Statistical Precoding for MIMO Systems With Channel Estimation Errors

Obtaining accurate instantaneous channel state information (CSI) is challenging for multiple-input-multiple-output (MIMO) systems, particularly if the channel fluctuates rapidly. A more practical assumption is statistical CSI as the channel statistics are likely to remain unchanged for a longer period. In this letter, we propose a precoder design, using only statistical CSI, to minimize error probability for practical MIMO systems with channel estimation errors. Our proposed precoder design is shown to achieve a significant improvement in error performance when compared with other precoding schemes in literature. For example, for a real 4 × 4 MIMO system with BPSK modulation and at a codeword error rate of 10^-3, coding gains of up to 12 dB can be achieved.