Keystroke biometric system for touch screen text input on android devices optimization of equal error rate based on medians vector proximity

Keystroke dynamics refers to the automated method of confirming the identity of an individual, based on the respective typing rhythm on a keyboard. An authentication with usage of this dynamics could increase the security of the system. Keystroke behavior on touchscreen based smart-phone enables additional features for the authentication. Therefore, the aim is to make the positive identification of a user more robust by analyzing the way in which a password is typed and not just on the content of what is typed. The touch-screen keyboard allows features ranging from pressure on the screen while typing or the area of keys covered by the fingers to the classical time-based features used for keystroke dynamics. In this research work, we have presented the effect of the equal combination of the touch-based and time-based based features on the identification and verification performance through a dataset of 7 users. An android application; on-screen soft keyboard, is developed to collect those keystroke details. The Median Vector Proximity classifier is applied on the collected keystroke data and the performance of the system is investigated using 47 features, which produces an average EER of 8.33% and an EER Standard deviation of 7.07%. The proposed system is compared against other systems and the results obtained in static authentication area have been found to be promising.