Convex Approximation-Based Joint Power and Admission Control for Cognitive Underlay Networks

Abstract

The joint power and admission control problem is considered for a cognitive underlay scenario, under Quality of Service (QoS) constraints for both primary and secondary users. Primary users must be guaranteed a premium service rate, measured by their signal to interference plus noise ratio (SINR); secondary users, if admitted, should be provided with at least a basic service rate, and the number of admitted secondary users should be maximized. The problem is NP-hard, but drawing upon recent results for joint beamforming and admission control for the cellular downlink, an efficient convex approximation algorithm is derived that yields close to optimal results at affordable complexity. The key step is a reformulation of the joint problem that naturally leads to a linear programming relaxation. Simulation results are included to illustrate the merits of the approach. Two scenarios are considered: with or without a primary user. In the latter, several good heuristic algorithms are available in the literature, and the prevalent one is used as a baseline. A brute- force enumeration algorithm is used in both cases to assess the gap to the optimal solution.