Optimal power-saving input covariance for MIMO wireless systems exploiting only channel spatial correlations

Most researches into multiple-input multiple-output (MIMO) antenna systems have aimed to determine the capacity-achieving input covariance given certain degree of channel state information (CSI) at the transmitter side. In practice, however, it is much preferred to find the input signal covariance that requires the least average transmit power to support a given rate based on only long-term channel information. In this paper, we analyze the characteristics of the optimal (in the sense of power-saving) input covariance for spatially correlated MIMO channels where only the long-term (slow-varying) channel spatial covariance information is available at the transmitter. Sufficient and necessary conditions of the optimal input covariance are derived. By considering the large-system regimes, we devise an efficient iterative algorithm to compute the asymptotic optimal power-saving input covariance. Remarkably, simulation results will demonstrate that the asymptotic solution is very effective in that it gives promising results even for MIMO systems with only a few antennas at the transmitter and the receiver.