Integrating PAM Clustering and GTO-Based Cluster Head Selection in WSNs with Multi-Metric Optimization

Wireless Sensor Networks (WSNs) are a vital component of modern data-driven applications, ranging from environmental monitoring to industrial automation. Clustering plays a pivotal role in enhancing the efficiency and scalability of WSNs by organizing sensor nodes into clusters, where each cluster is led by a Cluster Head (CH). To optimize cluster formation and CH selection, the study proposes a novel approach that combines the Partitioning Around Medoids (PAM) clustering algorithm with the Giant Trevally Optimizer (GTO) for cluster head selection. Our approach goes beyond traditional clustering methods by incorporating a comprehensive set of metrics to guide the clustering and CH selection processes. These metrics encompass Distance-Based Metrics, which consider proximity between nodes; Energy Metrics, accounting for energy availability and consumption; Energy Efficiency Metrics, aimed at prolonging network lifetime; Load Metrics, to balance the node workload; and Data Correlation Metrics, which capture data dependencies among nodes. The proposed clustering protocol implemented using MATLAB and verified for different network deployment conditions. Results shows that the proposed protocol outperforms all existing protocol in terms of lifetime and packet delivery ratio.