LAUSR

This study addresses the transceiver design method for multiple-input multiple-output multisource multidestination amplify-and-forward relay systems. The source, relay, and destination filters are designed to minimize the sum mean-squared-error under the per-antenna power constraints at the source and relay nodes. The joint optimization is challenging due to its non-convexity and multiple power constraints. To resolve these difficulties, we propose a source and relay filter design method and present an alternating algorithm that is based on the block-coordinate descent method. Specifically, by introducing transmit-centric modified MSEs and reformulating the problem, the non-convex problem is transformed into tractable forms with multiple power constraints. The multiple power constraints are then adjusted by semidefinite programming of the source filters and a 1-D line search method for the relay transceiver. Simulation results demonstrate the effectiveness of the proposed schemes under the per-antenna power constraints that are compared with the conventional method under the sum power constraints.