Software reliability engineering study of a large-scale telecommunications software system

Abstract

Software reliability is a crucial performance factor of telecommunications network elements and operational systems. This paper describes the state-of-practice software reliability engineering (SRE) methods that we selected and organized into an SRE framework for use at Bellcore. This framework comprises several SRE methods: determination of a reliability objective for a product, development and use of operational profiles, reliability modeling and estimation (prediction) to manage system testing, estimation of the product's reliability in the field, and subsequent validation of this estimate using actual field data. Reliability modeling involves assessment of several models according to their predictive accuracy and the use of the most accurate model for reliability estimation. We have successfully tested this framework on several pilot projects. As part of these projects, we tested the usefulness of three different reliability modeling tools [the Economic Stop-Test Model (ESTM), Computer-Aided Software Reliability Estimation (CASRE) and Software Reliability Modelling Programs (SRMP)], as well as several different system test time metrics. This paper describes one of these pilot projects, involving a large operational system for networks.