Decentralized multi-cell beamforming with base station cooperation

This paper presents a design strategy for base station (BS) coordinated multi cell joint beamforming. Our objective is to maximize the average cell throughput while maintaining quality of service (QoS) for individual users. The QoS in our analysis relates to minimum signal to interference plus noise ratio (SINR) achieved by every user. Since, throughput maximization is a non convex problem, an optimal solution can not be obtained using polynomial order algorithms. However, it can be observed that existence of inter cell interference (ICI) terms are the cause of non convexity. With this knowledge, we propose two decentralized approaches based on primal and dual decomposition of the throughput maximization problem. As ICI terms are the shared variables between the base stations (BSs), these terms are exchanged between the neighboring BSs and kept fixed during an iteration. This fixes the non convex (ICI) terms, resulting in convex subproblems. We demonstrate the quality of our sub optimal decentralized solution by simulations and comparing it with optimal exhaustive search algorithms (Genie-aided algorithm). A comparison analysis is also presented for both primal and dual solutions.