Pattern Analysis for Biometric Authentication using Electrocardiogram Signal

Recently, biometric authentication based on Electrocardiogram (ECG) has been actively studied as a potential alternative to traditional counterparts. Conventional authentication systems require passwords in alphanumeric or graphical format, which involves the risk of loss, theft, or even forgetfulness. ECG-based authentication is currently one of the hot topics of research to address such shortcomings. Here, we demonstrate a Biometric Authentication system based on ECG by using an infinite impulse response (IIR) filter. ECG signals from 36 volunteer subjects were obtained and de-noised using the IIR filtration technique. Then, we extracted 18 time domain, cepstral and morphological features from preprocessed ECG signals. Next, the extracted features were classified by utilizing various classifiers such as Support Vector Machine (SVMs), K-nearest neighbours (KNNs), and Kernel Naïve Bayes (KNBs). SVM-quadratic demonstrated an accuracy of 99.2%, which is the highest among the various classifiers employed in this work. We expect that our proposed method can be utilized for developing an embedded system, which can be employed for the authentication of individuals with higher accuracy.