Reinforcement Learning-Based Opportunistic Routing Protocol Using Depth Information for Energy-Efficient Underwater Wireless Sensor Networks

An efficient routing protocol is critical for the data transmission of underwater wireless sensor networks (UWSNs). Aiming to the problem of void region in UWSNs, this article proposes a reinforcement learning-based opportunistic routing protocol (DROR). By considering the limited energy and underwater environment, DROR is a receiver-based routing protocol, and combines reinforcement learning (RL) with opportunistic routing (OR) to ensure real-time performance of data transmission as well as energy efficiency. To achieve reliable transmission when encountering void regions, a void recovery mechanism is designed to enable packets to bypass void nodes and continue forwarding. Furthermore, a relative Q-based dynamic scheduling strategy is proposed to ensure that packets can efficiently forward along the global optimal routing path. Simulation results show that the proposed protocol performs well in terms of end-to-end delay, reliability, and energy efficiency in UWSNs.