Reinforcement learning-based approach for establishing energy-efficient routes in underwater sensor networks

Abstract

Underwater acoustic sensor networks find their applications in many areas including Environmental monitoring, Undersea explorations, Disaster prevention, Seismic monitoring, Assisted navigation, Mine reconnaissance, and many more. Many of the issues are addressed and resolved in underwater applications. One of the important issues to be addressed is routing. Routing is an essential task in all the networks. Finding the best path to send packets to the destination is of utmost importance. Routing in underwater networks is a difficult task due to invariant conditions of the underwater environment. Many of the algorithms have been designed to find the best path to the destination. In this paper, we propose a Reinforcement learning-based approach to establish the best path to the destination by considering the energy of the nodes and underwater environment. In RL based approach, a neighbor node is selected based on the underwater environment and the remaining energy of the nodes. The algorithm calculates the reward for every action and the best path is established based on total reward. Packets are then routed using the best path to the sink. The authors conclude RL based approach provides a better path to a destination by taking into consideration the energy of the nodes.