Toward Real-Time BCI Authentication for Enhanced Security in Collaborative Systems

Brain-Computer Interface (BCI)-based authentication offers a promising alternative to traditional passwords, especially in collaborative environments where close proximity increases the risk of spoofing. Since BCI systems rely on brain signals that cannot be externally observed, they enhance security by eliminating the need for visible password inputs. We present a longitudinal study using EEG signals in a closed-loop, real-time P300-based BCI authentication system. Volunteers completed multiple login sessions over two weeks using a consistent visual stimulus pattern. We evaluate the system’s performance across three aspects: (1) how authentication scores vary over time for the same user, (2) how accurately the system verifies genuine login attempts, and (3) how well it rejects impostor attempts using the same pattern. Pearson correlation was most effective for matching the same user over time, while Chebyshev distance best distinguished between different users.