MIMO Multiple Access Channels With Noisy Channel Estimation and Partial CSI Feedback

The authors consider correlated MIMO multiple access channels with block fading, where each block is divided into training and data transmission phases. They find the channel estimation and data transmission parameters that jointly optimize the achievable data rate of the system. Their results for the training phase are particularly interesting, where they show that the optimum training signals of the users should be non-overlapping in time. For the data transmission phase, they propose an iterative algorithm that updates the parameters of the users in a round-robin fashion.