Abstract
Channel estimation for multiple-input, multiple-output (MIMO) systems is studied in this paper. In particular, we present a simplified MIMO channel estimator based on orthogonal design. The performance of the proposed scheme is theoretically analyzed and compared to that of the optimum maximum likelihood estimator. The effect of non-orthogonality of the training sequences is investigated. Some modifications of the proposed estimator with sample stacking and averaging are introduced to further improve the estimation performance. This simplified scheme is evaluated in the context of the WiMAX MIMO systems in terms of mean square error for the channel estimation and bit error rate for the space-time turbo equalization. Both analytical and simulation results indicate that despite of its low computational complexity, this simplified estimator leads to minimum variance unbiased estimation and achieves identical performance to that of the maximum likelihood estimator.