ANALYSIS OF OFDM MIMO

_______________

A Thesis Presented to the Faculty of San Diego State University _______________

In Partial Fulfillment of the Requirements for the Degree Master of Sciences in Electrical Engineering _______________

by Jing Xiao Summer 2010

iii

Copyright 2010 by Jing Xiao All Rights Reserved

iv

DEDICATION
Dedicated to my husband, my son and my parents.

ABSTRACT OF THE THESIS

Analysis of OFDM MIMO) by Jing Xiao Master of Science in Electrical Engineering San Diego State University, 2010 Orthogonal frequency division multiplexing (OFDM) is a block transmission technique. In the baseband, complex-valued data symbols modulate a large number of tightly grouped carrier waveforms. The transmitted OFDM signal multiplexes several lowrate data streams and each data stream is associated with a given subcarrier. OFDM is implemented in broadband wireless access systems as a way to overcome wireless channel impairments and to improve bandwidth efficiency. OFDM is used today in wireless local area networks (LANs) as specified by the IEEE 802.11a and the ETSI HiperLAN/2 standards. It is also used for wireless digital radio and television transmissions, particularly in Europe. Multiple-input multiple-output (MIMO) communication refers to wireless communication systems using an array of antennas (i.e. multiple antennas) at either the transmitter or the receiver. Multiplexing would cause interference, but MIMO systems use smart selection and/or combining techniques at the receiving end to transmit more information and to improve signal quality. MIMO systems provide a significant capacity gain over conventional single antenna systems, along with more reliable communication. In this project, we will investigate MIMO techniques for OFDM systems in the context of WiFi and WiMax. We will conduct an extensive simulation-based study of MIMO-OFDM systems. We will analyze the performance of various MIMO techniques under different fading channel conditions. It is known that the complexity of OFDM with MIMO is overwhelming due to the necessity for multiple Fast Fourier Transforms (FFTs) at the receiver. In order to reduce system complexity, we also propose a novel combining technique that will exploit MIMO gains with only a single FFT at the receiver.

vi

TABLE OF CONTENTS
PAGE ABSTRACT ...............................................................................................................................v LIST OF FIGURES ............................................................................................................... viii ACKNOWLEDGEMENTS .......................................................................................................x CHAPTER 1 INTRODUCTION .........................................................................................................1 1.1 Introduction of Wireless Communication ..........................................................1 1.2 Overview of OFDM and MIMO ........................................................................2 1.3 Motivation of Research and Contributions of Thesis ........................................4 1.4 Outline of Thesis ................................................................................................4 2 OVERVIEW OF THE WIRELESS COMMUNICATION CHANNEL .......................6 2.1 Multi-path Spread ..............................................................................................7 2.2 Flat and Frequency Selective Fading .................................................................7 2.3 Rayleigh Fading .................................................................................................8 2.4 Doppler Fading ..................................................................................................8 2.5 Slow and Fast Fading .......................................................................................11 3 INTRODUTION OF OFDM MIMO ...........................................................................13 3.1 Overview of OFDM Techniques .....................................................................13 3.2 Overview of MIMO Techniques ......................................................................16 3.3 Introduction of MIMO OFDM.........................................................................18 4 DIVERSITY COMBINING TECHNIQUES ..............................................................20 4.1 Selection Diversity ...........................................................................................20 4.2 Alamouti Scheme .............................................................................................22 4.2.1 Two-transmitter-One-Receiver Alamouti Scheme .................................22 4.2.2 Two-Transmitter-Two-Receiver Alamouti Scheme ...............................24 4.3 Beamforming Technique .................................................................................25 4.4 Maximal-Ratio Receiver Combining (MRRC) ................................................28 5 CHANNEL ESTIMATION .........................................................................................31 5.1 Single Channel Estimation ...............................................................................31

vii 5.2 Joint Channel Estimation .................................................................................33 6 SINGLE FFT (FAST FOURIER TRANSFORM) RECEIVER COMBINING SCHEME .....................................................................................................................40 6.1 Single FFT Receiver Combining Scheme without Channel Estimation ..........41 6.1.1 Single FFT Receiver Combining Scheme with Known Channel Information ......................................................................................................41 6.1.2 Single FFT Receiver Combining Scheme In Unknown Channel Condition..........................................................................................................46 6.2 Convolutional Coding and Viterbi Decoding ..................................................49 7 CONCLUSIONS AND FUTURE WORK ..................................................................57 REFERENCES ........................................................................................................................59

viii

LIST OF FIGURES
PAGE Figure 2.1. Multi-path spread.....................................................................................................7 Figure 2.2. SISO simulations under Rayleigh fading channel, L is the number of tap in Rayleigh fading channel. ...........................................................................................9 Figure 2.3. Illustration of the Doppler fading effect. ...............................................................10 Figure 2.4. SISO simulation under Rayleigh and Doppler fading channel, fd is Doppler shift factor. .....................................................................................................11 Figure 3.1. Basic Structure of OFDM transmitter system. ......................................................13 Figure 3.2. Basic structure of OFDM receiver system. ...........................................................14 Figure 3.3. OFDM signal with cyclic prefix. ...........................................................................14 Figure 3.4. Modulated BPSK OFDM signals with 64 subcarriers in frequency domain and in time domain with cyclic prefix. ........................................................................15 Figure 3.5. With channel distortion demodulated BPSK OFDM signals with 64 subcarriers in frequency domain with cyclic prefix. ....................................................16 Figure 3.6. Equalized BPSK OFDM signal with 64 subcarriers. ............................................17 Figure 3.7. MIMO structure. ....................................................................................................17 Figure 3.8. MIMO OFDM structure. .......................................................................................19 Figure 4.1. The basic structure of transmit selection diversity. ...............................................21 Figure 4.2. Result of selection diversity when the number of the transmit antennas are 1, 2 and 4. .....................................................................................................................22 Figure 4.3. Performance of Alamouti Code Two-Branch Transmit Diversity with One Receiver. ......................................................................................................................24 Figure 4.4. Performance of Alamouti scheme Two-Branch Transmit Diversity with Two Receivers, the diversity is equal to One-Branch Transmit Diversity with four Receivers. 4th order diversity is obtained here. ...................................................26 Figure 4.5. Structure of two transmit antenna and one receive antenna Beamforming scheme..........................................................................................................................26 Figure 4.6. Performance of Beamforming scheme between 1 transmitter and 1 receiver, 2 transmitters and 1 receiver, 4 transmitters and 1 receiver..........................28 Figure 4.7. Basic structure of Maximal-ratio Receiver Combining.........................................29 Figure 4.8. Performance of Max Ratio Combining when the number of the receiving antenna are 1,2 and 4. ..................................................................................................30

ix Figure 5.1. Basic structure of Single Channel Estimation. ......................................................32 Figure 5.2. Performance about 1 transmitter 1, 2 and 4 receiver selection diversity scheme and 1 transmitter 1, 2 and 4 receiver selection diversity scheme with preamble. ......................................................................................................................34 Figure 5.3. Performance of single transmitter and single receiver when the rate between energy and noise of the preamble is 0dB, 5dB and 10dB..............................35 Figure 5.4. Basic structure of Joint Channel Estimation. ........................................................36 Figure 5.5. Joint channel estimation in two transmit antennas one receive antenna compare to single channel estimation 1 transmit antenna and 1 receive antenna. ........................................................................................................................38 Figure 6.1. Two transmitters and one receiver scheme with one FFT. ....................................40 Figure 6.2. Performance about single FFT combining sum abs 1Transmit antenna 1,2,3,4 Receive antenna. ..............................................................................................43 Figure 6.3. Performance about single FFT combining sum square 1Transmit antenna 1,2,4 Receive antenna. .................................................................................................44 Figure 6.4. Performance comparison of single FFT combing sum abs, sum square and sum square with weight. ..............................................................................................45 Figure 6.5. Performance comparison between sum square single FFT combing with weight and MRRC. ......................................................................................................47 Figure 6.6. Performance comparison between 1 transmit antenna 1,2,4 receive antenna single FFT combining Sum square with weight and Selection diversity........................................................................................................................48 Figure 6.7. Performance Comparison of 1Transmit antenna 1,2,4 Receive antenna single FFT combining sum square with weight and preamble. ...................................50 Figure 6.8. Convolutional code: k bits input, n bits output and L memory depth. ..................50 Figure 6.9. Performance comparison between single FFT combining sum abs with coding and without coding. ..........................................................................................51 Figure 6.10. performance comparison between single FFT combining sum square with coding and without coding. ..................................................................................52 Figure 6.11. Performance comparison between single FFT combining sum square plus weight with coding and without coding. ..............................................................53 Figure 6.12. Performance comparison between single FFT combining sum square with weight, preamble plus coding and sum square with weight and preamble..........54 Figure 6.13. Performance comparison between single FFT combining sum square with weight, preamble plus coding and MRRC with preamble and coding. ...............55 Figure 6.14. Performance comparisons between sum square with weight, preamble plus coding and selection diversity with preamble and preamble. ..............................56

ACKNOWLEDGEMENTS
My deepest gratitude to Professor Santosh Nagaraj for always being there offering suggestions and guidance. The thesis would not have been completed without his help. My special thanks to Professor Sridhar Seshagiri and Professor Samuel K. Kee S. Moon for their great patience to go through my thesis. Finally, I want to give my heartfelt thanks to my husband for his support and encouragement, also I want to give my thanks to my parents and my beloved baby Leo.

