Biometric Verification Based on ECG Signal using 1 Dimensional Convolutional Neural Network

Abstract

Biometric is an analysis of individual characteristics. For instance, fingerprint, voice, i ris, a nd face a re biometrics. Nowadays, those methods are often used; it still has the disadvantage of being easy to manipulate. Identification using Electrocardiogram (ECG) signal is one of the biometric methods developed to prevent individual manipulation since ECG signals are unique for each individual. This study designed a system using ECG signals for biometric verification. The ECG signals are unique since each individual has different physiological, geometric, and characteristics. The ECG-ID dataset used for evaluation contains 90 subjects. The One Dimensioanal Convolutional Neural Network is used in this research. We compared the difference using two ECG signal fragments, namely PQRST and PQRS waves. The best results show an accuracy of 91.57% using PQRST waves. This proposed study is feasible enough to be used as verification biometrics.