Dynamic User Verification Using Touch Keystroke Based on Medians Vector Proximity

In this paper a user verification system on mobile phones is proposed. This system is based on behavioral biometric traits which is a keystroke dynamics derived from a touchable keyboard. A mobile application is developed for collecting those touch keystroke dynamics. In contrast to other systems, no specific text or numbers are used to build our dataset. The Median Vector Proximity classifier is applied on the touch keystroke data (touchable keyboard) and the performance of the system is investigated using different number of features and we found that the system with 31 features gained an average EER=12.9%. While with an extra two features (average of finger size and pressure) the average EER=12.2%. This shows that the more features used results in more accurate systems. The proposed system is compared against other systems and shows promising results in dynamic authentication area.