CHAPTER 1 INTRODUCTION 1.1 INTRODUCTION OF WIRELESS COMMUNICATION

Wireless communications is a rapidly growing segment of the communications industry, with the potential to provide high-speed high-quality information exchange between portable devices located anywhere in the world. It has been the topic of study since 1960s, the tremendous development of wireless communication technology is due to a confluence of several factors. First, the demand of wireless connectivity is explosively increased. Second, the dramatic progress of VISL technology has enabled small-area and low-power implementation of sophisticated signal processing algorism and coding algorism. Third, second generation wireless communication standards, like CDMA, GSM, TDMA, make it possible to transmit voice and low volume digital data. Furthermore, third generation of wireless communications can offer users more advanced service that achieves greater capacity through improved spectral efficiency [1]. Potential applications enabled by this technology include multimedia Internet-enabled cell phones, smart homes and appliances, automated highway systems, video teleconferencing and distance learning, and autonomous sensor networks. However, there are two significant technical challenges in supporting these applications: first is the phenomenon of fading: the time variation of the channel due to small-scale effect of multi-path fading, as well as large-scale effect like pass loss by distance attenuation and shadowing by obstacles. Second, since wireless transmitter and receiver need communicate over air, there is significant interference between them. Overall the challenges are mostly because of limited availability of radio frequency spectrum and a complex time-varying wireless environment (fading and multipath). In nowadays, the key goal in wireless communication is to increase data rate and improve transmission reliability. In other words, because of the increasing demand for higher data rates, better quality of service, fewer dropped calls, higher network capacity and user

2 coverage calls for innovative techniques that improve spectral efficiency and link reliability, more technologies in wireless communication are introduced, like OFDM and MIMO.

1.2 OVERVIEW OF OFDM AND MIMO

Orthogonal frequency-division multiplexing (OFDM) is a method of digital modulation in which the data stream is split into N parallel streams of reduced data rate with each of them transmitted on separate subcarriers. In short, it is a kind of multicarrier digital communication method. OFDM has been around for about 40 years and it was first conceived in the 1960s and 1970s during research into minimizing interference among channels near each other in frequency [2].OFDM has shown up in such disparate places as asymmetric DSL (ADSL) broadband and digital audio and video broadcasts. OFDM is also successfully applied to a wide variety of wireless communication due to its high data rate transmission capability with high bandwidth efficiency and its robustness to multi-path delay [3]. OFDM has been proposed as a transmission method to support high-speed data transmission over wireless links in multipath environments. During the last forty years, OFDM has developed into a popular scheme for wideband digital communication, whether wireless or over wires, used in applications such as digital television and audio broadcasting, wireless networking and broadband internet access [4]. In wireless environments, transmitted signals follow several propagation paths. When refleted from surrounding objects these paths reach the receiver with different propagation delays that causes delay spread, inter-symbol interference (ISI), fading, and random phase distortion. For example, the delayed copies of the transmitted signal will interfere with subsequent signals, resulting in ISI. The transmitted symbol rate is therefore limited by the delay spread of the channel. OFDM has been used in wireless LAN standards such as American IEEE802.11a and the European equivalent HIPERLAN/2 and in multimedia wireless services such as Japanese Multimedia Mobile Access Communications [5]. 802.11 is a set of IEEE standards that govern wireless networking transmission methods. They are commonly used today in their 802.11a, 802.11b, and 802.11g versions to provide wireless system standards in the home, office and commercial establishments. The

3 802.11a standard uses the same core protocol as the original standard, operates at 5 GHz band, and uses a 52-subcarrier orthogonal frequency-division multiplexing (OFDM) with a maximum raw data rate of 54 Mbit/s, which yields realistic net achievable throughput in the mid-20 Mbit/s. Since the 2.4 GHz band is heavily used to the point of being crowded, using the 5 GHz band gives 802.11a a significant advantage [6].OFDM is also applied in 802.16 and WiMAX technologies. IEEE 802.16 standard is the IEEE Standard for Wireless Metropolitan Area Networks (MANs), also known as WiMAX and WirelessLAN [7]. Wimax is an acronym that stands for Worldwide Interoperability for Microwave Access. It is a telecommunications technology that provides wireless data in a variety of ways, from point-to-point links to full mobile cellular type access [8]. MIMO has been developed for many years for wireless systems. One of the earliest MIMO to wireless communications applications came in mid 1980 with the breakthrough developments by Jack Winters and Jack Saltz of Bell Laboratories [9]. They tried to send data from multiple users on the same frequency/time channel using multiple antennas both at the transmitter and receiver. Since then, several academics and engineers have made significant contributions in the field of MIMO. Now MIMO technology has aroused interest because of its possible applications in digital television, wireless local area networks, metropolitan area networks and mobile communication. Comparing to the Single-input-single-output (SISO) system MIMO provides enhanced system performance under the same transmission conditions. First, MIMO system greatly increases the channel capacity, which is in proportional to the total number of transmitter and receiver arrays. Second, MIMO system provides the advantage of spatial variety: each one transmitting signal is detected by the whole detector array, which not only improved system robustness and reliability, but also reduces the impact of ISI (inter symbol interference) and the channel fading since each signal determination is based on N detected results. In other words, spatial diversity offers N independent replicas of transmitted signal. Third, the Array gain is also increased, which means SNR gain achieved by focusing energy in desired direction is increased [10].On the other hand, MIMO also cost more energy including both the transmission energy and the circuit energy consumption. Energy-efficiency analysis of MIMO system is important topic in MIMO research.

1.3 MOTIVATION OF RESEARCH AND CONTRIBUTIONS OF THESIS

In order to manage the bandwidth and power more efficiently, multiple antennas may be used at either the transmitter or the receiver. Multiple antennas offer diversity to the communication and improve transmission reliability over a fading channel using. Some commonly used techniques with multiple antennas are: Maximal-ratio combining (MRC). In MRC, signals from each antenna for each subcarrier are individually weighted to provide maximum SNR at the subcarrier. Coherent channel state estimation (CSI) is necessary to get the weights for combining the received signals. Beamforming techniques. If the channel state information is available at the transmitter, by using Beamforming the capacity can be further increased and the decoding complexity can be reduced by the proper design of precoder and power/rate allocation. Beamforming has been proposed in the ongoing IEEE 802.11n standardization. All of the multiple antenna techniques mentioned above involve multiple FFT/IFFT implementations when applied to OFDM. When the number of transmit/receive antennas is large, we need more IFFT circuits to transfer the signals from frequency domain to time domain at the transmitter side and also more FFT structures to transfer signals from time domain to frequency domain at the receiver side, which substantially increases the system power budget and cost. In this thesis, we explore a new class of schemes for achieving multiple antenna gains in OFDM on fading channels wherein only one IFFT circuit is necessary at the transmitter and only one FFT circuit is necessary at the receiver. We study the performance of such 1-FFT schemes using analysis and simulations. We show that 1-FFT schemes, although inferior to MRC and Beamforming, offer significant benefits while keeping complexity very low.

1.4 OUTLINE OF THESIS

In the first chapter, we aim to introduce the basic structure of OFDM and MIMO, including transmitter and receiver structure of OFDM MIMO, concept and application of MIMO are also shown. The basic idea why and how MIMO and OFDM are combined together will be explained in the first chapter. In chapter 2 and chapter 3, channel impairment mechanisms such as multi-path effect and Doppler fading are explained, communication system schemes are introduced under

5 these impairment mechanisms and error probability of the different schemes are analyzed under the condition that CSI (channel state information) are fully or partially known over these fading channels. In chapter 4, if channel state information is unknown, there are two methods used in this thesis to get channel information. One is single channel estimation, another is joint channel estimation,, these two estimation methods could be applied to the schemes shown in chapter 2, then the error probability of analysis will be found and the results will be compared to previous ones. In chapter 5, new schemes will be introduced in this chapter. 1FFT schemes will be applied, and there are different ways to improve these schemes, like adding weight, convolutional coding and Viterbi decoding, then the comparison between the new scheme with MRRC and selection diversity are discussed. Finally, Chapter 6 summarizes the main conclusions and provides suggestions for further work in this area.

