MIMO Beamforming Design towards Maximizing Mutual Information in Wireless Sensor Network

This paper considers joint beamformer design towards maximizing the mutual information in a coherent wireless sensor network with noisy observation and multiple antennae. Leveraging the weighted minimum mean square error and block coordinate ascent (BCA) framework, we propose two new and efficient methods: batch-mode BCA and cyclic multi-block BCA. The existing batch-mode approaches require stringent conditions such as diagonal channel matrices and positive definite second-order matrices, and are therefore inapplicable to our problem. Our match-mode BCA overcomes the previous limitations via a general second-order cone programming formation, and exhibits a strong convergence property which we have rigorously proven. The existing multi-block approaches rely on numerical solvers to handle the subproblems and some render good performance only at high signal-to-noise ratios. Exploiting the convexity of the trust-region subproblem for the convex case, our multi-block BCA significantly reduces the complexity and enhances the previous results by providing an analytical expression for the energy-preserving optimal solution. Analysis and simulations confirm the advantages of the proposed methods.