Optimizing Underwater Acoustic Communication with Game Theory and Machine Learning

This paper addresses the challenge of resource allocation in multi-user Underwater acoustic (UWA) communication systems, focusing on cooperative power allocation for Orthogonal Frequency Division Multiplexing (OFDM) systems operating in interference-prone environments. The objective is to optimize overall system utility while ensuring fair power distribution. The resource allocation problem is modeled as a Nash Bargaining (NB) cooperative game, integrating Reinforcement Learning (RL) to enable dynamic and adaptive resource management. This hybrid approach balances cooperative fairness with autonomous learning, overcoming the limitations of traditional methods, particularly in high-interference conditions. Simulation results demonstrate the algorithm's effectiveness across different user scenarios. The proposed approach significantly reduces interference, enhances spectral efficiency, and ensures equitable resource allocation, outperforming non-cooperative strategies.