SMART BASE STATION ANTENNA Submitted in partial fulfillment for the degree of Master of Engineering in Electronics & Telecommunications S A S P u n c h i h e w a 072315 Umvcrsiiy ofMoraluwn June 2000 6 : ' t- »• 7 2 3 1 5 The work presented in this dissertation has not been submitted for the fulfillment of any other degree S A S Punchihewa Candidate Prof. (Mrs.) I J Dayawansa Supervisor DEDICATION To my Alma Maters St. Sebastian's College, Moratuwa and University of Moratuwa (Formally University of Sri Lanka, Katubedde Campus) for fine education I received... To my mother and late father for enabling that education against all the odds... To my wife Shanthi, and children for understanding, tolerance and encouragement... ...this really is your achievement. A C K N O W L E D G E M E N T S I hereby express my gratitude to Dr. (Mrs.) Dileeka Dias for allowing me to carry out research on 'Smart Base Station Antenna', thus enabling me to submit this report. I am indebted to all the academic and non-academic staff of the Department of Electronic & Telecommunication Engineering who in different and distinct manner helped me to complete this research. My special thanks to Mr. E. W. Perera, General Manager (Engineering), Mobitel (Pvt.) Ltd. for providing me with measured data on Sri Lankan mobile channels. My special gratitude to Prof. I. J. Dayawansa, who contributed immensely towards successful completion of this research by guiding, helping and encouraging me during the period stipulated for this activity. S. A. S. Punchihewa y CONTENTS Abstract i List of Figures ii List of Tables iii List of Abbreviations iv 1 Introduction 1 1.0 Spectrum and Smart Antennas t 1.1 Smart Antennas - An Overview 1.1.1 Switched beam antenna systems 1.1.2 Adaptive array systems 1.1.3 Comparison of two smart antenna systems 1.1.4 Signal processing in adaptive array systems 1.1.5 Functionality of an adaptive antenna array 1.2 Present Work 2 Adaptive Algorithms 8 2.1 Signal processing methods 2.2 Selection of an algorithm 2.3 Spatial Structure Methods 2.3.1 Subspace-Based Methods 2.3.1.1 Multiple Signal Classification (MUSIC) algorithm 3 Propagation Media 17 V 3.0 Overview of Mobile Propagation Media 3.1 Rayleigh Channel 3.2 Relevance of Rayleigh Channel 3.3 Propagation media with known delay characteristics 3.3.1 Overview 3.3.2 Rural Area 3.3.3 Typical case for non-hilly urban area 3.3.4 Worst case for Hilly Urban Area 3.3.5 Hilly Terrain 3.4 Application of knowledge of Propagation Media 4 Implementation of the Algorithm 22 4.0 Algorithm 4.1 Channel Model 4.2 Software Language 4.3 MATLAB implementation of the artificial channel 4.3.1 Model 1 4.4 Method to determine DOA 4.5 Testing of the method 4.5.1 Several attempts for the same model and array 4.5.2 For 10 attempts with several desired angles 4.6 Implementation for five-signal scenario 4.7 Implementation for rural area with pre-determined delay characteristics 4.8 Implementation for typical case for Non-hilly Urban Area 4.9 Implementation for Worst case for Hilly Urban terrain 4.10 Implementation for Hilly terrain 4.11 Modeling the Fort Area 5 Analysis of Observations 42 5.0 Non zeroing f(9) function 5.1 Accuracy of the method 5.2 Mean error and standard deviation of error 5.3 The channel 5.4 Problems with five-angle case 5.5 Rural area case 5.6 Success with other COST207 models 5.7 Colombo Fort Modeling 6 Results and Conclusions 48 6.0 Selection of algorithm 6.1 Method for DOA prediction 6.2 Testing the method 6.3 Validity of COST207 models 6.4 Future Work References 50 Appendix A v Appendix B xxviii Page i Telecommunications incur a strong impact on the society. Out of its many sectors, mobile communications experienced an unprecedented growth around the globe in recent times. Service providers will have to satisfy this increased customer need using a spectrum, which does not grow proportionately. Several multiple access systems such as frequency division multiple access, time division multiple access and code division multiple access are used at present to increase the efficiency of spectrum utilization. The smart antenna, consisting of