Successive convex approximation for simultaneous linear TX/RX design in MIMO BC

Weighted sum rate maximizing beamformer design is considered for MIMO broadcast channel. As opposed to the conventional alternating beamformer design, the transmit and receive beamformers are jointly optimized using successive semidefinite programming. The weighted sum rate maximization is approached via the corresponding mean-squared error minimization problem, which is further approximated to formulate a sequence of convex subproblems. The algorithm is shown to provide a monotonic improvement of the system performance objective in each iteration. Furthermore, the algorithm is shown to have an improved rate of convergence when compared to recently published weighted mean-squared error minimization algorithms with alternating transceiver design.