MIMO Transmitter Optimization with Mean and Covariance Feedback for Low SNR

Multiple-input multiple-output system capacity is improved using channel state information at the transmitter (CSIT). This work proposes a transmission optimization strategy that aims to maximize the mutual information using statistical CSIT. A second order approximation method is used to optimize the transmit covariance matrix.

The proposed second order scheme allocates all the power to the eigenvector of the channel with the highest SNR. With increasing SNR, optimum strategy distributes power in other directions as well and converges to equal power allocation.

The graph shown above demonstrate the higher capacity of the second order scheme at low SNR. The comparision is made with equal power allocation, first order approximation and waterfilling schemes.