Performance Analysis of Network-MIMO Systems : Luận văn ThS. Kỹ thuật điện tử - viễn thông: 60 52 70

VIET NAM NATIONAL UNIVERSITY, HA NOI UNIVERSITY OF ENGINEERING AND TECHNOLOGY PERFORMANCE ANALYSIS OF NETWORK-MIMO SYSTEMS A THESIS SUBMITTED IN FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF EECTRICAL ENGINEERING DUC-TUYEN TA 2010 Supervisor: Dr Trinh Anh Vu i ACKNOWLEDGMENTS First and foremost, I would like to express my gratitude to Dr Trinh Anh Vu for being a great mentor and for numerous technical discussions and suggestions that have found their way into this thesis I also very thank to all my colleagues at University of Engineering and Technology, VNU who have contributed greatly to provide a supportive and collaborative research atmosphere Many thanks to Phd Tran Duc Tan and Dinh Van Phong, with whom I have had opportunities to collaborate on various subjects I would like to sincerely thank my parents for their support, encouragement, and love throughout my life This thesis is dedicated to them ii ABSTRACT Network MIMO is a means of coordinating and processing the information gathered from multiple- input multiple- output (MIMO) communication systems to increase spectral efficiency, robustness, and data rates These properties make it a topic of great interest in the near future as the number of wireless users continues to grow and their individual demands on bandwidth climb Systems employing network MIMO capitalize on the fact that inter-cell interference, a major problem for dense wireless systems, is a superposition of signals With careful coordination between receivers (and transmitters), these super-positions can be decoupled and the information they contain can be utilized The goal of this thesis is to investigate the ability of network MIMO techniques to increase data rates in multi-user indoor wireless networks of various sizes with various channel schemes The simulation results also show that Network MIMO systems can be increase data rates and good through put than non- networked MIMO systems iii AUTHOR’S DECLARATION I declare that the work in this thesis was carried out in accordance with the Regulations of the University of Engineering and Technology, VNU The work is original except where indicated by special reference in the text and no part of the thesis has been submitted for any other degree Any views expressed in the dissertation are those of the author and not necessarily represent those of the University of Engineering, VNU The thesis has not been presented to any other university for examination either in Viet Nam or overseas Duc-Tuyen Ta 15 October 2010 iv TABLE OF CONTENTS Page LIST OF TABLES vii LIST OF FIGURES viii ABBREVIATIONS xi CHAPTER 1: INTRODUCTION 1.1 Wireless Communication 1.2 MIMO Techniques 1.3 Network-MIMO systems 1.4 Thesis’s Structure CHAPTER 2: BASIC MIMO THEORY 2.1 Wireless Background 2.2 MIMO Communications 2.2.1 MIMO systems Model 2.2.2 Theoretical MIMO Capacity Gains 10 2.2.3 Types of MIMO 12 2.3 Multi-user Communications 12 2.3.1 Limitations of Single-User view 13 2.3.2 Multi-User MIMO (MU-MIMO) 14 2.4 Multi-cell Communications 18 2.4.1 Limitations of Single-Cell View 19 2.4.2 Multi-Cell MIMO 19 3.1 Background 21 3.1.1 Inter-cell Interference 21 3.2 Theory behind Network MIMO 27 3.3 Network-MIMO systems Model 28 v 3.3.1 Uplink 29 3.3.2 Downlink 30 CHAPTER 4: SIMULATION AND RESULTS 34 4.1 Simulation Model 34 4.2 Simulation Diagram 36 4.3 Simulation Results 39 CHAPTER 5: CONCLUSION 45 REFERENCES 46 vi LIST OF TABLES Page Table Power Delay Profile 35 Table Simulation parameters 39 vii LIST OF FIGURES Page Figure MIMO communication from SISO to IA-MIMO (Source: www.wikipedia.org) Figure MIMO channel with M transmit and N receive antennas The sketched path, from transmitter and receiver, represent the channel which h11 is the channel between transmit antenna and receive antenna The transmit and receive signal are often presented by “black boxes” Figure From single- to multiuser communications, where all the users in the coverage area are simultaneously considered in the optimization The base station may choose to transmit data to a single or multiple user terminals at once 14 Figure Illustration of MU-MIMO: Downlink and Uplink 15 Figure MU-MIMO systems: MIMO Broadcast (Source: www.wikipedia.org) 16 Figure MU-MIMO systems: MIMO