Using keystroke dynamics in a multi-level architecture to protect online examinations from impersonation

As more people embrace online examinations, the need to protect their credibility becomes more crucial. Impersonation is a huge challenge when administering online examinations due to the anonymity of online users. In this paper, we address this problem through the use of keystroke dynamics which refers to the identification of users based on their typing pattern. Our results have supported the fact that it is possible to differentiate users based on their typing pattern. Even though there exist other solutions that use keystroke dynamics to discriminate between genuine and impostor users, our architecture promises a higher accuracy and robustness since it is 3-leveled. The architecture comprises of a statistical level, machine learning level, and a logical comparison level in hierarchical order. When a user signs up in the system, his typing data is automatically captured for use in templates generation. The templates are used as references to continuously authenticate the user while taking an online examination.