Intelligent Clustering in Wireless Sensor Networks

Abstract

Wireless Sensor Networks (WSNs) are limited by battery energy, making energy effciency and network lifetime critical challenges. Clustering helps reduce communication overhead and balance energy consumption through effcient Cluster Head (CH) management. This work proposes PUMA-GRID, an energy-effcient protocol that combines the Puma Optimization Algorithm (PUMA) for adaptive CH selection with a lightweight grid based multi-hop routing strategy inspired by k-nearest neighbor (k-NN) logic. During clustering, PUMA selects optimal CHs using a weighted fitness function based on residual energy, distance between nodes and CHs, and distance between CHs and the Base Station (BS), ensuring balanced energy usage and stable cluster formation. For routing, the sensing field is divided into grid cells where neighboring CHs are selected as relays using a k-NN-inspired mechanism to construct shorter and more energy effcient communication paths while reducing long range transmissions and communication overhead. The protocol was evaluated under three BS placement scenarios: central, edge, and external. For each scenario, optimal fitness function weights were first selected before conducting MATLAB simulations with randomly deployed nodes in a 200 × 200 m² sensing field containing up to 600 nodes. PUMA-GRID was then compared with LEACH, Atomic Energy Optimization-based approaches. Simulation results show that PUMA-GRID achieves better network lifetime, residual energy preservation, packet delivery, and communication efficiency, with approximately 35–50% improvement in network lifetime. This improvement is achieved through the combination of adaptive PUMA based CH optimization and effcient kNN inspired routing, which together reduce transmission cost and balance energy consumption across the network. Finally, the framework is mainly designed for static and homogeneous WSN environments, leaving opportunities for future improvements under more realistic conditions.