Improvements in Software Quality Assurance Using An Enhanced Software Reliability Model

Test coverage in software is an important issue since it aaects both the adequacy and effectiveness of software testing techniques as well as the estimated reliability of the software product under consideration. There are many notions of coverage, but there is no agreement on a common deenition for it. There are, however, analytical advantages to a common deenition, and we have tried to address this issue directly. Our notion of coverage is based on the concept of \potential fault-site" which may be interpreted to mean a speciication statement, or a source code statement, a block of code, etc. Coverage is then deened to be the fraction of potential fault-sites sensitized by a test, and it becomes a time-dependent parameter while testing is in progress. In general, test coverage is not an explicit parameter in existing software reliability models. We develop a reliability model which accounts explicitly for test coverage and through it for test adequacy and test eeectiveness. An illustration is provided to demonstrate the use of the model.