Semantic-Based Integrated Sensing and Communication in the IoUT

Oceans and seas are today mostly unexplored. In the aim of supporting their exploration, the vision of an Internet of Underwater Things (IoUT) emerged recently, indeed targeted at providing a degree of pervasiveness and connectivity in our seas, similarly to what happens in traditional terrestrial networks. This foreseen pervasiveness should however not be detrimental, from a sustainability point of view, to the environment. Accordingly, in this paper we present a semantic-based integrated sensing and communication solution specifically designed for the IoUT. We consider an acoustic underwater network where smart sensor nodes perform image collection and analysis to detect rare marine species. Based on the analysis of the collected data, the smart devices, equipped with a software radio, set physical layer parameters such as transmission gain, payload length, and modulation scheme, so that they communicate relevant data but minimize their impact on rare marine creatures identified in the area. This implies reduction in the sea noise level which could be dangerous for mammals, such as whales or dolphins. AI techniques are employed to classify images collected through sensors and accordingly tune transmission parameters. Performance analysis assess the effectiveness of the approach.