DYNAMIC CLUSTER HEAD SELECTION USING VARIOUS OPTIMIZATION ALGORITHMS IN INDUSTRIAL IOT AND WSN ENVIRONMENTS

Abstract

Selection of cluster heads (CHs) impacts both the performance and energy efficiency of the Internet of Things (IoT)-based Wireless Sensor Networks (WSNs) systems. Recently, metaheuristic algorithms such as Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), and Genetic Algorithm (GA) have been applied for dynamic CH selection. Despite being an improvement over static alternatives, these methods have drawbacks include poor exploration, rapid convergence, and an uneven energy allocation among the nodes. This research focuses on developing two new algorithms Sand Cat Swarm Optimization (SCSO) and Moth Flame Optimization (MFO) based on swarm intelligence for dynamic cluster head selection in WSNs deployed with IoT devices. The SCSO algorithm outperforms traditional approaches by emulating the hunting strategy of sand cats which maintains the balance between exploration and exploitation of sensor nodes, and thereby uniform energy consumption. Dynamic CH selection done by MFO algorithm uses adaptive flame control inspired from the way moths navigate around light sources. Through extensive simulations, both SCSO and MFO demonstrate superior performance compared to existing algorithms on important metrics such as the network lifetime, energy consumption, and the ratio of delivered packets. Furthermore, SCSO-based methods brought accuracy to 96%, demonstrating substantial improvements for prolonging operation while balancing node energy. Besides this, the MFO approaches stand out for attaining exceptional accuracy and outperforming other methods with a notable 97.50% due to their global search prowess and convergence speed. Evaluation shows that both methods improve cluster stability and communication overhead within ACO, PSO, and GA. To conclude, the application of sophisticated techniques SCSO and MFO further opens prospects toward effective and adaptable CH selection in WSNs, leading to stronger and more energy-efficient network architectures. These results will be further enhanced through dynamic changes of hybrid models to optimize network parameters.