Smart Antenna Engineering

Transcription

1

2 Smart Antenna Engineering

3 For a complete listing of recent titles in the Artech House Mobile Communications Series,turn to the back of this book.

4 Smart Antenna Engineering Ahmed El Zooghby a r techhouse. com

5 Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the U.S. Library of Congress. British Library Cataloguing in Publication Data El Zooghby, Ahmed Smart antenna engineering. (Artech House mobile communications series) 1. Antennas (Electronics) 2. Software radio I. Title ISBN-10: Cover design by Yekaterina Ratner The material covered in this book represents the views of the author, and does not necessarily reflect those of QUALCOMM Incorporated unless it is so indicated ARTECH HOUSE, INC. 685 Canton Street Norwood, MA All rights reserved. Printed and bound in the United States of America. No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the publisher. All terms mentioned in this book that are known to be trademarks or service marks have been appropriately capitalized. Artech House cannot attest to the accuracy of this information. Use of a term in this book should not be regarded as affecting the validity of any trademark or service mark. International Standard Book Number:

6 Contents Preface xiii Acknowledgments xvii 1 Introduction Wireless Mobile Communications Systems Global Mobile Market Growth Alternatives for Meeting Data Demand Technology Peak Rates and Throughput Why Smart Antennas? Benefits of Smart Antennas Types of Smart Antennas Switched and Fixed Beam Antennas Adaptive Arrays 10 References 11 2 Multiple Access Techniques for 2G and 3G Systems Introduction Multiple Access Wireless Communications FDMA Systems TDMA Systems 15 v

7 vi Smart Antenna Engineering Frequency Reuse Cochannel Interference CDMA Systems Fundamentals of CDMA Isolated Cell Capacity CDMA Codes IS-95 CDMA Systems Third Generation Systems CDMA WCDMA HSDPA Basic CDMA Procedures Acquisition State Idle State Access State and Call Setup Traffic or Dedicated State CDMA Embedded Cell Capacity Multipath Fading Coverage Versus Capacity Trade-Off Coverage-Capacity Trade-Off in the Uplink Conclusion 57 References 57 Selected Bibliography 59 3 Spatial Channel Modeling Introduction Radio Environments and Cell Types The Multipath Channel Channel Characterization Path Loss Models Okumura-Hata Propagation Models Spatial Channel Modeling Spatial Channel Model Parameters 68

8 Contents vii Number of Clusters Spatial Distribution of Clusters and Scatterers Base Station Azimuth Power Spectrum and Angle Spread Mobile Station Azimuth Power Spectrum and Angle Spread Spatial Channel Model Application in System Simulations Angle Spread Impact 77 References 80 Selected Bibliography 81 4 Fixed Beam Smart Antenna Systems Introduction Conventional Sectorization Limitations of Conventional Sectorization Antenna Arrays Fundamentals Broadside and End-Fire Arrays Impact of Number of Elements Impact of Element Spacing First Null Beamwidth Half-Power Beamwidth Array Directivity Array Gain Trade-Off Analysis Impact of Element Pattern Planar Arrays Beamforming The Butler Matrix Spatial Filtering with Beamformers Switched Beam Systems Multiple Fixed Beam Systems Adaptive Cell Sectorization in CDMA Systems 114 References 116

9 viii Smart Antenna Engineering 5 Adaptive Array Systems Uplink Processing Diversity Techniques Angle Diversity Maximum Ratio Combining Adaptive Beamforming Fixed Multiple Beams Versus Adaptive Beamforming Downlink Processing Transmit Diversity Concepts Transmit Diversity in 3G CDMA Standards Downlink Beamforming Spatial Signature-Based Beamforming DOA-Based Beamforming Maximum SNR Conclusion 149 References 151 Selected Bibliography Smart Antenna Receivers and Algorithms for Radio Base Stations Reference Signal Methods The Least Mean Square Algorithm The Recursive Least Squares Algorithm Blind Adaptive Beamforming Least Squares Constant Modulus Algorithm Decision-Directed Algorithm Cyclostationary Algorithms Conjugate Gradient Algorithm Lagrange Multiplier Method Comparison of Adaptive Algorithms Neural Network DOA-Based Beamforming Generation of Training Data Performance Phase of the RBFNN 174

10 Contents ix 6.3 Angle Spread Impact on Optimum Beamforming Downlink Beamforming Vector Rake Receivers Channel Estimation Beamforming Conclusion 185 References Coverage and Capacity Improvements in 3G Networks Introduction Link Budgets and Coverage Mobile Station Parameters Base Station Parameters System Parameters Margins Other Parameters Fade Margin Confidence (Cell Area) CDMA Traffic Loading Voice Services Uplink Budgets Downlink Budgets Data Applications Limiting Links for Coverage and Capacity Coverage Limited Scenarios Capacity Limited Scenarios Smart Antennas Impact on Uplink Coverage and Capacity Smart Antenna Impact on Downlink Capacity Conclusions 226 References 227

11 x Smart Antenna Engineering 8 Smart Antennas System Aspects Introduction Third Generation Air Interfaces and Protocol Stacks Physical Layer Data Multiplexing Transmit Chain UL/RL PN Scrambling/Spreading DL/FL Physical Channel Formatting Mobile Call States WCDMA CDMA Mobility Procedures to Support High-Speed Data Transfer Cell_FACH State or Control Hold Mode Idle, Cell_PCH, or URA_PCH States Procedures to Reestablish High-Speed Data Transfer Cell_FACH State or Control Hold Mode Idle Mode, Cell_PCH, or URA_PCH States Packet Data Services WCDMA Approach CDMA2000 Approach Pilot Channels CDMA WCDMA Channels Applicable for Downlink Beamforming Overview of Major Radio Network Algorithms Power Control Initial Power Setting Admission Control Congestion Control Soft/Softer Handoff Hard Handoff System Impact of Advanced Spatial Techniques Transmit Diversity 247

12 Contents xi Fixed Beam Approach Beam Steering/Adaptive Beamforming Channel Estimation at the Mobile Advantages and Disadvantages Uplink Beamforming Conclusion 261 References Mobile Stations Smart Antennas Introduction Multiple-Antenna MS Design Combining Techniques Selection (Switched) Diversity Maximal Ratio Combining Adaptive Beamforming or Optimum Combining RAKE Receiver Size Mutual Coupling Effects Dual-Antenna Performance Improvements Downlink Capacity Gains Conclusions 286 References MIMO Systems Introduction Principles of MIMO Systems SISO SIMO MISO MIMO Transmission Strategies Water Filling Uniform Power Allocation 296

13 xii Smart Antenna Engineering Beamforming Beam Steering MIMO Approaches MIMO Advantages and Key Performance Issues RF Propagation Characterization SINR Environment Spatial Multiplexing Conclusion 302 References 303 List of Acronyms 305 About the Author 311 Index 313

14 Preface Mobile and wireless communications systems are becoming increasingly more complex in an effort to cope with the growing demand for more supportable peak data rates, coverage requirements, and capacity objectives, as well as exciting new applications such as wireless multimedia and anywhere-anytime mobile Internet access. Although new air interface standards and access technologies such as code division multiple access (CDMA2000), wideband code division multiple access (WCDMA), and their evolutions, including evolution data optimized (EV-DO) and high-speed downlink packet access (HSDPA), promise to meet these requirements with data rates up to several megabits per second, this is often achievable only under ideal channel conditions assumptions are rarely encountered in real systems deployment. Smart antennas have great potential in overcoming the impairments of these systems by exploiting the spatial domain to reduce the effects of interference, extend the range and coverage of wireless networks, increase system capacity, and achievable data throughout. The area of smart antennas application in wireless communications has received increased attention both in the wireless industry and academia for the past few years. It is an interdisciplinary topic that requires knowledge and skills in areas such as antenna arrays, signal processing, digital communications, radio frequency (RF) engineering, and wave propagation. Today, a large body of literature about the topic exists, although much of this is in the form of complex research papers published across a multitude of technical journals, magazines, and conference proceedings, making it very difficult for a practicing engineer to develop the skills required for a successful design in a reasonable amount of time. With that in mind, this book attempts to close the gap by consolidating and presenting the principles of smart antennas along with the issues associated xiii

15 xiv Smart Antenna Engineering with their application in modern communications systems in an easy-to-follow format. The book s purpose is to explain the principles and techniques of smart antennas application in wireless and mobile communications systems. It presents topics and issues in the design of advanced antennas systems in an easy-to-follow methodology. The book is intended for graduate students in electrical engineering, practicing communications engineers, engineering and product managers, and wireless systems designers. It is intended to provide a useful and needed reference in one place and cover a collection of topics necessary for successful application of smart antennas in wireless systems. The book begins in Chapter 1 with a brief history of wireless communications systems and their drive to achieve increasing demands in terms of coverage and capacity. In Chapter 2, the effects of cochannel interference, multiple access interference, and other impairments affecting existing and future multiple access techniques of 2.5 and third generation (3G) wireless systems are discussed to show how they prevent these systems from achieving their full potential of range and system capacity. Models for the mobile radio propagation channel are integral tools that allow system designers to evaluate the benefits of different measures for enhancing system performance. The coverage of smart antennas would not be complete without addressing models that take the spatial domain into account. In Chapter 3, shortcomings of conventional models will be outlined, along with a description of spatial directional channel models adopted by the industry s standards bodies. Interference reduction with smart antennas offers an efficient way to reduce the interference in mobile communications systems through the use of narrow beams directed to a cluster of users or an individual user while, at the same time, steering nulls toward interfering users. Smart antennas could be divided into two major types, fixed multiple beams and adaptive array (AA) systems. A detailed explanation of these two approaches, along with their advantages and drawbacks, will be covered in Chapters 4 and 5. First, we will provide an overview of the fundamentals of antenna arrays and then show how these concepts tie into schemes like the Butler matrix and adaptive beamforming. We will also discuss diversity techniques and other methods applicable to both the uplink and downlink of wireless mobile communications systems. A daunting task facing any smart antennas developer is selecting the receiver structure and adaptive algorithms most suitable for the application in hand. Today, a large number of proposed methods and technical solutions exist. A comprehensive classification of smart antennas algorithms along with the main implementation issues and trade-offs is presented in Chapter 6, as well as some comparison between the different techniques. In Chapter 7, a section on system performance improvements demonstrates the impact of using smart antennas at the radio base station and potential improvements in terms of coverage and capacity of mobile communications networks. In Chapter 8, we will address the systems aspects of smart antennas and their interaction with various

16 Preface xv network control algorithms such as admission control, power control, and radio resource management. The application of antenna arrays in handsets is discussed in Chapter 9. Finally, the book concludes with a brief overview of multiple input multiple output (MIMO) systems, which combine antenna arrays at both the receive and transmit side to create parallel spatial channels that dramatically increase spectral efficiency and system capacity. Although practicing engineers and designers as well as engineering and product managers are the primary audience for this book, it can be easily adopted as a graduate course textbook in smart antenna applications in mobile communications systems.

17

18 Acknowledgments First, I would like to thank God for the knowledge and strength that made this project possible. I would also like to acknowledge and thank my family and friends for their support throughout this book. In addition, I would like to thank Bo Hagerman, Soren Andersson from Ericsson Research Corporate Unit, and Patrick Lundqvist from Ericsson Wireless Communications, Inc. for their valuable insights and numerous discussions in adaptive antennas for wireless mobile communications. Special thanks go to Professor Christos Christodoulou, chair of the electrical and computer engineering department at the University of New Mexico, for his encouragement and inspiration, which made this work possible. I would also like to thank Dr. Said El Khamy and Dr. Hassan El Kamshoushi from the University of Alexandria in Egypt for their guidance in my early work in adaptive antennas. In addition, I would like to express thanks to Qualcomm Inc. for permission to use some illustrations in this book. I would also like to acknowledge the publishing team at Artech House for their guidance and assistance, as well as the reviewer of this project. I welcome any comments and suggestions for improvement or changes that could be implemented in possible future editions and can be reached at a.elzooghby@att.net. xvii

19

20 1 Introduction Adaptive antennas have been used for decades in areas such as radars, satellite communications, remote sensing, and direction finding, to name a few. For instance, radar and secure communications systems take advantage of the ability of these antennas to adapt to the operating environment to combat jamming. Satellite communications systems have used multiple beam and spot beam antennas for years to tailor their coverage to specific geographic locations. Each of these applications is associated with its own unique set of challenges, such as the channel in which the system operates, the propagation environment, sources of interference, and noise or jamming. In addition, the end goal for which the adaptive antenna is used affects the selection of the type of array, size, adaptive algorithms, and integration with other system components. In this chapter we provide a summary of the status of current mobile cellular communications systems, their various evolution paths, mobile systems growth potentials, as well as an introductory discussion of the benefits and use of smart antennas in 3G cellular communications systems. 1.1 Wireless Mobile Communications Systems In the 1980s and 1990s, wireless cellular and personal communications systems (PCS) began to flourish with the advent of second generation mobile communications systems, or simply 2G, to cope with increasing demands. Early mobile communications systems were based on analog technologies that used frequency division multiple access (FDMA). In multiple access, a number of users access or share the resources of a common source. In FDMA systems, the available spectrum is divided into channels of specific bandwidth [30 khz in the case of 1

21 2 Smart Antenna Engineering advanced mobile phone service (AMPS), the North American analog standard] and users are assigned a pair of these channels for bidirectional communications with a base station (BS). In other words, the resource shared by all users is the bandwidth. Since the available spectrum is finite, there is a fundamental limit on the capacity or number of users that can be served by a cell. It is possible to reuse the whole available spectrum in each cell to maximize the capacity this is called reuse factor of one. However, the base station transmit power required to communicate with all these users plus additional margins to overcome fading caused by multipath creates so much cochannel interference to users in neighboring cells that the signal quality is significantly degraded. To reduce this interference to acceptable levels that support a given signal quality, the number of channels assigned to each cell must be decreased in other words, the reuse factor must be increased. This, of course, will lower the overall system capacity. Engineers then turned to technologies based on digital techniques to solve this trade-off between capacity and interference. In time division multiple access (TDMA), each user is assigned the entire resource at specific time slots. In this case, the shared resource is time. Global systems for mobiles (GSM) are based on this technology and it uses channels with bandwidth of 200 khz. In TDMA-based systems, frequency planning plays an important role in balancing system capacity versus cochannel interference. Another multiple access technique based on spread spectrum technology is CDMA, in which the code domain is shared among users as defined in the IS-95 standard. One major difference between CDMA systems and other multiple access technologies is their re-use factor of one, which enables them to offer higher capacities. This is possible because of the unique way in which CDMA handles interference. A combination of pseudonoise (PN) sequences and orthogonal codes are used to spread and channelize the base station and user s data. Spreading the signal to a much wider bandwidth helps reduce the power levels and makes each signal appear as background noise to other users. This scheme allows a large number of users to simultaneously share the same 1.25-MHz carrier. In addition to spreading, CDMA systems use power control techniques to maintain the interference in the system at the acceptable levels required to satisfy the signal or radio link quality. Furthermore, CDMA systems take advantage of multipath through the use of RAKE receivers to combat fading. Due to the explosion of mobile communications demand and the increasing shift to offer new and advanced services based on high-speed data rates, third generation technologies were developed. The main goals of 3G systems are to increase the voice capacity, improve mixed voice and data services, and offer peak data rates of up to 2 Mbps. There are currently two major 3G technologies, both based on CDMA. These are wideband CDMA or WCDMA, also known as universal mobile telecommunications system (UMTS) [1], and CDMA2000 [2]. Peak data rates of 384 Kbps are being achieved in commercially deployed WCDMA networks, whereas the WCDMA

22 Introduction 3 evolution path with HSDPA and high-speed uplink packet access (HSUPA) extends the peak rate to 14.4 Mbps on the downlink and more than 4 Mbps on the uplink, respectively, in a 5-MHz carrier. Similarly, peak data rates of Kbps in a 1.25-MHz carrier are being achieved on the currently deployed CDMA2000 1x networks. The CDMA2000 1xEV-DO standard further extends the peak rates to 3.1 Mbps and 1.8 Mbps on the downlink and uplink, respectively. Both 1xEV-DO and HSDPA technologies were developed to significantly increase the peak data rates to meet the rapidly growing demand for high-speed data applications. The basic concept behind both technologies is the same, namely the introduction of new features such as adaptive modulation and coding (AMC), short frames, multicode operation, fast L1 hybrid automatic repeat request (HARQ), and base station scheduling. In fact, these features replace the two basic CDMA features, namely variable spreading factor (VSF) codes and fast power control by adaptive rate control based on channel conditions. AMC is a fundamental feature of HSDPA and 1xEV-DO. It consists of continuously optimizing the code rate, the modulation scheme, the number of codes employed, and the transmit power per code based on the channel quality reported [channel quality indicator (CQI) feedback] by the mobile station. To achieve very high data rates, higher order modulation schemes such as 16 QAM is added to the existing quadrature phase shift keying (QPSK) modulation used for R 99 WCDMA and CDMA20001x channels. Different combinations of modulation and the channel coding-rate can be used to provide different peak data rates. Essentially, when targeting a given level of reliability, users experiencing more favorable channel conditions (e.g., closer to the base station) will be allocated higher data rates. According to industry bodies, at the beginning of 2005, global subscriptions to 3G/UMTS networks reached 16 million on more than 60 networks, whereas more than 180 million subscribers are using CDMA2000 on approximately 120 networks. 1.2 Global Mobile Market Growth At the end of 2004, worldwide cellular subscriptions passed the 1.4 billion mark and the rapid growth is expected to last for many years. The chart in Figure 1.1 shows that the number of worldwide cellular users is expected to reach nearly 2.5 billion by 2010 [3], while Figure 1.2 provides a breakdown of this forecast for CDMA technologies. Note that this breakdown does not include GSM and EDGE subscribers. This continued growth and evolution in mobile usage is driven by data services such as short message service (SMS), multimedia messaging service (MMS), downloadable ring-tones, images and games, news and information

23 4 Smart Antenna Engineering 3000 Wordwide cellular users (millions) Rest of world Central and Eastern Europe Central and Latin America Asia Pacific Western Europe North America Figure 1.1 Worldwide cellular users forecast [3] CDMA worldwide cellular users (millions) WCDMA CDMA2000 1xEV CDMA2000 CDMAOne Figure 1.2 Worldwide CDMA cellular users forecast [3]. sources, mobile chat sites, and Web portals. It is anticipated that voice services will still significantly contribute to revenue streams along with new 3G enabled services, including personalized access to information and entertainment services, mobile access to the Internet and corporate networks, location based services, and rich voice, which is the simultaneous transmission of photos, graphics, video, maps, documents, and other forms of data with pure voice. The chart in Figure 1.3 shows how the worldwide mobile voice traffic is expected to increase during the next few years to nearly three times the current levels by Alternatives for Meeting Data Demand Different wireless service providers have different evolution paths with different technology choices to upgrade their 2G networks to third generation systems defined in the IMT-2000 standard of the International Telecommunications

24 Introduction 5 350% 300% 250% Voice traffic growth 200% 150% 100% 50% 0% Figure 1.3 Projected voice traffic growth [3]. Union (ITU). The main evolution paths for GSM and CDMAOne operators are shown in Figure 1.4. A number of GSM operators have chosen a migration path that involves upgrading their networks to GPRS and EDGE as an interim step before a full WCDMA migration while others have chosen to evolve their networks directly to WCDMA. CDMAOne operators have a somewhat smoother migration path with CDMA2000. Eventually, to meet the growing demand for voice and data capacity, most current 2G networks will be upgraded to use CDMA. Figure 1.5 shows the global 3G cellular users forecast by technology until G 2.5G 3G Evolved 3G GSM Voice centric 9.6 Kbps Data 40 Kbps cdmaone Voice centric 9.6/14.4 Kbps GPRS EDGE Data 120 Kbps WCDMA R'99 Voice + data 384 Kbps Voice + data Kbps CDMA2000 1xRTT HSDPA Data DL: 14.4 Mbps/ UL: 384Kbps Data DL: 2.4 Mbps/ UL: Kbps 1xEV-DO Rev. 0 HSUPA Data DL:14.4/ UL: 4.3 Mbps Data DL: 3.1 Mbps/ UL: 1.8 Mbps 1xEV-DO Rev. A 1xEV-DV Figure 1.4 2G evolution paths toward 3G.

