An Empirical Study of Fault Detection by Static Units-Consistency Analysis
Judy A. Browning · Defense Technical Information Center (DTIC) · 1991
With increasing costs involved in software development, testing has become a more critical aspect of the software engineering process. Automatic methods, such as various static analysis techniques, may offer economic fault detection. This thesis analyzes a static analysis technique that allows users to associate units with variables in computer programs and to check that data transformations manipulate units in a consistent manner. A tool is designed and applied for this analysis. Its performance is measured by comparing the results in a previous study of other testing techniques in detecting faults. The results reveal that this technique consistently detected a narrow class of faults including some faults not found by other testing techniques. The results also show that application of this technique during the requirements and design phase of software development can identify faults associated with units-inconsistency early and reduce costs involved in developing a piece of software.