MAB-Learning-Based Hierarchical Defense for Secure Underwater Acoustic Networks

The underwater acoustic network (UAN) without defensive support is unreliable in an increasingly competitive marine environment. In this paper, a hierarchical defense algorithm that applies multi-armed bandit (MAB) learning is presented to combat the intelligent mobile attackers and protect the underwater relay transmissions. Specifically, the identities of the significant nodes of a network are changed periodically to hide the critical routing paths. The learning-based communication system can achieve stable links through the optimal spoofing scheme that is to mislead the attackers sensitive to the environmental feedback, and then alleviate the potential threat from invaders. Simulation results verify that the proposed defense strategy can fast reduce the outage probability of the protected objects and prolong the lifetime of underwater networks.