Authors:
Blaine Ayotte
;
Mahesh K. Banavar
;
Daqing Hou
and
Stephanie Schuckers
Affiliation:
Department of Electrical and Computer Engineering, Clarkson University, 8 Clarkson Avenue, Potsdam, U.S.A.
Keyword(s):
Behavioral Biometrics, Keystroke Dynamics, Principle Component Analysis, User Authentication.
Abstract:
Keystroke dynamics study how users input text via their keyboards. Having the ability to differentiate users, typing behaviors can unobtrusively form a component of a behavioral biometric recognition system to improve existing account security. However, because keystroke dynamics is behavioral biometric typing patterns can change over time. The temporal effects of keystroke dynamics are largely unstudied beyond empirically demonstrating that error rates will be higher for old or outdated profiles. In this paper, the effects on typing patterns over time is investigated in detail. Using a well-known fixed-text keystroke dynamics dataset, we show overall typing time for a provided password “.tie5Roanl” changes significantly over time, decreasing by almost 30%. Principal component analysis (PCA) is used to determine which monographs and digraphs tend to change throughout time. Rarely typed features, such as digraphs with a letter and number, are most likely to change over time, while com
monly occurring features such as common digraphs and monographs are much more stable.
(More)