On the problem of predicting meantime between failures on the basis of data from automated testing of on-board software components

Abstract

Abstract The article considers the problem of reducing costs for the development and maintenance of software components. The possibilities of reusing software components and as the result the need for their testing are considered. A method of the repeated testing is proposed. It allows achieving the required level of testability of software components that reduces time costs. This method is a more detailed investigation of the neighborhood of the identified failures and the elimination of already verified areas. The software implementation of the repeated testing method in the framework of the Integrated Development Environment developed by the authors is considered. The presented tool allows designers to automate the re-testing procedure partially. It is possible to predict the time of occurrence of the next error based on the data on the detected errors while testing. It is essentially the time between failures. A de-eutrophication model for predicting the mean time between failures and a software tool that implements the given model are proposed. The results obtained by testing on the known data set are presented. The models and approaches described in the article help to solve the problem that is confirmed by the results obtained using the developed software tools.