MODEL OF A CENTRALIZED SECURITY SYSTEM FOR OPERATING SYSTEMS RESISTANT TO LEAKS OF CONFIDENTIAL INFORMATION

Abstract

The construction of subsystem models of a decentralized and centralized OS security system designed to work as part of a protected system for processing confidential information in a multi-machine network computer system is considered. An analysis of publications on the construction of OSs resistant to leakage of confidential information and the protection of information processed in them in general is performed. Approaches to improving the main components of network OS security are summarized. Their protection mechanisms and methods for improving the efficiency of their operation within the framework of OS security systems are considered.

The main aspects of the construction of decentralized and centralized OS security systems and the principles of organizing the operation of their security mechanisms are considered. Mathematical models of decentralized and centralized OS security systems are presented, which take into account the impact of a set of threats on system resources and the counteraction of the corresponding OS protection mechanisms to them and can be used to study the developed OS security systems. At the same time, their set of parameters can be adapted to the requirements of the developed system.

A comparative analysis of the effectiveness of centralized security systems and decentralized systems has been performed.Their main disadvantages and advantages are given. The key aspect, according to the adopted approach, is finding a balanced architecture of the OS security subsystem, which can effectively ensure the OS's resistance to leaks of confidential information and its protection in general.

The built models and conducted tests have shown that centralized security systems have better indicators of incident detection and response, which reduces the risks of undetected information leaks, but have higher risks when compromising the central node due to the presence of a "single point of failure". Decentralized systems do not have it, but their resistance to leaks is much lower. Further research will be aimed at developing system models in order to inherit the best indicators of centralized and decentralized security systems.