Implicit authentication for mobile device based on 3D magnetic finger motion pattern

Touch pattern based implicit authentication has been proposed to defend against diverse attacks against mobile devices that aim to obtain credentials, e.g., passwords, in the process of user authentication. However, this defense technique cannot obtain a complete user operation pattern by merely deriving user operation data via a touch-enabled screen, since user operations, including on-screen and in-air finger movements, are performed in a three-dimensional space. In this paper, we propose a novel three-dimensional magnetic finger motion pattern based implicit authentication technique, referred to as FingerAuth. To use FingerAuth, a user first wears a magnetic ring on her finger and uses this finger to operate her mobile device, e.g., typing messages and surfing websites. By using a built-in three-axis magnetometer on the mobile device, we can derive the three-dimension (3D) magnetic finger motion pattern that is used as a human behavioral feature to implicitly authenticate the user. We construct robust 3D magnetic finger motion pattern detection model using machine learning techniques. Real-world experiments were conducted to demonstrate that our approach achieves high accuracy of 96.38% as well as low false acceptance rate of 4.06% and low false rejection rate of 3.18%.