Some new aspects on software reliability model comparisons

Tools and techniques for achieving and demonstrating high reliability are needed. One such is software reliability models. In order to make these usable in the industry, the area has to be structured. This calls for comparisons between models and a software reliability handbook based on the comparisons. The comparisons have to be done based on a number of criteria. The author adopts five criteria (predictive validity, capability, quality of assumptions, applicability and simplicity) presented by I.A. Musq et al. (1984) and complements them on some points. He also suggests two new criteria that have to be evaluated when comparing models: predictive quantities and coverage. The criteria have been structured into three separate parts of comparison to go through with a model; other models, environment and failure data