Mitigating jamming attacks in underwater sensor networks using M-Qubed-based opportunistic routing protocol

Abstract

Routing in underwater sensor networks (UWSNs) is highly challenging because of harsh underwater conditions, such as deep water, high pressure, and rapid ocean currents. Furthermore, UWSNs are vulnerable to jamming attacks because of their limited bandwidth and battery capacity. Advancements in machine learning enable numerous routing methods to address these problems. Accordingly, we propose a novel max or minimax Q-learning (M-Qubed)-based opportunistic routing method for UWSNs. The method uses an opportunistic routing protocol, in which nodes dynamically select the next relay node by considering the status of their neighbors. Moreover, M-Qubed can maximize the benefits for both players in a two-player repeated game through reinforcement learning. Hence, it can reduce the energy loss caused by jamming attacks during routing, thereby increasing the routing efficiency in UWSNs. Simulation results reveal that the proposed routing scheme is less affected by jamming attacks than existing state-of-the-art routing methods. In addition, it can balance energy consumption across the nodes in a UWSN.