Reinforcement Learning Based Energy Efficient Underwater Localization

Underwater localization with limited bandwidth and long propagation latency has to magnificently improve the localization accuracy and energy efficiency for applications such as resource reconnaissance and tactical monitoring. In this paper, we propose a reinforcement learning based energy efficient underwater localization scheme to improve the accuracy and energy efficiency, which depends on the two-way travel time of underwater acoustic signal and uses reinforcement learning to optimize the beacon selection policy without relying on the channel model between the beacon and the target. Simulation results based on the depth-dependent sound speed profile model verify the efficiency of the proposed scheme and provide the performance over existing time-based localization scheme.