Motion-senor behavior analysis for continuous authentication on smartphones

Existing smartphone authentication methods (e.g., PIN) typically provide one-time identity verification, but the verified user is still subject to session hijacking or masquerading attacks. This paper presents a framework and performance analysis of a sensor-based smartphone authentication system that continuously verifies the presence of a smartphone user. When a user touches the smartphone screen, motion-sensor data are extracted and analyzed to obtain descriptive features for accurately depicting users' touch habit and rhythm. Then a one-class learning algorithm is employed in the feature space to perform the continuous authentication task. Based on touch-tapping data collected from over 50 users, we conduct a series of experiments to validate the efficacy of our proposed approach. Our experimental results show that our verification system achieves a relatively high accuracy with an equal-error rate of 11.05%. Additional experiment on usability to the observation window size is provided to further examine the effectiveness. Our authentication system can be seamlessly integrated with extant smartphone authentication mechanisms, and is non-intrusive to users and does not need extra hardware.