EEG-based Person Authentication System in Different Brain States

Using EEG signals, which measure the electrical field generated by active neurons in a brain, to access a security system has received a considerable attention from researchers with a large number of publications. While elicited EEG-based credentials are known as sensitive to affective states, very few studies address that issue when utilizing brain-wave for authentication purpose. In this paper, we investigate the user-authenticating ability of an EEG-based person authentication (EBPA) system when clients are in a variety of brain states during performing mental tasks to login. The experimental results, which are supported by neuropsychological evidence, show that when people are in "like" or "unfamilar" brain states, the EBPA system has better accuracy compared to when they dislike or are familiar with stimuli. Some other issues involving human affective states are also introduced to help design a real-world EBPA system that is not only stable and high security but also comfortable.