Reinforcement Learning-Based ALOHA-Q Protocol with Slot Subdivision for Underwater Acoustic Sensor Networks

This paper presents SS-ALOHA-Q, a Reinforcement Learning-based MAC protocol that builds upon slotted ALOHA by introducing slot subdivision to effectively mitigate the impact of large propagation delays in Underwater Acoustic Sensor Networks (UASNs). The protocol leverages reinforcement learning to dynamically optimize channel access, enabling adaptive scheduling and responsive decision-making in highly variable underwater environments. Performance evaluation was conducted using the DESERT simulator, where a Hidden Markov Model (HMM) was employed to model underwater acoustic channel dynamics. Comparative analysis against benchmark MAC protocols demonstrates that SS-UW-ALOHA-Q significantly improves channel utilization while maintaining energy efficiency, demonstrating its potential for robust and efficient UASN operations.