The Impact of Password complexity level on Building a Reliable Model of Keystroke Features

Abstract

Electronic authentication is considered to be one of the most important problems in the field of accessing information sources. In an era that relies entirely on virtual accounts, the complexity of the password is no longer effective to protect it from penetration.  This necessitates the use of new technologies to enhance the level of security that passwords provide to protect users' data. Keystroke dynamics is a technique for identifying users based on their behavior on the keyboard. This study aims to identify the various methods of authentication and highlights the use of keystroke dynamics as a powerful tool that can be used alongside the password to verify the identity of the user. This paper provides a theoretical and analytical study of the use of the keystroke dynamics in producing a model for the behavioral user characteristics according to the type of password. A software application was developed by Java language to conduct an experiment to collect time data for users when they interact with the keyboard to enter three passwords of different complexity levels and then extract behavioral characteristics and the applying statistical classification algorithm. The study showed that the use of medium-complex and complex passwords give better results than the simple password, and also confirmed the results of previous studies. It was also found that the use of aggregated features (total typing time) associated with a complex password is the best to distinguish users from one another while taking into account their behavior when writing large letters and numbers.