QAAR: An Application-Adaptive Routing Protocol Based on Q-Learning in Underwater Sensor Networks

Underwater Wireless Sensor Networks (UWSNs) are promising for exploring ocean resources, which have attracted much attention from academia and industry in recent years. However, routing design for various underwater applications is difficult because a single route cannot meet specific requirements of each scenario, such as latency, throughput and network lifetime in UWSNs. In this paper, we propose an Application-Adaptive Routing protocol based on Q-Learning, called QAAR, to solve the difficulties above. Reward function of Q-Learning with consideration of residual energy, energy distribution, distance, and density that affect network performance when nodes choose next-hop forwarder, and employs Analytic Hierarchy Process (AHP) model to measure weights of parameters for our Q-Learning-based protocol, which can be applicable to different underwater network applications. A lot of simulation results tested on Aqua-Psim platform show that the proposed routing protocol has advantages on end-to-end delay, energy consumption and delivery rate in different network applications.