Optimal Training Sequences for MIMO Channel Estimation with Spatial Correlation

Optimal design of training sequences is considered for multiple-input multiple-output (MIMO) systems in the presence of spatial fading correlation. By the method of minimum mean square error (MMSE) channel estimation, we first evaluate the estimation accuracy and the training sequence length under correlated fading, based on which we then design optimal training sequences that have controllable length to make effective use of the correlation properties. It is investigated that the impacts of spatial correlation are helpful not only to improve channel estimation but also to decrease the training length, moreover, that the proposed training sequences have good performance via simulations.