per bit Higher Lower Data Rate Higher Lower BER Lower Higher Complexity Higher Lower
2.6.1 Coherent Modulation Schemes
In digital radios, phase shift keying (PSK) is a commonly used coherent modulation technique. As its name suggests, the phase of the signal carries the digital information. A PSK signal can be modeled as
si (t) = α (t) · cos (2 · π · fc · t + φi ) (2–6) where α (t) is the envelope of the signal, fc is the center frequency, φi is the phase of the signal corresponding to the ith modulation state of a single symbol. For instance, the popular binary-PSK(BPSK) scheme uses two phases, i.e. φ1 = 0 and φ2 = π, to represent a binary bit. For these phase values, s1 (t) = α (t) · cos (2 · π · fc · t) and
14 s2 (t) = −s1 (t). This is shown in Figure 2-10 where s1 (t) and s2 (t) represent a binary logic 0 and 1, respectively. Quadrature-PSK(QPSK) is another common digital modulation scheme employed in UWB systems. It follows the same principle as BPSK but it uses four distinct phases, i.e. φ1 = 0, φ2 = π/2, φ3 = π, and φ4 = 3π/2, to represent four modulation states, that is, a 2-bit symbol.
14 For φ1 = 0 and φ2 = π, the signals representing the binary states (0 and 1) are oppo- site in sign and, hence, BPSK is sometimes interpreted as ASK or PSM.
37 Figure 2-10: Examples of coherent modulation schemes for UWB communications
Although less common than PSK, pulse-shape modulation (PSM) is a coherent scheme that has been proposed for UWB communications [24, 43, 52]. This modulation method, instead of signal phases as in PSK, uses different pulse shapes to represent each modulation state of the symbol to be transmitted. Orthogonal-PSM(OPSM) is a typical way of implementing this type of modulation scheme[40, 89]. It utilizes pulses that are orthogonal to each other. An example of OPSM is shown in Figure 2-10.
2.6.2 Non-Coherent Modulation Schemes
In many cases, coherent digital modulation schemes derived from conventional narrowband systems (e.g. BPSK and QPSK) are not feasible to implement low-power UWB radios [86]. This has led researchers to shift towards non-coherent schemes due to the potential of very low-power radio implementations. Pulse-position modulation (PPM), probably the most common modulation technique found in the UWB literature [37], uses the position in time of a pulse to represent the modulation states of the symbol to be transmitted. With PPM, a pulse is located in
38 one of two time slots to represent each modulation state (0 or 1) of a 1-bit symbol. An example of PPM is shown in Figure 2-11(a). Another common and very simple modulation scheme used for UWB communica- tions is on-off keying (OOK). With this modulation technique, a pulse and its absence are used to represent each modulation state of a binary bit. An example of OOK is illustrated in Figure 2-11(a). Although less frequent, other non-coherent modulation schemes for UWB that can be found in the literature are frequency shift keying (FSK) [65, 78] and amplitude shift keying (ASK) [53]. FSK uses different center frequencies to represent two or more modulation states. Figure 2-11(b) shows an example for a 1-bit symbol in which the lower frequency represents a binary logic 0 while the higher frequency represents a logic 1. With ASK, the information is modulated in the amplitude of the signal, that is, each amplitude value represent a modulation state. Although ASK is not commonly in UWB applications, it is worth mentioning as OOK and even BPSK can be considered ASK. To
illustrate this, let the ASK signals s0 (t) and s1 (t) represent a binary logic 0 and logic 1, respectively, where αi · cos (2 · π · fc · t) , 0 ≤ t ≤ tp si (t) = (2–7) 0, otherwise
th fc is the center frequency and αi is the signal amplitude corresponding to the i modulation state (in this case, 0 or 1). If α0 = 0 and α0 = 1, then ASK resembles OOK.
