Energy‐Efficient Cluster Head Selection and Multipath Routing in WSN Using Heterogeneous‐Hyper Graph Neural Network and Velocity Paused Particle Swarm Optimization

ABSTRACT Wireless sensor networks (WSNs) are extensively utilized in applications such as environmental tracking and event monitoring without human intervention. The continual sensing and energy depletion limit the lifespan of sensor nodes in WSNs, particularly in high‐traffic areas close to sink nodes, resulting in the energy‐hole problem. The major challenges are the efficient energy utilization and routing. To address these, this paper proposes an energy‐efficient cluster head selection with multipath routing using heterogeneous‐hyper graph neural network and velocity paused particle swarm optimization (EECHS‐HHGNN‐VPPSO). The method employs a heterogeneous‐hyper graph neural network (HHGNN) for intelligent cluster head (CH) selection by considering factors like node proximity, communication cost, residual energy, and coverage. Multipath routing is further optimized using velocity paused particle swarm optimization (VPPSO), which improves load balancing and minimizes redundant energy use during data transmission. The proposed technique significantly enhances network performance by achieving higher throughput, improved packet delivery ratio (PDR), reduced energy consumption, prolonged network lifetime, and lower node mortality. Experimental results demonstrate that the proposed EECHS‐HHGNN‐VPPSO achieves a throughput of 0.9 Mbps and energy consumption of just 0.4 mJ with 500 sensor nodes, outperforming existing methods in scalability, efficiency, and robustness for energy‐constrained WSN environments.