Performance Assessment of Cognitive Radio Adaptation Engine based on Real-coded Genetic Algorithms

Abstract

Adaptation is one of the fundamental functionalities of Cognitive Radio Systems (CRS). Adaptation refers to theability of the radio to adapt its operating parameters in response to varying stimuli.Choice of the best parameter set of the radio to achieve certain objectives in shortest time possible remains one of the most challenging tasks in Cognitive Radio (CR) research.One possible approach to adaptation engine design is based on utilizing Genetic Algorithms (GA) which invoke a combination of exploration and exploitation processes to perform random and directed searches for semi-optimal solutions in the possible solution space. However, conventional Binary-coded Genetic Algorithms (BGA) based adaptation engines used frequently in CR research,are criticized for their slow convergence and response times. Accordingly, Real-coded Genetic Algorithms (RGA) -a specific type of GA -have been implemented in our work, to address this problem. RGA alleviates many of the disadvantages of conventional BGA based implementations. This paper focuses on RGA based adaptation engine implementations' performance assessment compared to conventional BGA based implementations. Performance assessment results indicate that RGA based implementation does demonstrate a superior performance over BGA based implementations; in achieving the best configuration to minimize the link BER with minimum possible transmitted EIRP levels; in the shortest time possible.