Predicting software reliability in a fuzzy-random field environment
2012 (English)Conference paper, Presentation (Refereed)
The development of a new complex software or industrial system produces a series of prototypes that may contain faults during the processes, including development, design, and production. Therefore, during the early stages of prototyping complex systems, reliability often faces a major challenge in meeting the desired requirements level. For these reasons, a typical reliability improvement process is carried out in order to achieve a specific software/system reliability level. Incorrect estimation of reliability could lead to an inappropriate system design and implementation of incorrect maintenance policies.Many software reliability models have been proposed to help software developers and managers to assess the level of the reliability and estimation of the development cost. Among these software reliability models, Non-Homogenous Poisson Process (NHPP) based models have been successfully applied to model the software failure processes, and predict the number of software failures. NHPP has been used also to determine “time to stop testing” and release the software.Usually, the testing-development phase has been carried out in a given control environment. However, the product will be used in different operating environment by different end-users, which is unknown to the developer. The operating environment may range from a very clean one up to a harsh environment. These uncertain operating environments will impact to the reliability and performance of the software which may differ from the testing phase reliability. Hence the effect of environmental factors on reliability should be considered for estimation of the operational phase reliability. In fact the effect of environment on reliability has a fuzzy nature and quiet random. On the other hand, it is well known that the probability distribution and its parameters cannot be univocally defined. To overcome this problem, the fuzzy probabilistic theory may be used in the processing of stochastic parameters, taking into account their fuzzy nature. In fact, the fuzzy random effects of the field environments can be captured by a unit-free environmental factor. Based on the fuzzy probability distribution and its properties, we can define a fuzzy reliability function. The aim of this paper is to introduce a Fuzzy random field environment (FRFE) reliability model that covers both the testing and operating phases in the development cycle. The proposed model is based on Weibull distribution. It should be noted that the testing costs is one the major concern in software/system development. Several researcher investigated software/System release policies to minimize development cost while satisfying a reliability objective. Although the length of testing phase directly relates to the number of errors removed, but leads to a significant financial loss by increasing testing cost and delay in delivery. Further, releasing software in the market before reaching its desired level of reliability (which is fixed by the manager) may increase the maintenance cost during operational phase as well as create risk to lose future market.For a critical software system, the penalty costs resulting from the software failures are much more significant than the software development costs themselves. Therefore, the total software/ system cost should consist of not only the development costs, but also the penalty costs resulting from the software failure in operational phase. For software developers and managers, the following questions need to be answered (1) How to allocate the resources to ensure the on-time delivery of a software product? (2) When to stop testing and release the software from current software testing activities? (3) Is the software product really reliable in field? To answer the above mentioned questions, the paper proposes a cost function using proposed FRFE reliability model, to determine the optimal release policies.
Place, publisher, year, edition, pages
Research subject Operation and Maintenance
IdentifiersURN: urn:nbn:se:ltu:diva-40032Local ID: f002006c-01fe-4ee3-8d36-736c2ea872e8OAI: oai:DiVA.org:ltu-40032DiVA: diva2:1013554
International Conference on Quality, Reliability, Infocom Technology and Industrial Technology Management : 26/11/2012 - 28/11/2012
Godkänd; 2012; 20130226 (ysko)2016-10-032016-10-03Bibliographically approved