A Data-Gathering Underwater Medium Access Control Scheme Using Carrier Sensing Associated Machine Learning

In this paper, we propose a medium access control method in underwater sensor networks aimed at gathering data from multiple sensor nodes to a sink node. Considering the long propagation delay of underwater acoustic channels, exchanging control packets between nodes is inefficient. In our work, sensor nodes are trained through a machine learning and determine the correct timing for data transmission to avoid possible collisions without exchanging control packets. To learn varying channel conditions, sensor nodes employ a carrier sensing. Simulation results reveal that among machine learning models, the proposed scheme utilizing the MLP model exhibits the most outstanding performance.