The Determination Method for Software Reliability Qualitative Indices

Abstract

The determination of software reliability indices is the primary task in the software reliability engineering. The indices are taken as not only the basis for the software reliability design and the constraints during the software development process, but also the foundation of the software's acceptance. Software reliability indices are usually divided into quantitative indices and qualitative indices. Quantitative indices are quantified software reliability parameters' values, such as software reliability is quantitatively defined as the probability of failure-free operation of a software program for a specified time under specified conditions. However, having a “number”, even with the appropriate accompanying evidence, is not generally sufficient to convince customers or even the system/software suppliers that the software satisfies its requirements. Thus, qualitative indices such as software reliability is also qualitatively defined as a set of attributes that bear on the capability of software to maintain its level of performance under stated conditions for a stated period of time. Attributes that relate to implementation of fault tolerance design, use of best engineering practices, application of specialized methods and techniques for ensuring reliability-critical requirements, and procedural methods to ensure mistake-proof loading and/or operation also provide evidence that improves the confidence that the software will not cause a system failure. So the qualitative indices can be regarded as the requirements for software reliability activities throughout the development process. Unfortunately, currently there is no systematic theory and approach for software reliability indices' determination. This paper proposes a method for determining the software reliability qualitative indices based on the two standards of SAE-JAI 003 and RTCA DO-I 78B which are widely used by the airworthiness and industrial sectors, as well as the best practices and management experiences of software reliability engineering. This paper proposes the method's principle, which determines the software reliability qualitative indices according to the profile formed by all stages of the software life cycle and the environment requirements, technique requirements, validation requirements and management requirements. Combined with the software's criticality levels, this paper also proposes a generic framework which recommends a variety of tailoring mechanisms and building guidelines to help users develop their demanded indices.