A secure routing protocol for underwater acoustic sensor networks using reinforcement learning

Underwater acoustic sensor networks are essential for underwater environment surveillance and monitoring and offshore exploration. Underwater acoustic sensor network experience challenges because of the hostile underwater environment, including bandwidth limitation, node mobility, propagation high propagation delays and security threats. Reinforcement Learning (RL) is a branch of machine learning where an agent learns to make decisions by interacting with an environment and receiving feedback in the form of rewards or penalties. The importance of reinforcement learning lies in its ability to handle complex decision-making problems where explicit supervision is difficult or impossible. This paper proposes a novel Reinforcement Learning-based Secured Routing Protocol (RL-SRP) for underwater acoustic sensor network. The proposed protocol integrates Q-learning with a trust management system to dynamically select secure and energy efficient routes while mitigating common attacks which consist of blackhole attack. Simulation results indicates that RL-SRP significantly improves packet delivery ratio, reduces end-to-end delay, and enhances network security and energy efficiency compared to existing routing protocols DBR and AODV.