Exploring Games for Improved Touchscreen Authentication on Mobile Devices

Mobile device theft is a growing problem. Yet due to usability issues and other concerns, people frequently choose not to use an authentication mechanism to protect their devices, putting the sensitive information that they store at risk. In order to provide mobile device owners with more usable authentication, we propose the study and development of mechanisms for authenticating users to mobile devices by modeling the manner in which they interact with games. We conducted an preliminary IRB approved study in which 12 users were asked to play 3 of the most popular games available in the Google Play Store on an Android device while their touchscreen interactions were logged. We then applied a Support Vector Machine to classify users based on 19 extracted touchscreen usage features. We were successfully able to classify over 90% of the samples for each game with a false reject rate of less than 1%. These results indicate that utilizing gameplay elements to encourage unique behavioral touchscreen features may be a promising direction of future research.