A novel convex approximation method to weighted sum outage rate maximization for multicell network

In this paper, we consider the transmit beamforming problem for multicell network, whereby only the channel distribution information (CDI) is available to the base stations (BSs). Our goal is to maximize the weighted sum outage rate subject to probability constrains on the rate outage. This problem is non-convex, and very difficult to solve. To address this problem, Semidefinite Relaxation (SDR) based methods are popular in the existing literatures. However, we note that SDR-based methods might incur considerable performance loss, especially when the total number of users is large, which is due to that optimal SDR solution tends to be high-rank, rather than rank one matrices when the total number of users is large. To circumvent this shortcoming, we first cast the original problem into the framework of difference of convex (DC) program, then a novel iterative algorithm based on first-order approximation is employed to solve the resulting DC program. Simulation results show that the proposed algorithm exhibits noteworthy performance improvement over the SDR-base methods, especially when the total number of users is large.