Fuzzy and Markov Models for Keystroke Biometrics Authentication

Keystroke biometrics authentication system is based on a password and keystroke biometric features captured when a user is typing in the password. The system offers a higher level of security and convenience for computers. The system does not require additional hardware as it can be used with any existing keyboard, making it relatively inexpensive and fairly unobtrusive to the user. There have been existing research publications on keystroke biometrics authentication that have solved problems in selecting appropriate keystroke features and modeling users. However methods for calculating score to reduce authentication error are not taken into account. Therefore we propose to use Markov modeling and fuzzy set theory-based normalization methods for keystroke biometrics authentication that can reduce both false rejection and false acceptance rates. Experiments showed better performance for the proposed methods