MAC (Source: www.wikipedia.org) 17 Figure Frequency reused in cellular network with the reuse factor is and Cells of same color are used with same frequency 18 Figure From multi-user to multi cell communication, where all the cells and all the users in the network are simultaneously considered in optimization The solid line marks the useful signals, where the interfering is dashed 20 Figure Coordination or Cooperation between all base stations in the wireless communication network under fast backhaul The central unit played an central network controller for control the coodination/cooperation between all the BS 20 Figure 10 Illustration of typical interference between users and access points in a cell-based wireless system The left image shows interference in down link and the right image shows interference in uplink 22 viii Figure 11 Illustration of traditional interference control between users and access points in a cell-based wireless system The left image shows down link and the right image shows uplink 23 Figure 12 Illustration of MIMO interference control between users and access points in a cell-based wireless system The left image shows down link and the right image shows uplink 24 Figure 13 Example of a small wireless communication with terminals, AP and the Central Network Controller 25 Figure 14 Network MIMO solution where all the signals are useful, i.e., interference is removed 25 Figure 15 Conventional vs Network MIMO average SINR and data rate improvements 26 Figure 16 Wireless network with two transmit and two receive antennas communicating through independent channels 27 Figure 17 Network-MIMO uplink channel: from m-th cell to all of base station 29 Figure 18 Network-MIMO downlink channel: from all base station to k-th user in the m-th cell 31 Figure 19 Block Diagram showing key functions that are to be implemented in MATLAB simulation 37 Figure 20 Simulation environment with cell, each cell include access point and end-user with randomly place 40 Figure 21 OFDM Pilot symbol to estimate the channel state information at both transmitter (AP/user) and receiver (user/AP) side with users 41 Figure 22 Compare between real channel and the estimated channel by using pilot symbol 42 ix Figure 23 Channel estimation between 4-th AP and 1-st User (in the different cell) and the channel between 1-st AP and 1-st cell (in the same cell) 43 Figure 24 Comparison between performance of Network-MIMO and non Network-MIMO communication system with the ranger of Signal-to-Noise Ratio (SNR) is 10 to 20 dB 43 x exponent, 𝑆𝑆𝑚𝑚,𝑘𝑘 is the lognormal shadowing between user k and BS m, and 𝜇𝜇 is the channel gain at a reference distance 𝑑𝑑𝑟𝑟𝑟𝑟𝑟𝑟 In other words, if there is no fading 𝑟𝑟𝑚𝑚,𝑘𝑘 = 𝑆𝑆𝑚𝑚,𝑘𝑘 = and 𝑑𝑑𝑚𝑚,𝑘𝑘=𝑑𝑑𝑟𝑟𝑟𝑟𝑟𝑟 , then ℎ𝑚𝑚,𝑘𝑘 = 𝜇𝜇 We refer to the parameter 𝜇𝜇 as the reference signal-to-noise ratio (SNR) To decision the transmitted signal, zero-forcing (ZF) beamforming is applied jointly across the BSs [11] By applying ZF beamforming across M transmit antennas, up to M users can receive data over mutually orthogonal beams under the assumption of ideal channel knowledge at both the transmitter and receiver and sufficient spatial separation of the users With the channel knowledged at both transmitter and receiver is not ideal, each user will experience some residual interference �𝑚𝑚 denote the estimation MIMO channel in the m-th cell Under the Let 𝐻𝐻 zero-forcing criterion, the transmitted signal vector in Eqt 3-9 is given by −1 �𝑚𝑚 ��𝐻𝐻 �𝑚𝑚 �𝐻𝐻 𝐻𝐻 �𝑚𝑚 � ) 𝑈𝑈𝑚𝑚 𝑋𝑋𝑚𝑚 = 𝐻𝐻 [3-11] Where 𝑈𝑈𝑚𝑚 is the data symbol vector for the K users in the m-th cell � 𝑚𝑚 = 𝐻𝐻𝑚𝑚 ), the received signal by Therefore, if the channel estimation is ideal (𝐻𝐻 each user is simply its desired data symbol plus additive noise: 𝑌𝑌𝑚𝑚 = 𝑈𝑈𝑚𝑚 + 𝑛𝑛𝑚𝑚 [3-12] 33 CHAPTER 4: SIMULATION AND RESULTS This chapter presented about simulation model of a network MIMO systems with many cell and user in indoor environment The simulated system was modeled after an OFDM multi-point access MIMO network to determine the performance of Network-MIMO vs nonNetwork-MIMO systems For simplicity, we assume that the MIMO link is only created by the multiple access schemes between APs and end-users 4.1 Simulation Model For our simulation study, we propose a multiple access network based on OFDM wireless access network in indoor channel environment