CHAPTER 2 OVERVIEW OF THE WIRELESS COMMUNICATION CHANNEL

This chapter summarizes some of the problems encountered during wireless transmission process. The first part of the chapter will provides an overview of channels encountered during mobile radio propagation and their effects on signal quality. The latter part of the chapter will focus on diversity schemes and beam forming schemes under the channel. In wireless communications, there are many distortion mechanisms in the communication channels. First, reflections from objects in the surroundings cause multipath effect; the signal received at the receive antenna are composed of many of its own replicas from different paths, which arrive at a receiving antenna at different time (called time dispersion), causing multipath fading and inter-symbol interference. Fading due to multi-path fading spread is called flat fading and frequency selective fading, which will be explained in detail in the second section of this chapter. Second, when transmitter and receiver move relative to each other with a constant velocity, time-varying Doppler effect occurs. Fading due to Doppler spread include fast fading and slow fading. Third, the received power attenuation depends strongly on path length the transmitted signal undergone. When there is large terrain between transmitter and receiver like hills and buildings, this attenuation is very likely to happen. Finally, there are system noise which comes from thermal noise in receiver equipment, atmospheric noise and various kinds of random noise like interference between wireless carriers, transmitters and systems. These noises are independent from the transmitting signals and the fading characteristic of the channel. Usually, it is described as additive white Gaussian noise (AWG). We will discuss the impairment mechanisms in detail in the following section.

2.1 MULTI-PATH SPREAD

In conventional wireless communications, one antenna is used at the source, and another antenna is used at the destination as the receiver. As we discussed, this structure sometimes gives rise to problems of multipath effects. When an electromagnetic field meets with obstructions such as hills, canyons, buildings, and utility wires, the wave fronts are scattered, and thus they take many paths to reach the receiver. Because those multiple waves arrive at random delays (phases), angles and amplitudes, problems such as fading, cutout, and intermittent reception occur [11]. In digital communications systems such as wireless Internet, it can cause reduction in data speed and increase in the number of errors. We illustrate the effects in Figure 2.1: the scattered wave from the transmitter to the receiver causing the multi-path effect, resulting in signal distortion and delay that can not be ignored.

Tx

Rx

multi-path fading

Figure 2.1. Multi-path spread.

2.2 FLAT AND FREQUENCY SELECTIVE FADING

There are in general two fading effects, namely, flat fading and frequency selective fading signals are transmitted to receivers, if all the spectral components of the transmitted signals are affected by the same amplitude gains and phase shifts, the channel is called flat fading channel. In this case, the transmitted signal bandwidth is much smaller than the channels coherent bandwidth. Flat fading channel is encountered in many wireless environments and causes deep fades. The amplitude distribution of flat fading is either Raleigh distribution or Ricean distribution. On the other hand, if the spectral components of the transmitted signals are affected by different amplitude gains and phase shifts, the fading is said to be frequency selective. In this case the transmitted bandwidth is larger than the channels coherence bandwidth.

8 Frequency selective fading will induce inter-symbol interference, which can be undone by digital processing. In other words, assuming a single transmitted impulse, whose time duration Tm is the duration between the first and last received component that possesses the maximum delay spread, therefore the coherence bandwidth f c is 1/ Tm. As we all know, symbol time is Ts. A channel is said to frequency selective fading if Tm>Ts and it is said to flat fading if Tm<Ts. [12].

2.3 RAYLEIGH FADING

The Rayleigh distribution is used to model multipath fading with non-line-of- sight (NLOS). If there is a line of sight (LOS), Rician fading is more applicable. In this case, the probability density function (PDF) of fading amplitude of i-th path i is given by

where i= E [i2] and E[]denotes the expectation of its argument [13]. Let us consider BPSK signals in a slow, Rayleigh fading channel with additive white Gaussian noise. We define the number of channel taps as L, meaning that there are L taps to represent the fading channel. After simulation with Matlab using simply one transmitter antenna and one receiver antenna, the results of which are shown in Figure 2.2, we see that when the tap is 1 in the fading channel, the bit error rate is the best; when the tap is 5, the bit error rate is the worst, which is simply because more taps mean more fading within the channel, and thus the worse the performance becomes.

2.4 DOPPLER FADING

Another major concern in wireless communications is the Doppler effect (shift). As we all know, this effect occurs due to the relative speed between the elements in the communication system. The Doppler effect is the change in frequency/wavelength of a wave as perceived by an observer moving relative to the source of the waves. The total Doppler effect may therefore result from either motion of the source or motion of the observer. The effect of the Doppler is directly proportional to the magnitude of the relative speed and is modeled here as a contribution to the carrier frequency.

Figure 2.2. SISO simulations under Rayleigh fading channel, L is the number of tap in Rayleigh fading channel. For waves which do not require a medium, such as light or gravity in special relativity, only the relative difference in velocity between the observer and the source needs to be considered [14]. For wireless communication, when electromagnetic wave is traveling towards or away from the receiver, the carrier frequency will be shifted, causing Doppler shift. It is noticed that Doppler shift is usually prominent when the transmit antenna is far from receive antenna. As we can see from Figure 2.3, the phase difference between two transmission paths is:

The Doppler shift is:

10

Figure 2.3. Illustration of the Doppler fading effect. Because the detected frequency increases as objects moving toward the observer, the source's velocity must be subtracted when motion is moving toward the observer. (This is because the source's velocity is in the denominator.) Conversely, detected frequency decreases when the source moves away, and so the source's velocity is added when the motion is away. If a pure sinusoid is transmitted, a range of frequencies adjacent to this sinusoid frequency will be received. Doppler shift broadens the spectrum of the received signal by spreading the basic spectrum in frequency domain. If the signal spectrum is wide enough compared to this spreading, the effect is not noticeable. Otherwise, distortion will occur. Let
2 us investigate the BPSK signals in slow, Rayleigh and Doppler fading channel with additive

white Gaussian noise with one transmitter antenna and one receive antenna. The performance when fd = 0, fd = 0.01, fd = 0.05 are shown in Figure 2.4 respectively. One can see that when fd is larger, the performance is getting worse because of the effect of Doppler fading. larger, the performance is getting worse because of the effect of Doppler fading. Both multi-path fading and Doppler effects can impair the reception of the transmitted signal. It is also well known that inter symbol interference happens when multiple paths are received with various delays and co-channel users create distortion to the target user. And thermal noise is electronic white noise that definitely need be counted in. Diversity, as stated by Proakis [15] is an effective way of improving error rate performance in fading channels. MIMO OFDM is introduced as a scheme in wireless communications to offer diversity, capacity and array gain [16] by using MIMO and also to

11

Figure 2.4. SISO simulation under Rayleigh and Doppler fading channel, fd is Doppler shift factor. avoid inter symbol interference by using OFDM.

2.5 SLOW AND FAST FADING

Fading due to Doppler spread includes both slow fading and fast fading. In wireless communication, a channel can be time varying and those dynamic channels are characterized as slow or fast fading channels. Fast fading channel changes significantly during the duration of a symbol. And when the channel varies rapidly, it distorts the symbols amplitude and phase erratically over its interval. On the other hand, slow fading occurs when the channel changes much slower than one symbol duration. This does not imply that the effects of the channel can be neglected, but it is possible to track the changes in the channel to appropriately compensate for channel dynamics. We define coherence time T c of the channel as the period of time over which the fading process is correlated. T c is closely related to Doppler spread fd as:

12 T c 1/ f d If the symbol time duration T s is smaller than T c, the fading is slow fading; otherwise, the channel fading is fast fading [17].

13

CHAPTER 3 INTRODUTION OF OFDM MIMO 3.1 OVERVIEW OF OFDM TECHNIQUES

In OFDM, a block of data is converted into a parallel form and mapped into each subcarrier in time domain. By transmitting the symbols in parallel, the interval between the signals becomes much larger and this effectively eliminates inter symbol interference in time dispersive channels. IFFT (Inverse fast Fourier transformation) is in turn used to transfer the signal from time domain to frequency domain. It takes in N symbols at one time where N is the number of subcarriers in the system. Each of these N input symbols has a period of T seconds. As we know, the basis functions for an IFFT are N orthogonal sinusoids. Each input symbol acts like a complex weight for the corresponding sinusoidal basis function. Since the input symbols are complex, the value of the symbol determines both the amplitude and phase of the sinusoid for that subcarrier. The IFFT output is the summation of all the N sinusoids. Thus, the IFFT block provides a simple way to modulate data onto N orthogonal subcarriers. The block of N output samples from the IFFT make up a single OFDM symbol [18]. The basic structure of OFDM transmitter is shown in Figure 3.1.

Figure 3.1. Basic Structure of OFDM transmitter system. After transmitted through the channel (in wireless communications, channel means the route through which the message is sent), when the frequency signals reach the receiver, the receiver has to perform Synchronization (both timing and frequency), Channel Estimation, Demodulation and Decoding. The data processing at receiver end

14 reverses that at the transmitter side, that is, at the receiver, an FFT block is firstly used to transfer the received time-domain signal into frequency-domain. Ideally, the output of the FFT block should be identical to the transmitted symbols before the IFFT block. Assuming channel information is known, the symbols will be demodulated and estimated based on these channel information. The basic structure of OFDM receiver is shown in Figure 3.2.

