Evolutionary Optimisation of Energy-Efficient Communication in Wireless Sensor Networks

Many real-world problems can be efficiently optimised using a multi-objective function—as these problems are simultaneously improved using multiple objectives, which most often preclude each other. A single-objective function incorporating all information required to solve the problem appears appropriate, but not without the penalties of slow convergence and difficulty in obtaining the best fitness function. This paper therefore implements a hybrid evolutionary system that minimises these penalties. We conscript two distance fitness functions, to improve communication distance between sensor nodes and cluster heads (CHs), and between CHs and the sink or base station. These functions are then mainstreamed into a globally defined fitness function using genetic algorithm (GA). Important parameters established by the GA topology are then preserved to serve a variety of modified particle swarm optimisation (PSO) models, to discover how suitable they reshape the optimisation process. Simulation results revealed the robustness of our proposed hybrid framework, as the framework enabled consistent coverage clustering topology. The GA multi-objective fitness function could maintain good genetic diversity and genealogy across the population generations, as the clustered topology of the sensor network presented a stable structure such that mobile sensor nodes do not unnecessarily exceed the global boundary. The PSO-fitness function guaranteed that particles maintained the shortest possible distance within the (population) cluster space. Furthermore, the modified PSO with Time Varying Inertia Weight and Constriction factor (PSO-TVIW–C) achieved tremendous improvements in the overall performance and is effective in solving optimisation problems of distance minimisation in wireless sensor networks (WSNs).