A NEW METHOD OF DIAGNOSTIC INFORMATION DECRIPTION BASED ON INTERVAL ASSESSMENTS

Abstract

The active development of computer systems and, as a result, the increase in the scope of their application presents scientific and technical progress with a number of new tasks, one of the most important of which is the organization of system diagnostics. Since the dimensions and complexity of computer systems are constantly growing, it is logical to try to organize their diagnostics in an autonomous mode, with minimal, or even without human intervention.

At the moment, such a method of diagnostics can be implemented, for example, by software in two stages: collection of diagnostic information and its decryption. Some problems can be encountered precisely at the decryption stage, since many currently proposed decryption of diagnostic information require quite a lot of calculations. Due to the fact that modern computer systems can contain a very large number of elements, the volume of diagnostic information can be quite large. As a result of this application of some existing methods, diagnostic information decription can take a long time. This is due to the fact that, in contrast to the collection of diagnostic information, the currently developed methods of its decription require a large number of calculations: some of them have an exponential complexity relative to the number of system elements. And this factor can be critical when developing, for example, large information systems or critical infrastructure systems. This creates the need to develop new, faster decryption methods.

This paper proposes a new method of diagnostic information decription, which is somewhat faster than many existing ones. This fact is achieved thanks to the combination of interval estimation and the Bernoulli distribution, which do not require an excessive number of calculations. The features of this method are also briefly described, one of which is clearly demonstrated on two specific examples of information systems.