A User Authentication Scheme Using Keystrokes for Smartphones while Moving

Recently, smartphones are so diffusing that they are indispensable in our daily life. They store various types of personal information and are used for sensitive applications such as electronic commerce or personal banking. Therefore, they are an easy target for criminals. One of important countermeasures against smartphone attacks is authentication and various authentication schemes has been proposed. One such scheme is a keystroke dynamics, which is a kind of biometrics. However, there is no literature to discuss with smartphone authentication using keystrokes while moving (walking, by car, and so on). The aim of this paper is to clarify that, by fully utilizing built-in sensors (touch screen, orientation, rotation, and so on) in smartphones, what combination of such sensor outputs and which matching algorithm for sensor metrics are the optimal for authentication while moving. We implement a prototype system in an Android smartphone and evaluate equal-error rate (EER) for it in real moving environment, that is, with walking and in a moving car. As a result, we firstly found that the EERs were unexpectedly good; it was comparatively the same with static situation.