On Rethinking Cognitive Access for Underwater Acoustic Communications

In this paper, we investigate how to reformulate the concepts of cognitive access, originally developed for radio communications, in the framework of underwater acoustic communications. A straightforward application of the classical energy-detection-based cognitive approach, such as the one employed for radio communications, would result in a reduced spectrum utilization in an acoustic scenario. Actually, in the underwater scenario, acoustic signals sensed by a network node are likely to be due to communication sources as well as natural/artificial acoustic sources (e.g., mammals, ship engines, and so forth), differently from classical cognitive radio access, where each signal at the receiver is generated by a communication source. To maximize the access probability for cognitive acoustic nodes, we focus on understanding the nature of sensed interference. Toward this aim, we try to discriminate among natural and communications sources by classifying the images representing the time and frequency features of the received signals, obtained by means of the Wigner-Ville transform. Two different classifiers are considered here. The first one is targeted on finding natural interference while the second one looks for communication. Simulation results show how the herein described approach drastically enhances the access probability in an acoustic scenario with respect to a direct rephrasing of classical cognitive access. A possible protocol for implementing cognitive access is also described and its performance evaluated.