User Behavior Continuous Authentication in Secure Transactional Application

The purpose of this paper is to find the appropriate physical and behavioral biometric as well as contextual indicators for continuous authentication in a secure transactional application. The indicators must have good accuracy in identifying a user and be able to form the basis of an anomaly detector. In addition, their combination must maintain the usability of the application. Therefore, a compromise between accuracy and usability will be made so that the user can use the application easily and at the same time be protected from a potential thief. The paper relies on research and results from over twenty-five articles in the specialized literature to identify the best indicators, classifiers, and algorithms that have demonstrated good accuracy and usability. Furthermore, additional indicators specific to a banking application were considered. The identified indicators will be collected during the use of the application from several volunteer users and subsequently analyzed to obtain the results. To decide whether a user is legitimate, a score is calculated for the current session based on the indicators that have a weight based on the importance that was established following an opinion questionnaire. If the score falls below a certain value, then the current session is stopped immediately, and the person is logged out. The study was conducted on a banking application and among the indicators are the preference for displaying or hiding the balance, dominant hand for using the application, the way in which they return to the application, either through the PIN number or by using the fingerprint, the location and time from which the application is accessed, the amounts transacted, the way in which they press buttons. Overall, this research paper contributes to the literature by identifying and testing a unique set of indicators suitable for continuous authentication using behavioral biometric data in the context of a banking application.