Weighted sum rate maximization in MIMO interfering broadcast channels exploiting virtual per-user power constraints

In this paper, we consider the weighted sum rate (WSR) maximization problem in a partially coordinated multiple-input multiple-output multicell broadcast channel with constraints on base stations' transmit powers where the communication is based on dirty paper coding. To be able to obtain a computationally efficient solution for this non-convex optimization problem, we solve it by a fast algorithm for WSR maximization under per-user power constraints. The main idea is to define virtual per-user power constraints that add up to the per-base station power budgets in each cell and optimize the per-user power constraints to maximize the WSR. We propose two computationally efficient power allocation algorithms to find the per-user power constraints, namely, the waterfilling-based power update and the gradient descent-based power update algorithms. In the latter one, we find a closed-form expression for the derivative of the maximum of WSR wrt the per-user power constraints resulting in a considerable reduction in the complexity. The resulting algorithms are computationally efficient, and our simulation results show that they achieve significantly higher bit rates than the current algorithms at the same signal-to-noise power ratio.