Figure 3.2. Basic structure of OFDM receiver system. When there are more than one transmission path between the transmitter and the receiver, or received signal is the sum of many versions of the transmitted signal with varying delay and attenuation, the multi-path effect occur, among which ISI effect is the most important. To reduce this effect, two methods are generally used in the OFDM scheme: parallel data transmission and cyclic prefix. Usually the length of the cyclic prefix is no shorter than the length of the channels impulse response. The basic idea is to replicate part of the OFDM time-domain waveform from the back to the front to create a guard period. The duration of the guard period Tg should be longer than the worst-case delay spread of the target multi-path environment. In Figure 3.3, the structure of cyclic prefix is illustrated [19].

Figure 3.3. OFDM signal with cyclic prefix.

15 We will better understand this concept through the following example, which shows OFDM signals in frequency domain (BPSK signals such as 1,-1). After cyclic prefix is applied, the number of subcarriers increase from 64 to 80, in which the last 16 subcarriers as we can see from Figure 3.4 is exactly the same with the first 16 subcarriers.

Figure 3.4. Modulated BPSK OFDM signals with 64 subcarriers in frequency domain and in time domain with cyclic prefix. For illustration purpose, let us consider BPSK signals in slow, frequency nonselective fading channels with additive white Gaussian noise. 64 subcarriers in frequency domain are set, illustrated in the top curve in Figure 3.4. The signals are then transferred to time domain and the cyclic prefix is added, the results of which are shown in the bottom curve in Figure 3.4. In Figure 3.5, we show the OFDM signals with cyclic prefix in frequency domain after channel distortion. , As we can see the last 16 subcarriers from Figure 3.5 is exactly the same with the first 16 subcarriers.

16

Figure 3.5. With channel distortion demodulated BPSK OFDM signals with 64 subcarriers in frequency domain with cyclic prefix. After we get the demodulated signals, we need equalize them. Figure 3.6 shows the equalized OFDM subcarriers in time domain without cyclic prefix.

3.2 OVERVIEW OF MIMO TECHNIQUES

Multiple Input/ Multiple Output (MIMO) is a technology that uses multiple antennas to transmit and receive signals. The basic MIMO structure is shown in Figure 3.7, where Tn stands for the nth transmitter antenna and Rn resembles the nth receiver antenna. In some special cases, for example, MISO is a system with multiple inputs and single output. On the other hand a system with single input and multiple outputs is called SIMO. In a MIMO system, data (x1, x2 xN) are transmitted with N transmitting antenna arrays. The receiver is constructed of M (M>=N) antenna arrays. Let rj(j= 1, 2,, M) represents the signal received by the jth antenna (see Figure 3.7), then the signals received at the receiver can be represented as: r1=h11x1+h12x2++h1NxN
2=h21x1+h22x2++h2NxN

rM=hM1x1+hM2x2++hMNxN

17

Figure 3.6. Equalized BPSK OFDM signal with 64 subcarriers.

Figure 3.7. MIMO structure. where hij is a weight coefficient that represents the impact of the jth transmitting signal xj on the jth receiver signal strength. We define a channel matrix H as:

Therefore, in MIMO system, the transmitted signals {xi } can be recovered by estimating the channel matrix H and the receiving signal vector R

18 MIMO system can provide two types of gains: diversity gain and spatial multiplexing gain. And in past work, it is known that there is a fundamental tradeoff between these two gains: higher spatial multiplexing gain comes at the price of sacrificing diversity [20]. Diversity is used in MIMO to combat channel fading. Since in MIMO each pair of transmitting and receiving antennas provides a signal path from the transmitter to the receiver and each path carry the same information simultaneously, the signal achieved in the receive antenna is more reliable and the fading can be effectively decreased. If the path gains between individual transmitreceive antenna pairs fade independently, in this case multiple parallel spatial channels are created. By transmitting independent information streams in parallel through the spatial channels, the data rate can be increased. This effect is also called spatial multiplexing [21]. So the benefit of diversity is lower error probability and the benefit of multiplexing is higher rate though the difference between them is that the requirement of diversity is sending the same information and the requirement of multiplexing is send independent information. Obviously, the conflicts between the two suggest fundamental tradeoff between benefits obtained from diversity and multiplexing.

3.3 INTRODUCTION OF MIMO OFDM

MIMO OFDM ( Multiple I n p u t M u l t i p l e Output, Orthogonal Frequency Division Multiplexing) is a technology that combines MIMO and OFDM together to transmit data in wireless communications in order to deal with frequency selective channel effect. The OFDM signal on each subcarrier can overcome narrowband fading, therefore, OFDM can transform frequency-selective fading channels into parallel flat ones. Then by combining MIMO and OFDM technology together, MIMO algorithms can be applied in broadband transmission [22]. A MIMO OFDM system transmits data modulated by OFDM from multiple antennas simultaneously. At the receiver, after OFDM demodulation, the signal are recovered by decoding each the sub-channels from all the transmit antennas [23]. MIMO OFDM will allow service providers to deploy a Broadband Wireless Access (BWA) system that has Non-Line-of-Sight (NLOS) functionality. Specifically, MIMO-OFDM takes

19 advantage of the multipath properties of environments using base station antennas that do not have LOS. By combining both techniques, MIMO-OFDM can offer both robustness and high throughput. In a multiuser scenario where many users communicate with a central station (base station or access point), MIMO-OFDM becomes even more appealing because it provides an additional opportunity to exploit due to many users. In Figure 3.8, the basic structure of MIMO OFDM is demonstrated. In this figure, the signals are modulated by OFDM modulator, then they are transmitted by MIMO system, finally, the signals are recovered by the OFDM demodulator.

Figure 3.8. MIMO OFDM structure. Therefore, MIMO OFDM achieves spectral efficiency, increased throughput and the inter-symbol interference (ISI) can thus be prevented.

20

CHAPTER 4 DIVERSITY COMBINING TECHNIQUES

In this chapter, the usage of diversity combining techniques to improve system performance will be presented. In nowadays, the most popular transmit diversity techniques in use are Selection Diversity, Space Time Coding (Alamouti scheme) and Beamforming techniques. On the other hand the main receive diversity techniques that will be discussed in this chapter is maximal-ratio combining in which the estimations of different channels are coherently combined to recover the transmit signals. In this chapter we will examine these schemes in greater detail and will compare the error probability through modulations in the case that the channel state information are known either by receiver or transmitter ( in Beamforming, channel state information need be known by the transmitter, in other cases, channel state information need be known by the receiver). We will focus on the performance of the MIMO system. In our simulation binary phase-shift keying (BPSK) modulation will be used, the impairments of the channel include Rayleigh fading and Doppler fading. In these modulations, each transmitter antenna carries 64 subcarrier, with cyclic prefix, it becomes 80, after convolving with channel and demodulation, we will use different combining scheme and compare them. In our simulation, perfect synchronization in time and frequency are available.

4.1 SELECTION DIVERSITY

In multiple input and single output systems, selection diversity selects the branch providing the largest magnitude of log-likelihood ratio (LLR). The LLR for BPSK signals in fading channels is found to be proportional to the product of the fading amplitude and the matched filter output after phase compensation. Channel state includes statistics information such as fading amplitudes, phases and delay. In selection diversity technique, none of these channel information is required. Selection diversity uses one receiver antenna which greatly reduces the complexity of the wireless systems. Here, the receiver simply looks at the outputs from each fading channel and

21 selects the one with the highest signal-to-noise ratio (SNR). In this case, the strongest signal is picked up and other signals that have undergone deep fades are unlikely to be picked by the receiver, which can avoid the deep fading effect. Therefore, the more the channels are, the more accurate the recovered would be. There is also no need for any addition of the fading channel outputs, which further decreases the complexity. Since Selection diversity does not require any knowledge of the phases, it is usually used in non coherent or differential coherent modulation schemes. The structure of selection diversity is shown in Figure 4.1: the signals from the transmitter antenna are sent through multiple channels, selected by the detector system and finally received by the receive antenna.

Figure 4.1. The basic structure of transmit selection diversity. Selection diversity can be viewed as the practical methods to reduce the implementation complexity of MIMO systems while still taking benefit of the use of multiple antennas. The performance of the BPSK signals in slow, Rayleigh channel with additive white Gaussian noise using one transmit antenna, two transmit antennas and four transmit antennas, respectively, and one receive antenna in all cases are calculated, the results of which are illustrated, in Figure 4.2. When there four transmit antennas are used, the performance is the best, the other extreme, that is, when one transmit antenna is used, the performance is the worst. This agrees with our above-mentioned principle that the more transmit antennas, the higher the received signal fidelity is.

22

Figure 4.2. Result of selection diversity when the number of the transmit antennas are 1, 2 and 4.

