Answers are grounded in this paper and cite their sources.
Wireless sensor networks (WSNs) are constrained by limited node energy, making balanced cluster‐head (CH) selection essential for extending network lifetime and maintaining reliable data delivery. This paper proposes an energy‐aware hierarchical clustering framework that integrates game theory (GT) and reinforcement learning (RL) for adaptive CH selection. The GT component employs a normalised utility function incorporating residual energy, local connectivity, communication distance and expected cluster load, thereby enabling decentralised and dimensionally consistent clustering decisions. The RL component uses a compact discrete state representation, executable node‐level actions, a transition‐based scalar reward and off‐policy Q‐learning to optimise long‐term energy balance, connectivity, delay and packet‐delivery performance. Four configurations—full GT, full RL, GT + RL and RL + GT—are evaluated under identical simulation conditions. The proposed GT + RL framework is additionally compared with DEEC, LEACH, Q‐learning‐based routing and Dijkstra‐based shortest‐path routing using 30 paired independent runs. Among the internal configurations, full RL achieves the highest final success rate and the lowest residual‐energy variance, whereas the hybrid approaches provide a favourable balance between structured decision‐making and adaptability. In the external comparison, GT + RL achieves mean first‐node‐death, half‐node‐death and last‐node‐death values of 35.1, 63.9 and 91.1 rounds, respectively, corresponding to improvements of approximately 27.6%, 12.3% and 7.3% over the Q‐learning routing baseline. These results demonstrate that combining utility‐driven clustering with adaptive learning can improve energy distribution, delay premature node failure and enhance WSN stability.
No takes yet. Share an insight, caveat, or question.
Eskandarpour et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: