On Component Dependencies in Compound Software

Predicting the reliability of software systems based on a component approach is in-herently difficult, in particular due to failure dependencies between the software compo-nents. Since it is practically difficult to include all component dependencies in a system’s reliability calculation, a more viable approach would be to include only those dependen-cies that have a significant impact on the assessed system reliability. This paper starts out by defining two new concepts: data-serial and data-parallel components. These con-cepts are illustrated on a simple compound software, and it is shown how dependencies between data-serial and data-parallel components, as well as combinations of these, can be expressed using conditional probabilities. Secondly, this paper illustrates how the components ’ marginal reliabilities put direct restrictions on the components ’ conditional probabilities. It is also shown that the degrees of freedom are much fewer than first antic-ipated when it comes to conditional probabilities. At last, this paper investigates three test cases, each representing a well-known software structure, to identify possible rules for selecting the most important component dependencies. To do this, three different techniques are applied: 1) direct calculation, 2) Birnbaum’s measure and 3) Principal Component Analysis (PCA). The results from the analyses clearly show that includ-ing partial dependency information may give substantial improvements in the reliability predictions, compared to assuming independence between all software components