Password policy characteristics and keystroke biometric authentication

Behavioural biometrics have the potential to provide an additional or alternative authentication mechanism those involving a shared secret (i.e. password). Keystroke timings are focus of this study, where key press and release acquired whilst monitoring user typing known phrase. Many studies exist in keystroke biometrics, but there is absence literature aiming understand relationship between characteristics password policies biometrics. Furthermore, benchmark datasets used biometric research do not enable useful insights into their capability policy. Herein, substitutions uppercase, numeric, special characters, combination passwords derived from English words considered. Timings for 42 participants same 40 acquired. A matching system using Manhattan distance measure with seven different feature sets implemented, culminating Equal Error Rate 6% 11% accuracy values 89% 94%, demonstrating comparable other threshold-based systems. Further analysis suggests that best containing all trigraph press. Evidence also phrases fewer characters greater accuracy, except character substitutions.