Exploring Touchscreen Biometrics for User Identification on Smart Phones

Abstract. The use of mobile smart devices for storing sensitive infor-mation and accessing online services is increasing. At the same time, methods for authenticating users into their devices and online services that are not only secure, but also privacy and user-friendly are needed. In this paper, we present our initial explorations of the use of lock pat-tern dynamics as a secure and user-friendly two-factor authentication method. We developed an application for the Android mobile platform to collect data on the way individuals draw lock patterns on a touch-screen. Using a Random Forest machine learning classier this method achieves an average Equal Error Rate (EER) of approximately 10.39%, meaning that lock patterns biometrics can be used for identifying users towards their device, but could also pose a threat to privacy if the users' biometric information is handled outside their control