MMSE transmitter design for correlated MIMO systems with imperfect channel estimates: power allocation trade-offs

We investigate the transmit precoder design problem for a multiple input multiple output (MIMO) link with correlated receive antennas, considering the effect of channel estimation. We work with the total mean squared error (MSE) as the performance measure, and develop transceiver structures considering the effect of channel estimation and the correlation of the MIMO link. The proposed transceiver structures are optimum in the sense of minimizing the total MSE and distributing the total MSE equally among the parallel data streams. Motivated by the substantial effect the channel estimation process can have on the system performance, we next investigate the problem of how the correlated MIMO link should distribute its total available power between power expended for channel estimation versus data transmission. The optimum power allocation problem between the training sequences for channel estimation and data transmission for the correlated MIMO link is shown to have a unique solution, that is different than the uncorrelated case. It is observed that the proposed transceiver achieves near-minimum MSE values via a relatively wide range of power allocation parameters. This is in contrast to the transceiver that is oblivious to the estimation errors when a more precise power allocation strategy is needed to achieve the best performance. Our results demonstrate that the correlation structure of the MEMO link has a profound effect on the performance, and that the transceiver optimization should be done by taking both the correlation and the channel estimation process into account