Enhanced genetic algorithm for energy efficient dynamic ad hoc wireless sensor networks

The paper proposes a new clustering approach based on genetic algorithms (GA) and devoted to improving energy efficiency in ad-hoc wireless sensor networks (WSN). A special designed MATLAB framework operates as test bench to evaluate different implementations. The solutions of the optimization algorithms provide the number of clusters along with the cluster structures. Real-time implementations of such algorithms justify the necessity to minimize their execution time. We have devised custom genetic operators in order to improve the GA convergence. The process of fine tuning of the GA parameters proved to be also extremely important. Intensive simulation studies have confirmed the validity and efficiency of the proposed solutions. Using our approach, we have improved the speed of convergence for the GA up to 50%, as compared with existing approaches, while reducing considerably the minimum communication distance in the ad-hoc WSN.