Redundant models of testable distributed real-time computing systems

Abstract

Diagnostic issues receive a lot of attention in the design of information processing and control systems since the systems’ reliability and fault tolerance depends on the quality of their solution. The article presents the results of the development of a synthesis algorithm for a model designed to solve the problem of test diagnostics and focused on distributed computing systems. The algorithm is integrated in the system and executed in parallel with the main software of the system, which makes it possible to simplify the process of testing the system. The description of a distributed computing system, complemented by an integrated diagnostic model, is a redundant model of the system. The proposed algorithm implies a reduced amount of diagnostic information. The diagnostic model has a hierarchical structure and involves two stages. At the first stage, the algorithm calculates the set of paths that make up the coverage of its edges for the graph of intermodular connections in the system. It matches a chain of dynamic links with each of the obtained paths, the number of the links being equal to the number of software modules through which this path passes. At the second stage, the type of dynamic links is determined. It is taken into account that the desired dynamic model of the system is used to generate tests. The test design procedure is simplified if the system model is linear, controllable, and observable. Based on this, the requirements for the links of the chains of the model are formulated. The proposed algorithm makes it possible to obtain a discrete-event model for the system characterized by a reduced amount of used diagnostic information.

Keywords

• discrete-event model
• test diagnostics
• testability
• non-stationary models
• observability
• controllability