Sink Attraction Q-Learning Routing Algorithm For UWSNs
Zhi Zhang, Yibing Li, Jialiang Gao, Fang Ye · 2024
In order to optimize the problems of routing detours, routing voids, and redundant transmission in static UWSNs data collection, this paper proposes the Sink Attraction Q-Learning (SAQL) Routing Algorithm for UWSNs. The proposed algorithm is trained to determine the optimal routing strategy based on the Q-Learning model. To optimize the balance between energy consumption and shortest forwarding paths of the forwarding policy to prevent nodes from prematurely decaying or forwarding into a dead loop, the SAQL algorithm proposes the sink attraction mechanism. This mechanism is designed based on the fixed V-value of sink nodes and the utilization of sink nodes’ coordinates to direct the training process of the Q-Learning model, thereby preventing excessive routing detours and voids. The efficacy of this approach has been rigorously tested and validated through simulations conducted on the NS3 network emulator. The outcomes from these simulations indicate that the proposed method enhances the packet delivery rates, reduces end-to-end delays, and optimizes the energy consumption during data forwarding.