4.2 ALAMOUTI SCHEME

Alamouti scheme is an orthogonal transformation scheme that gives the second order transmission diversity, which is applied to the receiver that can greatly reduce the complexity of the receiver. In this section, we will firstly investigate the two-transmitter-one-receiver Alamouti scheme. We then develop the two-transmitters-two-receiver scheme based on the results of the two-transmitter-one-receiver scheme.

4.2.1 Two-transmitter-One-Receiver Alamouti Scheme

This scheme uses two transmitting antennas and one receiving antenna and can be defined by the following three functions: At the transmitter of the Alamouti scheme, the two signals x1, x2 are transmitted at

23 the same time from the two transmitting antennas In this case, x1, x2 are baseband complex symbols carrying the information. The time and transmit sequence are shown as the following: Time T T+T Antenna1 X1 -x2* Antenna2 x2

x1* Symbols x1 and x2 are sent with the first beam at time t, while -x2*and x1* (* stands

for complex conjugate) are transmitted through the second beam at time t+T, where T is the symbol duration [24]. The channel model is h1(t) =h1(t+T)= h1=a1ej1 h2(t) =h2(t+T)= h2=a2ej2 At the receiver of the Alamouti scheme, the signals are respectively received at times t and t+T, where r1= h1x1+h2x2+n1 r2= -h1x2*+h2x1*+n2 Here r1, r2 are the received signals and n1, n2 are the complex random variables representing the noise and interference. The received symbols are then combined and processed by the decoder. The receivers obtain the signals s1, s2 through the following matrix after computing the estimates of the complex channel gains, that is, s1=h1*r1+h2r2* s2=h2*r1-h1r2* Figure 4.3 denotes the performance of BPSK signals using Two-TransmitterOneReceiver Alamouti scheme in slow, Rayleigh fading channel with additive white Gaussian noise. As we can see, the result is equivalent with the 1 transmitter 2 receiver MRRC (Maximum ratio receiver combining) scheme.

24

Figure 4.3. Performance of Alamouti Code Two-Branch Transmit Diversity with One Receiver.

4.2.2 Two-Transmitter-Two-Receiver Alamouti Scheme

When there are two transmitters and two receivers, Alamouti scheme can still solve the problems. In this case, the transmission sequence will subject to minor change from the two transmitters and one receiver Alamouti scheme. The transmitting signals at the transmitting antennas are shown as the following:: Time T t+T Antenna1 X1 -x2* Antenna2 x2 x1*

Where symbols x1 and x2 are sent with the first beam at time t, while -x2*and x1* (* stands for complex conjugate) are transmitted through the second beam at time t+T, where T is the symbol duration.

25 The receiving signals at the receiving antennas are shown in the following format:: Time T t+T Antenna1 R1 R3 Antenna2 r2 r4

There exist four communication channels between the two transmitter antennas and two receiver antennas, which are defined as h1, h2, h3, h4, respectively. The relationship between the receiving and transmitting signals can be represented as: r1=h1x1+h2x2+n1 r2=-h1x2*+h2x1*+n2 r3=h3x1+h4x2+n3 r4=-h3x2*+h4x1*+n3 Where x1, x2 represent the transmitted signals, r1 , r2 , r3, r4 are the received signals and n1 n2 n3 n4 are complex random variables representing the noise and interference. The receivers then obtain the signals s1, s2 after computing the estimates of the complex channel gains, that is, s1=h1*r1+h2r2*+ h3*r3+h4r4 s2=h2*r1-h1r2*+ h4*r3-h3r4* In this case, the results of this scheme are equivalent to those obtained by the 4-branch (MRRC) scheme, which is going to be discussed in Section 3.4. Figure 4.4 denotes performance of BPSK signals using Two Transmitters with Two Receivers Alamouti scheme in slow, Rayleigh fading channel with additive white Gaussian noise. As we can see, the result is equivalent to the one obtained with the 1 transmitter 4 receiver MRRC scheme [25].

4.3 BEAMFORMING TECHNIQUE

In Beamforming transmission scheme, usually the channel state information need to be known at the transmitter, so we use the known channel information to code the transmitting signals.. Then at the receiver side, the equalization process is greatly simplified. The structure of Beamforming is clearly shown in Figure 4.5.

26

Figure 4.4. Performance of Alamouti scheme Two-Branch Transmit Diversity with Two Receivers, the diversity is equal to One-Branch Transmit Diversity with four Receivers. 4th order diversity is obtained here.

Figure 4.5. Structure of two transmit antenna and one receive antenna Beamforming scheme.

27 For example, in a one-transmittertwo-receiver Beamforming scheme, assuming that the transmitted signal as x, the first channel as h1 and the second channel as h2 where h1= a1ej1 h2= a2ej2 Therefore the signal transmitted by the first transmitting antenna is xh1* and the signal transmitted by the second transmitting antenna is, xh2.* After going through the channels, signals that are received by the receivers could be represented as: r1= xh1*h1+n1= a12x+n1 r2= xh2*h2+n2= a22x+n2 Where r1, r2 are the received signals and n1, n2 are complex random variables representing the noise and interference. The combining scheme for two-branch Beamforming is:

Then after equalizing the a12+ a22 coefficient, we can get the final signal after decision making. The benefits of MIMO Beamforming is that the power gain and array gain get increased and that the co-channel inter-cell interference are reduced; also the diversity gain get increased, because Beamforming can combat the fading effects of the channel. With Per- subcarrier Beamforming reducing spectral nulls, the multipath effect mitigates. In Figure 4.6 the performance of BPSK signals using One Transmit antenna and One Receive antenna Beamforming scheme, Two Transmit antenna and One Receive antenna Beamforming scheme, Four Transmit antenna and One Receive antenna Beamforming scheme, in slow, Rayleigh fading channel with additive white Gaussian noise are illustrated. As we can see, the performance of Beamforming scheme is not as good as MRRC scheme.

28

Figure 4.6. Performance of Beamforming scheme between 1 transmitter and 1 receiver, 2 transmitters and 1 receiver, 4 transmitters and 1 receiver.

4.4 MAXIMAL-RATIO RECEIVER COMBINING (MRRC)

Maximal-ratio combining is a diversity technique over fading channels. While it is slightly more complex than Selection Diversity, the performance is especially good in the case of independent fading channels. MRRC selects all the branches and gets the average square value of each branch to decrease Rayleigh fading. The basic structure of Maximal-ratio Receiver Combining scheme is illustrated in Figure 4.7. Let us first consider a one-transmitter-two-receiver MRRC model. At a given time, a signal x is sent from the transmitter. The channel between the transmit antenna and the

29

Figure 4.7. Basic structure of Maximal-ratio Receiver Combining. receive antenna one is denoted by h1 and between the transmit antenna and the receive antenna two is denoted by h2 where h1= a1ej1 h2= a2ej2 Considering the noise and interference added at the two receivers, the resulting received signals could be written as: r1=h1x+n1 r2=h2x+n2 Where r1, r2 are the received signals and n1, n2 are complex random variables representing the noise and interference. The receiver combining scheme for a two-branch MRRC is as following: S=h1*r1+ h2*r2 = h1*(h1x+n1) + h2*(h2x+n2) = (a12+ a22) x+ h1*n1 +h2*n2
2 2 Then after equalizing the a1 2 + a2 coefficient, we can get the final signal after decision making.

Figure 4.8 illustrates the performance of BPSK signals using One Transmit antenna with One Receive antenna MRRC scheme, One Transmit antenna with Two Receive antenna MRRC scheme, One Transmit antenna with Four Receive antenna MRRC scheme,

30

Figure 4.8. Performance of Max Ratio Combining when the number of the receiving antenna are 1,2 and 4. in slow, Rayleigh fading channel with additive white Gaussian noise.

31

CHAPTER 5 CHANNEL ESTIMATION

The discussions and investigations in Chapter 4 are based on the fact that the channel state information is known either by transmitter or receiver, which is not always true in reality. In this chapter, we will discuss the channel estimation technique when the channel information is not available. We then study the performance of the communication schemes discussed in Chapter 3 with channel estimation technique. Channel information detector will be introduced. Some popular detection algorithms and the modulation result in OFDM MIMO will also be discussed in this chapter. A known sequence of bits is generally used in channel estimation. The bit-sequence is unique for a certain transmitter and is repeated in every transmission burst. Thus, the channel estimator is able to estimate channel impulse response for each burst separately by exploiting the known transmitted bits and the corresponding received samples. Dynamic estimation of channel is necessary before the demodulation of OFDM signals since the radio channel is frequency selective and time-varying for wideband mobile communication systems. In this chapter, binary phase-shift keying (BPSK) modulation in a Rayleigh fading and Doppler fading channel will be applied in Matlab simulations to illustrate and compare the performance difference of each communication scheme, using the above- mentioned channel estimation techniques.

