Efficient MSE balancing for the multi-user MIMO downlink

A novel MSE balancing paradigm is presented which is based on the unweighted sum-MSE minimization with individual power constraints in conjunction with an adaptive power allocation. By means of a counter-example, we show that even the two-user MSE region may be nonconvex in case of multiantenna users which is in contrast to the predominant opinion. Any algorithm utilizing the weighted sum-MSE minimization hence cannot achieve the complete MSE region for any channel realization. This resulting gap due to the nonconvex dent is closed by the presented approach. For the unweighted sum-MSE minimization with individual power constraints which represents the core of the balancing algorithm, we come up with an extremely fast converging alternating optimization outperforming all hitherto existing approaches.