A Comparison between Human Selected, Derived and System-Generated Passwords in terms of Keystroke Dynamics

Abstract

Pure hardware based biometric systems of user authentication have low rate of errors but increase  computational and economic cost. Alternate biometric methods based purely on software development are now booming. This paper presents our results on improving authentication of users using a password mechanism hardened with keystroke timing vectors. We develop a system that is trained with the keystroke timing vectors of the owners and then later used to differentiate between authentic users and impostors. Using a prototype implementation of our scheme, we compare the results of human-selected, derived passwords and system generated to reveal the practical viability of our approach in terms of results achieved, ease of implementation and use.

Keywords: Security, Biometric, Behavioral, Keystroke dynamics, Password