Study On NHPP Software Reliability Models

The study of the software reliability growth model plays an important role in thestudy of software engineering. At present, there are many software reliability growthmodels, and the bases of them are usually some random or non-random processes. So,it is necessary to study the software reliability growth models by the sorts ofprocesses.The first part of this thesis introduces the software reliability, the inner and outerhistory and actuality of software reliability and software reliability growth models.Then, expatiated on some concepts about software reliability and software reliabilitymodel, including the factors which impact software reliability, the features ofsoftware reliability, software reliability parameters, the meaning of the study ofsoftware reliability growth models. The part provides the theory basis for the study.The second part of this thesis firstly introduces the math basis of the study ofsoftware reliability growth models, such as random variable and its functionincluding binomial distribution, poisson distribution, geometric distribution, indexdistribution, Weibull distribution, etc. and some random process, such as poissonprocess, nonhomogenous poisson process and Markov process, and some methodsabout parameter estimation. This part provides the math basis for the study. On thebasis of above, the thesis import several classical software reliability growth models,such as Goel Okumoto models, Gompertz Growth Curve models, Logistic GrowthCurve models, Generalized Goel NHPP model, Yamada Delayed S-Shaped model,and Inflected S-Shaped model, Modified Duane model, Two-Error-Type model. Thisprovides the object for the study. Then, formulate a general weight function based onweight arithmetic, weight geometric and weighted harmonic mean, which providesmath mean for the study.The third part applies the general weight function in some models mentionedabove in the forms of discrete and continuous .The thesis proposes a more general NHPP models form the general weight function. Under this general framework, thethesis verifies the existing NHPP models, and shows that these approaches cover anumber of well-known models under different conditions.