Linear precoding for MIMO channels with non-zero mean and transmit correlation in orthogonal space-time coded systems

We study the problem of exploiting MIMO wireless transmitter channel knowledge in the form of the channel mean and transmit correlation from a coding perspective. A linear channel precoder is designed to capture the channel information, which is used in concatenation with an orthogonal space-time block code. Based on the pair-wise error probability criterion, the optimal precoding matrix is derived analytically. Depending on the channel mean and correlation, and the SNR, the solution may require mode-dropping, which resembles the water-filling principle. The difference is that both the "water level" and the mode directions change with each water-fill iteration, hence it is termed dynamic water-filling. Efficient binary search algorithms are proposed to carry out the dynamic water-filling process in solving for the optimal precoder. Numerical examples show that significant gain can be obtained using this linear channel precoder.