Fuzzy Clustering and Routing Protocol With Rules Tuned by Improved Particle Swarm Optimization for Wireless Sensor Networks

Fuzzy clustering and routing protocols have been proven to improve energy efficiency, extend network scalability, increase network throughput, balance network load as well as prolong network lifetime. However, rules defined manually according to field experts are impossible or impractical to achieve the optimal solution for a Fuzzy Inference System (FIS). Therefore, a Novel Fuzzy Clustering and Routing Protocol called NFCRP is proposed in this paper by using an improved Particle Swarm Optimization (PSO) algorithm to tune the fuzzy rules. Firstly, one FIS is used to complete clustering based on effective input parameters including residual energy, node degree deviation, and distance to centrality, thereby forming optimal clusters and minimizing the intra-cluster energy consumption. Secondly, the other FIS is adopted to perform routing with descriptors residual energy, distance to BS, and data load deviation, hence addressing the inter-cluster energy consumption. Finally, the rules of both FISs are tuned by an improved PSO algorithm whose parameters are updated by introducing chaotic mapping and adaptive inertia weight. Simulation experiments were conducted to verify the performance of NFCRP against LEACH, EFUCA, EEFUC, FBCR and FMSFLA. According to the results, the average network lifetime of NFCRP increased by 79.59%, 47.99%, 50.35%,15.66 and 13.04%, compared to LEACH, EEFUC, EFUCA, FBCR and FMSFLA. For the average standard deviation of CH’s traffic load, NFCRP decreased it by 29.29% over EEFUC, 31.42% over EFUCA, and 25.28% over FMSFLA. For network throughput, NFCRP outperformed LEACH, EEFUC, EFUCA, FBCR and FMSFLA by 16.87%, 46.52%, 48.18%, 29.97 and 71.79%. In addition, NFCRP also reduced energy consumption by 53.95%, 23.76%, 38.72%, 15.71 and 27.18% as compared to LEACH, EEFUC, EFUCA, FBCR and FMSFLA, respectively