ECG Based Biometric Identification System using EEMD, VMD and Renyi Entropy

Abstract

New biometric systems are being proposed to overcome the lack of conventional biometric systems, especially in high-security applications. One of the potential modality is an electrocardiogram (ECG) signal based biometric system. In this study, a biometric system has been simulated using one ECG lead signal. A total of 110 ECG waves from 11 subjects have been simulated. Ensemble Empirical Mode Decomposition (EEMD), Variational Mode Decomposition (VMD), and Renyi Entropy are proposed methods for feature extraction. EEMD and VMD decompose ECG signals into five levels. Then, signal complexity analysis using the Renyi Entropy approach is calculated for each of the decomposed signals. The values are then becoming the feature set that fed in the validation process. The highest accuracy of this proposed method for person identification is 96.4%, which achieved by using VMD and Cubic SVM. The proposed method can be considered to be implemented in real-world implementation by considering the use of low-cost devices.