Search

Telephone switching systems require using more and more complex and bulky software. Accordingly, before the operational phase France Telecom must evaluate software quality in particular reliability. This paper sets out to present an experiment conducted at the CNET-Lannion A centre. This experiment was based on the use of 3 conventional reliability growth models during the qualification phase of a product (analytical and overall tests) and also during the validation phase of a different product. The future prospects with regards the reliability data book and the use of the different models are also discussed.

To support software project managers for reliability evaluation, the company Mathix and the French telecommunications agency (France Telecom) have developed a CASE tool named M-elopee. The purpose of this tool is not only to provide some reliability growth model computations but, more generally, to propose an approach to do reliability studies, and therefore to offer all the features needed to undertake this approach. M-elopee typically promotes the transfer of software reliability technology to practice. Therefore it runs on PCs under Windows and benefits from the many facilities of this environment. This article describes the functionalites provided by M-elopee and its underlying methodology. Mainly, the following points are developed: the content and structure of the data processed by M-elopee; data processing; trend analysis; comparison of models; and the maximum number of failures to forecast.