The simulation environment is a multiple square cell, which each cell containing an access point as its center and the size of each cell is six-by-six meter In each drop of users, we sequentially generate random user locations with a uniform distribution within its perimeter All users were allowed to transmit over the entire systems frequency range This was done to mimic the interference patterns in a network with frequency reuse of one For simplicity, we assume that each user and access point processed a single transmit and receive antenna The simulation environment is set with the path loss exponent of up to m and 3.5 dB beyond, and shadow fading standard deviation of dB up to m and dB beyond There is no correlation across AP’s or users The power-delay profile is detailed in Table while the Doppler spectrum is Clarke-Jakes with a maximum frequency of 10 Hz 34 Table Power Delay Profile Delay (ns) Power (dB) 20 -2.5 10 30 40 50 60 70 80 -5.4 -5.9 -9.2 -12.5 -15.6 -18.7 -21.8 To compare the performance of conventional model and network-MIMO, we consider only two extreme cases: - Case1: full coordination between all the AP’s, it make the collaboration between each AP to make the network-MIMO systems With full coordination between AP’s, each user is detected jointly by all the AP’s using a linear ZF receiver spanning all the AP antennas - Case 2: There is no coordination between the AP’s That is the case of conventional wireless access network systems When there is no coordination between AP’s, each user is detected with a linear ZF receiver at the AP to which it has minimum average path loss In both cases, the ZF beamforming weights are computed as if the channel estimates were perfect at both end-user and access point’s side Channel Estimation In this simulation, we were using sounding pilot symbols to estimate the channel state information of each user or AP The first symbol of every frame was used to perform channel estimation By interleaving pilot symbols across various subcarrier frequencies, each user’s pilot signals were kept isolated in frequency 35 Then the values of intermediate channel coefficients were estimated using a simple zero-order-hold approximation To enhance the channel estimation, we can use the couple orthogonal symbols at begin of frame as the sounding pilot Another method is using training sequence to estimate the channel state information at both AP and user It was using in the WiMax IEEE 802.16e standard Power control At the start of each frame, the power at which each user is to transmit must be determined by the coordination cluster of APs serving it (recall that, without network MIMO, each AP constitutes a cluster by itself, while with network MIMO all APs belong to a single cluster) The choices of powers for the users are based on a target packet error rate, which we will take to be 10% The algorithm below presented the method to control transmit power of each user/AP Step Step Step Step Step Initialize all transmit powers to their maximum values Given the current power levels, compute the estimated SINR for each user This involves finding the SINR in each channel estimation tile resulting from a linear MMSE receiver, and averaging these SINR values over all tiles using an equivalent mutual information approach For each user, find the highest data rate that can be supported at the computed SINR while assuring the targeted packet error rate of 10% Lower each user’s power to just meet the SINR requirement for the selected data rate at the targeted packet error rate (computed as if all other users are maintaining their current power levels) Iterate until no user’s data rate changes between successive iterations 4.2 Simulation Diagram 36 Figure 19 Block Diagram showing key functions that are to be implemented in MATLAB simulation 37 The simulation diagram is shown in Figure 19 There are some key functions of simulation: Randomly Generate User Positions - This simulation will first generate a certain number of users and base stations (or APs) in square area Each AP will have a square cell coverage area and one randomly placed user within it This one-to-one ratio between APs and users is intended to mimic the interference that would occur in a cellular communication system with a universal frequency reuse scheme Determine Channel Matrix – The channel