A MACHINE LEARNING FRAMEWORK FOR BIOMETRIC AUTHENTICATION USING ELECTROCARDIOGRAM (ECG)

Abstract

Biometric authentication has emerged as a secure alternative to conventional authentication methods such as passwords, PINs, and smart cards. Among various biometric techniques, Electrocardiogram (ECG)-based authentication offers enhanced security because every individual's cardiac electrical activity is unique and inherently resistant to spoofing. This study presents a machine learning framework for biometric authentication using ECG signals. The proposed framework includes ECG signal acquisition, preprocessing through noise removal and normalization, R-peak detection, heartbeat segmentation, feature extraction, feature selection, and user classification. Four machine learning algorithms, namely Support Vector Machine (SVM), Random Forest, Extreme Gradient Boosting (XGBoost), and LightGBM, are employed to evaluate authentication performance. The models are assessed using accuracy, precision, recall, F1-score, ROC-AUC, False Acceptance Rate (FAR), and False Rejection Rate (FRR). Experimental analysis demonstrates that ensemble learning algorithms provide improved recognition accuracy, robustness, and faster processing compared with conventional approaches. The proposed framework offers a scalable, reliable, and real-time authentication solution suitable for healthcare, banking, Internet of Things (IoT), mobile devices, and intelligent access control systems.