Mimo-ofdm System, Frequency Offset And Channel Estimation Algorithm

OFDM (Orthogonal Frequency Division Multiplexing) has the strong immunity to multipath interference in wireless channel and significantly reduces receiver complexity due to the use of one-tap frequency-domain equalizer. MIMO (Multiple Input Multiple Output) can greatly increase the wireless system capacity without additional bandwidth and power requirements. Therefore, the combination of MIMO and OFDM is an attractive and competitive solution for future broadband wireless systems.This thesis focuses on frequency offset estimation and channel estimation. The main contributions are as follows:a) The traditional maximum likelihood estimator of carrier frequency offset based on virtual carriers suffers from huge computational amount. To reduce the computational amount, a low-complexity estimator with inverse fast Fourier transform (IFFT) and piecewise searching techniques is proposed. From simulation in additive white Gaussian noise (AWGN) and typical urban (TU) channels, we find that our estimator has far lower on complexity than the original one when providing the same performance as the original one.b) High peak-to-average power ratio (PAPR) is a disadvantage of comb-type frequency-domain training pattern in uncorrelated MIMO OFDM system. To deal with this problem, we derive the optimal condition based on maximum likelihood channel estimation from time domain. And an excellent orthogonal time-domain training pattern, based on circularly shifting Chu sequence, is constructed. By the simulation in TU channel, we obtain: our proposed time-domain training pattern can achieve the same mean square error (MSE) and bit error rate (BER) performances as the cyclic comb-type one.c) In wideband MIMO OFDM system with antenna correlation and multi-path fading, there is in general no closed-form optimal solution of training signal for linear minimum mean square error (LMMSE) channel estimator. To address this problem, we deduce the MSE lower bound of LMMSE estimator and the optimal condition of time-domain training pattern. When signal-to-noise ratio (SNR) is high, it degenerates to equi-powered allocation scheme. Computer simulation in correlated MIMO channel shows: equi-powered allocation pattern achieves the same performance as optimal power allocation pattern at medium and high SNR regions.