Skip to Main Content
In a multiuser interference channel, solving the optimal power and channel allocation for a weighted sum-rate maximization is a well-known non-convex problem, and has NP complexity. In this paper, we apply the recently developed deterministic channel model, and obtain a new formulation for this classic problem. Although the non-convex nature remains unavoidable, we exploit novel insights and techniques to significantly reduce the algorithm's complexity, while still guaranteeing its asymptotic optimality. For cellular structured networks with a fixed number of cells, our algorithm has a worst-case polynomial complexity. We provide simulation solutions of this non-convex optimization in a seven-cell network. The proposed algorithm also computes performance upper bounds in all simulation cases as a numerical verification of the solutions' optimality. The upper bounds demonstrate very small gaps from the maximum achieved objective values of the simulation solutions.
Information Theory and Applications Workshop (ITA), 2010
Date of Conference: Jan. 31 2010-Feb. 5 2010