Half Logistic Software Reliability Growth Model

Abstract

A software reliability growth model based on non- homogeneous Poisson process (NHPP) is derived with its mean value function satisfying conditions similar to those of Goel and Okumoto (1979) and Musa and Okumoto (1984). The model turned out to be the (folded) half logistic distribution of Balakrishnan (1985). Reliability performance measures of the model are presented. Its parameters are estimated by maximum likelihood and approximate maximum likelihood methods. Asymptotic dispersion ma-trix, interval estimation also are derived. The results are illustrated with software failure data. The performance of the proposed model is evaluated in relation that of Goel and Okumoto (1979)

Keyword: Count process, mean value function, intensity function, software reliability, maximum likelihood estimation.