Biometric Authentication in Galvanic Coupling Communications: A Physical Layer Approach for Secure Wearable Medical Systems

This paper presents a biometric authentication framework based on Galvanic Coupling (GC) signals for secure communication in Wireless Body Area Networks (WBANs). Unlike traditional biometric systems that rely on handcrafted features, our approach leverages a Long Short-Term Memory (LSTM) neural network to classify raw GC signals directly, exploiting the subject-specific characteristics of the propagation channel. The system is designed for on-body configurations and integrates a lightweight authentication protocol that operates at the physical layer. Experimental results demonstrate the effectiveness of the proposed method in mitigating both over-the-air and direct contact attacks, achieving high classification accuracy and robustness through confidence-based rejection mechanisms. The solution is optimized for low-resource environments, making it suitable for real-time medical applications.