Precoding for Cooperative MIMO Channels with Asymmetric Feedback

The problem of optimally precoding over cooperative MIMO channels when the transmitters are endowed with different noisy channel state information is a long standing and challenging open problem. Recently an information theoretic result was obtained which characterized the common message capacity of a channel with two transmitters and a single receiver with such distributed channel state information (DCSIT) generated from different feedback links. While classical MIMO precoding with centralized CSIT implies the transmission of a number of spatial streams bounded by the number of transmit and receiver antennas, the above result suggests that, surprisingly, the transmission of additional streams may be beneficial. In this work, we explore the operational implications of the above intuition to optimally tackle the problem of ergodic rate optimization under distributed feedback. In particular, we propose a method for joint distributed precoding and feedback design under asymmetric feedback rate constraints. In doing so, we also optimize the number of spatial data streams under practical complexity constraints. Finally, we provide numerical simulations and illustrate the performance gains compared to conventional precoder design.