Abstract:

In this paper，an algorithm of channel estimation based on space-time correlation in MIMO OFDM systems is proposed，which adopts an LS channel estimation scheme of time domain for initial estimation and exploits full information of the space-time correlation known at the receiver. The optimal weight of the space-time filter for the initial estimation is obtained by minimizing the MSE of channel estimation in average sense of multiple symbols in the FIR window. Further，the complexity of the algorithm is greatly decreased by the SVD method. Finally，simulations show that this algorithm derived outperforms the conventional MA algorithm of channel estimation in the performances of the MSE and BER when the optimal length of the FIR window reaches under the same conditions. And at some conditions，the performance of the algorithm can approach the performance of the ideal channel estimation，and can be improved with the space-time correlation increased.

Key words: MIMO OFDM system, least square channel estimation of time domain（TD-LS-CE）, space-time correlation, moving average（MA）, normalized mean square error of time domain（TD-NMSE）