Deceive Mouse-Dynamics-Based Authentication Model via Movement Simulation

Biometric security is one of the fundamental problems in user authentication. The authentication model based on mouse dy-namics and biometrics has been widely researched in recent years. However, the model security against potential attacks is largely ignored. In this paper, we propose a mouse movement simulation method to inspect the vulnerability of this authenti-cation technique. The authenticity of the mouse movement data generated by the simulation method is validated by five benchmark classifiers. Experimental results show the simula-tion method possesses a high authentication pass rate for most users. It indicates the proposed method can simulate real user mouse actions and the persuasive assumption that mouse be-havior is difficult to imitate might be too optimistic.