Abstract:  To improve accuracy of sparse channel estimation and reduce the pilot number in MIMO-OFDM systems, we use the sparse prior information of the channel impulse response in the time domain, and model the estimation of frequency selective fading channel for MIMO-OFDM systems as the reconstruction of complex sparse signal interfered by noise in compressed sensing. Two methods of sparse channel estimation in MIMOOFDM systems are proposed, based on sparsity adaptive matching pursuit (SAMP) and sparse reconstruction by separable approximation (SRSA), respectively. Simulation shows that, under the same signal-to-noise ratio and for the same performance of MSE and BER without prior information of the sparsity, the two proposed methods can reduce pilot signals by 40% as compared to the conventional least square method. In the two methods, the one based on SAMP runs faster and is closer to the Cramer-Rao bound, while parameters of the one based on SRSA are easier to be set in practical applications.

Key words: sparse channel estimation, sparsity adaptive matching pursuit (SAMP), sparse reconstruction by separable approximation (SRSA), multiple input multiple output (MIMO), orthogonal frequency division multiplexing (OFDM)