Robust Resource Allocation

We propose a framework for robust power allocation based on worst-case interference functions, which are determined by a parameter-dependent coupling matrix. The power allocation is optimized with respect to the worst-case interference. One main contribution of this paper is to show that these worst-case interference functions have useful properties, which can be exploited for the design of an efficient iterative algorithm. The proposed iteration minimizes the total transmit power while the worst-case SINR of each user achieves a given target value. By analyzing the properties of the interference functions, we show monotonicity and super-linear global convergence, even though the original problem is non-convex. The iteration always requires less steps than an alternative fixed-point iteration, which has only linear convergence.