Per-antenna power constrained rate optimization for multiuser MIMO systems

This paper investigates the problem of linear transceiver design for rate optimization in multiuser Gaussian multi-input-multi-output (MiMO) systems. We focus on a scenario with per-antenna power constraint, which is of importance for a realistic implementation. Two design criteria are considered: (1) maximizing the weighted sum rate subject to per-antenna power constraints; (2) maximizing the minimum user rate subject to per-antenna power constraints. Non-convexity and the complicated mathematical structure of the problems make it difficult to derive efficient, globally optimal algorithms. Even the power allocation problem for sum-rate optimization leads to a problem formulation involving ratios of posynomials, which is generally an intrinsically intractable NP hard problem. The problems studied here are even more complicated, since we consider a joint optimization over powers, transmit filters and receive filters. For these problems, we derive iterative algorithms and prove the monotonic convergence of the algorithms. Additionally, we show that the results can be extended to other variations of the problems and the problems with other kinds of power constraints.