Utility-based power control with QoS support

Abstract

Listen

Translate article

This paper addresses the issue of incorporating QoS requirements expressed in terms of some minimum SIR targets into the traditional utility-based power control problem. As suitable projection methods seem to be not amenable to distributed implementation, we first focus on a primal-dual algorithm to solve the utility-based power control problem subject to the SIR requirements. We prove a global convergence of the algorithm for a large class of utility functions and show that it can be implemented in a distributed wireless environment. However, the approach has an important drawback: An optimal solution may not exist as the SIR targets may be infeasible due to, for instance, channel effects. This motivates a reformulation of the problem so that an optimal solution always exists. We consider the possibility of using a barrier method to closely approach the desired SIRs of the users and combine this approach with the conventional utility-based power control problem to incorporate best effort users. We prove relevant properties of optimal solutions and propose a distributed recursive algorithm with global convergence. Finally, the performance of the proposed approaches is verified by simulations.