Fuzzy Rule Generation Using Modified PSO for Clustering in Wireless Sensor Networks

Clustering is one of the popular methods for improving energy efficiency in wireless sensor networks. In most of the existing fuzzy approaches, the CHs are selected first, and then clusters are generated, but this may lead to uneven distribution of the sensor nodes in the clusters. In this article, the clusters are generated using the famous Fuzzy C-means (FCM) algorithm and the Cluster Head (CH) from each cluster is selected using the Sugeno fuzzy system. FCM generates load-balanced clusters and the proposed approach named SF-MPSO selects the suitable CH from each cluster. The local information of the sensor node such as residual energy, its distance from cluster centroid and the distance from the BS is provided to SF-MPSO. In the existing algorithms, the fuzzy rules are manually designed, whereas, in this article, the modified Particle Swarm Optimization (PSO) algorithm is applied to generate optimum Sugeno fuzzy rules. A novel fitness function is designed to identify the effectiveness of the generated solution. The simulations are performed under three scenarios where SF-MPSO outperforms existing EAUCF, DUCF, FGWO and ARSH-FATI-CHS when evaluated under the parameters such as energy consumption and network lifetime.