25 6 Smart Antenna Engineering G worldwide users (millions) WCDMA CDMA2000 1x CDMA2000 1xEV Figure 1.5 3G cellular users forecast by technology. 1.4 Technology Peak Rates and Throughput As we can see from Figure 1.4, different technologies support different peak data rates. The peak rate is the maximum transmission speed an individual user may experience under ideal conditions (i.e., it only affects the user experience). Data throughput, on the other hand, is a far more important metric for performance. Sector throughput is the average total capacity available to multiple users, whereas user throughput is the average data rate a user may experience. As the sector throughput increases, each sector can handle higher volumes of data, the network requires fewer sites, and, consequently, the capital and operational expenses are also reduced. Table 1.1 compares the peak data rates and throughput for different 3G technologies. Table 1.1 3G Technology Comparisons Technology Carrier Bandwidth/ Spectrum (MHz) Downlink Peak Data Rate (Kbps) Average User Throughput (Kbps) CDMA2000 1x 1.25/ CDMA2000 1xEV-DO Rev /1.25 2, WCDMA 3.84/ HSDPA /- 5 14,

26 Introduction 7 For CDMA2000 1x and CDMA2000 1xEV-DO, user throughputs listed in Table 1.1 are based on promotional material from North American operators and on real network deployments. WCDMA and HSDPA user throughputs are based on results from [4-8]. Unlike EV-DO systems, there are no commercial deployments of HSDPA systems yet; these systems are expected to deploy in late 2005 and into The user throughput for HSDPA is based on simulation data [5]. Moreover, the choice of scheduler significantly affects the throughput of both 1xEV-DO and HSDPA because of the adaptive modulation and coding nature of the technologies. For instance, one popular scheduler called the proportional fairness (PF) schedules users according to the ratio between their instantaneous achievable data rate and their average served data rate. This results in all users having equal probability of being served even though they may experience very different average channel quality. This scheme provides a good balance between system throughput and fairness. Other schedulers will be discussed in Chapter Why Smart Antennas? Achieving the peak data rates specified in each standard in a real system remains very unlikely because it would require an unloaded system serving a single user to be extremely close to the base station. This leads to two questions: why the increased interest in smart antennas a more attractive name for adaptive antennas and how are they being considered as a viable technology for applications such as mobile communications? As we have seen, operators are faced with increasing capacity demands for both voice and data services. Although various 3G technologies offer higher data rates and double voice capacity compared with their 2G counterparts, their actual performance is still susceptible to interference, and adverse channel conditions created by multipath propagation and system loading. As such, smart antennas techniques can complement 3G systems and improve their performance by alleviating and reducing the degradation caused by the aforementioned factors. In fact, because of their nature, technologies such as HSDPA and 1xEV-DO can greatly benefit from smart antennas since any improvement in the SNR experienced by the users would directly translate to better throughput for individual users as well as increased sector throughput that can support higher capacities. 1.6 Benefits of Smart Antennas It is a fact that current technologies have nearly maximized the use of temporal and spectral techniques to improve capacity and data transfer speeds. This leaves

27 8 Smart Antenna Engineering an additional parameter that has not been fully tapped yet, namely space. In space division multiple access (SDMA), a user or cluster of users are assigned a dedicated narrow beam that tracks their movement across the cell, adapting to the constantly changing radio environment. The obvious advantage of this approach is its applicability to any multiple access technique. Wireless system design and planning involve the optimization of two major components, coverage and capacity through the manipulation and control of power, interference, and noise. To that extent, smart antennas offer substantial benefits to the design of wireless mobile communications systems, which can be summarized as follows: Increased antenna gain: this helps increase the base station range and coverage, extends battery life, and allows for smaller and lighter handset designs. Interference rejection: antenna pattern nulls can be generated toward interference sources. On the reverse link or uplink this reduces the interference seen by the base station. It also reduces the amount of interference spread in the system on the forward link or downlink. Such improvements in the carrier to interference ratio C/I lead to increased capacity. Diversity: composite information from the array can be used to minimize fading and other undesirable effects of multipath propagation. In addition to spatial and polarization diversity, antenna arrays also allow the use of angular diversity. As with any other adaptive antennas application, the nature of the system in which they are employed, the conditions under which they operate, and the results they are intended to achieve all have to be considered when a smart antenna system design is incorporated in a specific wireless system. Figure 1.6 shows a system overview that describes some of the involved factors when we consider a smart antenna design for mobile communications systems. Subsequent chapters will provide more details and analysis regarding these areas and how they affect the selection, design, and performance of a smart antenna system. 1.7 Types of Smart Antennas Sectorization schemes, which attempt to reduce interference and increase capacity, are the most commonly used spatial technique that have been used in current mobile communications systems for years. Cells are broken into three or six

28 Introduction 9 Network planning Structure and algorithms Structure and algorithms Network dependent parameters Air interface parameters Radio network protocols Transmitter Propagation environment spatial channel modeling interference environment Channel Radio network control Network dependent parameters Air interface parameters Radio network protocols Receiver Figure 1.6 Smart antenna system overview. sectors with dedicated antennas and RF paths. Increasing the amount of sectorization reduces the interference seen by the desired signal. One drawback of current sectorization techniques is that their efficiency decreases as the number of sectors increases due to antenna pattern overlap. Furthermore, increasing the number of sectors increases the handoffs the mobile experiences while moving across the cell. Compare this technique to that of a narrow beam being directed towards a desired user. It is clear that some interference that would have been seen by the existing 120 sector antenna will be outside the beamwidth of the array. Any reduction in the interference level translates into system capacity improvements. Smart antennas could be divided into two major types, fixed multiple beams and AA systems. Both systems attempt to increase gain in the direction of the user. This could be achieved by directing the main lobe, with increased gain, in the direction of the user, and nulls in the directions of the interference [9, 10]. 1.8 Switched and Fixed Beam Antennas The switched beam method is considered an extension of the current cellular sectorization scheme. The switched beam approach further subdivides the macro-sectors into several micro-sectors. Each micro-sector contains a predetermined fixed beam pattern with the greatest gain placed in the center of the

29 10 Smart Antenna Engineering beam. When a mobile user is in the vicinity of a micro-sector, the switched beam system selects the beam containing the strongest signal. During the call, the system monitors the signal strength and switches to other fixed beams if required. Better performance can be achieved with integrated embedded systems of fixed multibeam antennas, which can enhance signal detection on the uplink by making use of the signals from all the available paths in the beams followed by maximum ratio combining (MRC) [11]. The beam receiving the most power in the uplink can be used to transmit to the desired mobile on the downlink. 1.9 Adaptive Arrays The main advantage of adaptive antenna arrays compared with switched beam antennas is their ability to steer beams towards desired users and nulls toward interfering signals as they move around a sector. Several beamforming approaches exist with varying degrees of complexity. A conventional beamformer or delay-and-sum beamformer has all the weights of equal magnitudes. To steer the array in a particular direction, the phases are selected appropriately. In order to be able to null an interfering signal, the null-steering beamformer can be used to cancel a plane wave arriving from a known direction producing a null in the response pattern at this direction. When the number of interferers becomes large, such as in the case of IS-95 based systems, this beamformer might not be a practical approach. The well-known minimum variance distortionless response (MVDR) beamformer attempts to minimize the total output noise while keeping the output signal constant in the direction of the desired user. This is the same as maximizing the output SNR. For an M-element array with M degrees of freedom, the number of interferers must be less than or equal to M 2, since one has been used by the constraint in the look direction. This may not be true in a mobile communications environment with multipath arrivals, and the array beamformer may not be able to achieve the maximization of the output SNR by suppressing every interference source. Some a priori knowledge of the desired signal such as the direction of arrival (DOA) is required by the MVDR beamformer. Since in the MVDR approach the weight vector that minimizes the output power is a function of the spatial correlation matrix, some degree of coherency between the uplink and downlink is needed to provide an estimate of the correlation matrix for transmission. In the minimum mean square error (MMSE) approach a minimization of the square of the difference between the array output and a reference signal results in the weight vector that maximizes the signal quality. Since this approach relies on the inversion of the covariance matrix, its complexity is very high. The maximum likelihood (ML) principle attempts to estimate the data sequence that was most likely sent based on the received or observed data. Other spatial techniques

30 Introduction 11 include transmit diversity and MIMO systems. In MIMO systems, antenna arrays are used in the transmitter as well as in the receiver, and the system creates multiple parallel channels that significantly increase the supportable data rates. Figure 1.7 compares the performance improvement expected from major smart antenna techniques with their complexity. MIMO systems Performance improvements Transmit diversity Higher order sectorization Basic sectorization Fixed multi-beam antennas Beam steering (beam shaping, adaptive nulling) System complexity Figure 1.7 Comparison of major spatial techniques. References [1] Third Generation Partnership Project, [2] Third Generation Partnership Project2, [3] Worldwide Cellular User Forecasts ( ), Strategy Analytics, December [4] The Economics of Wireless Mobile Data, Qualcomm Inc, [5] Data Capabilities: GPRS to HSDPA, Rysavy Research, September 2004, [6] Holma, H., and A. Toskala, WCDMA for UMTS: Radio Access for Third Generation Mobile Communications, 3rd ed., New York: John Wiley & Sons, [7] HSDPA for Improved Downlink Data Transfer, Qualcomm CDMA Technologies, October 2004, [8] Nokia High Speed Packet Access Solution, ZD Net UK, uk/.

31 12 Smart Antenna Engineering [9] Rappaport, T. S., (ed.), Smart Antennas: Adaptive Arrays, Algorithms and Wireless Position Location, New York: IEEE Press, [10] Tsoulos, G.V., (ed.), Adaptive Antennas for Wireless Communications, IEEE Press, [11] Göransson, B., B. Hagerman, and J. Barta, Adaptive Antennas in WCDMA Systems Link Level Simulation Results Based on Typical User Scenarios, IEEE Vehicular Technology Conference, Boston, MA, September 2000.

32 2 Multiple Access Techniques for 2G and 3G Systems 2.1 Introduction Evaluating the various design choices of different smart antennas architectures, algorithms, and performance trade-offs when applied to modern mobile cellular communications systems requires knowledge and understanding of core access technologies as well as the impairments facing different systems. This chapter presents the concepts of FDMA, TDMA, and CDMA and describes the main differences between these access technologies. An overview of the frequency reuse concept and cochannel interference, critical to the network design of some second generation mobile communications systems, is provided. Since all 3G technologies are based on CDMA, there will be greater emphasis on this technology. When evaluating performance issues, two main components are usually considered, the link level performance and system level performance. In link level performance, we are mainly concerned with a single link between a mobile station and the base station; this link is typically based on the physical layer structure of the air interface. The physical layer is the layer that carries the actual RF transmissions. On the other hand, in system level performance the impact of the upper layers and their interactions with the physical layer has to be taken into consideration. Functions performed by the upper layers include radio resource management, admission control, and so on. This chapter has been divided as follows. First, in Section 2.2 we discuss the concepts of FDMA and TDMA systems and how frequency reuse is applied in the design of mobile networks also briefly cover cochannel interference. The fundamentals of CDMA technologies are discussed in Sections 2.3 and 2.4, 13

33 14 Smart Antenna Engineering along with systems aspects such as RAKE receiver, power control, and soft handoff, as well as an overview of the IS-95 air interface. In Section 2.5, we introduce third generation systems and summarize the CDMA2000 and WCDMA standards. An overview of how a CDMA phone works and the different procedures employed to acquire the system and complete a mobile call are introduced in Section 2.6. Since the main motivation for using smart antennas with 3G systems is to improve their coverage and capacity performance, the chapter concludes with Section 2.7, in which the factors affecting CDMA capacity are presented and the coverage versus capacity trade-off is discussed using simple models. In a later chapter, a more complex and specific discussion involving this trade-off will be presented to provide tools to evaluate the different smart antennas gains. 2.2 Multiple Access Wireless Communications In cellular and PCS wireless communications systems a multitude of users access and share network resources (frequency bandwidth) to obtain different types of services, including voice, messaging, and data. The goals of these multiple access communications systems are to provide communications services in a near-universal geographical coverage while minimizing both subscriber stations and network equipment, deployment, and operational costs. Because regulatory agencies have allocated limited bandwidth to these services, a crucial goal of these solutions is to achieve high spectral efficiency, traditionally measured in Erlangs/megahertz/unit service area for voice applications and in bits/second/megahertz/unit service area for data applications. The cellular concept pioneered by Bell Labs in the 1970s makes use of multiple fixed stations, or cells that each serve a number of mobile subscribers within a limited geographical area. When a subscriber moves between cells, over-the-air messaging is used to handoff the call between cells, ensuring its continuity. The first such system in North America was called AMPS. Similar analog systems were also deployed in different parts of the world, including the Nordic Mobile Telephone (NMT) in Scandinavia, and the Total Access Communications System (TACS) used in the United Kingdom, China, and other countries. The spectrum chosen for these systems was in the MHz band. The frequency band allotted to each system was then divided according to a scheme called FDMA FDMA Systems In wireless mobile communications systems subscribers share a common resource such as time, frequency spectrum, power, or code. This is referred to as access technology or channelization. This leads to the generation of interference in the system, which affects signal quality. The degree to which system

34 Multiple Access Techniques for 2G and 3G Systems 15 performance is affected by interference actually depends on the access technology used to separate the users in the network. In FDMA, the available spectrum is divided among users by assigning different frequencies to various users, as shown in Figure 2.1. With FDMA systems, a user is assigned a 30-kHz or a 25-kHz pair of frequencies for the forward link (downlink) and the reverse link (uplink) throughout a call. To maintain the interference between the two links at a minimum, the frequency pair is separated by, for example, 45 MHz and 80 MHz in North American cellular and PCS systems, respectively. The FDMA scheme could be equally applied to analog and digital communications systems TDMA Systems TDMA is a digital transmission technology that allows a number of users to access a single RF channel while reducing interference by allocating unique time slots to each user within each channel. In TDMA systems channelization is provided first by dividing the frequency among the users, just like in FDMA, and then again by dividing users in time by assigning users different time slots. This transmission scheme multiplexes three signals over a single channel. The TDMA standard for cellular divides a single channel into six time slots, with each signal using two slots, providing a 3 to 1 gain in capacity over AMPS. Each caller is assigned a specific time slot for transmission, shown in Figure 2.2. In US TDMA (IS-54), a 30-kHz channel is further divided into three time slots, Time Power f 1 f 2 f 3 f 4 f n Frequency Figure 2.1 The FDMA concept.

35 16 Smart Antenna Engineering Time Power Frequency Figure 2.2 The TDMA concept. which increases the number of simultaneous users per channel to three. In the European TDMA version, or GSM, a 200-kHz channel is divided among eight users. TDMA relies on the fact that the audio signal has been digitized; that is, divided into a number of milliseconds-long packets. It allocates a single frequency channel for a short time and then moves to another channel. The digital samples from a single transmitter occupy different time slots in several bands at the same time. One of the disadvantages of TDMA is that each user has a predefined time slot and users handing off from one cell to another are not allotted a time slot. Thus, if all the time slots in a cell are already occupied, no additional calls are allowed. This represents a hard limit on the cell capacity. Another problem with TDMA is that it is subjected to multipath distortion Frequency Reuse In cellular and PCS systems, a cell s coverage is typically represented by a hexagon when omnidirectional antennas with constant transmit power are used at the base station. As we have seen with FDMA and TDMA systems, the available frequency spectrum is divided among the users in the network. Now, let us assume two adjacent cells with two users assigned frequency f 1. As these mobile stations move closer together, their use of a frequency f 1 will begin to create interference.to overcome this problem, a process called frequency planning is implemented, where a group of frequencies are reused in cells that are separated from one another by distances large enough to maintain the interference at

36 Multiple Access Techniques for 2G and 3G Systems 17 acceptable levels. Frequency reuse is the term that describes how frequencies are allocated throughout the system as a result of frequency planning. Assume a cellular system has F total frequency pairs or duplex channels available for users. By allocating each cell a group of k channels and dividing the F channels among N cells, we get F = kn (2.1) It follows that the cluster of N cells use the complete available band of frequencies. By replicating this cluster several times across the whole system, we can see that the system capacity will be proportional to N, which is also referred to as cluster size. Since each cell is assigned 1/N of the total channels, this factor is called the frequency reuse factor. Since the available spectrum is finite, there is a fundamental limit on the capacity or number of users that can be served by a cell. It is possible to reuse the whole available spectrum in each cell to maximize the capacity; this is called reuse factor of one. However, the base station transmit power required to communicate with all these users plus additional margins to overcome fading caused by multipath creates so much cochannel interference to users in neighboring cells that the signal quality is significantly degraded. To reduce this interference to acceptable levels that support a given signal quality, the number of channels assigned to each cell must be decreased; in other words, the reuse factor must be increased. This, of course, will lower the overall system capacity. Typical cellular reuse assumes N = 7 sets of channels are used, one set in each cell. This seven-cell building block is then repeated over the service area, as shown in Figure 2.3. The design ensures that there are no adjacent cells using the same channel (frequency). Several N-way reuse patterns have been deployed in different networks, including the above seven-way reuse. To calculate the capacity of an N-way reuse pattern, let us consider a 12.5-MHz band in which we need to deploy a cellular AMPS system. The total number of available channels with K = 7 becomes Capacity 12. 5MHz = = 57 channels 30KHz 7 That is, there are approximately 57 AMPS channels available per cell. TDMA systems use the same frequency reuse concept as well but their capacity is higher than that provided by analog systems. The capacity derived above assumes that the cells are using omnidirectional antennas. In practice, cell sites are sectorized, usually into three sectors (i.e., each site is equipped with three sets of directional antennas, with their azimuths separated by 120 ). In practice, sectorization does not lead to an

37 18 Smart Antenna Engineering R D Figure 2.3 N = 7 frequency reuse plan. increase in a sector s capacity in AMPS. This is because the sector isolation, often no more than a few decibels, is insufficient to guarantee acceptably low interference. However, an increase in coverage is possible with sectorization because of the increased gain of the directional antenna but there is no gain in the reuse. The total cell capacity remains at 57 and the sector capacity becomes 19 channels. With this scheme the overall reuse factor (sector-based) becomes K =7*3= Cochannel Interference In FDMA and TDMA-based systems, when signals from cells using the same frequency group interfere with each other they create cochannel interference, which affects the signal quality and system performance. Therefore, these cells must be separated by some distance, which is referred to as cochannel separation D and is given by [1] D = 3 NR (2.2) Under the assumption that the cell sizes and cell transmit powers are the same, cochannel interference becomes a function of the ratio of the separation distance to the cell s coverage distance or D/R, where R is the cell radius [2, 3].

38 Multiple Access Techniques for 2G and 3G Systems 19 This shows that reducing the cochannel interference requires larger cochannel separations. Let K be the number of cochannel interfering cells, then the signal to interference ratio (SIR) could be approximated as SIR S = = I K i = 1 1 ( D R) i n (2.3) where n is the path loss exponent. It can also be shown that most of the cochannel interference results from cells in the first tier. Based on the hexagonal cell shape, we get K = 6, assuming that the cochannel separations are the same, and using (2.2) we can rewrite (2.3) as follows S 1 1 = = I 6 3N N n ( ) 63 ( ) n 2 (2.4) From (2.4) we can clearly see the trade-off that exists between the system capacity and cochannel interference. To illustrate this trade-off let us assume that we have a 12.5-MHz spectrum available and a 30-kHz channel bandwidth. Figure 2.4 shows the relation between the cell s capacity in terms of the number of voice channels and the SIR versus N for n = 4. We can clearly see the trade-off between achieving a high-capacity design versus maintaining an acceptable SIR. Thus, smart antennas become a crucial tool in dealing with such issues, as we will see in subsequent chapters. Figure 2.4 Capacity and SIR versus cluster size.