On the other hand, if α0 = −1 and α1 = 1, then ASK appears as BPSK. 2.6.3 Other Coherent and Non-Coherent Modulation Schemes
Over the last decade, several modulation schemes have been developed for UWB communications in order to achieve higher data rates and improve the BER performance. These go beyond the scope of the discussion in this work. Nevertheless,
39 (a) Most common schemes
(b) Other schemes
Figure 2-11: Examples of non-coherent modulation schemes for UWB communications
40 a few of them are mentioned next with references that the reader may look up for further information. In the non-coherent techniques domain, transmitted-reference (TR) signaling [45, 69] is an often used technique to achieve a non-coherent phase comparison of the carrier signal at the receiver. Similar to PSK schemes, it uses the phase of the carrier signal to modulate binary information. However, in contrast to PSK, a reference signal (sometimes called signal template) is transmitted along with the signal carrying the information. At the receiver, the signals can be correlated to perform the phase comparison eliminating the need for coherent carrier recovery. Another modulation scheme able to compare phases using a non-coherent de- modulation is differential-PSK(DPSK) [11, 44]. 15 With this modulation technique, the information is modulated using the difference in phases of the carrier signal. At the receiver, similar toTR signaling, the current signal and the previous signal can be correlated to determine the change in phase and, hence, demodulate the signal. In the coherent techniques domain, some interesting modulation schemes are bi-orthogonal keying (BOK) [59] and the recently proposed dual-carrier modulation (DCM) [70]. Similar to OPSM, BOK utilizes different pulse shapes to modulate binary information. However, the set of pulses are bi-orthogonal to each other rather than orthogonal as in OPSM. On the other hand, DCM uses two QPSK symbols and a dual-frequency carrier to modulate the binary information.
2.7 UWB Applications
The vast benefits offered by UWB communications qualify this technology as a promising alternative to existing and future short-range wireless applications. As shown in Figure 2-12, UWB can be used to implement a variety of today’s wireless technologies
15 For PSK, both coherent [19] and non-coherent [44] implementations can be found in literature.
41 Figure 2-12: Some UWB wireless applications. such as WSN, WBAN, WPAN, RFID, and radio localization systems. Table 2-3 shows some of these wireless technologies and how they can mainly benefit from the UWB advantages discussed in section 2.5. For instance, WBAN and RFID require simple architectures that consume low energy and produce very small interference to other wireless systems. Therefore, they benefit mainly from the low energy operation, low complexity, and low interference that UWB communications offer. UWB technology has been used in the last few years to realize wireless systems that were traditionally accomplished with narrowband (sometimes with wideband) communications. This demonstrate the feasibility of UWB as the alternative technology to current wireless systems. Take for example HDR wireless systems (e.g. Wi-Fi). In [97, 98], two UWB transceivers are reported to achieve data rates of up to 400 Mbps which is far more than the 150 Mbps offered by the popular 802.11n Wi-Fi standard and close to the 450 Mbps offered by the recent 802.11ac standard. Similarly, in [5, 17], [32, 83], [33, 62], and [81, 96] (see Table 2-3) wireless radios were implemented for WSN, WBAN, RFID, and positioning systems, respectively, demonstrating the ability of UWB communications to be an alternative to current technology in different wireless applications.
42 Table 2-3: Wireless applications and their potential benefits from UWB technology
Examples Multipath Low Low Low Low High Application in HDR Robust- Comple- Interfe- Energy SNR PG Literature ness xity rence WSN[5, 17] WBAN[32, 83] RFID[33, 62] WPAN[97, 98] Localization [81, 96]
2.8 UWB Channel Modeling
2.8.1 Large-Scale Path Loss for Indoor UWB Channels
Wireless propagation models often use analytical expressions or fitting curves to recreate empirical data measured in different environments. Indoor propagation models in particular have been extensively studied over the years and can be frequently found in the literature (e.g. [21, 35, 79, 93]). All of them agree that the average received power decreases exponentially with distance. Therefore, the average path loss PL can be approximated by the log-distance model given by [66]
PL (d) = PL0 + 10 · γPL · log10 (d/d0) , [dB] (2–8) and it is commonly used to estimate the average path loss as a function of distance d .
