Reinforcement Learning Based Relay Selection for Underwater Acoustic Cooperative Networks

In the complex and dynamically varying underwater acoustic (UWA) channel, cooperative communication can improve throughput for UWA sensor networks. this paper, we design a reasonable relay selection strategy efficient cooperation with reinforcement learning (RL), considering characteristics of channel variation long transmission delay. The proposed scheme establishes effective state reward expression to better reveal relationship between RL environment. Meanwhile, simulated annealing (SA) algorithm is integrated performance selection, where exploration rate adapted by SA optimization through temperature decline rate. Furthermore, fast (FRL) pre-training process practical network implementation. whole SA-FRL has been evaluated both simulation experimental data. results show that converge more quickly than classical random increase number iterations. reward, access delay data at highest value are close ideal optimum value. All in all, efficiency nodes high link quality low