39 20 Smart Antenna Engineering CDMA Systems As we have seen in previous sections, the most fundamental issue in wireless mobile systems design is how to deal with interference between users. One approach to mitigating interference is using the concept of slotting, in which each mobile user is assigned a frequency or time slot that he, and he alone, uses while he is active, such as in FDMA- and TDMA-based systems. The drawback of this approach is the reduced spectral efficiency inherent in the frequency reuse approach because only a portion of the available spectrum can be used in a given cell at any given time. Another drawback is the need to change the frequency plan when new base stations are added to cope with increased capacity demands. In CDMA, users are divided by the assignment of a unique code to each. Because users can be identified by their unique code, there is no need to divide the spectrum in either frequency or time and all users in a CDMA system are given access to the system at the same time and on the same frequency. This is shown in Figure 2.5, where a number of users share the same RF band using different codes. One major difference between CDMA systems and other multiple access technologies is their reuse factor of one, which enables them to offer higher capacities. This is possible because of the unique way by which CDMA handles interference. A combination of PN sequences and orthogonal codes are used to spread and channelize the base station s and user s data. Radio receivers based on other digital technologies separate channels by filtering in the frequency domain. CDMA receivers separate channels by means of the pseudo-random modulation that is applied and removed in the digital domain, not on the basis of frequency. Spreading the signal to a much wider bandwidth helps reduce the t C n C 3 C 1 C 2 f Figure 2.5 CDMA access technology.

40 Multiple Access Techniques for 2G and 3G Systems 21 power levels and makes each signal appear as background noise to other users. This scheme allows a large number of users to simultaneously share the same 1.25-MHz carrier. In addition to spreading, CDMA systems use power control techniques to maintain the interference in the system at the acceptable levels required to satisfy the signal or radio link quality. Furthermore, CDMA systems take advantage of multipath through the use of RAKE receivers to combat fading. There are currently two major 3G technologies, both based on CDMA, namely, WCDMA and CDMA2000. Let us consider the link between a mobile station and a base station in a CDMA system communicating using a unique code. Because of the characteristics of these codes, namely, orthogonality, the communication is successful despite the interference generated in the system from other mobiles. This is possible because of the way CDMA is designed, where the signals from the other links are filtered out as background noise. So in a way, CDMA mitigates interference between users by accepting the fact that interference is present and optimizing the system to operate in an environment of interference. To achieve this goal, CDMA uses spread spectrum technology. One form of spread spectrum is direct sequence spread spectrum, in which special spreading codes are used to spread out the signal over a wide bandwidth while reducing its power at the same time, as shown in Figure 2.6. A spreading code is applied to the narrowband data at the transmitter, resulting in a signal with a much wider bandwidth. Since the total signal power remains the same, the signal level drops to the noise floor level. After passing through the channel, the signal at the receiver will consist of the wanted signal, multiple access interference, and noise. By applying the same spreading code used in the transmitter to the combined signal, a pulse-like peak results for the wanted signal and a small residual signal level for all interferers. The major advantage of CDMA technology is the potential of extraordinary capacity increase over narrowband multiple access wireless technologies. Idealized models show that the capacity improvement may be as high as 20 times that of the narrowband cellular standards, such as AMPS in North America, NMT in Scandinavia, TACS in the United Kingdom, and 13 times that of TDMA. However, in practice coverage areas are highly irregular, the load is not spatially uniform and is time variant throughout the day, leading to less but still significant capacity improvements. 2.3 Fundamentals of CDMA The key to CDMA high capacity is the use of noise-like carrier waves. Instead of assigning frequency or time slots, different users are assigned different nearly orthogonal instances of the noise carrier. This alters the system sensitivity to

41 22 Smart Antenna Engineering Interference + noise + signal Spreading code I C Data Spreading Channel Despreading C C I C Spread signal Figure 2.6 Direct sequence spread spectrum fundamentals. interference, from having to design a system based on the worst-case interference to the average interference. Traditional time or frequency slotted systems must be designed with a reuse ratio that satisfies the worst-case interference scenario, which is experienced by only a small fraction of users. Use of pseudonoise carriers, with all users occupying the same spectrum, makes the effective noise the sum of all other-user signals. The CDMA receiver correlates its input with the desired noise carrier, enhancing the signal-to-noise ratio at the detector and overcoming the summed noise enough to provide an adequate SNR at the detector. Because the interference is summed, the system is sensitive to the average interference instead of the worst-case interference. Frequency reuse is universal, that is, multiple users use the same CDMA carrier frequency. Capacity is determined by the balance between the required SNR for each user, and the spread spectrum processing gain, defined as the ratio between the carrier chip rate to the user s data rate. The figure of merit of a well-designed digital receiver is the dimensionless E b /N t, defined as

42 Multiple Access Techniques for 2G and 3G Systems 23 E N b t = Energy per bit Noise Power Spectral Density + Interference Power Spectral Density (2.5) The noise part of E b /N t, in a spread spectrum system is the sum of thermal noise and all the other-user interference. Assuming the spectrum of the signals is rectangular, with a bandwidth W, then the noise + interference power spectral density is N t = N + o W P i otherusers (2.6) where the first term represents the thermal noise level of the receiver. We can then rewrite E b /N t in terms of the data rate and the spread-spectrum bandwidth as: E N b t j = N o P + j R W P i otherusers (2.7) The interference in this equation is the sum of the signals from all users other than the one of interest. This equation is the key to understanding how and why CDMA works. Early arguments against CDMA were centered on what is termed the near-far problem. In the mobile radio environment some users may be located near the base station while others may be located at the cell edge. The propagation path loss difference between those extreme users can be on the order of several tens of decibels. Consequently, the difference in the received power and the SNR at the base station from users in those two extreme cases could be as high as 50 or 60 db, if the users are all transmitting at the same constant power. Hence, for the base station to accommodate users at the cell edge, the spreading bandwidth would have to be on the order of 40 db or so, that is 10,000 times the data rate. Using a bandwidth of 100 MHz to support a data rate of 10 Kbps would lead to a much worse spectral efficiency than compared with a narrowband system. Choosing a more reasonable bandwidth would deny service to remote users. The key to the high capacity of commercial CDMA was a simple solution; instead of using constant power, the transmitter s power can be controlled in such a way that the received powers from all users are roughly equal. This works because by controlling the received power, the total interference seen at the base station cannot be dominated by any single user as long as all users have similar data rates. Assuming perfect power control, the interference can be

43 24 Smart Antenna Engineering given by I o =(N 1)Pwhere N is the total number of users and P is the received signal power from each user. The uplink E b /N t now becomes E N b PR = = N + N P W W R ( 1) NW P+ ( N 1) t o o (2.8) N N pole W R N o = +1 (2.9) E b P N t W R = as P (2.10) E N b t Equation (2.10) shows the fundamental dependence of CDMA capacity not only on power control but also on interference reduction techniques such as smart antennas. Capacity can be maximized if we adjust the power control, or more broadly P, so that the SNR is exactly what is needed for an acceptable error rate Isolated Cell Capacity Using (2.10) to solve for N with the assumption that power in unlimited P and a nominal SNR target of 4.5 to 5 db for IS-95 CDMA with 9.6-Kbps data rate, we obtain an uplink pole capacity of 46 to 42, respectively. The pole capacity of a cell is defined as the maximum number of users a cell can support if there is no constraint on the peak received power. In practice, the pole capacity cannot be reached since it implies that the interference is allowed to grow to such high levels that the coverage shrinks to zero. Typically CDMA networks are designed and planned to operate at uplink loads of 50% 60%, levels considered to provide good coverage versus capacity trade-off. Ideally, that leads to users on the uplink with IS-95A CDMA. The actual number of subscribers that 50% or 60% translates to in real networks may vary depending on the data rate selected and fade margin expected, among other factors. Note that since capacity and SNR are reciprocal, a reduction in the required SNR or E b /N t leads to improvement in capacity, and vice versa. CDMA capacity will be discussed in more details in the next sections, along with additional factors that contribute to the actual performance, where we will see that overall there is major improvement over narrowband technologies. Recall that in the same

44 Multiple Access Techniques for 2G and 3G Systems MHz bandwidth, a single sector of a single AMPS cell has only two channels available CDMA Codes Since in CDMA systems all mobiles need to share the same frequency carrier, orthogonal codes called Walsh codes are used to separate between users and different communications channels within a cell; that is, they provide channelization on the forward link. This is essential in CDMA to avoid or at least minimize multiple access interference in the forward link. Walsh codes are orthogonal binary sequences generated using the Hadamard matrix as follows [4, 5]: WN WN W 2 N = WN W (2.11) N Figure 2.7 shows how Walsh codes are generated based on (2.11). Similarly, Walsh codes of any length 2 N where N is an integer can be generated. By changing 0s to -1s, Walsh codes can be rewritten as W 2 1 [ 1 1], W 2 2 [ 1 1] = = where W mn denotes the mth Walsh code of length n. To illustrate how Walsh codes are used in CDMA, let us consider three users with messages given by 1 W 1 = 0 W 2 1 = 0 0 W 2 2 = 0 1 W 4 1 = W 4 2 = W 4 3 = W 4 4 = Figure 2.7 Walsh code generation.

45 26 Smart Antenna Engineering m m m [ 1 1 1] [ 1 1 1] [ 1 1 1] = = = (2.12) Now let us assign each of the users a Walsh code of length eight, respectively, W W W [ ] [ ] [ ] = = = (2.13) Since the chip rate for the Walsh code in this case is eight times the message bit rate, spreading each signal with its assigned code will result in widening the band from 1/T b to 1/T c where T b and T c are the bit and chip periods, respectively. The spread spectrum signals of the three users S n (t) and the combined signal C(t) are then given by, respectively, 8 1( ) = 1( ) 2 8 2( ) = 2( ) 4 8 3( ) = 3( ) 6 ( ) = ( ) + ( ) + ( ) S t m t W S t m t W S t m t W C t S t S t S t The resultant signals are shown in Figures 2.8 through 2.11, respectively. Now, in order to recover a user s original message, the receiver spreads the Figure 2.8 User 1 spread spectrum signal.

46 Multiple Access Techniques for 2G and 3G Systems 27 Figure 2.9 User 2 spread spectrum signal. Figure 2.10 User 3 spread spectrum signal. received composite signal with the Walsh code assigned to that user. This operation is shown in Figure 2.12 for user 1, where the receiver integrates all the values over a bit period. The original message is reconstructed using the following decision criterion ( ) ( ) ( ) ( ) n mt $ = 1 if Ct W > 0 n mt $ = 1 if Ct W < 0 m m (2.14)

47 28 Smart Antenna Engineering Figure 2.11 Composite spread spectrum signal. (a) (b) Figure 2.12 (T b =8T c ). (a) Effect of spreading received signal with first user s code; (b) User 1 recovered signal

48 Multiple Access Techniques for 2G and 3G Systems 29 If the receiver attempts to spread the composite signal with a code that was 8 W 8 = ,we not assigned to the user, (e.g., with [ ] get all zeros after the integration, as we can see in Figure 2.13, which means the signal cannot be recovered. In addition to using Walsh codes to separate different users and different channels on the forward link within a sector, a CDMA system needs to separate transmissions from different sectors within a network. This is accomplished using PN codes, as described in Figure Some important key differences between Walsh codes and PN codes, which greatly impact the interference level in a CDMA system, are illustrated in Table IS-95 CDMA Systems The TIA IS-95 CDMA system is a 2G mobile wireless system that operates in the cellular 800-MHz band [6, 7]. Another version of this system that operates in the PCS 1,900-MHz band is defined in J-STD-008 [8]. Both systems use a 1.25-MHz wide carrier and a chip rate of Mcps. On the forward link, a family of 64 Walsh codes is used to separate the different channels and different users. Short PN codes of length chips with a period of chips or ms are used to separate transmissions from different sectors. This is accomplished by using the same PN sequence for all sectors and then identifying each sector by a unique time offset in increments of 64 chips, resulting in 512 possible PN sequences. On the reverse link, long PN codes of length chips are used for channelization, that is, to distinguish different users. In addition, Figure 2.13 Effect of spreading received signal with wrong Walsh code.

49 30 Smart Antenna Engineering Walsh 1 Data stream 1 Sector specific PN code Walsh 2 Data stream 2 Filtering Modulator Walsh n Data stream n CDMA transmitter Sector specific PN code Walsh 1 MS1 Demodulator Filtering Data stream 1 Sector specific PN code Walsh n MS2 Demodulator Filtering Data stream n CDMA receiver Figure 2.14 CDMA transmitter and receiver block diagrams. the reverse link signal is further spread by short PN codes of length chips to identify the sector to which the transmission is intended Forward Link Channels The IS-95 and J-STD-008 standards define two types of forward link channels, namely common channels broadcast to all mobiles in a sector and dedicated channels to specific mobiles. Note that in addition to assigning different Walsh

50 Multiple Access Techniques for 2G and 3G Systems 31 Table 2.1 Comparison Between Walsh and PN Codes Transmit and Receive Walsh Codes Correlation PN Codes Correlation Same codes, same time offsets 100% 100% Different codes 0% Low noise-like Same codes, different time offsets > 0% < 100% Low noise-like codes to each channel, all channels are spread with the same PN sequence associated with the transmitting sector. The set of channels defined in the standard are listed here: Pilot channel: The pilot channel is continuously transmitted sector-wide to provide timing and phase references to all users to aid in system acquisition, signal strength comparison, and demodulation operations. The pilot channel is assigned Walsh code 0 or W 0, which is a sequence of 64 zeros with Mcps chip rate. Note that no baseband information is carried by the pilot channel. Sync channel: The sync channel is also continuously transmitted sector wide to provide timing information to the mobile during system acquisition and power up. Baseband information contained in the sync channel message is used to inform mobiles of system synchronization information and other system parameters. The sync channel is assigned Walsh code W 32 and it is transmitted in groups of superframes at the bit level. Each superframe lasts for 80 ms and consists of three ms sync channel frames that are synchronized with each period of the short PN sequence. Hence, once the mobile acquires synchronization with the pilot channel, the sync channel frame boundaries are immediately known. Paging channel: The paging channel is used to transmit overhead information to a mobile such as pages and other commands. Call setup commands and traffic channel assignments are also sent over the paging channel. Based on the standard specifications, there can be up to seven paging channels but there must be at least one. At the bit level, each paging channel frame lasts for 20 ms, four of which are combined into an 80-ms paging channel slot.

51 32 Smart Antenna Engineering Forward traffic channels: Forward traffic channels carry voice, data, and signaling once a call has been established. There are two rate sets defined in the standard. Rate set 1, or RS1, with data rates 1.2, 2.4, 4.8, and 9.6 Kbps, and RS2 with data rates 1.8, 3.6, 7.2, and 14.4 Kbps. In systems with only one paging channels, there are 61 available Walsh codes that could be assigned to traffic channels. Traffic channel frames last for 20 ms Reverse Link Channels In IS-95 and J-STD-008 standards there are only two reverse link channels: Access channel: This channel is used by the mobiles to access the system for registration, call origination, page responses, and overhead transmission to the base stations. Reverse traffic channels: Similar to the forward link, the reverse traffic channels carry voice, data, and signaling once a call has been established RAKE Receiver The presence of buildings, trees, hills, and other objects in the areas served by mobile systems cause signal reflection, diffraction, and scattering. This creates multiple replicas of the transmitted signal with different attenuations and time delays at the receiver. The interaction of the incoming waves at the receiver antenna results in deep and rapid fading or fluctuations in the signal strength. This significantly degrades the system performance. IS-95 based CDMA systems actually take advantage of the multipath components through the use of RAKE receivers. Multiple correlators are used to detect the strongest multipath components using a searcher finger designed to compare the incoming signals with the PN code used. This operation detects multipath arrivals by producing a series of correlation peaks at different times. The magnitude of each peak is proportional to the envelope of the signal in a particular path, whereas the time of each peak relative to the time of arrival of the first path gives that path s delay. With the amplitudes and time delays of the strongest multipath components known, a RAKE receiver compensates for the delays and combines the signals based on their strengths. This produces a diversity gain at the CDMA receiver, which helps combat fading. The block diagram of a CDMA RAKE receiver is shown in Figure Power Control Recall that in CDMA systems all users share the same RF carrier through the use of PN codes, therefore each user appears like random noise to other users and

52 Multiple Access Techniques for 2G and 3G Systems 33 Path 1 Path 2 Path 3 Strongest multipath components Delay element (chips) Path 1+ interference Correlator A 1 Delay element (chips) Correlator Path 2+ interference A 2 Sum Delay element t (chips) Correlator Path 3+ interference A 3 Figure 2.15 RAKE receiver block diagram. contributes to the system noise. If the power of each user is not properly controlled and allowed to increase unnecessarily, other users would suffer from interference that could severely degrade system performance. Consider a CDMA system where all users transmit at the same power. A user close to the base station will result in a high SNR 1 at the receiver, whereas another user further away from the base station would yield a lower SNR 2. Obviously, this disparity results in different signal quality between users. This is the classical near-far problem. Assume that the required SNR necessary to maintain the desired signal quality is given by SNR req. When new users are added to the cell, the interference level in the cell increases, thus reducing the SNRs of existing and new users up to the point at which the SNR of a new user would not be able to reach SNR req. Therefore, no more users can be added to the cell and the capacity is reached with only a few users. Hence, power control is essential to overcome the near-far problem and maximize the capacity. Power control is the

53 34 Smart Antenna Engineering process by which the transmit power of each user is controlled such that the received powers at the base station are equal. The capacity is then maximized by only allowing each user to transmit just enough power to achieve SNR req Reverse Link Open Loop Power Control As in any communications system, there is always a propagation loss that impairs the signal on the forward and reverse links. In addition to the regular distance-dependent path loss, other factors such as shadowing and multipath produce fading in mobile communications systems. Basically, there are two types of fading, slow fading and fast fading. Slow fading is modeled by a lognormal distribution and it manifests itself by slow power variation over several wavelengths, as shown in Figure This type of fading is typically caused by the signal being partially blocked by buildings, trees, and other obstacles. On the other hand, when multipath components with different amplitudes, phases, and arrival times add up at the receiver, they combine constructively and destructively, forming a standing wave pattern with a half wavelength period. As the mobile moves through this pattern, the received power will experience fast fading with an envelope distribution characterized by a Rayleigh distribution. When a mobile is in idle state, that is, a state where it monitors the overhead channels but no call has been established yet, the base station cannot control the power of the mobile. To solve this problem, the IS-95 standard defines the open loop power control process, which ensures that each mobile starts its initial transmissions, also called access probes, with a power level that depends on the received power from the base station p r or Slow fading Fast fading Signal strength in db Time Figure 2.16 Fading as function of time.

