Developing a keystroke dynamics based agent using rough sets

Abstract

Software based biometrics, utilising keystroke dynamics has been proposed as a cost effective means of enhancing computer access security. Keystroke dynamics has been successfully employed as a means of identifying legitimate/illegitimate login attempts based on the typing style of the login entry. In this paper, we collected keystroke dynamics data in the form of digraphs from a series of users entering a specific login ID. We wished to determine if there were any particular patterns in the typing styles that would indicate whether a login attempt was legitimate or not using rough sets. Our analysis produced a sensitivity of 98%, specificity of 94% and an overall accuracy of 97% with respect to detecting intruders. In addition, our results indicate that typing speed and particular digraph combinations were the main determinants with respect to automated detection of system attacks.