Two proposals for improving the image-based authentication system: H-IBAS-H

The paper describes a flexible image-based authentication system developed at the University of Portsmouth and proposes two possible additions to the existing system - an additional knowledge-based (KB) authentication stage and an intrusion detection (ID) feature. The knowledge-ased part of the authentication procedure will include a real-time question and answer session before the users are allowed to proceed to the second stage and identify the correct images from a number of challenge-sets presented to them. The proposed ID algorithm will employ a statistical data classification technique based on the real-time tracking of the user behavior. Two statistical algorithms are proposed for the implementation of these additional features of the system. Importance sampling technique will be employed to select the attributes for the knowledge-based authentication stage while the expectation maximization (EM) algorithm will form the basis of the intrusion detection part of the system. The paper outlines some issues that need to be resolved before proceeding with the implementation of the additional system features and lists some contributions and insights that might be gained from this work. The paper also reports on a recent pilot experiment conducted in relevance with the work.