Personal authentication using timed keystrokes
Date
1998
Authors
Anothayanon, Worapa
Major Professor
Advisor
Davis, James A.
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
As the world becomes more global, user authentication methods are becoming increasingly important in computer security. Conventional authentication approaches are vulnerable to exhaustive attacks. This thesis proposes a way to enhance the performance of authentication systems by employing biometric techniques. The authentication software developed here combines a conventional password scheme with an analysis of a user's personal typing habits. The major components of a user's profile include characteristic keystroke durations and keystroke latencies.
The Student's t standard distribution is used to statistically compare the similarity between the user's reference profile and future typing samples. The effectiveness of the algorithms are compared using the false alarm rate (FAR) and the impostor pass rate (IPR). Initial experimental results show that this simple method of authentication is able to correctly identify real users and reject impostors in most cases. It is more likely that analysis of timed keystrokes will be used to strengthen other authentication schemes, like passwords.
In this thesis, we direct our research work towards the design and evaluation of keystroke analysis for authentication. Most contemporary computer systems carry out the authentication process using passwords. A password is considered as a secret knowledge to verify the user's identification. However, using only a password becomes problematic when the real user cannot be distinguished from an unauthentic hacker. To enhance the security, we combine the use of password with keystroke dynamics. We investigate the efficiency of several authenticating techniques based on user personal typing characteristics, also called typing rhythms. The typing rhythms authentication software was developed on PC platform. Extensive experiments were set up to evaluate the effectiveness of this authentication approach.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
thesis