Enhanced Wireless Sensor Network Lifetime with Optimized Multi-hop Clustering in LEACH Protocol Using WOA and Reinforcement Learning

Abstract

Wireless Sensor Networks (WSNs) are critically constrained by limited energy resources and finite node lifetimes, necessitating highly efficient clustering and routing mechanisms. This paper proposes and evaluates two independent optimization frameworks integrated with the multi-hop Adaptive Multi-hop Dynamic Clustering (AMDC) protocol built upon the Low-Energy Adaptive Clustering Hierarchy (LEACH) foundation. The first framework employs the Whale Optimization Algorithm (WOA) for intelligent Cluster Head (CH) selection by jointly minimizing residual energy consumption, intra-cluster distance, and CH-to-Base Station (BS) distance through a weighted multi-objective fitness function. The second framework applies Q-learning-based Reinforcement Learning (RL) to enable adaptive, experience-driven routing decisions that respond dynamically to time-varying network topology and energy conditions. Extensive MATLAB simulations conducted over a 200 m × 200 m and 500 m × 500 m deployment area with 100 heterogeneous sensor nodes demonstrate that WOA-AMDC achieves a 61% improvement in First Dead Node (FDN) round and RL-AMDC achieves an 84% improvement compared to conventional LEACH. Additionally, both methods exhibit superior Packet Delivery Ratio (PDR), throughput, and energy efficiency over baseline and state-of-the-art protocols. The results substantiate the effectiveness of metaheuristic and machine learning paradigms in solving the NP-hard CH selection and routing optimization problems in large-scale WSNs.