Energy-efficient clustering and mobile sink routing in wireless sensor networks using quantum-inspired optimization and deep learning

Abstract

Wireless sensor networks (WSNs) require energy-efficient clustering and adaptive routing to maintain reliable communication and prolong network lifetime under varying node density, initial energy, communication range, packet size, residual energy, routing distance, latency, and mobile-sink conditions. However, conventional clustering and routing methods generally optimize these aspects separately, resulting in inefficient energy distribution, increased routing overhead, unstable sink mobility, and reduced network sustainability. To address these limitations, this work proposes an integrated framework for energy-efficient clustering and adaptive mobile-sink routing. The framework employs Quantum-based Beetle Swarm Optimization (QBSO) for exploratory cluster-head (CH) selection, Animated Oat Optimization (AO) for cluster refinement and energy balancing, a Quantum Convolutional Neural Network (QCNN) for spatial-energy feature extraction, a Deep Quantum Transformer Network (DQN) for adaptive sink-mobility and routing decisions, and Supercell Thunderstorm Algorithm (ScTsA) for QCNN-DQN parameter optimization. The ScTsA optimization evaluates candidate parameter configurations using packet delivery, energy consumption, routing distance, latency, and sink-mobility smoothness and refines them through spiral rotation, tornado perturbation, and jet-stream alignment. Simulation results demonstrate a 98.4% packet delivery ratio (PDR), 14 units of routing overhead, 22 J energy consumption, 1380 s network lifetime, 3.6 Mbps throughput, 8 ms jitter, 28% buffer occupancy, 180 units of computational complexity, and 0.52 J average residual energy. The component analysis further shows that the complete framework achieves 1.10% packet loss ratio (PLR), 62 ms latency, 0.26% bit error rate (BER), and 86% energy efficiency, compared with 2.4% PLR, 78 ms latency, 0.48% BER, and 82% energy efficiency without ScTsA. Thus, the proposed strategy reduces PLR by 54.17%, latency by 20.51%, and BER by 45.83%, while improving energy efficiency by 4.88%, demonstrating its effectiveness for reliable, energy-conscious, and adaptive WSN communication.