Double Q Learning Multi-Agent Routing Method for Maritime Search and Rescue

Maritime search and rescue missions rely heavily on effective communication networks to ensure timely and accurate response. However, traditional network routing cannot meet the communication requirements of large networks, especially in the context of maritime search and rescue missions. To address this challenge, this paper proposes a double Q-learning routing protocol to meet the requirements and overcome the positive bias estimation problem of the Q-learning based routing protocols. DQR uses a novel Reinforcement Learning (RL) model to reduce computational and communication overheads. It is combined with an effective trust management system to ensure a reliable data transfer and defeat packet dropping attacks. The experimental results demonstrate robust performance under various attacks with minimal resource footprint and efficient energy consumption.