54 Multiple Access Techniques for 2G and 3G Systems 35 Ptinitial, = pr K + NOM _ PWR + INIT _ PWR (2.15) where K is a constant equal to 73 in the cellular band and 76 in the PCS band. NOM_PWR and INI _PWR are system parameters that are broadcast from the base station to all mobiles. The reason this is called open loop power control is that if the mobile does not receive an acknowledgement from the base station after sending an access probe, it waits for a random time period before sending the next access probe with a slightly higher power. The mobile repeats this process until an acknowledgement is received but there is no feedback from the base station about the signal quality. Since this process is slow, it only compensates for the slow lognormal fading Reverse Link Closed Loop Power Control The closed loop power control attempts to balance losses between the link due to Rayleigh fading or fast fading at slow mobile speeds and interference variations due to loading once the mobile is on a traffic state. It also improves the performance of mobiles at the cell edge where the signal is weak and the interfering signals from other cells are strong. As briefly described previously, power control adjusts the transmit power of each mobile to maintain the required SNR given a specific signal quality. To achieve that, the power control process must be able to determine the value of the SNR req to maintain the signal quality. The outer loop power control performs this function by adjusting the target SNR according to the prevailing environment to achieve the desired end-user quality of service. Let us define E b as the energy per bit and N o as the interference plus noise power spectral density, then we get E N b o P = R ( N + I) W W P = R N + I W = R SNR (2.16) where W is the RF carrier bandwidth, R is the signal data rate, and W/R is defined as the processing gain. It is clear from (2.16) that adjusting E b /N o is equivalent to adjusting the SNR. The closed loop power control is summarized in Figure Based on the target E b /N o, the base station controls the mobile transmit power. The power control commands are sent from the base station on the forward link in the form of power control bit (PCB); each power control group lasts for 1.25 ms in IS-95 based systems. Hence, the power of the mobile can be adjusted up to 800 times per second. This is performed using the inner loop power control as follows: The base station monitors the reverse link E b /N o and compares it to (E b /N o ) Target.IfE b /N o >(E b /N o ) Target, the base station commands the mobile to decrease the transmit power by sending a power down command,

55 36 Smart Antenna Engineering Is received signal Decrease b N o Target Yes quality better than required quality No Increase (E b N o ) Target Figure 2.17 Closed loop power control mechanism. PCB = 0. If E b /N o <(E b /N o ) Target, the base station commands the mobile to increase the transmit power by sending a power up command, PCB = Soft Handoff One major advantage of having all users in a CDMA system on the same RF carrier is the ability of maintaining simultaneous connections. When a mobile maintains simultaneous traffic channels with sectors belonging to different base stations, it is said to be in soft handoff. On the forward link, the mobile s RAKE receiver demodulates the signals received from separate sectors and combines them to produce a signal with a better quality. On the reverse link, multiple base stations demodulate the mobile s signal and the demodulated frames are sent back to the base station controller (BSC) to select the best frame. This operation provides some diversity since the signals on different links are typically uncorrelated and do not fade at the same time with the same depth. This results in a soft handoff gain, which improves the air interface capacity. As the mobile moves around the system, it keeps a list of all active pilots from the soft handoff links in a set called the active set. Other pilots with raw SNR E c /I o strong enough to be candidates for soft handoff are kept in a set called the candidate set. Another important set kept by the mobile is the neighbor set, which contains those pilots that are neighbors to the current serving sector. When the sectors in the active set belong to the same base station, the mobile is said to be in softer handoff. The procedure by which these sets are maintained and the pilots are processed is defined in the IS-95 standard. 2.4 Third Generation Systems As second generation systems started to reach their limits in terms of spectral efficiency along with the increasing demands for higher data rate services, a need

56 Multiple Access Techniques for 2G and 3G Systems 37 emerged for improved networks that can provide these future requirements. This led to the development of 3G systems [5, 9], with the following main objectives: Provide data rates from 144 Kbps up to 384 Kbps for mobility scenarios; Provide data rates up to 2 Mbps for limited mobility and fixed wireless scenarios; Provide higher spectral efficiency compared with 2G systems; Support multiple simultaneous services (e.g., speech, high-speed data). There are currently two major standards adopted for 3G systems, both of which are based on CDMA, namely CDMA2000 and WCDMA. Another emerging technology also based on CDMA is the time division synchronous CDMA (TD-SCDMA) CDMA2000 The CDMA2000 family of standards is a wideband spread spectrum radio interface that uses CDMA technology to meet the objectives of 3G systems while maintaining backward compatibility with IS-95 based systems. This means that mobile handsets designed according to the IS-95 standard are capable of operating in a CDMA2000 system and vice versa. The first component of the CDMA2000 standard is called 1X radio transmission technology (1X RTT) because it uses an RF carrier of 1.25 MHz just like IS-95 based systems, hence the 1X, which is also referred to as spreading rate (SR)1. The key benefits of the 1XRTT technology standardized under the name of IS-2000 [10, 11] compared with IS-95A/B standards [7] can be summarized as follows: Better forward error correction (FEC). This is achieved through the use of higher convolutional coding rates as well as turbo codes for high data rates. The coding rate refers to the number of symbols produced by the encoder for every bit of input data. The greater this number is, the more protection we get against errors because of the increased correction power. A direct impact of the improved coding is a reduction in the required E b /N o, which directly translates into higher capacity or higher data rates. A coding gain of up to 2dB can be achieved with IS-2000 systems compared with the IS-95 standard. Fast forward link power control mechanism. As described earlier, a power control mechanism is defined for the reverse link of the IS-95 standard, whereby the transmit power of the mobile is controlled up to 800 times

57 38 Smart Antenna Engineering per second. IS-2000 extends the use of this power control process to the forward link as well, where the mobile station can control the power transmitted by the base station with speeds of times per second through the use of the forward link closed loop power control mechanism. This allows the power resources on the forward link to be optimized and used much more efficiently than in IS-95 systems, yielding significant improvements in capacity. Multimedia services and improved data services support. In addition to the improvements listed above, the IS-2000 standard introduces new dedicated channel and common channels to support high data rate applications as well as improved diversity techniques. Moreover, the battery life is extended through the use of a new quick paging channel. The combination of these improvements results in a voice capacity increase of 1.6 to 2 times compared with IS-95A/B as well as data rates of up to 307 Kbps Overview of IS-2000 Forward Link Physical Channels As we have seen in the IS-95 standards, there are two rate sets, RS1and RS2, with data rates of up to 9.6 Kbps and 14.4 Kbps, respectively. In IS-2000, a wider range of data rates are available and are defined in terms of radio configurations (RC), which can be summarized as: RC1, which supports IS-95A/B backward compatibility for all rate set 1 (RS1) based services up to 9.6 Kbps; RC2, which supports IS-95A/B backward compatibility for all rate set 2 (RS2) based services, up to 14.4 Kbps; RC3, which supports data rates from 1,500 bps up to Kbps, using rate 1/4 FEC encoding; RC4, which supports data rates from 1,500 bps up to Kbps, using rate 1/2 FEC encoding; RC5, which supports data rates from 1,800 bps up to Kbps, using rate 1/4 FEC encoding. Table 2.2 provides a summary of the forward link physical channels of the IS-2000 standard and a brief description of their functions Overview of IS-2000 Reverse Link Physical Channels Since CDMA networks based on the IS-95 standard have been launched in the 1990s it became apparent over the years that there are certain inefficiencies in

58 Multiple Access Techniques for 2G and 3G Systems 39 Table 2.2 IS-2000 Forward Link Physical Channels Channel Common pilot channel (F-CPICH) Common auxiliary pilot channel (F-CAPICH) Dedicated auxiliary pilot channel (F-DAPICH) Sync channel Fundamental channel (F-FCH) Supplemental channel (F-SCH), (RC 3-9) Supplemental code channel (F-SCCH), (RC1-2) Dedicated control channel (F-DCCH) Common assignment channel (F-CACH) Common power control channel (F-CPCCH) Paging channel IS 95 A/B Broadcast control channel (F-BCCH) Common control channel (F-CCCH) Quick paging channel (F-QPCH) Function Used to broadcast pilot for the entire cell/sector, for channel and phase estimation (coherent demodulation), initial acquisition, and handoffs. Used for beamforming applications for a group of mobiles. Used for beam steering and beamforming applications for a single mobile. This is the same channel as in IS-95A/B, containing system synchronization information. This is identical to the 95A/B traffic channel. Used for voice, data, and control. There can be 0-1 channels. This channel was introduced for supporting high data rates. There can be 0 2 channels. This channel was introduced in IS-95B for medium data rate service option. There can be 0 7 channels. Introduced for MAC, data, and signaling. The power control subchannel can also be punctured here when F-FCH is absent. Designed to provide fast response reverse link channel assignments to support transmission of random access packets on the reverse link. Used by the base station for transmitting common power control subchannels. One common control paging channel (PCH) where broadcast and mobile station directed messages are transmitted. Broadcasts only cell-specific overhead messages (e.g., CDMA channel list, extended systems parameters message and neighbor list) at (4.8, 9.6, or 19.2 Kbps). The sync channel is used to let mobiles know if F-BCCH is supported. Broadcasts mobile station specific messages (e.g., extended channel assignment message, general page message, order message) at 9.6, 19.2, and 38.4 Kbps in discontinuous transmit mode. Helps decrease the wake time of a mobile station, that is the time the mobile has to periodically demodulate the PCH or F-CCCH. This improves MS standby time and reduces battery consumption. The support of this channel is optional.

59 40 Smart Antenna Engineering the design of the reverse link. This led to an IS-2000 physical layer design that adds several enhancements to improve the performance of the reverse link. Table 2.3 provides a summary of the reverse link physical channels of the IS-2000 standard and a brief description of their functions. One of the major drawbacks of the IS-95 reverse link design is the noncoherent demodulation by rake receiver, which requires a high received signal-to-noise ratio for good performance. The IS-2000 standard solves this problem by introducing the reverse link pilot or R-PICH, which allows the base station to estimate the carrier phase and makes coherent demodulation possible. This improves both searching and tracking of mobiles. Another major improvement is the introduction of the forward link power control by which the mobile can adjust the forward link power. This is performed by time-multiplexing forward power control (PC) information on the reverse pilot channel, as shown in the frame structure in Figure CDMA2000 1x EV-DO Data services are expected to have a significant growth over the next few years and will likely become the dominant source of 3G traffic and revenue. The Table 2.3 IS-2000 Reverse Link Physical Channels Channel Reverse pilot channel (R-PICH) Access channel (R-ACH) and enhanced access channel (R-EACH) Reverse common control channel (R-CCCH) Reverse dedicated control channel (R-DCCH) Reverse fundamental channel (R-FCH) Reverse supplemental code channel Reverse supplemental channel (R-SCH) Function Used for searching, tracking, and coherent demodulation. Also used by the forward link channels to adjust forward link power and maintain the quality of the link. Used by mobiles to access the system for registration, call origination, and page responses. Used to support efficient access procedures of packet data services. Used for the discontinuous transmission of user traffic, control, and signaling information to the base station while the mobile is in the traffic state. Used to carry user traffic for RC1 and RC2. Used to support medium data rates based on services for RC1 and RC2. There can be up to seven such channels. Used to support high data rates for RC3 and RC4. There can be zero to two channels.

60 Multiple Access Techniques for 2G and 3G Systems 41 1 Frame (20ms) PCG (1.25ms) Reverse pilot channel Reverse power control subchannel Figure 2.18 IS-2000 reverse link pilot frame structure. current 3G operators in Japan and Korea as well as in the United States are already experiencing great success with their data services. KTK in Korea has reported that 34% of its ARPU was related to data usage for the third quarter of 2003, particularly after the deployment of 1xEV-DO. The 1xEV-DO standard [12 20] is optimized for wireless high-speed packet data services. Because of the typical asymmetric characteristics of IP traffic, the downlink is the more critical of the two links [12]. Thus, several techniques were introduced in 1xEV-DO to optimize the downlink throughput. The 1xEV-DO downlink uses time-division-multiplexed (TDM) waveform, which eliminates power sharing among active users by allocating full sector power and all code channels to a single user at any instant. This is in contrast to code-division-multiplexed (CDM) waveform on the IS-95 downlink, where there is always an unused margin of transmit power depending on the number of active users and power allocated to each user. Through power control, this margin is used to account for large variations of the required mobile station transmit power in fading channels to guarantee a given target frame error rate. Figure 2.19 shows the sector power usage of the IS-95 and 1xEV-DO downlinks. Each channel in IS-95 is transmitted the entire time with a certain fraction of the total sector power, while the equivalent channel in 1xEV-DO is transmitted, at full power, only during a certain fraction of time. The efficient usage of sector power resource in 1xEV-DO improves cell coverage as well as signal-to-interference and noise ratio (SINR) for noise-limited users. Similar to the IS-95 concepts, every mobile station reports to the network the strongest downlink pilots it can measure. In turn, the network selects an active set for each terminal. Each sector in the terminal s active set maintains a connection with the terminal. The active set of sectors for any given terminal is also the set of power controlling sectors for its uplink. However, instead of transmitting equal power on all downlink traffic channels in the active set as adopted

61 42 Smart Antenna Engineering Unused power margin Max transmit power Max transmit power Sector transmit power Traffic channels Sync channel Sector transmit power Pilot channel Control channel Totaldatatraffic channels Paging channel Pilot channel Time Time Figure 2.19 Sector power usage comparison between 1xRTT(left) and 1xEV-DO (Right). in IS-95, the 1xEV-DO network only transmits on the best link and allocates no power on the others. To accomplish this procedure, a mobile terminal monitors the SINR of all the sectors in its active set and informs the network, via a feedback channel, of the identity of the selected serving sector. Due to the TDM waveform of the 1xEV-DO downlink, a terminal is allocated the full sector power whenever it is served, thus no power adaptation is needed. Instead, rate adaptation is used on the downlink. The highest data rate that can be transmitted to each terminal is a function of the received SINR from the serving sector, which is typically a time-varying quantity. To achieve the highest data rate at each time of transmission, each terminal predicts the channel condition over the next packet for its serving sector based on the correlation of the channel states. It selects the highest data rate that can be reliably decoded based on the predicted SINR, and then informs the serving sector of its selected rate over an uplink feedback channel. Whenever the network decides to serve a terminal, it transmits at the most recent selected rate fed back from the terminal. Since a sector transmits traffic data to a single user at any instant of time, a scheduling algorithm is implemented in each sector to fairly allocate the available time slots among the active users, thus maximizing capacity by exploiting the channel dynamics. Because different users experience independent fading processes, it is unlikely that all users SINR will fall into deep fades at the same time. In other words, when some users experience a deep fade, others reach peaks of their received signal strength. As a mobile user goes through periods of varying fades, the data rate allocated to it by the network will vary. However, since Internet

62 Multiple Access Techniques for 2G and 3G Systems 43 protocol (IP) traffic can tolerate relatively longer and variable time delays, unlike voice services, this can be tolerated. The standard does not specify the type of scheduler to be used. A smart scheduler will attempt to serve an active user near its peak SINR while maintaining a certain degree of fairness. For instance, the PF scheduling algorithm represents a good balance because it incorporates the two important features of a capacity enhancing scheduler: multiuser diversity gain and fairness. The algorithm selects the terminal based on a metric equal to the ratio of the instantaneous channel state to the long-term average of the served throughput. Thus, it attempts to serve each terminal at their local peaks of channel conditions and maintain higher average served throughput when the terminal is in better coverage. Another critical concept in the design of the 1xEV-DO standard is link adaptation. Link adaptation is achieved by combining several mechanisms designed to improve spectral efficiency while achieving the required simplicity and robustness for effective operation in a wireless cellular environment. The idea behind link adaptation is to optimize spectral efficiency by matching the transmit data rate, modulation, and coding to the time varying received SINR at the terminal. A variety of modulation schemes, including QPSK, 8 PSK, and 16 QAM as well as coding rates that best matches the fading channel, are defined in what is commonly called adaptive modulation and coding techniques. To fully exploit these concepts, the system includes a collection of techniques that consist of incremental redundancy and repetition coding, time diversity adaptation, and HARQ [15 18] WCDMA WCDMA is another 3G air interface based on direct-sequence CDMA (DS-CDMA). WCDMA uses a chip rate of 3.84 Mcps, compared with Mcps in both IS-95 and IS-2000 standards and requires an RF carrier with 5-MHz bandwidth [21 25]. There are two modes of operation in the WCDMA air interface, a frequency division duplex (FDD) mode, where a pair of 5-MHz carriers are used, and a time division duplex (TDD), where only one carrier is used. Similar to IS-95 and IS-2000 systems, channelization is achieved through the use of orthogonal codes referred to as orthogonal spreading factor codes (OVSF), which are Walsh codes of variable lengths, whereas source separation is achieved using gold codes. Table 2.4 shows a comparative summary of the WCDMA, IS-95, and IS-2000 air interfaces FDD-WCDMA Forward Link (Downlink) Physical Channels In this section we will summarize the physical channels associated with the WCDMA downlink. The physical channels are those channels that perform the actual transmission of data bits and are distinguished by an RF carrier, a channelization code, a spreading code, and modulation parameters. WCDMA

63 44 Smart Antenna Engineering Table 2.4 Main Differences Between IS-95, IS-2000, and WCDMA Link Function IS-95 A/B IS-2000 WCDMA Forward link (Downlink) Channelization 64-chip Walsh codes 4~256-chip Walsh codes 4~512-chip OVSF codes Source separation (2 15 1)-chip short PN codes (2 15 1)-chip short PN codes chips of 2 18 Gold code Power control rate Slow 800 Hz 1,500 Hz Reverse link (Uplink) Channelization None 4~256-chip Walsh codes 4~256-chip OVSF codes Source separation (2 42 1)-chip long PN codes (2 42 1)-chip long PN codes 38,400 chips of 2 25 Gold code Power control rate 800 Hz 800 Hz 1,500 Hz physical channels can be grouped into four categories: common channels broadcast to all mobiles in the cell or sector, channels that carry paging information, channels used for random- and packet-access, and dedicated connection channels. Common pilot channel (CPICH): This channel is a cell-wide channel, which provides a coherent phase reference for the downlink channels, and it uses the gold code specific for that cell. It also aids channel estimation for cell selection and reselection as well as handoff procedures 256 for the mobiles. The CPICH uses orthogonal codec 0. Primary common control physical channel (P-CCPCH): This channel is used to broadcast cell information; that is, cell system frame number (SFN) and timing reference for all downlink channels necessary for synchronization operations. That is why the P-CCPCH is continuously transmitted over the entire cell and it always uses the same 256 channelization codec 1. Secondary common control physical channel (S-CCPCH): This channel is used to transmit information related to the forward access channel (FACH) and the PCH and is mainly monitored by the mobiles in idle mode.

64 Multiple Access Techniques for 2G and 3G Systems 45 Paging indication channel (PICH): This channel is used in conjunction with the PCH to provide mobiles with a sleep mode operation, which saves the battery in idle mode. Basically, the PICH is used to alert mobiles of an incoming page. Dedicated physical channel (DPCH): The DPCH consists of two separate channels, the dedicated physical data channel (DPDCH) and the dedicated physical control channel (DPCCH), which are time multiplexed onto one time slot. The DPCCH carries control bits such as pilot bits, which are used by the receiver to measure the channel quality, and transmission power control (TPC) bits, which are used to adjust the power of the mobile. The DPDCH is mainly used to carry user traffic as well as some overhead and signaling data FDD-WCDMA Reverse Link (Uplink) Physical Channels As with the downlink, there are two categories of channels on the WCDMA uplink, common uplink physical channels and dedicated uplink physical channels. Physical random access channel (PRACH): This channel is used to carry access requests (i.e., control information and short data bursts) and does not contain any pilot or TPC bits since it uses only open loop power control. Physical common packet channel (PCPCH): This channel is used to carry connectionless packet data. Dedicated physical data channel (DPDCH): The uplink DPDCH is used to carry dedicated user traffic data generated at an upper layer. There may be zero, one, or up to six uplink DPDCHs. Dedicated physical control channel (DPCCH): The uplink DPCCH is used to carry control information consisting of pilot bits to support channel estimation for coherent detection, TPC commands, and some feedback information. Unlike the downlink case, the DPDCH and DPCCH are not time multiplexed; instead, they are fed into the I and Q inputs of a complex spreader HSDPA To meet the increasing demand for high data rates in multimedia services over networks supporting WCDMA, the Third Generation Partnership Project (3GPP) has released a new high-speed data transfer protocol named HSDPA [26 29]. HSDPA is expected to provide significant improvements over the basic

65 46 Smart Antenna Engineering WCDMA R 99 for downlink asymmetrical and bursty packet data services. HSDPA will offer a peak data rate up to and in excess of 10 Mbps (as opposed to the currently deployed 384 Kbps), as well as at least threefold sector throughput. For end users, HSDPA will mean lower delays and faster connection and response times, particularly for high data rate applications in loaded systems. The substantial increase in data rate and throughput is achieved by implementing a fast and complex channel control mechanism based on a short and fixed packet transmission time interval (TTI, fast HARQ, and fast scheduling performed at the Node B (base station) instead of the radio network controller (RNC), the equivalent of the BSC. The TTI indicates how often data arrives from higher layers to the physical layer and could take any of the values of 10, 20, 40, or 80 ms in R 99 WCDMA. R 99 WCDMA already includes three different channels for downlink packet data transmission: dedicated channel (DCH), downlink shared channel (DSCH), and FACH. The FACH is a common channel offering low latency. However, it is not spectraly efficient since it does not support fast closed loop power control. It is therefore limited to carrying only small data traffic. The DCH is the primary data channel and can be used for any traffic class. The DCH is allocated a certain orthogonal variable spreading factor (OVSF: 4-512) according to the connection peak data rate, whereas the block error rate (BLER) is controlled by inner and outer loop power control. The DCH code and power allocation are therefore inefficient for bursty and low duty cycle data applications since channel reconfiguration can be very slow (in the range of 500 ms) [21]. The DSCH provides the possibility to time-multiplex different users and improve the channel reconfiguration time and packet scheduling procedure (in the order of 10 ms) [5]. The HSDPA concept can be seen as an extension of the DSCH with the introduction of new features such as AMC, short packet size, multicode operation, and fast L1HARQ. In fact, these features replace the two basic WCDMA features, namely variable spreading factor VSF and fast power control [21]. Next, we will provide an explanation of key HSDPA features. A new transport channel named high-speed downlink shared channel (HS-DSCH) has been introduced as the primary radio bearer. Similarly to the DSCH, the HS-DSCH is shared between all users in a particular sector. The primary channel multiplexing occurs in the time domain, where each TTI consists of three time slots (or 2 ms). The TTI is also referred to as a subframe. The TTI has been significantly reduced from the 10, 20, 40, or 80 ms TTI sizes supported in R 99 to better achieve short round-trip delay between the mobile station and the Node B and improve the link adaptation rate and efficiency of the AMC. Within each 2 ms TTI, a constant spreading factor (SF) of 16 is used for code multiplexing with a maximum of 15 parallel codes for the HS-DSCH. These codes may all be assigned to one user during the TTI, or may be split amongst several users. Note that the more codes allocated to a user, the higher peak data rate it can achieve. The number of parallel codes allocated to each user