an array of elements, monitors its signal environment and forms a beam towards the wanted signal. Thus, on top of the existing access methods it provides an additional multiple access method namely space division multiple access in which several users access portions of space simultaneously. There exist different methods or algorithms for formation of the beam towards the desired signal. Some of them form a beam and rotate while monitoring the satisfaction of certain conditions, which indicate the correct formation of the beam. Some others find the directions of arrival of signals (DOA) and then form the beam towards the desired direction of which the resolution is higher. In spite of high-resolution capability, these algorithms demand knowledge of the propagation characteristics of the mobile channel. This necessitates modeling of the channel after theoretical or empirical considerations. This dissertation presents the work carried out to determine the DOA of a desired signal which is to be used in an adaptive antenna in a variety of propagation channels. The suitability of MUSIC (Multiple Signal Classification) algorithm was investigated. It was necessary to find the ability of the algorithm to estimate the DOA of impinging signals. However, the channel modeling was also a necessity. To determine the accuracy of the estimation, the error between the actual and estimated DOA was determined and analyzed. MATLAB was uaed for simulations because of its capabilities to handle large amount of matrix related computational activities efficiently. An artificial channel with free space conditions was initially used to test the method of estimating the DOA, and to check the suitability of error analysis as a method of determining the accuracy. In this artificial channel, estimation of several DOA was performed for different conditions of environment monitoring and different antenna array geometry. Different number of signals was used with different angles of arrival. Hence, the dependence of errors on the above different conditions was determined and thereby the suitability of the error analysis to determine the accuracy was examined. The algorithm was tested for different channel models. The COST207 models developed by European Union were used and the performance of the MUSIC algorithm in different channel conditions was analyzed. Using the measured signal value data in the Colombo Fort area, the channel was mathematically modeled and MUSIC algorithm was tested for Colombo Fort. MUSIC algorithm was found to be suitable for use in adaptive cellular base station antenna. ABSTRACT Page ii LIST OF FIGURES 1.1 Switched Beam System Coverage Patterns 3 1.2 Adaptive Antenna Coverage : Main lobe extending towards 4 a user with a null directed toward a co-channel interferer 1.3 Uplink and Downlink in adaptive array processing 5 1.4 Functional diagram of N element 'smart' antenna 6 2.1 A signal arriving at an antenna array 10 4.1 Flow chart for creation of a snap shot 23 4.2 Flow chart for determination of DOA 24 4.3 Plot of f(6) Vs angle for artificial channel 27 4.4 Exploded view of fig 4.3 near 30° 28 4.5 Exploded view of fig 4.3 near 120° 29 4.6 Plot for increased number of elements for artificial channel 30 4.7 Plot for increased number of snap shots for artificial channel 31 4.8 Plot for artificial channel several attempts 32 4.9 Plot for five-signal case in artificial channel 34 4.10 Plot for five signal case (Modified algorithm) 35 4.11 Plot for rural area model 36 4.12 Plot for non-hilly urban area 37 4.13 implementation for Worst Case for Hilly Urban Area 38 4.14 Plot for Hilly Terrain 39 4.15 Plot for Colombo Fort Modeling 41 5.1 The Results for Artificial Channel with increased number of 43 elements and 10 attempts Page iii LIST OF TABLES Mean error and standard deviation of error for several theta(l) 33 V > Page iv LIST OF ABBREVIATIONS CDMA Code division multiple access DOA Direction of arrival FDMA Frequency division multiple access MUSIC Multiple signal classification SDMA Space division multiple access TDMA Time division multiple access SNR Signal to noise ratio >