Understanding secure and usable gestures for realtime motion based authentication

One promising approach to achieving strong authentication is using gesture based behavioral biometrics, that is, user authentication is based on how a gesture is performed. With the advent of smart wearable devices, 3D motion gesture based authentication becomes increasingly appealing. Understanding 3D motion gestures especially their security and usability is fundamentally important but remains to be conducted. Towards this goal, we perform an empirical study on the security and usability of user-created 3D gestures using a realtime free-form motion gesture authentication scheme we have developed. To create a real-world authentication experience, in our experiment, a participant sees the authentication result from his or her wearing smartwatch in real time for each gesture test. Our experiment consists of not only a set of self-tests but also a series of attacks from easily launched random guessing attacks to much more sophisticated and dangerous targeted mimicry attacks. Our experimental results reveal several interesting findings on 3D gestures' security and usability including the correlation between gesture categories and their performance and attack resistance, and the effect of posture and psychological factor, which we believe shed light on the future design of 3D motion gestures for smart wearable devices.