Personalized user authentication system using wireless EEG headset and machine learning

In the realm of authentication, biometric verification has gained widespread adoption, especially within high-security user authentication systems. Although convenient, existing biometric systems are susceptible to a number of security vulnerabilities, including spoofing tools such as gummy fingers for fingerprint systems and voice coders for voice recognition systems. In this regard, brainwave-based authentication has emerged as a novel form of biometric scheme that has the potential to overcome the security limitations of existing systems while facilitating additional capabilities, such as continuous user authentication. In this study, we focus on a data-driven approach to Electroencephalography (EEG)-based authentication, guided by the power of machine learning algorithms. Our methodology addresses the fundamental challenge of distinguishing real users from intruders by training classification algorithms to the unique EEG signatures of every individual. The system is characterized by its convenience, ensuring real-time applicability without compromising its efficiency. By employing a commercially available single-channel EEG sensor and extracting a set of 8 power spectral features (delta [0–4 Hz], theta [4–8 Hz], low alpha [8–10 Hz], high alpha [10–12 Hz], low beta [12–20 Hz], high beta [20–30 Hz], low gamma [30–60 Hz], high gamma [60–100 Hz]), a commendable mean accuracy of 85.4% was achieved.