Decentralized coordinated downlink beamforming for cognitive radio networks

Abstract

In this paper, we consider a decentralized downlink beamformer design problem in a multiantenna cognitive radio network where multiple primary and secondary transmitters serve multiple single antenna primary and secondary users, respectively. The design goal is to minimize the total transmitted power across the secondary transmitters subject to the constraints of the received SINR at each secondary user and the received interference power at each primary user.We propose two alternative decentralized approaches where iterative beamformer designs are based on primal and dual decomposition methods. Beamformers are calculated locally relying on the limited backhaul message passing between secondary transmitters. Since the original problem is convex, both decentralized approaches converge to the globally optimal solution for static users' channels. Moreover, a feasible set of beamformers are guaranteed at each iteration. The proposed beamformer designs apply directly to the case of hierarchical femtocell networks where multiple macro- and femtocells co-exist. Convergence behavior of the algorithms is demonstrated through simulations. In addition, sum power performance of the optimal decentralized beamforming is compared with zero-forcing beamforming in a quasi-static flat fading scenario.