CDAS

Abstract

The rapid development of smartphones has greatly facilitated our lives. At the same time, securing the data stored and accessed from smartphones makes it important to authenticate the user. However, current smartphones perform one-time authentication at the entrance while they don't authenticate users continuously when in use, which brings serious privacy and security issues, such as collisions and social engineering to bypass the authentication. This paper introduces CDAS (Continuous Dynamic Authentication System), which uses the Support Vector Machine (SVM) to construct user's behavior model by collecting his touch data to judge him authorized whether or not. CDAS works independently in the background without interacting with users most time. Therefore, CDAS is featured with security, efficiency and continuity. We conducted a two-week experiment involving more than 20 users which shows that the system we design achieves a high accuracy, a low False Accept Rate (FAR) and a low False Reject Rate (FRR), which indicates that CDAS ensures the security and enjoys a promising prospect.