Improving UWSN Performance Using Reinforcement Learning Algorithm QENDIP

The deployment of underwater wireless sensor networks (UWSN) in aquatic environments allows man to better expose and explore the riches and opportunities presented in these environments. At this regard, several researches and studies have been conducted with the aim of improving UWSN networks in terms of performance, efficiency and quality of service. Our research is included in the same paradigm that illustrates artificial intelligence's role in boosting network performance through optimization of energy consumption and network lifetime extension. In effect, this work essentially proposes a method called Q- Learning, based energy and distance related routing path (QENDIP), which finds the best route for each data forwarding from a randomly selected node sender to the sink node. QENDIP is an off-policy type of reinforcement learning technique that works well in stochastic processes. We assessed the performance in this study in terms of network lifetime, throughput, and energy consumption.