Adaptive Behaviometrics using Dynamic Keystroke for Authentication System

The increasing number of information systems requires a reliable authentication system. Authentication system using behavioral biometrics of human or so-called behaviometrics can be the simplest and the best choice. Therefore, Dynamic Keystroke Authentication System (DKAS) becomes an effective solution which can be easily implemented to gain a high security information system. DKAS verify users based on their typing rythm on keyboard. Two main stages of DKAS is enrollment stage to register the user into the system, and the authentication stage to check the validity of the user. Verification is done by comparing the input feature vectors with template references. Feature matching employed is the combination of statistical method, measure of disorder and direction similarity measure. This combination method is chosen because it is simple and requires only small training data. In order to create an adaptive system we use a local threshold which has different value for different users. From the experiment conducted, the accuracy rate in distinguishing genuine and impostor users is 91.72%. This shows that the adaptive method of DKAS has a promising result.