On the average rate of precoded QSTBC in MIMO channels with non-zero mean

It was shown recently, that in case of Rayleigh fading and an uninformed transmitter the average rate achieved with quasi-orthogonal space-time block codes (QSTBC), which use an orthogonal space-time block code (OSTBC) as building element, is equal to the average rate of the underlying OSTBC. In this work, we assume that the transmitter knows the mean channel matrix in a more general scenario either by feedback or channel estimation and the receiver has perfect CSI. We then analyze the optimal transmit strategy with respect to the average rate. Interestingly, at first the optimization problem is non-convex. However, after choosing the optimal directions, the remaining power allocation problem turns out to be convex. We analyze the impact of the mean component on the average rate, the optimal transmit strategy including the optimal power allocation. We show that in case of mean knowledge the QSTBC outperform the OSTBC with respect to the average rate and illustrate our theoretical results using numerical simulations.