Adaptive Authentication Method for Underwater Acoustic Sensor Networks Based on Time Varying Distribution Embedding

Underwater acoustic sensor networks (UASNs) increasingly serve as vital infrastructure for scientific exploration, industrial operations, and national security, yet they remain susceptible to malicious interference. This paper proposes a novel Reproducing Kernel Hilbert Space (RKHS) based authentication method that enhances security without imposing additional communication overhead. The proposed approach extracts channel impulse response features, embeds them into an RKHS as distribution signatures, and develops a predictive operator that authenticates packets by measuring their deviation from expected legitimate signal patterns. A notable advantage of this approach is its capability to adapt dynamically to varying underwater acoustic conditions without requiring labeled malicious samples. Through Bellhop simulations, we demonstrate the method's high detection accuracy under both static and varying sound speed profiles, confirming its suitability for robust authentication in resource constrained environments.