condition is determined to the based for calculate the necessary user transmit power The channel is determined by input parameters such as the path loss exponent, the variance of log normal shadowing in dB, etc In simulation, the channel state information at transmitter is estimated by a feedback link from receiver side Calculate Necessary User Transmit Power – This function will calculate the user transmit power based on the channel matrix apply the algorithm that is presented in section 4.1 Generate Message to be Sent – creates random messages of length L for each of the M users and stores it in a matrix m with m is the generated message consisting of +/-1s with length l These messages are going to the modulation and transmit function block to send Modulate message and Transmit – modulate message based on OFDM transmission method and transmit them after that Channel Estimation- Estimate the channel state information at receiver (CSIR) by using sounding pilot method The channel information is then using for decode the received message 38 Decode received message and calculate BER – received message is decoded The result is used to calculate the Bit Error Rates (BER) of the simulation system 4.3 Simulation Results Using Monte-Carlo method to simulated the Network-MIMO and nonNetworkMIMO wireless access system The simulation scenario is detailed in Table below Table Simulation parameters Parameter Value Average noise amplitude Number of subcarriers 64 channels Subcarrier spacing in frequency 10kHz Amplitude gain on pilot symbols relative to other P=30 information carrying symbols Path-loss exponent p=2.5 Variance of lognormal shadow fading constant dB Configure of antenna system Tx=1; Rx=1 Length of random message from each user L=64*2*20 Square area in meters of a cell 36 m2 Number of wireless cell Number of user Number of APs Center frequency 7.5 MHz 39 The simulation environment is shown in Fig 20 with access point and End-user (1 user/ AP/ cell) The results here are for the reference SNR is 15 db The reference SNR represents the average SNR at which a user at the midpoint between two adjacent AP’s would be received at either of those AP’s in the absence of shadow fading and interference from other users It is a composite measure of the transmitter power available to the user, the carrier frequency and bandwidth of operation, propagation characteristics of the environment, all antenna gains, noise figures, etc Clearly, as the reference SNR is made higher, interference between users becomes more significant relative to receiver thermal noise, and therefore the mitigation of such interference through Network MIMO becomes more beneficial Figure 20 Simulation environment with cell, each cell include access point and end-user with randomly place 40 The most important when simulation Network-MIMO systems is perfect channel knowledge in both transmitter and receiver, which are Access Point (AP) and enduser in case of Network-MIMO communication system In both case of conventional wireless access network and Network-MIMO systems, the channel information at both transmitter and receiver are estimated by using sounding pilot method In simulation, we added the pilot symbol as the first symbol of each data transmission frame Fig 21 show the OFDM pilot symbol for user 1, 2, and From the Fig 21, we can seen that the amplitude gain on pilot symbols relative to other information carrying symbols is P=30 That makes the result of sounding pilot symbol after channel can be using for estimate the channel state information Figure 21 OFDM Pilot symbol to estimate the channel state information at both transmitter (AP/user) and receiver (user/AP) side with users 41 By using sounding pilot symbol, we can estimate the channel between APs and End-user Fig 22 presented an example of channel estimation by using pilot Figure 22 Compare between real channel and the estimated channel by using pilot symbol In this simulation, we categorized the channel sate information in two groups: The channel information between the AP m-th and the user m-th at the same cell and this of AP m-th and the user n-th (n≠m) of other cell Fig 24 is presented the channel between AP and end-user in communication cell and the channel between AP in 4-th cell and the user in 1-st cell That results shown that there is more difference in channel