5.1 SINGLE CHANNEL ESTIMATION

Consider the selection diversity scheme with single transmit antenna and single receiver antenna. If channel state information is unknown, a known preamble sequence will be sent first, the obtained preamble sequence is then demodulated and compared with the original one to obtain the channel state information. Figure 5.1 shows us this process. Now consider the one-transmitter-two-receiver-selection diversity model with unknown channel state. In this case, a preamble sequence p is sent from transmitter to the both receivers. The channel between the transmitter antenna and the receiver antenna 1 is

32

Figure 5.1. Basic structure of Single Channel Estimation. denoted as h1 and between the transmit antenna and the receive antenna 2 is denoted as h2 where h1= a1ej1 h2= a2ej2 After convolving with the channel, noise and interference are added to the preamble symbols. The resulting received preamble signals are p-received1= h1p+n1 p-received2= h2p+n2 Where p-received1, p-received2 are the received preamble signals and n1, n2 are complex random variables representing the noise and interference. Since the receiver has already known the preamble sequence p, the channel can be estimated as following: estimated-h1= p-received1/ p estimated-h2= p-received1/ p Then at a given time, a signal x is sent from the transmitter after is channel estimation is completed. After convolving with the channel, noise and interference are added to the symbols. The resulting received baseband signals are r1=h1x+n3 r2=h2x+n4 Where r1, r2 are the received signals and n3, n4 are complex random variables representing the noise and interference.

33 Consider the estimated channel state information estimated-h1 and estimated-h2, The final signals after decision making are S= estimated-h1 *r1+ estimated-h2*r2 = estimated-h1*(h1x+n1) + estimated-h2*(h2x+n2) = (a12+ a22)x+ estimated-h1*n1+ estimated-h2*n2 Then after equalizing, we can get the final signal after decision making. As we can see in the Figure 5.2, we simply simulate the performance of BPSK signals using One Transmit antenna with One Receive antenna selection diversity scheme, Two Transmit antenna with Two Receive antenna selection diversity scheme, One Transmit antenna with Four Receive antenna selection diversity scheme, through slow, Rayleigh fading channel with additive white Gaussian noise. And also in the same figure, we simulate the performance of BPSK signals with preamble using One Transmit antenna with One Receive antenna selection diversity scheme, Two Transmit antenna with Two Receive antenna selection diversity scheme, One Transmit antenna with Four Receive antenna selection diversity scheme, in slow, Rayleigh fading channel with additive white Gaussian noise. Apparently, the performance is much worse when BPSK signals begin having preambles because the channel statement information is unknown. In the same condition with one transmit antenna and one receive antenna, we also simulate the performance when Eb/N0 of the preamble sequences are changed to 0dB, 5 dB, 10dB, which can be seen in the following figure, and here is what we find out: when the energy noise ratio is higher, the preamble sequence received by the receive antenna are more accurate and the channel state information is estimated better, therefore the performance is getting better. The result is clearly illustrated in Figure 5.3.

5.2 JOINT CHANNEL ESTIMATION

In this section, a different channel estimation technique called joint channel estimation will be discussed. Consider the two-transmitter-one-receiver configuration, two preamble sequences are needed to estimate the communication channels. Assuming that the same preamble sequences are sent from the two transmitter antennas, the two channels can then be estimated at the same time. This channel estimation scheme is called joint channel estimation. Assuming that 64 symbols in the preamble sequences are sent from the transmit

34

Figure 5.2. Performance about 1 transmitter 1, 2 and 4 receiver selection diversity scheme and 1 transmitter 1, 2 and 4 receiver selection diversity scheme with preamble.

35

Figure 5.3. Performance of single transmitter and single receiver when the rate between energy and noise of the preamble is 0dB, 5dB and 10dB.

36 antennas, namely, p(1), p(2), p(3), p(4), p(5)p(64) , which can be splitted into two parts: odd part and even part:

Then 0 will be inserted in the rest even and odd part of these two sequences, so the preamble sequences become p(1), 0, p(3), 0, p(5)p(63), 0 and 0, p(2), 0, p(4), 0, p(6)0, p(64). Then these two preamble sequences will be individually transmitted through the channel and be estimated at the same time to increase the efficiency. The procedure is shown at Figure 5.4.

Figure 5.4. Basic structure of Joint Channel Estimation. Let us consider applying the joint channel estimation technique in the Beamforming scheme.. Here preamble sequence p is separated into two parts: p1 and p2, where p1 is sent from the first transmitter to the receiver, p2 is sent from the second transmitter to the receiver. The channel between the transmit antenna one and the receive antenna is denoted as h1 and between the transmit antenna two and the receive antenna is denoted as h2 where h1= a1ej1 h2= a2ej2 After convolving with the channel, noise and interference are added to the preamble symbols. The resulting received preamble signals are p-received1= h1p1+n1 p-received2= h2p2+n2

37 Where p-received1, p-received2 are the received preamble signals and n1, n2 are complex random variables representing the noise and interference. Since the receiver has already known the preamble sequence p, the channel can be estimated as following: estimated-h1=odd- p-received1/odd- p1 estimated-h2= even-p-received2/even- p2 Where odd- p-received1 is the odd part of p-received1, even-p-received1 is the even part of p-received2, odd- p1 is the odd part of p1, even- p2 is the even part of p. After being resampled, estimated-h1 and estimated-h2 will turn from 32 to 64, which can be denoted as resample- h1 ,resample- h2. Then at a given time, a signal x is sent from the transmitter after is channel estimation is completed. After convolving with the channel, noise and interference are added to the symbols. The resulting received baseband signals are r1=h3x+n3 r2=h4x+n4 Where r1, r2 are the received signals, h3, h4 are channel information and n3, n4 are complex random variables representing the noise and interference. Consider the estimated channel state information resample-h1 and ressample-h2, The final signal after decision making is = resample-h1 *r1+ resample-h2,*r2 = resample-h1 *(h1x+n1) + ressample-h2,*(h2x+n2) = (a12+ a22)x+ resample-h1 *n1+ ressample-h2*n2 The extracted preamble sequences are denoted as the following with the. Odd extracted preambles are on left and even extracted preamble are on right.

In the Figure 5.5, we compared the performance between one transmit antenna one receive antenna using single channel estimation and two transmit antennas one receive antenna using joint channel estimation through slow, Rayleigh fading channel with additive

38

Figure 5.5. Joint channel estimation in two transmit antennas one receive antenna compare to single channel estimation 1 transmit antenna and 1 receive antenna. white Gaussian noise. As we can see from the figure, when Eb/N0 are low, one transmit antenna and one receive antenna using single channel estimation is better than two transmit antennas one receive antenna using joint channel estimation; when Eb/N0 are high, two transmit antennas one receive antenna using joint channel estimation is better. When Eb/N0 are lower, since little signal information is received and the two branches transmission system has two transmit antennas, apparently, the signals sent by the two branches transmit antennas are more accurate, at this time, the bit error rate is lower, therefore, the performance of joint channel estimation two transmit antennas one receive antenna system is better. And when Eb/N0 is getting higher, the performance is much more dependent on the channel state information estimation, since most signal information are clearly received and the channel information are much better estimated in the single channel estimation system, therefore, at

39 this time, the performance of singlet channel estimation one transmit antenna one receive antenna system is better.

40

CHAPTER 6 SINGLE FFT (FAST FOURIER TRANSFORM) RECEIVER COMBINING SCHEME

In the previous chapters, we discussed several OFDM MIMO transmission schemes and channel estimation methods. In those schemes the number of the FFT (fast Fourier transform) structure is the same as the number of channels, which is complex and costly. To address this issue, we investigate using only one FFT structure at the receive antenna side to deal with many-channel communications, which requires at the receiver the best channel or channel combination to be chosen in frequency domain instead of in time domain as we did in pervious chapters. The goal is to find the receive signals that best resembles the transmitted signal in frequency domain. This is done by finding the optimum threshold to find the channel with the minimum distortion and degradation on the signal. One of the solutions is to compare the magnitude of sum of all the subcarriers in each channel then find the largest one as the best channel since it has the maximum energy. Simulations (as we shall see later) show that the new scheme can achieve excellent diversity gains and bit error performance. The new scheme also has obvious advantages on its simplicity and flexibility. Figure 6.1 shows the basic structure of 1-FFT transmission.

Figure 6.1. Two transmitters and one receiver scheme with one FFT.

41

6.1 SINGLE FFT RECEIVER COMBINING SCHEME WITHOUT CHANNEL ESTIMATION

In this section, we are elaborate on how to find the optimum channel by using one FFT structure at the receiver antenna. When multiple FFT structures are used as in previous chapters, subcarriers with the highest energy at each channel are found in frequency domain. However, when only one FFT block is used, we are forced to find the optimum channel in time domain and then transform the selected channel to frequency domain. One of the general solutions is to sum up all the subcarriers at each channel and select the one with the largest magnitude (energy), which presumably provides the largest signal/noise ratio. In the following context we will explore this idea and demonstrate several viable schemes. The performance difference of these schemes will also be discussed. Again, we divide our discussions into two parts: channel information known and unknown.

