Keypress Interval Timing Ratios as Behavioral Biometrics for Authentication in Computer Security

Many different types of keystroke dynamics approaches have been explored to protect personal data in networked systems. Keystroke patterns are behavioral biometrics, and are considered to be as unique to an individual as a signature. This paper presents a new approach to keystroke analysis that uses key press interval ratios to authenticate users. Participants in this study registered their passwords into a specially-designed analysis program. Keypress ratios were calculated, and neural network techniques were employed to obtain a mapping between patterns and the correct user. Results indicate that authentication through keypress ratios achieves high true acceptance rates, while also maintaining low false acceptance rates, which are particularly important in high-security applications. The approach presented here is suitable for incorporation into agent-based networked security systems.