Liveness-Verified Dynamic Time Warping-Based Authentication and Hybrid Adaptive Neuro-Fuzzy Inference System Identification for Single-Channel Diaphragmatic Breathing Surface Electromyography Biometrics

Abstract

This article introduces a novel hybrid biometric identification framework that harnesses respiratory-induced surface electromyography signals recorded from the diaphragm using a single-channel electrode. The proposed system capitalizes on the unique, dynamic muscle activation patterns elicited during deep-normal-deep breathing sequences. In this framework, robust statistical features are first extracted and reduced via principal component analysis, then a streamlined parallel adaptive neuro-fuzzy inference system structure, designed to capture individual-specific patterns with minimal training error, is employed for feature vector generation. Finally, dynamic time warping is incorporated as a supportive tool to align temporal respiration patterns, refining decision thresholds and enhancing intersubject discrimination. Experimental results demonstrate that this integrated approach achieves high recognition accuracy, underscoring its potential for secure, real-time biometric authentication.