A distributed algorithm for cooperative relay beamforming

We consider the problem of cooperative beamforming in relay networks. Assuming knowledge of the second-order statistics of channel state information (CSI), the optimal beamforming weights are determined so that the total transmitted power at the relays is minimized, while meeting signal-to-interference-plus-noise-ratio (SINR) requirements at the destinations. Our formulation utilizes the Semidefinite Relaxation (SDR) technique to produce an efficient, tractable convex programming approximation of the problem in centralized form. We propose a distributed optimization algorithm to calculate the optimal beamforming weights for scenarios where multiple clusters of source-destination node pairs, along with their dedicated relays, coexist in space. Our method relies on dual decomposition techniques with regularization, that can significantly improve on the inherent disadvantages of simple dual subgradient methods. Numerical anaysis demonstrates that the proposed algorithm exhibits very fast convergence rates.