An Efficient Cluster Head Selection Approach in MANET using Enhanced Elephant Herding Optimization based on Chaos Theory and the Simplex Method

Abstract

In Mobile Ad Hoc Networks (MANETs), choosing an efficient cluster-head (CH) is very important to ensure network stability, scalability and communication efficiency. However, existing clustering algorithms are prone to premature convergence, unstable cluster formation, high re-affiliation rates and high control-message overhead under dynamic network conditions. In order to solve these problems, this paper presents a Chaos-Simplex Elephant Herding Optimization (CS-EHO) algorithm for the cluster-head selection in MANETs. The proposed approach employs chaotic population initialization to increase the diversity of search and avoid the local optima, and the simplex-based local search to improve the exploitation capability and the convergence accuracy. The performance of the proposed CS-EHO is compared with GWOCA, EHO, HBA, MPSO, ABC and DGA using four clustering metrics, namely, average number of clusters, cluster lifetime, re-affiliation rate and control-message overhead, for different network sizes, node speeds and transmission ranges. The experimental results show that CS-EHO always outperforms all benchmark algorithms. For a network size of 500 nodes with a transmission range of 100 m, CS-EHO produced only 26 clusters in comparison with 32, 37, 54, 59, 65 and 102 clusters for GWOCA, EHO, HBA, MPSO, ABC and DGA respectively. CS-EHO achieved cluster lifetime of 79% and 80% for transmission ranges of 100 m, 200 m respectively with a low re-affiliation rate of 0.0372 and control overhead of 0.0032 under high mobility (80 km/h). The hybridization of chaotic exploration and simplex based exploitation significantly improves the cluster-head selection ability of EHO. The proposed CS-EHO framework gives rise to more stable clusters, reduces the cost of cluster maintenance, lowers the communication overhead and improves the scalability of the network. Therefore, CS-EHO is an effective and robust clustering solution for highly dynamic MANET scenarios.