Decentralized beamforming for weighted sum rate maximization with rate constraints

A novel decentralized beamformer design for weighted sum rate maximization with user specific rate constraints is proposed for interfering multiple-input multiple-output (MIMO) broadcast channel (BC). In order to reduce the inherent complexity of the rate maximizing beamformer design, alternating receive and transmit beamformer generation together with successive convex approximation are applied. The rate maximization problem is reduced to mean-squared error (MSE) minimization with spatial data stream specific weighting factors that imply the inter-stream interference costs, where the weights depend additionally on the violation of the rate constraints. The iterative process is based on the Lagrangian relaxation of the rate constraints, and, thus, each intermediate beamformer solution is not required to be strictly feasible in terms of the rate constraints but rather steers the algorithm towards a solution with feasible rate constraints. Exchange of the weighting factors among the adjacent cells provides decentralized problem formulation for TDD based systems with a very low over-the-air signaling overhead.