Decentralized Robust Beamforming for Coordinated Multi-Cell MISO Networks

Abstract

In this letter, we consider a multi-cell multiuser MISO network. We aim to minimize the sum power over base stations (BSs) while guaranteeing the worst case SINR for each user. We propose a decentralized robust beamforming design which relies only on local imperfect channel state information and limited backhaul signaling. First, the non-convex problem is approximated by a convex one via the semidefinite relaxation and S-Procedure methods. Then, we propose a primal decomposition method to equivalently turn the approximated problem into a network-level master problem and BS-level subproblems, which can be optimally solved using an iterative projected subgradient method and a convex optimization solver, respectively. The proposed algorithm is applicable when it yields a rank-one solution providing an optimal solution also for the original problem. Computational and backhaul signaling loads per iteration are reduced as compared with the existing algorithm.