66 Multiple Access Techniques for 2G and 3G Systems 47 depends on cell loading, quality of service (QoS) requirements, and the mobile station code capabilities (5, 10, or 15 codes). To support the HS-DSCH operation, two control channels have been added: the high-speed shared control channel (HS-SCCH) and the high-speed dedicated physical control channel (HS-DPCCH). The HS-SCCH is a fixed rate channel used for carrying downlink signaling between the Node B and the mobile station before the beginning of each scheduled TTI. This includes the mobile station identity, HARQ-related information, and the parameters of the HS-DSCH transport format selected by the link-adaptation mechanism. The HS-DPCCH carries uplink cyclic redundancy check (CRC)-based ACK/NACK signaling for the physical layer retransmission as well as CQI to be used in the link adaptation mechanism. The CQI is based on the CPICH and is used to estimate the transport block size, modulation type, and number of channelization codes for downlink transmission. The feedback cycle of the CQI can be set as a network parameter in predefined steps of 2 ms Adaptive Modulation and Coding Adaptive modulation and coding is the fundamental feature of HSDPA. It consists of continuously optimizing the code rate, the modulation scheme, and the number of multicodes employed as well as the transmit power per code according to the channel quality experienced (CQI feedback) by the mobile station. To achieve very high data rates, HSDPA adds a higher order modulation (16 QAM) to the existing QPSK modulation in R 99. Different combinations of modulation and channel encoding can be used to provide data rates ranging from 119 Kbps/code with QPSK and 1/4 code rate to 712 Kbps/code with 16 QAM and 3/4 code rate (SF = 16). Users with the most favorable channel condition (close to the Node B) will get the highest data rates, whereas users with the least favorable channel condition will get the lowest data rates (located at the cell edge). HSDPA supports the use of 5, 10, and 15 multicodes. A single user can receive up to 15 multicodes, resulting in a peak data rate of 10.8 Mbps. However, the maximum specified peak data rate with HSDPA is 14.4 Mbps (or 960 Kbps/code) when 16 QAM modulation is used with no coding (effective code rate of 1) and 15 multicodes. This rate remains very unlikely to achieve since it corresponds to an unloaded system where the served user is extremely close to the node B. Another benefit of AMC is better utilization of the Node B power. If no power constraints are specified, the leftover power from the dedicated channels (R 99) can be allocated to HS-DSCH, resulting in near-maximum power utilization Hybrid-ARQ with Soft Combining The retransmission mechanism selected for HSDPA is HARQ with Stop and Wait protocol (SAW). HARQ allows the mobile station to rapidly request

67 48 Smart Antenna Engineering retransmission of erroneous transport blocs until they are successfully received. HARQ functionality is implemented at the medium access control (MAC) layer, as opposed to the radio link control (RLC) layer, as is the case in R 99 WCDMA. Therefore, the retransmission delay of HSDPA is much lower than for R 99 because it does not involve the RNC. In normal circumstances, a NACK may require less than 10 ms at the MAC layer, while it can take up to 100 ms at the RLC layer when Iub signaling is involved [9]. This reduces significantly the delay jittering for Transmission Control Protocol/Internet Protocol (TCP/IP) and delay-sensitive applications. During retransmission, the mobile station does not discard the original transmission but rather combines it with the following retransmission(s) to increase the probability of successful decoding. This is called soft combining. HSDPA supports both chase combining (CC) and incremental redundancy (IR). CC is the basic combining scheme. It consists of the Node B simply retransmitting the exact same copy of the original packet. With IR, additional redundant information is incrementally retransmitted, providing additional coding gain. This can result in fewer retransmissions than for CC. However, the disadvantage of IR over CC is the much higher memory requirement for the phone Fast Scheduling The scheduler is a key element of HSDPA that determines the overall behavior of the system and, to a certain extent, its performance. For each TTI, it decides which terminal (or terminals) the HS-DSCH should be transmitted to and, in conjunction with the AMC, at which data rate. One important change from the R 99 implementation is that the scheduler is located at the Node B as opposed to the RNC. This, with the short TTI (2 ms), enables the scheduler to quickly track the user equipment (UE) channel condition and adapt the data rate allocation accordingly. Three main scheduler algorithms have been proposed for HSDPA: round robin (RR), maximum C/I, and PF. RR schedules users according to a first-in first-out approach. It allows achieving a high degree of fairness between the users at the expense of the overall system throughput (and therefore spectral efficiency) because some users can be served even when they are experiencing a destructive fading. The maximum C/I schedules users with the highest C/I during the current TTI. This naturally leads to the highest system throughput because most of the served users will likely sustain a high peak data rate with a low probability of error. However, the fairness between the users is minimal. In fact, users at the cell edge will be largely penalized by experiencing excessive service delays and significant outage. The PF offers a good trade-off between RR and the maximum C/I. The PF schedules users according to the ratio of their instantaneous data rate to average served data rate. This results in all users getting an equal probability of being served even though they may experience very

68 Multiple Access Techniques for 2G and 3G Systems 49 different channel quality. This allows having a good balance between the system throughput and the user fairness. 2.5 Basic CDMA Procedures As previously described, each CDMA base station transmits a different PN sequence (gold code). In WCDMA, there are 512 primary scrambling codes available. Each scrambling code has low cross-correlation with any other scrambling code regardless of the timing offset between the two scrambling codes. This allows the base stations to be deployed asynchronously. In CDMA2000, base stations transmit the same PN sequence (M-sequence) offset by different amounts of time. There are 512 PN offsets available. An M-sequence transmitted with a given PN offset has low autocorrelation with any other PN offset. To guarantee that each base station transmits a PN offset that is distinct from those in its vicinity, base stations need to be synchronized so that they have a common sense of timing. Figure 2.20 shows a typical CDMA cell layout where the color of each cell indicates a different PN offset. This approach is necessary so users could identify the different base stations or sectors and to enable a CDMA phone to acquire the system. This acquisition process is one of several possible states for a mobile phone or UE. Other states shown in Figure 2.21 include idle, access, and dedicated Acquisition State Acquisition means acquiring the system. It is done upon power up or loss of service. For example, in WCDMA acquisition consists of a three-step process Figure 2.20 CDMA cell layout.

69 50 Smart Antenna Engineering Power up/initialization state Phone acquires system. CPICH, SYNC. Idle state Phone receives overhead information on the paging channel. PCH,P-CCPCH,S-CCPCH,PICH Traffic/dedicated state A dedicated channel is allocated to the phone. FCH,SCH,DPCH Access state Phone accesses the network for call origination. RACH,PRACH Figure 2.21 CDMA call states. designed to simplify the 512 x 38,400 search space, where there are 512 primary scrambling codes (PSCs) and 38,400 possible chip offsets of the PSC. The UE acquisition begins by searching and finding the synchronization channel (SCH). On the SCH, the primary SCH (P-SCH) and the secondary SCH (S-SCH) are sent simultaneously, as shown in Figure These two channels are coded differently. The P-SCH is coded with the primary synchronization code, which is the same for every cell in the system. The S-SCH is coded with the secondary synchronization code, comprising 64 different code groups. When the code group is determined (step 2), the PSC is determined on the CPICH. There are eight PSCs per code group; the UE tries these eight combinations. 256 chips One slot (2560 chips) PSC PSC PSC PSC SSC SSC SSC SSC Common pilot channel One frame (38400 chips, T f =10 ms) Figure 2.22 WCDMA cell search signals.

70 Multiple Access Techniques for 2G and 3G Systems 51 When the PSC is found, the UE resolves a 10 ms ambiguity on the broadcast channel (BCH). Finally, the UE can start to demodulate, or read, the transport BCH on the PCCPCH. This process, also know as cell search, is summarized as follows. Step 1: Slot synchronization During the first step of the cell search procedure, the UE uses the SCH s primary synchroniszation code to acquire slot synchronization to a cell. This is typically done with a single matched filter matched to the primary synchronization code, which is common to all cells. The slot timing of the cell can be obtained by detecting peaks in the matched filter output. Figure 2.23 shows a depiction of this step. Step 2: Frame synchronization and code-group identification During the second step of the cell search procedure, the UE uses the SCH s secondary synchronization code to find frame synchronization and identify the code group of the cell found in the first step. This is done by correlating the received signal with all possible secondary synchronization code sequences, and identifying the maximum correlation value. Since the cyclic shifts of the sequences are unique, the code group as well as the frame synchronization is determined. Step 3: Scrambling-code identification During the third and last step of the cell search procedure, the UE determines the exact primary scrambling code used by the found cell. The primary scrambling code is typically identified through symbol-by-symbol correlation over the CPICH with all codes within the code group identified in the second step. After the primary scrambling code has been identified, the primary CCPCH can be detected and the system- and cell-specific BCH information can be read. This operation is shown in Figure If the UE has received information about which scrambling codes to search for, steps 2 and 3 above can be simplified. Indicates timing of PSC 2560 chips Matched filter 2560 chips Figure 2.23 WCDMA acquisition step 1.

71 52 Smart Antenna Engineering Common pilot 10 ms Correlator bank 10 ms Correct code Figure 2.24 WCDMA acquisition step 3. In CDMA2000, the mobile first gains some idea of system timing by searching for usable pilot signals. The mobile can first align its own timing by correlating with the pilot. When this correlation is found, the mobile synchronizes with the synchronization channel. It reads the sync channel message, and changes to absolute time. When the mobile is synchronized to absolute time, it can then read the broadcast messages on the F-PCH or F-BCCH Idle State Once the phone has acquired the system upon power up or loss of service, it enters into an idle state except if a call needs to be placed. During this state, the phones monitors the PCH or the P-CCPCH to get updated system information or to receive pages. To save power, the mobile periodically enters a low-power or sleep state, during which it cannot receive data from the network. To page a mobile while it is in this state, the network assigns times when the mobile should wake up and receive any paging messages. Both WCDMA and CDMA2000 use a paging indication protocol. A paging indicator indicates to the mobile whether it has a paging channel message. If the paging indicator indicates that the mobile has a paging message, the mobile demodulates the entire paging channel message. Otherwise, it returns to sleep until the next paging occasion Access State and Call Setup The mobile enters the access state whenever it needs to get dedicated network resources, respond to a page, or establish a voice or data call. An access channel (ACH) is used by the mobile to send requests to the network to set up a dedicated connection. The same procedure is used in both CDMA2000 and WCDMA, although the channel structure and messages contents and sequence are different. Let us assume that the mobile phone needs to place a call; the following summarizes what happens to establish this call: The first step in call setup is called origination, in which the phone sends a message to the network over the ACH or PRACH to request a connection.

72 Multiple Access Techniques for 2G and 3G Systems 53 The network then responds to the phone to acknowledge the message receipt and inform the mobile about the connection setup and code channels to be assigned. This is done over the PCH or S-CCPCH. Once the network authenticates the mobile, both the phone and the network start negotiating the type of service and data rate required. When the service has been negotiated, the network assigns RF and hardware resources to the call and this information is exchanged with the mobile so that a dedicated channel [fundamental channel (FCH), SCH, or DPCH] could be established Traffic or Dedicated State Call setup is concluded by assigning a dedicated traffic channel to the mobile. In CDMA2000 this could be the FCH, the SCH, or a combination of both, depending on the type of call and data rate. In WCDMA, the network assigns a DPCH for the same purposes. Although the procedures used for call setup are similar, the GSM layering structure (RRC, MM, CC) of WCDMA requires layer 3 signaling to be exchanged at many different layers, resulting in more signaling messages. In CDMA2000, the messaging is more streamlined. 2.6 CDMA Embedded Cell Capacity The isolated CDMA cell capacity derived in Chapter 1 assumes only a single cell and ignores the interference from users in neighboring cells. The capacity of an isolated cell in a narrowband system would also be very high since a reuse factor of one can be employed and all channels can be assigned in the 1.25-MHz bandwidth. In fact, CDMA makes a big difference when the impact of neighboring cells is taken into account. Let us rewrite the isolated cell pole capacity given by N pole W R = as P (2.17) E N b t in which the interference was averaged only over the users in the same cell. To quantify the potential improvements of smart antennas in CDMA systems the characteristics of the interference must be understood. On the downlink, several base stations are radiating and a mobile unit suffers from interference from other cells as well as from its home cell. On the uplink of a direct sequence CDMA system, the capacity is related to the E b /N t, as was shown in Chapter 1. If E b /N t is

73 54 Smart Antenna Engineering too low, the frame error rate (FER) or BLER will be high and the system performance will degrade. If E b /N t is too high, the interference level will increase and this will decrease the reverse link capacity. In TDMA and GSM systems, for every time slot there is one desired signal and a very small number of cochannel interferers, which makes the application of adaptive interference cancellation practical. On the other hand, in low-rate CDMA systems (i.e., for voice-dominated services), due to the large number of users sharing the channel, the interference in these systems is typically assumed to be statistically close to white Gaussian noise. And since in most cases the users can be assumed to be uniformly distributed across the cell, interference can also be considered spatially white in most operating scenarios. This means that isolating individual users requires arrays of large size. In CDMA2000 and WCDMA systems, a large number of voice users are mixed with a smaller number of high-speed data users. Since high data rate users have a lower processing gain, in order for them to maintain the same required E b /N t as voice users, their transmit power must be much stronger than voice users. As a result, on the reverse link high data rate users will present strong directional interference in the reception of voice users and the interference observed by voice users will be colored by data users and is no longer approximated as white Gaussian. In these mixed voice and high data rate systems an interference cancellation/reduction algorithm can be effectively used to null/reduce the impact of the limited number of high bit rate users, thereby increasing the overall system capacity. In CDMA systems, the uplink pole capacity of an embedded cell is given by [4] N pole = E N b t W R G 1 ν ( + f ) s (2.18) where v is the voice activity factor, G s is the sectorization gain, N t is the total noise + interference power spectral density, and f is the reuse efficiency defined as f I oc = (2.19) I sc In (2.19), I oc denotes the other cell interference power and I sc is the same cell interference power. From (2.18) it can be readily seen that reducing the interference level increases the number of maximum supportable users in a sector. It turns out that the fraction of the uplink interference that comes from the neighbor cells is about 60% of the own-cell interference and this ratio is not very sensitive to the parameters of the model, provided the assumption that the

74 Multiple Access Techniques for 2G and 3G Systems 55 mobiles are power-controlled in a sensible way still holds. The effective frequency reuse factor in CDMA can be calculated as F = 1 + f (2.20) F plays the same role in the CDMA capacity equation as that of the narrowband frequency reuse factor K in TDMA and GSM systems Multipath Fading Just as system capacity is affected by interference, it is also affected by propagation phenomena. Fading in a moving vehicle is more rapid than for pedestrians being caused by motion of the vehicle through stationary interference patterns, where the spatial scale of the interference pattern is the wavelength. We can address the impact of multipath fading on the performance of CDMA by first understanding under what circumstances will fading affect CDMA and what is that effect on the CDMA channel. When the multipath components delays separated by at least the decorrelation time of the spreading, they can be resolved by the CDMA waveform and can be separated by the despreader in the receiver because each component correlates at a different delay. When the multipath components are separated by less than the decorrelation time, then they cannot be separated in the receiver, and they interfere with one another, leading to flat fading. The duration of one spreading chip is 1/ Mcps = 814 ns in CDMA2000 and 1/3.84 Mcps = 260 ns in WCDMA. Multipath differences less than those will lead to flat fading, whereas greater separations will lead to resolvable multipath, which will be diversity combined by the receiver. The effects of fading depend mainly on the fading rate, which in turn depends on the velocity of the mobile station. Fading increases the average E b /N t needed for a particular error rate, which in turns causes capacity degradation. Coverage is also affected because a certain fading margin has to be built into the link budget. Power control mitigates the effects of fading at low speed and, to a less degree, at high speed. At high speed, the forward error correction coding and interleaving becomes more effective in combating fading as the characteristic fade time becomes less than the interleaver span. 2.7 Coverage Versus Capacity Trade-Off There is an inherit trade-off in CDMA between the capacity and coverage because of the way interference affects performance. On the uplink, interference increases as the load is increased and follows the expression of 1/(1 η), where η is fractional loading defined as N/N pole. Figure 2.25 shows how the interference

75 56 Smart Antenna Engineering rises over thermal noise as the fractional loading increases. To reach the pole capacity, the power that the mobiles are required to transmit goes to infinity to overcome this interference. As the required power increases, mobiles at the edge of coverage will begin to run out of transmitter power. That is, they will be asked to transmit more than their capability allows. It then follows that the system load should be controlled so that the planned service area never experiences coverage failures because of this phenomenon. This trade-off is not so much a problem or a limitation of CDMA systems as it is a system design consideration, which implies that maximum capacity and maximum coverage cannot be simultaneously achieved Coverage-Capacity Trade-Off in the Uplink Simple capacity models of the reverse link show that RF power rises with loading to overcome interference, as previously discussed. Real systems must operate below the pole capacity because real user stations have an upper bound to their transmitter power. As the load is increased, the average interference level also increases. Because mobile stations are transmitting at their maximum power, S, their corresponding received power at the base station is fixed and N is growing rapidly; that is, SNR degrades. This in turns means that those users would need to move closer to the base station to a point where a smaller path loss allows their received power and, consequently, SNR to be restored to the target set point necessary to achieve the required quality of service. In effect, the cell coverage shrinks by the same range. The effect of traffic loading on the range from a cell site is referred to as cell breathing. For example, for a loading of 50% of the pole capacity η = 0.5, the loss of coverage on the uplink is 1/(1 0.5) = 2, or 3 db loss. The impact of the interference rise over thermal noise on the CDMA coverage is illustrated in Figure CDMA networks are typically planned with this Figure 2.25 CDMA cell breathing.

76 Multiple Access Techniques for 2G and 3G Systems 57 fractional loading level. That is, the network is designed with the appropriate margins, and sufficient number of sites and the site locations are optimized so that there is no performance degradation when the system is loaded. In a poorly planned CDMA system, increasing this loading would lead to range reductions and open coverage holes under high loads. The pole capacity of a cell depends only on the average E b /N t target, the processing gain, and voice activity factor. The coverage area of a cell defined as the area over which all users obtain the target E b /N t, depends on the fractional loading relative to the pole capacity. Detailed analysis of the interaction of coverage and capacity is a complex process involving power control, soft handoff, fading, the mobility mix of subscribers, and other factors, as well as the differences between downlink and uplink. 2.8 Conclusion Third generation mobile communications systems are already offering significant improvements over their 2G counterparts in peak data rates and throughput. However, impairments caused by the propagation channel such as multipath and interference caused by other users in the system still represent challenges for the network and system design. As we have seen, in CDMA systems the performance of most users except those at the cell edge is interference limited. Techniques to reduce the average interference would therefore significantly improve performance in terms of capacity and coverage. As such, the technology of smart antennas can be considered as complimentary to existing multiple access techniques and an extra tool at the disposal of the system designer. References [1] Faruque, S., Cellular Mobile Systems Engineering, Norwood, MA: Artech House, [2] Garg, V. K., and J. E. Wilkes, Principles and Applications of GSM, Upper Saddle River, NJ: Prentice Hall, [3] Rappaport, T., Wireless Communications, Principles and Practices, New York: IEEE Press and Prentice Hall, [4] Yang, S. C., CDMA RF System Engineering, Norwood, MA: Artech House, [5] Garg, V. K., IS-95 CDMA and CDMA2000 Cellular/PCS Systems Implementation, Upper Saddle River, NJ: Prentice Hall, [6] TIA/EIA IS-95A, Mobile Station-Base Station Compatibility Standard for Dual-Mode Wideband Spread Spectrum Cellular System, 1995.

