An improved statistical keystroke dynamics algorithm

Abstract

User identification and authentication are major security concerns on all kind of computational systems. Especially when information is critical, security policies enforce the need for these security preventive measures. Passwords implement a simple mechanism that rarely satisfies those requirements. Biometry is a better alternative and, from all possible techniques, Keystroke Dynamics, a behavior biometry that tries to characterize the user by the way he/she uses a keyboard, appears as an non intrusive technique that can capture user cooperation easily. However, most of the algorithms presented so far show poor performance, or lack of confidence due to a very restricted test process. In this paper we present a new Keystroke Dynamics algorithm, and describe the strategy followed to test it and to compare it with some other solutions. We conclude that this algorithm is a feasible solution for a flexible and easy to use identification and authentication process.