Multi-Agent Reinforcement Learning Based Channel Access Scheme for Underwater Optical Wireless Communication Networks

In this paper, we investigate the channel access problem in underwater optical wireless communication (UOWC) networks, wherein multiple transmitters randomly select receivers to send data. The objective is to find a joint multi-transmitter strategy that maximizes the network throughput while guaranteeing the reliability of the data transmission. Considering that the underwater channel conditions are dynamically changing, we propose a multi-agent Deep-Q network (DQN) algorithm based on the reinforcement learning. In this design, the transmitters, each acting as an agent, simultaneously interact with the communication environment. These agents receive observations and evaluate the reward, then learn to choose a receiver using the gained experiences. Simulation results reveal the effectiveness of the proposed scheme, and demonstrate that the multiple agents can learn to generate appropriate access strategies to improve the total transmission capacity while retaining the reliability performances.