Personal authentication based on unconscious arm shaking action

In this paper, we propose a behavior-based personal authentication method, arm shaking action, for smartphones that remains robust over time, a crucial factor in smartphone authentication. We investigate the authentication accuracy using unconscious arm-shaking actions, analyzed through a support vector machine (SVM). Two types of experiments are conducted: the first involves subjects shaking a smartphone with their own individual actions over time, while the second involves subjects imitating legitimate users' actions via video recordings. In the first experiment, the best results showed both false acceptance rate (FAR) and false rejection rate (FRR) at 0% after one week, while the worst results were 0.2% for FAR and 18% for FRR. In the second experiment, the best results also showed 0% for both FAR and FRR, while the worst were 1.3% for FAR and 0% for FRR. Thus, personal authentication method relying on unconscious arm shaking action is practical on smartphone, is robust over time, and is also resilient against impersonation attacks.