A study on the reproducibility of biometric authentication based on electroencephalogram (EEG)

This paper introduced a method for biometric authentication using EEG signals. Especially, we focused on the performance of reproducibility for personal authentication. Four healthy subjects participated in the experiment. EEG was measured from only one bipolar channel (O1A2) during resting state with closed eyes. EEG was also recorded with same protocol from same subjects on different days to verify reproducibility (the interval between 1st and 2nd recording: 10 days or 5 months). Three features were extracted: the spectral power, the maximum power, and the frequency of maximum power in alpha band. Linear discriminant analysis was used as a classifier. The authentication accuracy for reproducibility was 98.33% with 20s data length and 100% with 50s data length.