Capacity bounds for MIMO multiple access channel with imperfect channel state information

This paper investigates the effect of channel estimation errors in presence of spatial correlation for fading MIMO multiple access channels (MACs). Specifically, the capacity is derived under the notion of reliable communication based on the average of the error probability over all channel estimation errors. Although not necessarily optimality in presence of channel estimation errors, we restrict the channel inputs to be Gaussians, allowing a closed-form expression for the mutual information, which leads to an inner bound of the capacity region. For different types of transmitter information (e.g. instantaneous and covariance information), we study optimal transmission schemes that permit to maximize the sum-rate. Numerical results show that the performance of covariance feedback is quite sensitive to channel accuracy while the gap between the open-loop and the closed-loop scenario increases with the channel uncertainty. Simulations also assess the performance degradation of transmit/receive correlation and reveal that their impact behaves independently of the estimation errors.