Software reliability models for critical applications

Abstract

This report presents the results of the first phase of the ongoing EG&G Idaho, Inc., Software Reliability Research Program. The program is studying the existing software reliability mc:lels and proposes a state-of-the-art software reliability model that is relevant to the nuclear reactor control environment. This report consists of three parts: (1) summaries of the literature review of existing software reliability and fault tolerant software reliability models and their related issues, (2) proposed technique for saitware reliability enhancement, and (3) general discussion and future research.

Suitability analysis of software reliability models for its applicability on NPP systems

“…In Pham and Pham, an exhaustive survey on the existing SRMs. The SRMs have been broadly classified into deterministic and probabilistic models.…”

“…Shooman was the first to use execution path model, and his approach is known as Shooman decomposition model …”

“…The reliability as well as the errors remaining in a software program is estimated by the error‐counting models. The models falling in the category of the deterministic models are JMD model, Musa model, Shooman model, and Schick‐Wolverton model . Littlewood model is a type of Bayesian model which assumes that larger errors are detected before the smaller errors.…”

“…The JMD model discussed in the failure rate model group is a special case for this model, its explanation along with the 3 models, namely, linear death with imperfect debugging, nonstationary linear death with perfect debugging, and nonstationary linear birth‐and‐death can be found in Pham and Pham …”

“…Many authors argue that sometimes error occurs in clusters . A Fourier series was used to signify the time interval between faults to check whether the clusters of errors exist or not .…”