In Equation 2–8, d0 is the reference distance (usually chosen as 1 or 3 meters), PL0 is
the path loss at d0 , and γPL is the path loss exponent which indicates the slope of the average increase in path loss. One problem with the log-distance path loss model is that it does not take into account that the environmental clutter differs from one location to another resulting in different path losses even when the T-R separation distance is the same. This phe- nomenon is often called shadowing (or log-normal shadowing). Empirical observations
43 Table 2-4: Path loss parameters for UWB channels in residential and commercial build- ings
Residential Commercial Parameter LOS NLOS LOS NLOS
d0 (m) 1 1 1 1 PL0 (dB) 45.9 50.3 43.7 47.3 γPL (dB) 2.01 3.12 2.07 2.95 XPL std. dev., σ (dB) 3.02 3.8 2.3 4.1
have shown that the path loss has a random component and follows a log-normal distri- bution [15]. Thus, the path loss PL for indoor environments is better represented by the log-normal shadowing model described by
PL (d) = PL0 + 10 · γPL · log10 (d/d0) + XPL, [dB] (2–9)
where XPL is a zero-mean Gaussian random variable with standard deviation (std. dev.) σ that models the shadowing effect. Using Equation 2–9, an UWB indoor path loss model for residential and commercial buildings was presented in [35]. Table 2-4
shows the values of the model parameters PL0 , γPL , d0 , and XPL for LOS and NLOS measurements in residential and commercial buildings. This log-normal shadowing model using the values in Table 2-4 is illustrated in Figure 2-13. Figure 2-13(a) shows PL as a function of T-R separation distance in residential buildings for both LOS and NLOS. Similarly, Figure 2-13(b) also shows PL but this time using the parameters values for commercial buildings. In both figures, the solid lines represent the average path loss PL and the dashed lines enclose the 98% confidence-interval region (i.e. ±2.33 · σ).
2.8.2 Small-Scale Fading Model for Indoor UWB Multipath Channels
In contrast to narrowband systems, in UWB systems the sampling period is much smaller due its wideband nature and, hence, the number of resolvable multipath
44 (a) Residential (b) Commercial Figure 2-13: Path loss for UWB channels in residential and commercial buildings components within this period is too small to justify its approximation under the CLT. Therefore, it is often argued that Rayleigh and Rice fading16 are not good small-scale fading models for UWB wireless channels. Despite of this argument, there are empirical measurements that support the Rayleigh [60] and Rice [48] distributions to model UWB multipath fading scenarios. Furthermore, according to [74], the Akaike Information Criterion (AIC) [2] supports Rayleigh and Rice amplitude distributions to adequately model the UWB channels measured by the authors. Although there are measurements that support Rayleigh and Rice distributions to model small-scale fading for some UWB scenarios, extensive work can be found in the literature using other distributions to model UWB channels more accurately. Some of these distributions are Nakagami [13], Weibull [36], and log-normal [29]. Among these, the most commonly found in literature is the log-normal distribution probably because it is the one adopted by the IEEE P802.15.3a task group in its 2003 final report [27].
16 For narrowband systems, Rayleigh and Rice are probably the most commonly used distributions to model small-scale fading.
45 Therefore, to model multipath fading, the IEEE P802.15 model is used in this work and it is presented next. The IEEE P802.15.3a model is derived from the Saleh-Valenzuela model [72] with minor modifications. It consists of the discrete-time channel impulse response (CIR) given by
L K X X i i i hi (t) = Xi · αk,l · δ t − Tl − τk,l (2–10) l=0 k=0 i i th where αk,l are the multipath gain coefficients, Tl is the delay of the l cluster,
i th th τk,l is the delay of the k multipath component relative to the l cluster arrival time
i 17 th (Tl ), δ (·) is the delta function, i refers to the i realization, L is the total number of clusters, and K is the number of rays (multipath components) within the lth cluster. The
random variable Xi represents shadowing and follows a log-normal distribution such that