Phase synchronization of alpha activity for EEG-based personal authentication

Abstract

There has been a growing interest in the EEG-based biometric system as an alternative approach to personal authentication (PA). In this study, we focused on the potentialities of functional connectivity, especially phase synchronization in the alpha rhythm represented by the phase locking value (PLV) as a novel EEG signature for PA. We analyzed an EEG dataset of 39 trials from 7 subjects who participated in the 5–7 sessions repeatedly on different days. In the sessions, a total of 16 EEG signals were acquired by a portable EEG device, when the subjects were in a resting state with their eyes closed for 2 min. The characteristics of alpha phase synchronization were estimated by the following procedures. (1) The alpha rhythm was extracted from the EEG signals using a band-pass filter of 8–13 Hz. (2) We randomly selected 20 2-s time segments of the alpha rhythm and calculated the mean phase coherence [1] between channels within each time segment. From all possible pairs of 16 EEG channels, a total of 120 mean alpha phase coherence values were extracted. 3) From these mean alpha phase coherence values, we calculated a criteria index (CI) of each of them where the CI calculated the ratio of an inter-subject variability to an intra-subject variability, which was developed to discriminate critical EEG features for PA in our previous study [2]. 4) Using the mean alpha phase coherence and its CI values, we constructed the association matrix of phase coherence and extracted the 12 top-ranked CI connections (Fig. 70). The topographical result showed that there were apparently two functional connectivity networks of the alpha rhythm in the brain for PA. The first network was distributed over the anterior regions including the pre-frontal, frontal and left central regions. The second one was located in the posterior region covering the occipital region. It should be noted that two regions have been well known as main sources of the alpha rhythm from many studies on EEG alpha rhythms. Our results suggest an important role of the alpha rhythm in the EEG-based biometrics system.