Research In Synchronization And Channel Estimation For MIMO-OFDM

Recent research has shown that Multiple-Input Multiple-Output (MIMO) is becoming an attractive technology which boosts channel capacity and spectral efficiency greatly; while Orthogonal Frequency Division Multiplexing (OFDM), on the other hand, can cope with multi-path fading and is easy to implement. Therefore, the combination of these two - MIMO-OFDM, which has been developed into a key technology for B3G/4G physical layer transmission - is revealing its potential to dramatically increase both channel capacity and spectral efficiency as well as reducing the receiver complexity in a multi-path communication environment.Key technologies involved in MIMO-OFDM include: MEMO channel modeling, system synchronization, channel estimation and space-time multiplexing. This thesis focuses on a MIMO-OFDM system, highlights the analyses of system synchronization and channel estimation. Systems based on OFDM are extremely sensitive to phase deviation caused by errors in synchronization. Inter-Carrier Interference (ICI) caused by the carrier frequency deviation and sampling frequency deviation greatly deteriorate system performance. In coherent MIMO systems, the coefficients of channel response (either time-domain or frequency-domain) are essential to signal processing modules such as space-time coding and equalization. The accuracy of channel estimation is therefore crucial to system performance. On the other side, each channel in a MIMO system carries a combination of signals transmitted from multiple antennas, and therefore implies a much more complicated system configuration and estimation scheme compared to SISO systems.This thesis first discusses a variety of MIMO channel models, their capacities and signal model, system framework and key technologies for MIMO-OFDM systems. Then it focuses on system synchronization and channel estimation. The work being done is briefly stated as follows:I. Starting from the time-domain signal model of MIMO-OFDM, this papar analyzes the influences of all kinds of errors - which occur in synchronization acquisition - on the received signal. Then an improved orthogonal pilot design method was proposed, which adopts a time-frequency domain structure, to resist the influences of nonideal base-band digital filters. A synchronization acquisition algorithm using such pilots is also given. This orthogonal pilot sequence inherits the excellent orthogonality from the time-domain structure and a nice spectrum property from the frequency-domain structure. It remains a satisfactory cross-correlation property even after being processed by the digital filters at both the transmitter and receiver, which is quite favorable in improving the accuracy in parameter estimation of synchronization acquisition.II. In view of the residual carrier frequency deviation and residual sampling frequency deviation, analyzes the received signal model with the presence of both the two errors and puts forward a pilot-based tracking algorithm for synchronization error extraction over frequency-selective fading channels. The major merit of this algorithm is its insensitivity to the residual carrier-frequency deviation. It can extract the sampling frequency deviation correctly with the presence of residual carrier-frequency deviation by using the pilot sequence and therefore track the synchronization error in the received signals timely and efficiently.III. The system model of MIMO-OFDM over time-varying channels which takes advantage of the quasi-linear property of time-varying channels under certain conditions is propesed. And then the impact of ICI (generated by time variability of the channel) on the received SNR is analyzed and an interpolation-based MIMO-OFDM channel estimation algorithm as well as the constraints on estimation validity are discussed. Basic criteria for the design of MIMO pilot matrix and examples are given. This MIMO-OFDM-oriented algorithm uses interpolation to reduce the number of parameters required in time-varying channel response estimation. Estimation of channel parameters is realized by using pilot sequences and therefore is robust to the time variability of the channel. Also, calculating the equivalent transition matrix of channel response coefficients offline greatly reduces the implementation complexity.IV. A (semi-) blind channel estimation algorithm based on the noise sub-space is proposed. It is designed for certain Orthogonal Space-Time Block Codes (OSTBC) whose code rates are smaller than 1, and is well-fit for the frequency-selective fading channel. This algorithm takes full advantage of the inherent redundancy and orthogonality of STBC and successfully avoids using signal processing methods which lower down the system efficiency, such as pre-coding, virtual sub-carrier insertion and so on. Meanwhile, there is no such need of using more receive antennas than transmit antennas as required by some kinds of blind estimation algorithms which are based on the second order statistical properties of the signals such as the sub-space blind estimation algorithm and linear prediction algorithm. Therefore it is well-fit to be used in the downlink of wireless communication systems. The algorithm adapts the overall computational complexity by changing the dimension of the objective of singular value decomposition. Simulation results show that appropriate adjustment of the dimension can improve the performance of the algorithm.