77 58 Smart Antenna Engineering [7] TIA/EIA-95B, Mobile Station-Base Station Compatibility Standard for Wideband Spread Spectrum Cellular Systems, [8] ANSI J-STD-008, Personal Station-Base Station Compatibility Requirements for 1.8 to 2 GHz CDMA Personal Communications Systems, [9] Ojanpera, T., and R. Prasad, Wideband CDMA for Third Generation Mobile Communications, Norwood, MA: Artech House, [10] TIA/EIA/IS A, Introduction to CDMA2000 Standard for Spread Spectrum Systems, [11] TIA/EIA/IS A, Physical Layer Standard for CDMA2000 Spread Spectrum Systems, [12] Esteves, E., M. Gurelli, and M. Fan, Performance of Fixed Wireless Access with CDMA2000 1xEV-DO, IEEE 58th Vehicular Technology Conference, Vol. 2, October 6 9, 2003, pp [13] Kuenyoung, K., K. Hoon, and H. Youngnam, A Proportionally Fair Scheduling Algorithm with QoS and Priority in 1xEV-DO, Personal, Indoor, and Mobile Radio Communications, 13th IEEE Intl. Symp., Vol. 5, September 15 18, 2002, pp [14] Huang, C. Y., et al., Schedulers for 1xEV-DO: Third Generation Wireless High-Speed Data Systems, 57th IEEE Semiannual Vehicular Technology Conference, Vol. 3, April 22 25, 2003, pp [15] Yavuz, M., and Paranchych, D. W., Adaptive Rate Control in High Data Rate Wireless Networks, IEEE Wireless Communications and Networking, Vol. 2, March 16 20, 2003, pp [16] Sindhushayana, N. T., and P. J. Black, Forward Link Coding and Modulation for CDMA2000 1XEV-DO (IS-856), 13th IEEE Intl. Symp. on Personal, Indoor, and Mobile Radio Communications, Vol. 4, September 15 18, 2002, pp [17] Yonghoon, C., and H. Youngnam Han, A Channel-Based Scheduling Algorithm for CDMA2000 1xEV-DO System, 5th Intl. Symp. on Wireless Personal Multimedia Communications, Vol. 2, October 27 30, 2002, pp [18] Yavuz, M., et al., Performance Improvement of the HDR System Due to Hybrid ARQ, IEEE VTS 54th Vehicular Technology Conference, Vol. 4, October 7 11, 2001, pp [19] Chung, W., W. L. Hong, and M. Jungbae, Downlink Capacity of CDMA/HDR, IEEE VTS 53rd Vehicular Technology Conference, Vol. 3, May 6 9, 2001, pp [20] Qualcomm, Inc., [21] Holma, H., and A. Toscala, WCDMA for UMTS, New York: John Wiley & Sons, [22] 3GPP Technical Specification , Physical Channels and Mapping of Transport Channels onto Physical Channels (FDD). [23] 3GPP Technical Specification , Multiplexing and Channel Coding (FDD). [24] 3GPP Technical Specification , Spreading and Modulation (FDD). [25] 3GPP Technical Specification , Physical Layer Procedures (FDD). [26] Kolding, T. E., et al., High Speed Downlink Packet Access: WCDMA Evolution, IEEE Vehicular Technology Society News, February 2003, pp

78 Multiple Access Techniques for 2G and 3G Systems 59 [27] Parkvall, S., et al., WCDMA Evolved High Speed Packet Data Services, Ericsson Review, No. 2, 2003, pp [28] Helmersson, K. W., and G. Bark, Performance of Downlink Shard Channels in WCDMA Radio Networks, Proc. IEEE Vehicular Technology Conference, Vol. 4, Spring 2001, pp [29] 3GPP TS25.855, High Speed Downlink Packet Access; Overall UTRAN Description Selected Bibliography CDMA Andersen, N. P., M. Pecen, and I. Gonorovsky, GSM/EDGE Evolution, Based on 8-PSK Circuits and Systems, Proc. of the 2003 International Symposium on ISCAS, Vol. 3, May 25 28, 2003 pp. III-598 III-601. Cai, J., and D. J. Goodman, General Packet Radio Service, GSM Communications Magazine, IEEE, Vol. 35, No. 10, October 1997, pp Christensen, G., et al., Wireless Intelligent Networking, Norwood, MA: Artech House, The Evolution of Digital Wireless Technology from Space Exploration to Personal Communication Services, IEEE Trans. on Vehicular Technology, Vol. 43, No. 3, August Four Laws of Nature and Society: The Governing Principles of Digital Wireless Communication Networks, Wireless Communications: Signal Processing Perspective, H. V. Poor and G. W. Wornell, (eds.), Upper Saddle River, NJ: Prentice Hall, 1998, pp Gilhousen, K. S., et al., On the Capacity of a Cellular CDMA System, IEEE Trans. Veh. Tech., Vol. 40, No. 2, 1991, pp Glisic, S., and V. Branka, Spread Spectrum CDMA Systems for Wireless Communications, Norwood, MA: Artech House, Hallmann, E., and R. Helmchen, Investigations on the Throughput in EDGE and GPRS Radio Networks, IEEE VTS 53rd Vehicular Technology Conference, Vol. 4, May 6 9, 2001, pp Jakes, W. C. Jr., Microwave Mobile Communications, New York: John Wiley & Sons, 1974; reprinted by IEEE Press, Lee, J. S., and L. E. Miller, CDMA Systems Engineering Handbook, Norwood, MA: Artech House, Lee, W. C. Y., Lee s Essentials of Wireless Communications, New York: McGraw-Hill, Lee, W. C. Y., Mobile Cellular Telecommunications, 2nd ed., New York: McGraw-Hill, Molkdar, D., W. Featherstone, and S. Larnbotharan, An Overview of EGPRS: The Packet Data Component of EDGE, Electronics & Communication Engineering Journal, Vol. 14, No. 1, February 2002, pp Parsons, D., The Mobile Radio Propagation Channel, New York: John Wiley & Sons, Peterson, R. L., R. E. Ziemer, and D. E. Borth, Introduction to Spread Spectrum Communications, Englewood Cliffs, NJ: Prentice Hall, Prasad, R., CDMA for Wireless Personal Communications, Norwood, MA: Artech House, 1996.

79 60 Smart Antenna Engineering Pribylov, V. P., and I. I. Rezvan, On the Way to 3G Networks: The GPRS/EDGE Concept, Proc. of the 4th IEEE-Russia Conference on MEMIA, December 23 26, 2003, pp Proakis, J. G., Digital Communications, 2nd ed., New York: McGraw-Hill, Simon, M. K., et al., Spread Spectrum Communication Handbook, New York: McGraw-Hill, Ross, A. H. M., and K. S. Gilhousen, CDMA Technology and the IS-95 North American Standard, in The Mobile Communications Handbook, Boca Raton, FL: CRC Press and IEEE Press, 1996, pp Scholtz, R. A., The Origins of Spread Spectrum Communications, IEEE Trans. Commun., COM-30, May 1982 (Part I), pp Shannon, C. E., Communication in the Presence of Noise, Proc. IRE 37, January 1949, pp Turin, G. L., Introduction to Spread Spectrum Antimultipath Techniques and Their Application to Urban Digital Radio, Proc. IEEE 68, 1980, pp Viterbi, A., CDMA: Principles of Spread Spectrum Communication, Reading, MA: Addison-Wesley, Viterbi, A. J., Error Bounds for Convolutional Codes and an Asymptotically Optimum Decoding Algorithm, IEEE Trans. Inform. Th. IT-13, 1967, pp Viterbi, A. J., A. M. Viterbi, and E. Zehavi, Performance of Power-Controlled Wideband Terrestrial Digital Communications, IEEE Trans. on Comm., Vol. 41, No. 4, 1993, pp Viterbi, A. J., et al., Soft Handoff Extends CDMA Cell Coverage and Increases Reverse Link Capacity, IEEE J. Selected Areas in Communications, Vol. 12, No. 8, 1994, pp Viterbi, A. M., and A. J. Viterbi, Erlang Capacity of a Power Controlled CDMA System, IEEE J. on Selected Areas in Communication, Vol. 11, No. 6, 1993, pp Yallapragada, R., V. Kripalani, and A. Kripalani, EDGE: A Technology Assessment, IEEE International Conference on Personal Wireless Communications, December 15 17, 2002, pp Yang, S. C., CDMA RF System Engineering, Norwood, MA: Artech House, GSM/GPRS/EDGE Bates, R. J., GPRS: General Packet Radio Service, New York: McGraw-Hill, Mehrotra, A., GSM System Engineering, Norwood, MA: Artech House, Redl, S. M., et al., GSM and Personal Communications Handbook, Norwood, MA: Artech House, Redl, S. M., et al., Introduction to GSM, Norwood, MA: Artech House, Seurre, E., et al., EDGE for Mobile Internet, Norwood, MA: Artech House, Steele, R., et al., GSM, cdmaone and 3G Systems, New York: John Wiley and Sons, Timo, H., GSM, GPRS, and EDGE Performance, 2nd ed., New York: John Wiley & Sons, 2003.

80 3 Spatial Channel Modeling 3.1 Introduction The detailed knowledge of radio propagation characteristics is essential to develop a successful wireless system. Measurement studies have been carried out to identify propagation loss, spatial distribution of power, wideband and narrowband statistics concerning the random variables of received signals at a fixed location due to any surrounding movement, and delay spread. A radio channel is a generally hostile medium in nature. Transmitted signals undergo several propagation phenomena such as reflection, diffraction, and scattering before they reach the receiver. This is mainly caused by the existence of objects in the physical channel between a transmitter and a receiver such as buildings, trees, mountains, hills, and moving objects. Therefore it is rather difficult to predict the channel behavior. Traditionally, radio channels are modeled in a statistical way using real propagation measurement data. Classical propagation models, commonly used for narrowband transmission systems, represent a signal in the radio environment using a large-scale path loss component together with a medium-scale slow varying component having a lognormal distribution, and a small-scale fast varying component with a Rician or Rayleigh distribution, depending on the presence or absence of the line-of-sight (LOS) component between the transmitter and receiver [1, 2]. Accordingly, conventional radio propagation models describing a wireless cellular environment have focused on: Area mean power depending on the path loss characteristics between the transmitter and receiver. Local mean power within that area, which is slow varying. This can be represented by a lognormal distribution. 61

81 62 Smart Antenna Engineering Superimposed fast fading instantaneous power, which follows a Rayleigh or nonline-of-sight (NLOS) or Rician distribution (LOS). Figure 3.1 illustrates a typical propagation environment. Large variations in the transmission path between the transmitter and receiver can be found, ranging from direct LOS to severely obstructed paths due to buildings, mountains, or foliage. The phenomenon of decreasing received power with distance due to reflection, diffraction around structures, and refraction within them is known as path loss. Propagation models have been developed to determine this path loss and are known as large-scale propagation models because they characterize the received signal strength by averaging the power level over large transmitter-receiver separation distances, in the range of hundreds or thousands of meters. On the other hand, medium-scale propagation models determine the gradual changes in the local-mean power if the receiving antenna is moved over distances larger then a few tens or hundreds of meters. The medium-scale variation of the received signal power is called shadowing, and it is caused by obstruction by trees and foliage. The term local-mean power is used to denote the power level averaged over a few tens of wavelengths, typically 40λ. Finally, small-scale propagation models characterize the fast variation of the signal strength over a short distance on the order of a few wavelengths or over short time duration on the order of seconds. Small-scale fast fading, also known as short-term fading or multipath fading, is due to multipath reflections of a transmitted wave by local scatterers such as houses, buildings, and man-made Direct path Multipath Multipath Multipath Figure 3.1 Mobile radio propagation channel.

82 Spatial Channel Modeling 63 structures, or natural objects such as forests surrounding a mobile unit. Typically, detailed models are needed for a complete coverage and capacity analysis in a certain region. On the other hand, if only rough capacity and range calculations are needed, simple and easy-to-use models are sufficient. In addition to the modeling of the propagation environment, the mobility of the wireless terminals also needs to be understood in each radio environment. Mobility modeling has a significant impact on the analysis of radio resource management, channel allocation, and handoff performance. 3.2 Radio Environments and Cell Types There are a large number of environments where mobile radio systems can operate. These include large and small cities, with variations in building construction, as well as tropical, rural, desert, and mountainous areas. Moreover, antenna design and height impacts the radio environments. Since it is impossible to consider all possible radio environments in the design of a mobile radio system, more general models that will consider the essence of different radio environments are required. Therefore, the large number of possible radio environments has to be condensed into a finite set of generic radio environments [3 6]. One approach is to classify a radio environment based on the typical cell size, which leads to: Macrocells: In a macrocell the base station antenna is placed above the rooftops and is much higher than the mobile users. Usually macrocells have a radius of more than 1 km and can be found in rural as well as urban areas. Minicells: A minicell can be considered as a small macrocell where the base station antenna is placed at the same height as the rooftops. This type of cell is only used in urban environments with cell radius ranging from 700m to 1 km. Microcell: In a microcell the base station antenna is placed in street level with a typical antenna height in the range of 5m. It has a cell radius of typically 200 to 500m and is mainly used for increasing the coverage and capacity in a dense urban environment. Picocell: A picocell is mainly for indoor usage. The cell radius is about 10 to 20m and is limited by the building itself due to high penetration losses in the walls, number of floors, and their compositions. Another approach is to classify radio environments based on the nature of the mobile users being served by the system. This leads to the following classification:

83 64 Smart Antenna Engineering Vehicular radio environment: The vehicular environment is characterized by large macrocells and large transmit powers as well as high mobile speeds (fast-moving vehicles). There is typically no LOS component, and the received signal is mostly composed of reflections. In these environments, the average power of the received signal decreases with distance raised to some exponent, referred to as the path loss exponent, which varies depending on the environment but is typically between three and five. In addition, shadowing is caused by obstruction from trees and foliage, and the resulting medium-scale variation in the received signal power can be modeled with a lognormal distribution. The standard deviation of that power varies considerably; for example, 8 to 10 db is generally used in urban and suburban areas, whereas a lower value is used in rural and mountainous areas. In addition, small-scale fading is characterized by Rayleigh distribution. Typical delay spreads in this case are on the order of 0.8 µs but can be as high as tens of microseconds. Outdoor to indoor and pedestrian radio environment: This radio environment is characterized by small microcells and low transmit powers with the antennas usually located below rooftops. Both LOS and NLOS multipath components exist. Indoor coverage can also be provided from this outdoor base station. The path loss exponent varies quite a bit and can be anywhere from two in areas with LOS up to six with NLOS cases due to trees and other obstructions along the path. Furthermore, a mobile station can experience a sudden drop of 15 to 25 db when it moves around a corner. The standard deviation of the shadowing in these environments varies from 10 to 12 db and the small-scale fading is either Rayleigh (NLOS) or Rician (LOS), with typical delay spreads on the order of 0.2 µs. Indoor office radio environment: In the indoor office radio environment transmit powers are small and base stations and users are located indoors. Path loss attenuation exponent varies from two to five depending on the scattering and attenuation by walls, floors, and metallic structures. Note that each one of these environments has individual characteristics with respect to path loss attenuation, shadowing, and small-scale fading. 3.3 The Multipath Channel The short-term fluctuations of the received signal caused by multipath propagation are called small-scale fading. The different propagation path lengths of the multipath signal give rise to different propagation time delays. A multipath

84 Spatial Channel Modeling 65 channel can be represented by a power-delay profile consisting of different sets of distinct paths, which are also called multipath taps. Depending on the phase of each multipath signal when arriving at the receiver, they sum either constructively or destructively. Consequently, the power of each multipath tap is time varying, resulting in fading dips. The depth of the fading dips depends on the channel type. Using a baseband complex envelope representation and modeling the RF channel as a time-variant channel, we can represent the classical channel impulse response as a sum of L multipath components given by [1] L j ϕ l ( t) ht (, τ) = A( t) e δ( t τ ( t) ) l = 0 l l (3.1) where A l (t) is the amplitude, ϕ l (t) is the phase, and τ l (t) is the time delay of the signal component. The distribution of the instantaneous power of the channel taps can be described by a distribution function, which depends on the radio environment. A so-called Rayleigh fading channel is the most severe mobile radio channel, with deep fading dips every λ/2. In a Rayleigh fading channel all multipath taps are independent and there is no dominant path. The envelope of individual multipath components in this case can be characterized by a Rayleigh distribution given by: r pr ( ) e = 2 σ ( r 2 2 σ 2 ) 0 r 0 r < 0 (3.2) In a Rician fading channel, the fading dips are shallower due to a dominant path in addition to the scattered paths. This is usually the case in microcell and picocell environments due to the existence of LOS. 3.4 Channel Characterization The multipath fading channel can be characterized based on delay spread, coherence bandwidth, Doppler spread, and coherence time. The root-meansquare (rms) value of the delay spread is a statistical measure that describes the spread of the multipath components around the mean delay of the channel. The maximum delay spread tells the delay difference between the first and last multipath components in the power-delay profile. The coherence bandwidth is the maximum frequency difference for which the signals are still strongly correlated. The coherence bandwidth is inversely proportional to the delay spread (i.e., the smaller the delay spread the larger the coherence bandwidth). If the transmission bandwidth of the signal is larger than the coherence

85 66 Smart Antenna Engineering bandwidth, the signal will undergo frequency selective fading. On the other hand, a flat fading channel results if the transmission bandwidth of the signal is smaller than the coherence bandwidth. The coherence bandwidth can be thought of as being a measure of the diversity available to a RAKE or equalized receiver. A smaller coherence bandwidth means a higher order diversity. If the coherence bandwidth is as large as the transmission bandwidth, then the entire received spectrum would be observed to fade. The maximum delay spread can be used to calculate how many resolvable paths exist in the channel that could be used in the RAKE receiver. In addition, the movement of the mobile station gives rise to a Doppler spread, which is the width of the observed spectrum when an unmodulated carrier is transmitted. If there is only one path from the mobile to base station, the base station will observe a zero Doppler spread combined with a simple shift of the carrier frequency (Doppler frequency shift). The Doppler frequency varies depending on the angle of the mobile station movement relative to the base station. The range of values when the Doppler power spectrum is nonzero is called the Doppler spread. The reciprocal of the Doppler shift is a measure of the coherence time of the channel. The coherence time is the duration over which the channel characteristics do not change significantly. 3.5 Path Loss Models Typically, path loss models are derived using a combination of analytical and empirical methods. In the empirical approach, the measured data is modeled using curve fitting or analytical expressions. The validity of empirical models in other environments and frequencies can only be validated by comparing the model to data measured from the specific area and for the specific frequency. It should be noted that these models present only a snapshot of the real radio environment Okumura-Hata Propagation Models The empirical Okumura method [2] is based on exhaustive measurements that were performed in the Tokyo metropolitan area. The results were a series of curves, plotting recorded field strength as a function of distance from the transmitting antenna. The model is valid for distance ranges of 1 km to 100 km, frequency bands from 150 MHz to 2,000 MHz, and base station effective antenna heights from 30m to 1,000m Hata s Model Since Okumura s curves and tables were intended to be used manually as a look-up resource, Hata was able to derive equations from Okumura s work. This allowed for accurate computation of path loss without having to peer

86 Spatial Channel Modeling 67 through any set of graphs. The standard Okumura-Hata model for an urban city is given by: Max. Path Loss = ( log H b ) log R log H b log f a( H m ) (3.3) where H b is the radio base station (RBS) effective antenna height, H m is the mobile subscriber antenna height, f is the operating frequency, and R is the distance between the RBS and the mobile station (MS), or the radius to RBS from the measurement point. The propagation model is valid under the following conditions: 150 < f < 1,500 MHz, 1 < R < 20 km, 30 < H b < 200m and 1 < H m < 10m. The mobile height correction factor a(h m ) can be computed as follows: a(h m ) = 3.2 ( log (11.75 H m )) 1/2 4.97, when f 400 MHz for urban environments a(h m ) = (1.1 log f 0.7) H m 1.56 log f for suburban or rural areas The COST-231 Model (Suburban) This model has been developed by the European Union s Forum for Co-operative Scientific and Technical Research (COST). Since the traditional Okumura-Hata model is restricted to application in the frequency band below 1,500 MHz, it is not applicable in either the PCS or IMT-2000 spectrum regions. The COST-231 model [2] was developed based on analysis of Okumura s propagation curves in the higher frequency regions with the aim of implementing a suitable formula that characterizes radio wave propagation in the PCS and IMT-2000 bands. The results led to the following adaptation of the Okumura-Hata equation: where: Max. Path Loss = ( log H b ) log R log H b log f a(h m ) (3.4) a(h m ) = 3.2 ( log (11.75 H m )) 1/2, when f 400 MHz for large cities a(h m ) = (1.1 log f 0.7) H m 1.56 log f for medium small cities or rural areas. 3.6 Spatial Channel Modeling As we can see from previous sections, classical propagation models focus on the power delay profile without taking into account the angular distribution of the

87 68 Smart Antenna Engineering multipath components. To analyze the performance impact of smart antennas at the link and system level, the spatial domain must also be considered. Channel models that characterize the DOA of multipath components are referred to as spatial channel models. Taking into account the angular dependence of the channel, the directional channel impulse response can be written as [7]: ( ) ( ) L j ϕ l ( t) ( ( ) ( )) ( ( )) ht, τθφ,, = A t e a θ t, φ t δt τ t l = 0 l l l l (3.5) where α(θ, φ) is the array response vector given by: [ ] jψ j( M 1 ) ψ T a( θφ, ) = 1 e K e (3.6) As can be seen from (3.5) the spatial channel can be characterized by a number of multipath components L, each with a complex amplitude A l, elevation angle A l, azimuth angle θ l, and delay τ l. We have previously seen how classical channel models define envelope probability distribution functions, delay and Doppler spread ranges for different radio environments, and cell types. The same approach can also be adopted for spatial channel modeling to define the characteristics of the parameters that make up the directional channel impulse response. In order for (3.5) to be used in link level or any other type of simulation intended to study the performance of smart antennas, we either have to define the amplitude, time delay, and angular spread distributions of the different multipath components or the spatial distribution of the different scatterers in the channel that can then be used to generate different multipath components Spatial Channel Model Parameters It has been known that multipath components tend to cluster in groups that could be exploited in modeling the structure of the directional or angular impulse response. For instance, a cluster can be viewed as a collection of multipath components that experience the same small-scale variations since small scale variations such as fast fading, caused by the instructive or destructive interactions of multipath components, occur on the scale of a wavelength. This clustering is mainly caused by the fact that the physical structures that cause scattering, reflections, and shadowing of the radio signals can be grouped into those in the vicinity of the mobile, those located near or around the base station, and a third group of distant objects that might exist in the channel. This spatial distribution of objects in the channel will cluster the multipath components into groups of signals with similar time delay and angular properties.

