Abstract
It is of paramount importance that software reliability requirements not only can be formulated, but also enforced and evaluated before accepting a software product for release. This requires sound acceptance criteria for software, which implies that methods for evaluating the software reliability are needed during the testing phase. Thus, theoretical models which are practically applicable are needed. These models should support software procurers in their difficult task of objectively accepting software products prior to releasing them for operation. This paper addresses this issue through studying two possible models for when to stop testing and accepting the software. The models are discussed with one particular software reliability growth model in mind, but the approach can be enlarged to take other software reliability growth models into account. The proposed stopping rules are illustrated on a set of failure data. It is concluded that it is indeed possible to evaluate and hence accept software products based on a software reliability estimate, including confidence in the estimate.
| Original language | English |
|---|---|
| Title of host publication | Workshop on Issues in Software Reliability |
| Publication status | Published - 1996 |
| Externally published | Yes |
Bibliographical note
Fourth Bellcore/KPN/Purdue Workshop on Issues inSoftware Reliability", Leidschendam, The Netherlands, 1995
Subject classification (UKÄ)
- Software Engineering