6.1.1 Single FFT Receiver Combining Scheme with Known Channel Information
In the following simulation example, we consider one-transmitter two-receiver transmission model with known channel state information. In this case, the transmitting symbol is denoted as x, the channel between the transmit antenna and the receive antenna one is denoted as h1 and the one between the transmit antenna and the receive antenna two is denoted as h2 where h1= a1ej1 h2= a2ej2 After convolving with the channel, noise and interference are added to the transmitting symbols. The resulting received preamble signals are r1= h1x+n1 r2= h2 x+n2 Where r1, r2 are the received signals and n1, n2 are complex random variables representing the noise and interference. Assuming that 64 subcarriers are used in each channel, which are denoted as

42 h1 = [h10, h11 ,h12 h163,h164 ] h2= [h20 ,h21 ,h22 h263,h264 ] We define the received time main signals are r1 = [r10, r11 ,r12 r163,r164 ] r2= [r20 ,r21 ,r22 r263,r264 ] Sum up the absolute value of the subcarriers sum1= abs(h10)+ abs(h11)+ abs(h12)+ abs(h163)+ abs(h164) sum2= abs(h20)+ abs(h21)+ abs(h22)+ abs(h263)+ abs(h264) Compare the sum and find the largest magnitude which is Maximum (sum1, sum2) So if sum1= maximum (sum1, sum2) Then choose h1 as channel information, the final signal after decision making are S= r1/h= (h1x+n1)/ h1 =x+ n1/ h1 If sum2= max(sum1,sum2) Then choose h2 as channel information, the final signal after decision making are S= r2/h= (h2x+n2)/ h2 =x+ n2/ h2 We then apply the above scheme to the cases of 1 transmitter with 1 receiver, 2 receivers, 3 receivers, 4 receivers, the results of which are shown in Figure 6.2. It is demonstrated that even with 1-FFT block at the receiver, spatial diversity scheme still shows the improved performance. A minor variation to previous scheme is to sum the square value of all the 64 subcarriers at the same channel for all channels and select the one with largest sum-up value. We apply this scheme to the same example of Figure 6.3, the results of which are shown in Figure 6.4. Again, it is confirmed that spatial diversity showing improved performance even in one FFT structure configuration. For example, in a two-channel communication system, the weight of each channel is defined as W1= sum1/( sum1+sum2) W2= sum2/( sum1+sum2)

43

Figure 6.2. Performance about single FFT combining sum abs 1Transmit antenna 1,2,3,4 Receive antenna.

44

Figure 6.3. Performance about single FFT combining sum square 1Transmit antenna 1,2,4 Receive antenna.

45

Figure 6.4. Performance comparison of single FFT combing sum abs, sum square and sum square with weight.

46 The final signal after decision-making is therefore S= (W1* r1+ W2* r2)/ (W1* h1+ W2* h2) =( sum1/( sum1+sum2)* r1+ sum2/( sum1+sum2)* r2)/ (sum1/( sum1+sum2)* h1+ sum2/( sum1+sum2)* h2) = (sum1* r1+ sum2* r2)/ (sum1* h1+ sum2* h2) =( sum1*(h1x+n1) + sum2*(h2x+n2))/ (sum1* h1+ sum2* h2) ( sum1*h1x+ sum1*n1 + sum2*h2x+ sum2*n2) / (sum1* h1+ sum2* h2) =(sum1* h1+ sum2* h2) *x /(sum1* h1+ sum2* h2)+ (sum1*n1 + sum2*n2) / (sum1* h1+ sum2*h2) =x+ (sum1*n1 + sum2*n2) / (sum1* h1+ sum2* h2) In Figure 6.4, we compare the performance of the above three decision-making schemes, namely, sum of absolute receiver-signal magnitude, sum of square of receiversignal and weighted sum of all receiver-signal, in a 1-transmitter-4-reciever system. It is clearly demonstrated that weighted-sum scheme show the best performance. We also compare our new scheme to the selection diversity and MRRC. In Figure 6.5 performance comparison of MRRC and square-sum weight is shown. In Figure 6.6 we show the performance comparison of Selection diversity and square-sum with weight. As we can see, the performance degrades with one FFT structure, which is not surprising. It is a trade of between system performance and complexity and cost.

6.1.2 Single FFT Receiver Combining Scheme In Unknown Channel Condition

Again, in the unknown channel condition, preamble has to be introduced for channel estimation. Consider the one-transmitter-two-receiver model and assume the channel between the transmit antenna and the receive antenna one as h1 and between the transmit antenna and the receive antenna two as h2, where h1= a1ej1 h2= a2ej2

47

Figure 6.5. Performance comparison between sum square single FFT combing with weight and MRRC.

48

Figure 6.6. Performance comparison between 1 transmit antenna 1,2,4 receive antenna single FFT combining Sum square with weight and Selection diversity. After convolving with the channel, noise and interference are added to the preamble symbols. The resulting received preamble signals are p-received1= h1p+n1 p-received2= h2p+n2 Where p-received1, p-received2 are the two received preamble signals and n1, n2 are complex random variables representing the noise and interference. The channel can be estimated as following based on the received preamble signals and the know preambles:estimated-h1= p-received1/ p estimated-h2= p-received1/ p. On the other hand, the received signals when transmitting a general signal x are r1=h1x+n3

49 r2=h2x+n4 Where n3, n4 denote noise and interference. The signal estimation of the originally transmitted signal could then be defined during the following process: S= (W1* r1+ W2* r2)/ (W1* h1+ W2* h2) =( sum1/( sum1+sum2)* r1+ sum2/( sum1+sum2)* r2)/ (sum1/( sum1+sum2)* h1+ sum2/( sum1+sum2)* h2) = (sum1* r1+ sum2* r2)/ (sum1* h1+ sum2* h2) =( sum1*( estimated-h1x+n1) + sum2*( estimated-h2x+n2))/ (sum1* h1+ sum2* h2) =( sum1* estimated-h1x+ sum1*n1 + sum2* estimated-h2x+ sum2*n2) / (sum1* h1+ sum2* h2) =(sum1* estimated-h1+ sum2* estimated-h2) *x /(sum1* h1+ sum2* h2)+ (sum1*n1 + sum2*n2)/ (sum1* h1+ sum2* h2) Where W1 and W2 are the channel weight as defined in the previous section, note that the sum1 and sum2 in the formula can be both magnitude-sum or square-sum We apply the above signal estimation scheme in the 1 transmitter 1 receiver, 2 receivers and 4 receivers with square- sum weight principle, the results of which are shown in Figure 6.7.

6.2 CONVOLUTIONAL CODING AND VITERBI DECODING

In communication technology, forward error correction (FEC) technique is often used to improve the channel capacity by adding specifically designed redundant information to the data being transmitted through the channel. The process of adding this redundant information is known as channel coding. There are two major types of channel coding: convolutional coding and block coding. Convolutional coding operates on serial data, not static block and it has memory that uses previous bits to encode or decode following bits. Block coding operates on relatively large message blocks. It is denoted by (n,k,L), where L is code memory depth [26]. The basic process of convolutional encoder is illustrated in Figure 6.8: the input signal is k-bit long, output signal is n-bit long after encoder whose memory length is L. The Viterbi Decoder block decodes the signal using the Viterbi algorithm, which iswidely applied in wireless communications and in 1964 it was developed by Andrew J. Viterbi, a founder of Qualcomm Corporation. Viterbi decoding is one of the two types of decoding algorithms used with convolutional encoding; the other type is sequential decoding.

50

Figure 6.7. Performance Comparison of 1Transmit antenna 1,2,4 Receive antenna single FFT combining sum square with weight and preamble.

Figure 6.8. Convolutional code: k bits input, n bits output and L memory depth. Viterbi decoding has the advantage that it has a fixed decoding time, which is well suited to hardware decoder implementation. Viterbi decoding eliminates least likely trellis path at each transmission stage and reduces decoding complexity with early rejection of unlike path [27]. So it is more efficient than Maximal Likelihood via concentrating on survival paths of the trellis.

51 Convolutional encoding and viterbi decoding is an effective method to improve bit error performance of the whole wireless system. The following two figures depicts the performance improvement after using encoding and decoding in the transmission process, which shows that after coding ,SNR increases by 3 dB. Figure 6.9 compares the performance of 1-FFT receiver combining with and without coding using abs-sum in a 1-transmitter-1receiver configuration Figure 6.10 compares the performance of 1-FFT combining scheme with and without coding using square-sum in the 1-transmitter-1-receiver, 2 receivers, 4 receivers configuration respectively.

Figure 6.9. Performance comparison between single FFT combining sum abs with coding and without coding.

