A Trust-Aware Hybrid Q-AHA Energy-Efficient secure Clustering for Wireless Sensor Networks with a Mobile Sink Node

Abstract Mobile Wireless Sensor Networks (MWSNs) have important function in dynamic data collecting, but suffer from severe limitations, such as limited energy, topology instability, and susceptibility to malicious routing attacks. Most traditional clustering approaches regard energy efficiency and security as independent objectives, which may cause high energy consumption and low data dependability. In this study, to overcome these challenges, we propose a Trust-Aware Unified Hybrid Q-Learning and Artificial Hummingbird Algorithm (Q-AHA) clustering algorithm for MWSNs with mobile sink. At the local level, distributed Q-learning agents evaluate the quality of sensor nodes based on their residual energy, mobility, distance from the sink, and a dynamic Beta-reputation-based trust metric. At the global level, the Artificial Hummingbird Algorithm (AHA) is used in the mobile sink to optimize the CH selection based on the learnt node quality scores. Furthermore, an event-triggered control technique is proposed to avoid superfluous control packet broadcasts. Under the same simulation environment, the performance of Q-AHA is tested and compared with RL-PSO, LEACH, LEACH-RLC and M-TAAHA with node numbers 50, 100 and 200 with 10% malicious-node condition. It achieved LND values of 3720, 3723, and 3727 rounds, PDR values of 93.55%, 94.86%, and 95.44%, throughputs of 13.2045, 23.9824, and 45.7374 kbps, and delays of 9.55, 10.1543, and 11.4244 ms, respectively. Q-AHA has good trade-offs between network lifetime, reliability, throughput and delay, showing that it is suitable for application to scalable, energy-efficient MWSNs.