EXPERIMENTAL STUDY OF AN ADAPTIVE CONTEXTUAL ACCESS CONTROL MODEL FOR ENHANCING THE RESILIENCE OF CRITICAL INFORMATION SYSTEMS

Abstract

In the context of the growing number and increasing complexity of cyber threats, ensuring secure access to critical information systems (CIS) is one of the key tasks in protecting the state’s critical information infrastructure. The use of adaptive access control mechanisms capable of accounting for changes in endpoint device posture, user privilege levels, and access environment parameters in accordance with the principles of the Zero Trust architecture is becoming increasingly important. This article presents an experimental study of an adaptive contextual access control model aimed at enhancing the resilience of CIS. To conduct the experiment, the model parameters were specified taking into account the characteristics of energy-sector CIS, a set of access scenarios was formed, and a software-based evaluation of the proposed approach was performed in comparison with existing access control models. The experimental results confirmed the adequacy and operability of the proposed model, as well as its ability to adaptively make access decisions in response to changes in contextual parameters. For scenarios related to endpoint device posture, the model demonstrated the highest average accuracy among the approaches considered, reaching 0.742, while under the reference policy prioritizing endpoint device posture, its accuracy reached 0.756. The results confirmed the feasibility of applying the proposed model to CIS in which endpoint device posture is one of the priority risk factors. The practical significance of the obtained results lies in the possibility of applying the developed model to build adaptive contextual access control systems for CIS while taking into account the threat profile of a specific critical information infrastructure facility. Further research will focus on improving the method for aggregating contextual parameters, expanding the set of experimental scenarios, and validating the proposed model using real-world CIS data.