Investigative analysis of different mutation on diversity-driven multi-parent evolutionary algorithm and its application in area coverage optimization of WSN

Almost all evolutionary algorithms suffer from the problem of premature convergence and stagnation in local optima. An approach based on an evolutionary algorithm is presented in this work with different mutation schemes to address these issues. The mutation process used is an adaptive one which utilizes fitness variance and space aggregation concept. The mutation used in the technique is wavelet mutation, Levy flight, particle swarm optimization-based mutation, Chaotic, and non-uniform mutation. Levy flight is a random walk process which determines the step size based on Levy distribution, whereas chaotic and non-uniform mutation is based on logistic map and Gaussian distribution, respectively. In the wavelet mutation, Morlet wavelet is used as a mutation operator. The experimentation is carried out with each mutation strategy, and the results are obtained in terms of standard deviation and average. Also, the effectiveness of the proposed work is tested by performing a statistical analysis named Wilcoxon’s rank-sum test. The results from each mutation are compared with each other, and the results from the best mutation method are further compared with other optimization techniques. Moreover, the best strategy, i.e. DDMPEA with the chaotic mutation, is applied to the area coverage optimization problem of WSN.