Dynamic risk-based decision methods for access control systems

In traditional multi-level security systems, trust and risk values are pre-computed. Any change in these values requires manual intervention of an administrator. In many dynamic environments, however, these values should be auto-adaptive, and auto-tunable according to the usage history of the users. Moreover, occasional exceptions on resource needs, which are common in dynamic environments like healthcare, should be allowed if the subjects show a positive record of use towards resources they acquired in the past. Conversely, access of authorized users, who have negative record, should be restricted. These requirements are not taken into consideration in existing risk-based access control systems. In order to overcome these shortcomings and to meet different sensitivity requirements of various applications, we propose two dynamic risk-based decision methods for access control systems. We provide theoretical and simulation-based analysis and evaluation of both schemes. Also, we analytically prove that the proposed methods, not only allow exceptions under certain controlled conditions, but uniquely restrict legitimate access of bad authorized users.