estimated in the same cell and the difference cell 42 Figure 23 Channel estimation between 4-th AP and 1-st User (in the different cell) and the channel between 1-st AP and 1-st cell (in the same cell) Figure 24 Comparison between performance of Network-MIMO and non Network-MIMO communication system with the ranger of Signal-to-Noise Ratio (SNR) is 10 to 20 dB 43 With the nearly perfect channel knowledge between transmitter and receiver, we can decision the transmission signal Using Monte-Carlo simulation method, we get the BER of Network MIMO and non Network-MIMO systems as in Figure 24 Fig 24 presented the comparison between conventional system and NetworkMIMO systems in uplink It so that the new system get lower Bit-Error ratio (BER) than the conventional in both SNR value Recall that these results depend on channel estimation method, time, and frequency variations in the channel, and the deviation from a Gaussian distribution of the residual interference affecting each user after the linear ZF beamforming 44 CHAPTER 5: CONCLUSION Network MIMO is a means of coordinating and processing the information gathered from multiple- input multiple- output (MIMO) communication systems to increase spectral efficiency, robustness, and data rates These properties make it a topic of great interest in the near future as the number of wireless users continues to grow and their individual demands on bandwidth climb Systems employing network MIMO capitalize on the fact that inter-cell interference, a major problem for dense wireless systems, is a superposition of signals With careful coordination between receivers (and transmitters), these super-positions can be decoupled and the information they contain can be utilized This thesis investigated the ability of network MIMO techniques to increase data rates in multi-user indoor wireless networks of various sizes with various channel schemes The simulation results also show that Network MIMO systems can be increase data rates and good through put than non- networked MIMO systems In this thesis, the system was simulated in indoor environment with the limitation of APs and end-users However, due to the small-scale simulation environment, we cannot evaluate the potential of Network-MIMO in data rates increase Therefore, the future works in large-scale system such as mobile network are necessary 45 REFERENCES [1] S Dekleva, J Shim, U Varshney, and G Knoerzer, “Evolution and emerging issues in mobile wireless networks,” Communications of the ACM, vol 50, no 5, pp 38–43, June 2007 [2] H R Anderson, Fixed Broadband Wireless System Design Chichester, West Sussex, England: John Wiley & Sons Ltd, 2003 [3] T K Sarkar, R J Mailloux, A A Oliner, M Salazar-Palma, and D L.Sengupta, History of Wireless Wiley Interscience, John Wiley & Sons, 2006 [4] K Kyung-Ho, “Key technologies for the next generation wireless communications,” Proc of the 4th Intern Conf Hardware/software codesign and system synthesis, 2006 CODES+ISSS ’06., pp 266–269, Oct 2006 [5] G J Foschini “Layered space-time architecture for wireless communication in a fading environment when using multiple antennas” Bell Labs Technical Journal, 1(2):41–59, 1996 [6] D Tse and P Viswanath, Fundamentals of Wireless Communication Cambridge University Press, 2005 [7] D Gesbert, M Kountouris, R W Heath, Jr., C.-B Chae, and T Salzer, ”Shifting the MIMO Paradigm: From Single User to Multiuser Communications”, IEEE Signal Processing Magazine, vol 24, no 5, pp 36-46, Oct., 2007 [8] S Venkatesan, A Lozano, R A Valenzuela, "Network MIMO: Overcoming Inter-cell Interference in Indoor Wireless Systems", Asilomar Conference on Signals, Systems and Computers, Pacific Grove, CA, USA, November 2007 46 [9] Sivarama Venkatesan, Howard Huang, Angel Lozano, and Reinaldo Valenzuela, ”A WiMAX-Based Implementation of Network MIMO for Indoor Wireless Systems”, EURASIP Journal on Advances in Signal Processing, 2009, July, Vol 2009, 1, [10] Lin Hui, Wang Wen Bo,”Networked MIMO with Clustered Linear Pre- coding”, Asia-Pacific Conference on Communications, 2009, volume, 6, [11] “Network-MIMO: Coherently-Coordinated Base Stations”, Bell Labs Research Project, 2008 [12] T Yoo and A Goldsmith, “On the optimality of multi-antenna broadcast scheduling using zero-forcing beamforming,” IEEE J Select Areas Commun., vol 24, no 3, Mar 2006 47