52

Figure 6.10. performance comparison between single FFT combining sum square with coding and without coding. Also we can find better performance when weight is added. Figure 6.11 compares the performance of 1-FFT combining scheme using weighted square-sum technique in the 1 transmitter and 1 receiver, 2 receivers, 4 receivers configurations respectively, with and without coding. In this set of simulation, we assume known channel condition. We conduct the same simulation in the unknown channel condition. In this case, channel estimation is conducted based on sending known preambles. Simulation results are shown in Figure 6.12. We note that when convolutional encoder and Viterbi decoder are introduced, the performance increases by more than 5dB. Finally, as a reference, we also compare MRRC to the single FFT combining scheme using coding + square-sum weight + preamble. The performance is shown in Figure 6.13. It is also illustrated in Figure 6.14 about the performance comparison of selection diversity with the single FFT combining scheme using coding + square-sum weight + preamble. As we can see, the performance degrades when using single FFT structure, which is, again, the trade-off

53

Figure 6.11. Performance comparison between single FFT combining sum square plus weight with coding and without coding. between transmission fidelity and system complexity and cost. To summarize, we demonstrate in this chapter the feasibility of using one FFT structure in a MIMO OFDM communication system. Reasonable transmission performance (BER) was obtained. We investigated and designed the scheme for channel selection and signal estimation in time domain for the implementation of one FFT structure. We also notice the one FFT structure is, in general, has degraded performance than multi-FFT structure schemes, such as MRRC and selection diversity schemes, which is simply a trade-off between transmission quality and system complexity and cost. Further research can focus on the possibility of improving the channel selection and signal estimation scheme in one FFT structure to approach the performance of the multi-FFT structures.

54

Figure 6.12. Performance comparison between single FFT combining sum square with weight, preamble plus coding and sum square with weight and preamble.

55

Figure 6.13. Performance comparison between single FFT combining sum square with weight, preamble plus coding and MRRC with preamble and coding.

56

Figure 6.14. Performance comparisons between sum square with weight, preamble plus coding and selection diversity with preamble and preamble.

57

CHAPTER 7 CONCLUSIONS AND FUTURE WORK

In this thesis, we focused on investigating OFDM MIMO communication systems. Different OFDM MIMO schemes were explored. We discussed standard multi-FFT structure schemes and then researched on the feasibility of using one-FFT structure for OFDM MIMO system in order to reduce the system complexity and cost. In OFDM MIMO transmission system, there are a lot of different schemes that can be applied. Beamforming, MRRC and selection diversity are generally used in MISO and SIMO; while in second order transmission system, Alamouti scheme is often introduced. All these above schemes are under the environments that need consider multipath effect and Doppler effects. Chapter Four discusses all these schemes in detail. In the unknown channel condition, preamble is applied to the transmission system in order to estimate the channel information. There are two ways to estimate channel information: single channel estimate and joint channel estimate, which is described in Chapter four. In Chapter Six, major contribution of this thesis work is described, that is, to use only single FFT structure at the receiver to reduce the system complexity and cost. Since the subcarriers are in frequency domain, using only single FFT to decode all the subcarriers require new signal estimation/equalization technique. We used the best channel identification as such technique. The basic idea is to add all the subcarrier together to get the branch of largest magnitude as the optimal channel, leveraging the receiver signals signal-to- noise ratio. As improvement to this basic idea, channel weighting and coding are used to further minimize the bit error rate (BER). We performed complete sets of simulations and the results are illustrated to demonstrate the merits of the proposed schemes. Given the work finished in this thesis, there are still a lot of opportunities for future research on OFDM MIMO that related to this thesis. We list below several topics that might be interesting. Explore the possibility of further improving or designing new signal estimation/equalization method for signal FFT combining structure OFDM MIMO system based on the current method in this thesis. For example, we may design new

58 space -time trellis according to the characteristic of the channel to improve the signal estimation process. Research on the possibility of further reducing MIMO system complexity and developing corresponding coding and decoding schemes, without significantly impact the system performance. Design some pre-coding/framing mechanism which can be employed to translate a MISO or SIMO system into a mathematically equivalent SISO channel, furthermore to translate a MIMO channel into a mathematically equivalent SiMO or MOSI channel. Explore more channel estimation methods to get better and more accurate channel state information, these methods may include coherent detection, equalization or pilot-symbol-aided estimation.

59

REFERENCES
[1] [2] D. Tse and P. Viswanath, Fundamentals of wireless communication, Cambridge, MA: Cambridge University Press, 2005. A. C Brooks and S. J, Hoelzer, Design and Simulation of Orthogonal Frequency Division Multiplexing signaling, Department of Electrical Engineering, Bradley University, Peoria, IL, Project Report, May, 2001. S. J. Vaughan-Nichols (2001, November 14), OFDM: Old Technology for New Market, Tutorial article [Online], Wi-FI Planet. Available: http://www.wifiplanet.com/tutorials/article.php/1500641/OFDM-Old-Technology-for-NewMarkets.htm. R. V. Nee, OFDM for Wireless Multimedia Communications, Boston, MA: Artech House Publishers, 1999. A. Doufexi, S. Armour, M. Butler, A. Nix, D. Ull, J. McGeehan and P.Karlsson, A Comparison of the HIPERLAN/2 and IEEE 802.11a Wireless Standards, IEEE Communications Magazine, vol. 40, pp. 172-180, 2002. IEEE 802.11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications. IEEE-SA Standard. 2007. Y. Loucheryavy, J. Harju and V. B. Iversen, Next Generation Teletraffic and Wired/Wireless Advanced Networking, Berlin, Germany: Springer, 2006. IEEE 802.16: Wireless Broadband Standards, IEEE Standard, 2001. J. Winters ,Optimum Combining in Digital Mobile Radio with Cochannel Interference, IEEE Trans. on Vehicular Technology, vol. 33, pp.144-155, 1984. A. Katalinic, R. Nagy and, R. Zentnerl, Benefits of MIMO Systems in Practice: Increased Capacity, Reliability and Spectrum Efficiency in 48th International Symposium ELMAR, Zadar, Croatia, June 2006. L. Hanzo, M. Minster, B.J. Choi and T. Keller, OFDM and MC-CDMA for Broadband Multi-User Communications, Hoboken, NJ: Wiley-IEEE Press, 2003. B. Sklar, Digital Communications: Fundamentals and Applications,2nd edition, Upple Saddle River, NJ: Prentice Hall, 2001. B. Sklar, Rayleigh Fading Channels in Mobile Digital Communication Systems Part I: Characterization. IEEE Communications Magazine, vol. 35, pp. 90100, 1997. A. B. do Paula, An Introduction to Doppler effect and fading in mobile Communications, Masters thesis, Report Number A679062, Naval Postgraduate School, Montreal, CA, 1992. J. G. Proakis,, Digital Communications, New York, NY: McGraw-Hill Book Company, 1989.

[3]

[4] [5]

[6] [7] [8] [9] [10]

[11] [12] [13] [14]

[15]

60 [16] [17] [18] [19] C. Oestges and B. Clerckx, MIMO Wireless Communications: From Real-world Propagation to Space-time Code Design, San Diego, CA: Academic Press, 2007. M. K. Simon and M. S. Alouini, Digital Communication Over Fading Channels, 2nd Edition, Hoboken, NJ: Wiley-IEEE Press, 2004. K. Baum, B. Classon and P. Sartori, Principles of Broadband OFDM Cellular System Design, Hoboken, NJ: John Willey & Sons, 2010. X. Li and F. Ng , Using cyclic prefix to Mitigate Carrier Frequency and Timing Asynchronism in Cooperative OFDM Transmissions in ACSSC 06, Fortieth Asilomar Conference on Signals, Systems and Computers, 2006. L. Zheng and D. N. C. Tse , Diversity and Multiplexing: A Fundamental Tradeoff in Multiple-Antenna Channels, IEEE Transactions on Information Theory, vol. 49, pp. 1073-1096, 2004. R. Heath, Jr. and A. Paulraj, Switching between multiplexing and diversity based on constellation distance in Proc. Allerton Conf. Communication, Control and Computing, Monticello, IL, 2000. D.Agrawal, V Tarokh, A Naguib and N Seshadri,, Space time coded OFDM for highdata rate wireless communication over wide band channels in Proc of 48th IEEE Vehicular technology Conference, vol.3, pp. 22322236, Ottawa, Canada, 1998. Y S. Cho, J. Kim, W. Y. Yang and C.G. Kang, MIMO-OFDM Wireless Communications with MatLab, Hoboken, NJ: Wiley Press, 2010. S. M. Alamouti, A simple transmit diversity technique for wireless communications, IEEE journal on select areas in communications ,vol. 16, pp. 1451-1458, 2008. T. M. Duman and A. Ghrayeb, Coding for MIMO Communication Systems, Hoboken: NJ: Wiley Press, 2008. S. Lin and D. J. Costello, Error Control Coding,2nd edition, Upple Saddle River, NJ: Prentice Hall, 2004. A. J. Viterbi, Error Bounds for Convolutional Codes and an Asymptotically Optimum Decoding Algorism, IEEE Transactions on Information Theory, vol. IT13, pp. 260-269, 1967.

[20]

[21]

[22]

[23] [24]

[25] [26] [27]