Transmitter design in partially coherent antenna systems

In this paper, we study the effect of channel estimation error on transmitter design in a multiple input multiple output (MIMO) channel. We assume that only knowledge of either the channel correlation or mean is available at the transmitter and find the transmitting strategy. Under covariance feedback, at the transmitter the channel is modelled as a matrix of zero mean circularly symmetric complex Gaussian (ZMCSCG) random variables with known covariances. We assume that only rows of channel matrix are correlated. Under mean feedback the covariance of channel is modelled as white. We determine the necessary and sufficient conditions under which a unit rank input covariance matrix, i.e. eigen-beamforming, can achieve capacity lower bound. Numerical simulations are conducted to corroborate theoretical results.