The Research and Implementation of the Fine-Grained Implicit Authentication Framework for Android

Nowadays, in order to protect sensitive information in Android apps, plenty of identity authentication techniques were developed, including the password, graphical-password, fingerprinting, etc. Unfortunately, these schemes have many disadvantages. For example, the graphical-password could be reproduced by the trace on the screen. Different from the explicit authentication above, the implicit authentication scheme silently collects user behavior patterns for authentication, without the actions like inputting password. In this paper, firstly, we realized a more fine-grained implicit authentication scheme for the first time, which has refined the unit for authentication from App-level to Activity-level. Secondly, we improved the feature extraction and applied the classification algorithm called SVDD. Thirdly, we developed a no-buried-point library to enhance the usability. Finally, we recruited 21 volunteers for experiments. The experimental results reveal that the accuracy of proposed scheme can double the previous work and the no-buried-point feature can greatly improve the efficiency of app development.