Identity Authentication Using Portable Electroencephalography Signals in Resting States

Abstract

Biometric identity authentication technology is widely used in the field of information security. As a new biological method, electroencephalography (EEG) is gradually applied in biometric recognition, because scientists believe that practicability and portability are the development direction of EEG identity authentication. This work investigates the feasibility of using resting state EEG signals recorded by single-channel portable device for identity authentication. Single-channel EEG classification are effectively improved by using mixed-method in three feature domains (i.e., time, frequency, and time-frequency), feature selection (Rayleigh quotient) and classifier design (ensemble classifier). We invited 46 subjects to participate in the EEG identity authentication experiment. Experimental results show that the open-eyed resting state is an ideal authentication method, and the average classification accuracy of the authentication algorithm can reach 95.48% in 2 seconds, validating that the new method of processing single-channel EEG signals is especially useful in extrapolating EEG identity authentication to realistic field contexts.