88 Spatial Channel Modeling Number of Clusters Clearly, there must be at least one cluster present in the channel that occurs due to local scattering around the mobile or base station. The existence of more clusters will depend on whether there is any scattering due to other distant objects in the channel, such as buildings, hills, and so on. Therefore, the number of clusters, N C, has to be determined based on measurements. In [8] the appropriate values for the number of clusters for different radio environments for the COST 259 model has been identified based on power delay profile measurements throughout Europe. It was found that in macrocell urban environments, N C ranges from one to two. This is due to the presence of scattering from local as well as far objects. In suburban and rural environments, N C is typically around one, implying that only those objects in the vicinity of the receiver contribute to the channel. On the other hand, [9] considers the number of clusters in any urban environment to be six Spatial Distribution of Clusters and Scatterers Since the local objects in the vicinity of the MS or BS will significantly contribute to the multipath, it is reasonable to seek a model for their spatial distributions from which the components of the channel impulse response could be derived. Let us consider the macrocell case where the BS antenna is mounted on a rooftop higher than any of the local scatterers; as a result, the scattering is dominated by those objects around the MS. In fact, most of the relevant scatterers are those located closer to the MS since they will have the greatest impact on the channel. In addition, the signals received at the BS will mainly arrive from a certain angular region, as shown in Figure 3.2. A distribution function that approximates this physical behavior is the Gaussian distribution given by f ( r) r r MS 2 R r r ζ = R e 2 2 2π 2 (3.7) where r is the position vector, r MS is the MS location vector, assuming the BS as the origin, ζ is a normalization constant, and R is the radius of the scattering circle, shown in Figure 3.2. Recall that with this Gaussian shape, f (r) will be larger for objects closer to the mobile (i.e., for small r r MS ) and will decrease for increasing r or for objects further away from the MS. r MS Base Station Azimuth Power Spectrum and Angle Spread Once the spatial distribution of the scatterers is known, it is possible to compute the azimuth power spectrum, that is, the distribution of the received power versus the azimuth angle. Various PDFs have been proposed in the literature for the

89 70 Smart Antenna Engineering Scatterers MS BS Figure 3.2 Macrocell base station model. azimuth distribution, including a cos n (φ) distribution, a uniform distribution, and a normal distribution [10]. However, based on field measurements, it was found that Laplacian distribution is a better representation for the azimuth power spectrum [9, 11]. From the azimuth power spectrum and angle of arrival (AOA) distribution, we can derive an expression for the azimuth angle spread. Let us consider the geometry shown in Figure 3.3. Here we assume that each cluster will contribute N p paths to the channel. To characterize each of these paths, we need to define their AOA as the mean angle with which an arriving path s power is received by the BS array with respect to the bore site and their angle spread defined as the rms of angles with which an arriving path s power is received by the base station array and azimuth power spectrum. The Laplacian distribution was adopted by the COST 259 model [8] and the spatial channel models jointly developed by the third generation partnership programs Third Generation Partnership Project 2 (3GPP2) and 3GPP to model the azimuth power spectrum of both paths and clusters. On a cluster basis, the azimuth power spectrum can then be written as N p ( φ ) P l l Pφ ( φ) = = 1 e 2σ φ 2 φl φo σ φ (3.8)

90 Spatial Channel Modeling 71 Cluster γ MS σ φ φ c LOS Path φ o BS Figure 3.3 Spacial channel model parameters. N p where P( φ ) l = 1 l is the total power received from the cluster. On a path basis, we can write the azimuth power spectrum as σ p ( φ ) = ( φ) P N e G φp l norm 2 φl φp (3.9) where N norm is a normalization constant, φ p is the mean path AOA, σ p is the angle spread, and G(φ) is the BS antenna gain at angle φ. The normalized azimuth power spectrum is shown in Figure 3.4 for various spread angles for a mean path AOA of 45. Now, let us look at the angle spread given by [12] { } [ {}] Sφ = E φ E φ 2 2 (3.10) Using (3.8) we can write E 2 ( φ ) = N c N p i = 1 l N N c p i = 1 φ l P P ( φi l) 2 i, l φ, φ ( φi, l) (3.11)

91 72 Smart Antenna Engineering PDF Angle spread:10 Angle spread:20 Angle spread: Azimuth Angle Figure 3.4 Azimuth power spectrum as a function of φ and σ p for φ p of 10, 20, and 30. From Figure 3.3 we can see that the path AOA is given by Substituting (3.12) in (3.11) we then get φi = φo + φc + γ (3.12) E 2 ( φ ) = = N N c p 2 ( φ ο +φ c,i +γi,l ) Pφ( φi, l ) i = 1 N l N c i = 1 N p l P φ ( φi, l) c p 2 ( φo φc, i) Pφ( φi, l) N i = 1 N + + N l N c p c p 2 ( γi, l) Pφ( φi, l) + 2 ( φo+ φc, i) Pφ( φi l) i = 1 l N c i = 1 i = 1 N p l N P l φ ( φi, l) γ, i, l (3.13) Similarly

92 Spatial Channel Modeling 73 [ E ( φ) ] 2 = = N N c p ( φo + φc, i + γi,l) Pφ( φi, l) i= l l N c i = 1 N p l P N φ ( φi, l) Nc p c p ( φo + φc, i) Pφ( φi, l) + γi, lpφ( φi, l) i= l l N c i = 1 N p l P N i = 1 ( i l) φ φ, N l 2 2 (3.14) We can see from (3.13) that S φ is a random variable that is a function of the LOS angle φ o, the clusters angles φ c, the path angles within a cluster γ, the number of clusters, the number of paths N p, and the azimuth power spectrum. Note that the AOA of the clusters and paths have PDFs that could, for example, be given by [13]: f ( φ ) c = 1 2πσ c e φ c 2 σ c (3.15) and f ( γ) = 1 2πσ p e γ 2 σ p (3.16) where σ c is the cluster s AOA spread and σ p is the path AOA spread. Let us consider the special case where all clusters have the same AOA spreads σ c and all paths have the same power. It follows that the expected value of S φ is given by [11] ( φ) E S N c 2 Pi 2 2 i = σ c + σ = 1 p 1 (3.17) N N c p Pφ ( φ) In the spatial channel model adopted by the 3GPP and 3GPP2 [9], two values are considered for the path angle spread at the BS, 2 and 5 corresponding to path AOAs of 50 and 20, respectively. i = 1 l = 1

93 74 Smart Antenna Engineering Mobile Station Azimuth Power Spectrum and Angle Spread Due to the nature of the scatterers that are mostly located around the mobile station, it is reasonable to assume that the azimuth power spectrum has a uniform distribution [ 180, 180 ] for which the angle spread is given by [10] S φ = = 104 However, in some cases where the scatterers are not uniformly distributed around the mobile, a Laplacian power azimuth spectrum can be a better approximation for the path s power as follows: ( ) P φ = N e φp l norm 2 φl φp σ p (3.18) where the MS is assumed to have an omnidirectional antenna. In this case, the angle spread is be given by [10] π 3 σ p Sφ = N σ e π + 4σ + 8πσ 4 ( ) norm p p p (3.19) In such a case, the path angle spread is expected to be lower compared with the uniform azimuth power spectrum. 3.7 Spatial Channel Model Application in System Simulations To evaluate the performance of a given smart antenna algorithm or implementation, a system level simulation needs to be carried out. In this approach, multiple base stations or sectors and mobile stations are considered in addition to mobility modeling to account for the fast fading resulting from a user s motion. A general procedure to generate a spatial channel model for this type of system level simulation may consist of the following steps: 1. Select a radio environment such as urban, suburban or rural. 2. Select a cell type, such as macrocell, microcell, or picocell. 3. Define the path loss model based on the radio environment and cell type selected above. For instance, the COST 231 model can be used for suburban macrocells.

94 Spatial Channel Modeling Define the antenna pattern and gain both at the base station and mobile station. This will directly impact the azimuth power spectra for the multipath components. 5. Determine the locations and orientations of all sectors and users in the system. Based on this, LOS AOAs as well as distances between the users and sectors can be computed. 6. Determine the delay spread, angle spread, and lognormal shadow fading parameters. The distributions of these parameters are assumed to have the form σ DS ( αε DS + µ DS ) = 10 for the delay spread (3.20) where µ DS is the logarithmic mean of the distribution of the delay spread and ε DS is the logarithmic standard deviation of the distribution of the delay spread. Similarly, for the angle spread we assume the distribution σ AS ( βε AS + µ AS ) = 10 (3.21) where µ AS is the logarithmic mean of the distribution of the angle spread and ε AS is the logarithmic standard deviation of the distribution of angle spread. The shadow fading distribution is given by σ LN = 10 ( σ γ 10) SF (3.22) The parameters α, β and γ are defined in [9, 14] and σ SF is the well-known lognormal shadow fading standard deviation. 7. Determine the number of clusters N C and their random delays τ n such that τn > τn > > τ c c These delays are assumed to follow the classical exponentially decaying profile based on the Laplacian distribution: 1 Pτ( τ) = e σ τ 2 τ τm σ τ (3.23) where σ τ is the well-known rms delay spread. 8. Determine the random average power for each cluster. It is expected that the cluster power is a function of the distance between that cluster and the BS or MS. Furthermore, the longer the cluster s delay is

95 76 Smart Antenna Engineering (implying additional path loss), the more attenuation there is. A simple model similar to that adopted by the COST 259 would then set the power of the first cluster equal to the transmit power attenuated by the path loss computed from the appropriate Hata or COST 231 empirical formulae, taking into account shadow fading as P 1 = P PL σ LN n and additional cluster powers as P = P σ 10 n PL LN n k τ min ( τn τ1, τmax ) 10 (3.24) where k is shown in Figure 3.5 and the S n is the shadow fading gain defined above. This implies that we have an exponentially decaying power profile up to a maximum delay after which the power received is almost negligible. 9. Determine the powers and phases of the N p paths within each cluster. Here we can assume that all paths have identical powers (P path = P n /N c ). The phases can be drawn from a uniform distribution [0 to 360 ]. 10. Determine the AOA of the clusters and paths. This is achieved by first locating the clusters according to their spatial distribution and then computing their AOA relative to either the BS or MS. The paths AOA are calculated based on offsets from the clusters AOA such that the desired angle spread is obtained. 11. Associate the clusters and paths with the BS and MS. This will allow the calculation of the antenna gains at the respective AOAs of each of the multipath components. P τ k τ k τ τ max τmax τ Figure 3.5 Power delay profile.

96 Spatial Channel Modeling Angle Spread Impact Due to multipath, shadowing, and mobile speed the wireless propagation channel causes the transmitted signal to appear spread in time, frequency, and angle at both the base station and mobile user sides. As we discussed earlier, this dispersion can be characterized in terms of delay spread, Doppler spread, and angle spread [15]. These three parameters determine frequency fading, temporal fading, and spatial fading, respectively. The impact of the angle spread in terms of spatial fading is illustrated in Figures 3.6 through 3.9. In a channel with a (a) (b) Figure 3.6 (a, b) Fading envelope, Laplacian Azimuth power spectrum, AS = 10, angle of arrival 90.

97 78 Smart Antenna Engineering (a) (b) Figure 3.7 (a, b) Fading envelope, Laplacian Azimuth power spectrum, AS = 60, angle of arrival 0. narrow angle spread the fading envelope across an antenna array s elements is relatively constant in space, as can be seen in Figure 3.6 (AS = 10 ). This is due to the fact that the signals across the antennas are correlated. This scenario is beneficial to the performance of beamforming techniques. A narrow angle spread helps maintain a focused and narrow beam for better interference reduction. On the other hand, in Figures 3.7, 3.8, and 3.9 we see the impact of

98 Spatial Channel Modeling 79 (a) (b) Figure (a, b) Fading envelope, Laplacian Azimuth power spectrum, AS = 360, angle of arrival wide angle spread (60 and 360 ). In these cases, the signal experiences fading in space, where we can clearly see peaks and valleys in the fading envelope. This results from paths with low cross-correlation and is beneficial for spatial diversity applications that result in higher diversity gain as the signals become uncorrelated, but it will degrade the performance of transmit beamforming.

99 80 Smart Antenna Engineering Figure 3.9 Fading envelope, AS = 360, angle of arrival 0. References [1] Rappaport, T., Wireless Communications, Principles and Practices, NJ: IEEE Press and Prentice Hall, [2] Siwiak, K., Radiowave Propagation and Antennas for Personal Communications, Norwood, MA: Artech House, [3] Lee, W. C. Y., Mobile Communications Engineering, New York: McGraw-Hill, [4] Parsons, J. D., The Mobile Radio Propagation Channel, New York: John Wiley & Sons, [5] Lee, W. C. Y., Mobile Cellular Telecommunications Systems, New York: McGraw-Hill, [6] Fleury, B. H, and P. E. Leuthold, Radiowave Propagation in Mobile Communications: An Overview of European Research, IEEE Communications Magazine, February [7] Ertel, R. B., et al., Overview of Spatial Channel Models for Antenna Array Communication Systems, IEEE Personal Communications, February [8] Correia, L. M., Wireless Flexible Personalized Communications, COST 259: European Co-operation in Mobile Radio Research, New York: John Wiley & Sons, [9] 3GPP-3GPP2 SCM-121, Spatial Channel Model Text Description, March 14, 2003.

100 Spatial Channel Modeling 81 [10] Fuhl, J., A. F. Molisch, and E. Bonek, Unified Channel Model for Mobile Radio Systems with Smart Antennas, IEE Proc. Radar, Sonar Navigation, Vol. 145, No. 1, February [11] 3GPP-3GPP2 SCM-027, Note on the Angle Spread Distribution, Motorola, May 22, [12] Fleury, B. H., Direction Dispersion and Space Selectivity in the Mobile Radio Channel, IEEE VTS Fall 52nd Vehicular Technology Conference, 2000, Vol. 2, [13] 3GPP-3GPP2 SCM-025, RMS Angle Spread, Motorola, May 3, [14] 3GPP-3GPP2 SCM-029, Correlated System Level Spatial Channel Model, June 5, [15] Buehrer, R. M., Generalized Equations for Spatial Correlation for Low to Moderate Angle Spread, Proc. 10th Va. Tech. Symp. Wireless Communication, Blacksburg, VA, June 2000, pp Selected Bibliography Beach, M., B. Allen, and P. Karlsonn, Correlation of Power Azimuth Spectrum for Varying Frequency Division Duplex Spacings, EPMCC 2001, Centre for Communication Research, University of Bristol, Vienna, February 20 22, Bertoni H. L., et al., Sources and Statistics of Multipath Arrival at Elevated Base Station Antenna, IEEE 49th Vehicular Technology Conference, Vol. 1, Chen, M, and Asplund, H., Measurements and Models for Direction of Arrival of Radio Waves in LOS in Urban Microcells, 12th IEEE Int. Symp. on Personal, Indoor, and Mobile Radio Communications, Vol. 1, September Greenstein, L. J., et al., A New Path-Gain/Delay-Spread Propagation Model for Digital Cellular Channels, IEEE Trans. on Vehicular Technology, Vol. 46, No. 2, May Hata, M., Empirical Formula for Propagation Loss in Land Mobile Radio Services, IEEE Trans. on Vehicular Technology, Vol. VT-29, No. 3, August Kuchar, A., J. P. Rossi, and E. Bonek, Directional Macrocell Channel Characterization from Urban Measurements, IEEE Trans. on Antennas and Propagation, Vol. 48, No. 2, February Liberti, J. C., and T. S. Rappaport, A Geometrically Based Model for Line-of-Sight Multipath Radio Channels, Proc. IEEE VTC 96, Atlanta, GA, May Lu, M., T. Lo, and J. Litva, A Physical Spatio-temporal Model of Multipath Propagation Channels, Proc. IEEE VTC 97, Phoenix, AZ, May Pajusco, P., Experimental Characterization of D.O.A at the Base Station in Rural and Urban Area, Proc. IEEE VTC 98, Ottawa, Canada, May Pedersen, K. I., P. E. Mogensen, and B. H. Fleury, A Stochastic Model of Temporal and Azimuthal Dispersion Seen at the Base Station in Outdoor Propagation Environments, IEEE Trans. on Vehicular Technology, Vol. 49, No. 2, March Pedersen, K. I., P. E. Mogensen, and B. H. Fleury, Spatial Channel Characteristics in Outdoor Environments and Their Impact on BS Antenna System Performance, 48th IEEE Vehicular Technology Conference, Vol. 2, 1998.

101 82 Smart Antenna Engineering Pedersen, K. I., P. E. Mogensen, and B. H. Fleury, Spatial Channel Characteristics in Outdoor Environments and Their Impact on BS Antenna System Performance, Proc. IEEE VTC 98, Ottawa, Canada, May Petrus, P., J. H. Reed, and T. S. Rappaport, Geometrically Based Statistical Model for Macrocellular Mobile Environments, Proc. IEEE Globecom 96, London, November Saleh, A. A. M., and R. A. Valenzuela, A Statistical Model for Indoor Multipath Propagation, IEEE Journal on Selected Areas in Communications, Vol. SAC-5, No. 2, February Turin, G. L., et al., A Statistical Model of Urban Multipath Propagation, IEEE Trans. on Vehicular Technology, Vol. VT-21, No. 1, February 1972.

