Performance of GA in Power Allocation for Underlay Cognitive Radio Systems

Abstract

Power allocation plays a significant role in performance enhancement of cognitive radio (CR) system as it needs to not only limit the interference to primary users (PUs) but also maintains a minimum quality-of-service (QoS) requirements of secondary users (SUs) with a limited available resource. In many cases, traditional analytical optimization methods require a specific structure of objective function and constraints, hence, may not be applied directly to power allocation and also demand high computational complexity. This paper investigates the effectiveness of classical genetic algorithm (GA) in power distribution of sensing-free, location-aware underlay spectrum sharing based CR system models. GA has been implemented to optimize the various performance parameters of the system models. The performance of GA has been validated and compared with that of interior point method (IPM) and successive approximation solved by CVX toolbox. Simulation results depict that GA reaches closer to optimal value with a reduced computational time.