Machine Learning for Self-Adaptive Internet of Underwater Things

Abstract

Internet of Underwater Things (IoUTs) has gained increased momentum thanks to the advancements in underwater nodes, sensing, and communication technologies. This novel paradigm has tremendous potential to empower smart ocean applications. However, the harsh and dynamic nature of the underwater environment and underwater communication, the stringent requirements of underwater applications, and the difficulty and cost for IoUT management and maintenance have limited the development and application of IoUTs. In this regard, machine learning has been proposed to create self-adaptive IoUTs and boost the performance of smart oceans applications. In this paper, we shed light on the design of machine learning models for the on-the-fly intelligent and autonomous management of IoUT networking parameters and configurations aimed at boosting data delivery. We discuss the recent proposals for IoUT network management and how machine learning algorithms can improve such solutions at different networking layers. Finally, we point out some future research directions in need of further attention.