102 4 Fixed Beam Smart Antenna Systems 4.1 Introduction In wireless system design and planning, one has to deal with two main problems, coverage and capacity. Coverage designs include, among other parameters, the selection of the number of base station sites, locations, heights, antenna orientation, and transmit power required to provide service in a given geographic area. When the system is first deployed, the number of subscribers is low and the aim is to maximize the coverage area of each site. As the number of network users grows, the number of users per site and sector also grows. As more users are added to a sector, the base station transmit power as well as the total power transmitted by those users increases, which effectively increases the interference on both the forward and reverse links. The system then becomes capacity or interference limited. One of the most common ways to deal with this interference problem is sectorization. 4.2 Conventional Sectorization In base stations employing omnidirectional antennas, the transmit power is equally radiated in all directions. The equal distribution leads to a portion of the power being transmitted throughout the cell but not received by the user. This wasted power then becomes forward link interference to other base stations or users in other cells. Similarly, each new user added to a cell increases the interference and noise levels on the reverse link. This results in a reduction in the signal-to-noise ratio, which in turn degrades the performance of the detection and demodulation operations. One way to reduce interference is to divide the cell 83

103 84 Smart Antenna Engineering into a number of smaller sectors using directional antennas. The most common scheme is the three 120 -sectors; however, two and six-sector cells also have some practical applications. As we can see from Figure 4.1, in a three-sector site the radiation pattern of the directional antenna allows it to receive substantially higher power levels from its own sector compared with that received from the other two sectors. It is obvious that sectorization is an effective technique that can increase capacity. In fact, since CDMA capacity is noise or interference limited, assuming ideal antenna radiation patterns, the capacity of an N s -sector cell will be N s times that of an omnidirectional cell. This capacity gain is often referred to as sectorization gain (SG). In practice, overlapping sector coverage areas due to nonideal antenna radiation patterns will increase the multiuser interference, reducing SG [1 3]. It is well known that CDMA has a soft capacity, which is determined by the balance between the required SNR for each user and the spread spectrum processing gain (PG) given by PG = W (4.1) R where W is the bandwidth of the spreading signal and R is the user s data rate. The figure of merit of the digital receiver is the dimensionless SNR given by Figure 4.1 Three-sector patterns.

104 Fixed Beam Smart Antenna Systems 85 E N b 0 = Energy per bit Noise plus Interference power spectraldensity (4.2) where the energy per bit is given by P E = b R (4.3) To derive an approximate SG for a CDMA system, let us consider the reverse link capacity of an omni system given by N omni = E I b o W R ( + f ) 1 ν omni (4.4) where v is the voice activity factor, N omni is the number of users, I o is the total interference density, and f is the reuse efficiency f I oc = (4.5) I sc In (4.5) I oc denotes the other cell interference power and I sc is the same cell interference power. From (4.4) it can be readily seen that reducing the interference level increases the number of maximum supportable users in a cell. Now, let us consider the capacity of a single sector given by: N sect = E I b o W R ( + f ) 1 ν sect (4.6) Hence the sectorization gain for an N-sector cell becomes SG N N f sect 1+ = s = N s N 1+ f omni omni sect (4.7) It is clear that the sectorization gain is highly dependent on the amount of reduction in interference provided by the antenna, which in turn is a function of the antenna beamwidth and the size of the overlap region. Based on simulation results for reuse efficiency provided in [4], Figure 4.2 shows SG as a function of

105 86 Smart Antenna Engineering Sectorization gain Number of sectors Figure 4.2 Sectorization gain. N S. For three-sector sites, SG becomes 2.4, which is comparable to 2.55, the sectorization gain measured in actual CDMA network deployments. To show the effect of the overlap area on the sectorization efficiency, consider Figure 4.3, where a three-sector site is shown with the areas of overlap or softer handoff determined by the angle θ SH. Recall that in CDMA systems, there are mainly two types of handoffs, namely hard handoffs and soft handoffs, as discussed in Chapter 2. Soft handoff can be further divided into handoff between two or more sectors of different cells and handoff between two sectors of the same cell, which is referred to as softer handoff. As a result of the three overlapping sectors, instead of reducing the interference by three times, as in the ideal case, only (120 + θ SH )/360 of the interference is blocked. Hence, we can define the sectorization efficiency by E sect 120 = θ SH (4.8) Figure 4.4 shows the relation between the sectorization efficiency and θ SH. Sectorization and soft/softer handoffs also affect the capacity of the forward link of a CDMA system. On the forward link, a simplified form for capacity is given by N FL 1 Poverhead = P H ν traffic (4.9)

106 Fixed Beam Smart Antenna Systems 87 θ SH Softer handoff Region Figure 4.3 Softer handoff overlap areas Efficiency Overlap angle Figure 4.4 Sectorization efficiency.

107 Tx Rx Rx 88 Smart Antenna Engineering where P overhead denotes the total power in the common channels, P traffic is the average traffic channel power that depends on the forward link SNR or E b /N t, and H denotes the handoff reduction factor. When mobile users are in soft or softer handoff, additional power is required on the forward link, hence the forward link capacity is reduced. However, the E b /N t required by those mobiles to achieve a given FER will be lower than that required without soft/softer handoff, so sectorization will also provide some capacity gain on the forward link. [2 6] provide detailed analysis of the impact of sectorization and soft/softer handoff on CDMA systems. 4.3 Limitations of Conventional Sectorization The most common form of sectorization uses spatial diversity antennas for signal reception and a single antenna for transmission, as shown in Figure 4.5, where all antennas use vertical polarization. Another form of diversity, called polarization diversity, uses either 0 /90 antennas, otherwise known as vertical/horizontal polarization or ±45, also known as cross polarization. In either case, two replicas of the received signal are available to the receiver for diversity combining to combat fading. One drawback of conventional sectorization is that the signals cannot be separated in the spatial domain, which makes spatial interference cancellation or reduction impossible to carry out. Another fundamental problem at the heart of network optimization is that of traffic loading imbalance when cellular traffic is distributed unevenly among different geographical areas of the network or among the sectors of a site leading to increased blocking. This imbalance is often time dependent, for instance during rush hour traffic on highways, business districts, or sport avenues. Rx Tx Rx Duplexers and LNAs Figure 4.5 Conventional sectorization scheme.

108 Fixed Beam Smart Antenna Systems 89 Alleviating such imbalance would require sectors with flexible orientations or beamwidths, which are not available with conventional sectorization. As a result, unused capacity on other sectors/sites is locked and wasted. As discussed earlier, on the forward link handoff zones have an immediate impact on capacity. Reducing the size of handoff zones and shifting those zones from high- to low-traffic areas can minimize the negative impact of soft/softer handoff on capacity. However, with conventional sectorization both the size and orientation of these handoff zones are fixed. One way to deal with these issues and provide means for spatial signal separation for further processing is to replace the conventional base station antennas with antenna arrays. The additional degrees of freedom provided by antenna arrays can offer more effective techniques to deal with multipath and interference and improve signal quality, leading to improved coverage and/or capacity. The fundamental advantage of arrays is their ability to generate one or more main beams with tailored beamwidths, with radiation pattern nulls and increased gain. Two main approaches exist fixed multiple beam antennas and fully adaptive antennas. The remainder of this chapter is devoted to the first approach. 4.4 Antenna Arrays Fundamentals Assume that we have a linear array composed of M identical antenna elements arranged along some axis, with interelement spacing d, as shown in Figure 4.6. In the simplest form, the array elements are fed with equal amplitudes A m and constant phase delay β. The radiation pattern of the array excluding the element pattern is referred to as the array factor. A general form for the array factor is given by j ( kd cos γ+ β) j 2( kd cos γ+ β) j ( M 1) ( kd cos γ+ β) ( 1 ) AF = A + e + e + L + e (4.10) where k = 2π λ. Without loss of generality, we can assume the signal amplitudes as follows: A = 1 m m = 12,,K M (4.11) Hence, the array factor can be rewritten as M ( ) ψ m = 1 AF = e j m (4.12)

109 90 Smart Antenna Engineering M z d 3 2 a r 1 θ y φ x Figure 4.6 Coordinate system for linear antenna arrays. where ψ= kd cos γ+ β (4.13) The total radiation pattern of an antenna array in the far field E(θ, ϕ) is represented by a product between two factors, the array factor AF (θ, ϕ) and the element factor EF (θ, ϕ). The element factor depends on physical dimensions and electromagnetic characteristics of the radiating element, whereas the array factor depends on the amplitude, phase, and position of each of the elements in the array antenna. In (4.13) γ is the angle between the array axis and the vector from the origin to the observation point. For an array along the z-axis, we have: ( ) cos γ = a$ $ = $ $ sin θcos φ+ $ sin θsin φ+ $ z ar az ax a y az cos θ = cos θ (4.14) It follows that ψ= kd cos θ+ β (4.15) When the array is placed along the x-axis, we get:

110 Fixed Beam Smart Antenna Systems 91 ( ) cos γ = a$ a$ = a$ a$ sin θcos φ+ a$ sin θsin φ+ a$ cos θ x r x x y z = sin θcos φ (4.16) and ψ= kd sin θcos φ+ β (4.17) Finally, for an array along the y-axis, we get: ( ) cos γ = a$ a$ = a$ a$ sin θcos φ+ a$ sin θsin φ+ a$ cos θ y r x x y z = sin θsin φ (4.18) and ψ= kd sin θsin φ+ β (4.19) When the reference point or origin is chosen at the physical center of the array, it can be shown [7, 8] that the normalized array factor of a uniformly excited, equally spaced linear array is reduced to AF sin = M sin 1 2 M ( ψ) ( 2 ) ψ (4.20) Broadside and End-Fire Arrays Note that the AF in (4.20) has a maximum at ψ = 0. To find the conditions under which the maximum radiation occurs, let us assume we have a linear array placed along the z-axis; it follows that ψ= kd cos θ + β= 0 (4.21) therefore β = kdcosθ o, where θ o is the direction of the maximum radiation. When θ o =90, the array is called broadside, and it follows that β = 0. The maximum of the radiation pattern of broadside arrays is always directed normal to the array axis. From the above we can see that a broadside array requires equal magnitude and phase excitation. For θ o =0 or θ o = 180 the resulting array is called an end-fire array. The maximum of the radiation pattern in this case is directed along the array axis. For θ o =0 we get ψ = kdcosθ + β = kdcos(0) + β = kd + β =0. Hence, the progressive phase shift required for an end-fire array with maximum radiation directed at 0 is o

111 92 Smart Antenna Engineering β = kd (4.22) As can be seen, the array factor given by (4.20) is a function of the number of elements M, element spacing d, and the phase shift β. Therefore, it is important to investigate the impact of these parameters on the radiation pattern of an antenna array Impact of Number of Elements Figures 4.7 and 4.8 show the effect of increasing the number of elements M on the radiation pattern of a broadside array along the z-axis. θ Figure 4.7 Radiation patterns of broadside array along the z-axis, d = λ/2. θ Figure 4.8 Radiation patterns of broadside array along the z-axis, d = λ/2.

112 Fixed Beam Smart Antenna Systems 93 We can observe that increasing M has the following effects on the radiation pattern: The width of the main lobe decreases; in other words, it becomes narrower. This is crucial for the applications of smart antennas when a single narrow beam is required to track a mobile or cluster of mobiles. The number of sidelobes increases. In addition, the level of the first and subsequent sidelobes decreases compared with the main lobe. Sidelobes represent power radiated or received in potentially unwanted directions. So in a wireless communications system, sidelobes will contribute to the level of interference spread in the cell or sector by a transmitter as well as the level of interference seen by a receiver when antenna arrays are used. The number of nulls in the pattern increases. In interference cancellation applications, the directions of these nulls as well as the null depths have to be optimized Impact of Element Spacing The element spacing d also has a significant impact on the shape of the radiation pattern. It is evident that the more elements an array has or alternatively the larger the array gets, the better the characteristics of the radiation pattern as far as its shape and degrees of freedom. Another way of achieving a larger array would be by increasing d. The major drawback of this approach lies in the behavior of the array factor function in (4.20), namely the appearance of replicas of the main lobe in undesired directions, referred to as grating lobes. Figure 4.9 shows the polar radiation pattern of a broadside six-element array along the z-axis with element spacing of d = λ/2. We can see that for this element separation, aside from a few sidelobes, we only have a main lobe directed toward 90. When we increase the spacing to d = λ, we get the radiation pattern shown in Figure Notice the appearance of a grating lobe at 0. Not only have we wasted power in the grating lobe, we also spread or receive more interference from the broader lobe. In practice, the optimum element spacing for beamforming and adaptive interference cancellation applications is d = λ/2. However, in specific applications such as transmit diversity, we intentionally design an array with much larger spacing to combat fading effects, as will be described in detail in a later chapter. A typical transmit or receive diversity antenna array has two elements separated by up to 10λ. The radiation patterns of a two-element array with element spacing of λ/2, 5λ, and 10λ are shown in Figures 4.11, 4.12, and 4.13, respectively. The

113 94 Smart Antenna Engineering Figure 4.9 Polar pattern, broadside array, M = 6, d = λ/ Figure 4.10 Polar pattern, broadside array M = 6, d = λ.

114 Fixed Beam Smart Antenna Systems Figure 4.11 Polar pattern, broadside array, M = 2, d = λ/ Figure 4.12 Polar pattern, broadside array, M = 2, d = 5λ.

115 96 Smart Antenna Engineering Figure 4.13 Polar pattern, broadside array, M = 2, d = 10λ. sidelobes and grating lobes generated in Figures 4.11 and 4.12 makes this design unsuitable for applications seeking to improve system performance when the degradation is mainly caused by interference or jamming First Null Beamwidth The null-to-null beamwidth (NNBW) of the array has a significant impact on the performance of a smart antenna system and is considered one of the important parameters that need to be considered in the antenna design. For a broadside array on the z-axis, the null-to-null beamwidth is given by [7] θ N π 1 λ = 2 cos 2 Md (4.23) The behavior of the NNBW is shown in Figures 4.14 and 4.15 as a function of d and M, respectively. Note that the larger the array, the smaller the NNBW becomes and the narrower the main lobe gets.

116 Fixed Beam Smart Antenna Systems 97 θ λ Figure 4.14 NNBW as a function of element spacing d. θλ λ λ λ Figure 4.15 NNBW as a function of number of elements M Half-Power Beamwidth Another very important beamwidth measure to consider is the half-power or 3-dB beamwidth. The 3-dB beamwidth of a broadside array on the z-axis is given by [7] θ H π λ = 2 cos 2 πmd for πd λ<< 1 (4.24)

117 98 Smart Antenna Engineering A more general formula for the 3-dB beamwidth of a linear-phased array antenna is θ λ λ = θ θ Md cos 1 cos cos 1 cos Md H o o (4.25) This is valid for a range of scanning angles but not for end-fire arrays. In Figures 4.16 and 4.17, we notice the same behavior for the 3-dB beamwidth when we increase M or d. Figure 4.18 demonstrates that the 3-dB beamwidth of a linear-phased array of a given size is not constant but rather it depends on the scanning angle. θ λ Figure dB bandwidth as a function of element spacing d. θλ λ λ λ Figure dB bandwidth as a function of number of element spacing M.

118 Fixed Beam Smart Antenna Systems 99 θ θ Figure 4.18 Effect of the scanning angle on the 3-dB beamwidth Array Directivity The radiation intensity of an antenna array can be defined as: U ( θ ) = [ AF] 2 (4.26) Antenna arrays have the ability to direct or concentrate the radiated power in a particular angular direction in space. This ability is measured by what is called the directive gain, defined as [9]: power radiated per unit solid angle in the directio D( θϕ, ) = 4π n( θϕ, ) Total power radiated by the antenna (4.27) The directive gain in the direction of the maximum radiation density is referred to as the directivity and is given by D o U = 4π max P rad (4.28) For a broadside array and small element spacing (d < λ), the directivity can be approximated by [7] D M d o = 2 λ (4.29)

119 100 Smart Antenna Engineering Array Gain The gain of an antenna array is the ratio of the radiation density in a particular angular direction in space to the total input power to the array or power radiated per unit solid angle in the directio G( θϕ, ) = 4 π n( θϕ, ) Total input power to the antenna (4.30) Note that we can define the antenna array efficiency as η = P rad P in (4.31) It follows that G = Dη (4.32) Trade-Off Analysis We have seen from previous discussions the effect of different parameters on the characteristics of the array and potential system performance impacts. Table 4.1 presents a trade-off analysis summarizing the impact of increasing each of these parameters. Table 4.1 Impact of Array Parameters on System Performance Parameter Pros Cons Smart Antenna Performance Impact Number of elements M Lower sidelobe levels More and deeper nulls Narrower beams Higher gain More sidelobes Larger arrays may be more costly Physical limitations on installation Better interference cancellation capabilities Improved performance because of higher gain and narrower beams Element spacing d Narrower beams Higher gain Grating lobes Grating lobes have negative impact on interference nulling Scanning angle θ o Smaller 3-dB beamwidths. Improved performance because of narrower beams

120 Fixed Beam Smart Antenna Systems Impact of Element Pattern As we indicated earlier, the total radiation pattern of an antenna array in the far field E(θ, ϕ) is represented by a product between two factors, the array factor AF(θ, ϕ) and the element factor EF(θ, ϕ). Consider two of the most widely used antennas, namely, the short dipole and a half-wave patch antenna, with element patterns given by cos( π 2cos( θ) ) E ( θϕ, ) = half wave dipole 2 sin ( θ) πl E ( θϕ, ) = cos sin( θ) λ E plane, half wave patch where L 049. λ ε r is the resonant length of a half-wave patch [10]. The complete radiation pattern of the array is obtained by pattern multiplication, as shown in Figures 4.19 through 4.22, where we have considered two- and four-element arrays of half-wavelength dipoles as well as half-wave patches. It is obvious how the element pattern can affect the total array pattern by changing the shape of the pattern and the direction of the main beam, as in Figure 4.19, by changing the beamwidth, as in Figure 4.21, and by changing the maximum gain, as in Figure Planar Arrays In planar arrays, elements are placed in a planar or rectangular grid. Let us consider a general planar array antenna with elements located at arbitrary positions (y n,z n ) in the yz-plane, as shown in Figure Array factor Element pattern Total pattern Figure 4.19 Total pattern, two-element array of half-wave dipoles, β = 180, d = λ/2.

121 102 Smart Antenna Engineering Array factor Element pattern Total pattern Figure 4.20 Total pattern, two-element array of half-wave dipoles, β =0,d = λ/ Array factor Element pattern Total pattern Figure 4.21 Total pattern, four-element array of half-wave patch, β = 180, d = λ/ Array factor Element pattern Total pattern Figure 4.22 Total pattern, four-element array of half-wave patch, β =0,d = λ/2. The array factor for this array antenna is given by AF = ( θφ, ) = M m = 1 a e n ( n sin θ sin φ+ β y + n cos θ+ βz) jk y z (4.33) where a n is the complex excitation of element n. Assuming we have M elements in the y-direction and N elements in the z-direction, we can rewrite (4.33 ) as

122 ... Fixed Beam Smart Antenna Systems 103 N z d y 3 d z 2 θ M y φ x Figure 4.23 Planar array geometry. M N jk[ ( m 1)( d y sin θ sin φ+ β y) + ( n 1) ( dz cos θ+ βz )] AF ( θφ, ) = a mn e (4.34) m = 1 n = 1 The above array factor can then be separated into the product of two terms as follows: N M jk n d jk m d z cos θ+ β z y sin θ sin φ+ β y AF ( θφ, ) = e 1 a mn e 1 (4.35) n = 1 ( ) ( ) ( )( ) where d y and d z are the element separations in the y and z directions, respectively, and a mn is the complex excitation of the element at position (md y ;nd z ). For simplicity, let us assume the excitation distribution over the array to be uniform and equal to a o, we can then rewrite (4.35) as a product of two sums as N m = 1 jk n d jk m d z cos θ+ β z y sin θ sin φ+ β y AF ( θφ, ) = a o e 1 e 1 (4.36) n = 1 M ( ) ( ) ( )( ) m = 1 Comparing (4.36) and (4.12) and using (4.20), it follows that

123 104 Smart Antenna Engineering AF ( θφ, ) = Nψ sin 2 ψ z sin N M z Nψ y sin 2 ψ y sin 2 (4.37) To get an expression for the radiation pattern of the planar linear array, we multiply the array factor by the element factor: ( θφ, ) ( θφ, ) ( θφ, ) E = AF EF (4.38) An example of the 3D array factor of a planar array of 16 elements is plotted in Figure Directivity of Planar Arrays Using the expression in (4.27) we can write the directivity as D( θφ, ) = π π π 0 4πE ( θ, φ) 2 E( θ, φ) sinθd θd φ 2 (4.39) Figure x 4 planar array pattern, d z= d y = λ/2