Smartphone continuous authentication based on keystroke and gesture profiling
Jain-Shing Wu, Wan-Ching Lin, Chih-Ta Lin, Te-En Wei · 2015
Recently, smartphones have become increasingly popular, whereas data leakage continues to be a serious problem for many large organizations. Consequently, smartphone applications containing sensitive personal or company data are at risk when targeted by attackers. Continuous and passive authentication is a popular scheme for secretly classifying users' identities based on their own unique touch motions (i.e., keystrokes and gestures). However, previous methods are inadequate when classifying users' singular touch motions. In this paper, we propose a novel continuous authentication method. The proposed method not only profiles behavioral biometrics from keystrokes and gestures, it also acquires the specific properties of a one-touch motion during the user's interaction with the smartphone. We demonstrate that the manner by which a user uses the touchscreen - that is, the specific location touched on the screen, the drift from when a finger moves up and down, the area touched, and the pressure used - reflects unique physical and behavioral biometrics. Moreover, the speed of the Gesture Segment (GS) is defined to extract a meaningful velocity segment. Experiments conducted to evaluate the proposed method for combining keystroke and gesture behavior demonstrate its effectiveness and accuracy.