Software reliability growth model with normal distribution and its parameter estimation
Hiroyuki Okamura, Tadashi Dohi, Shunji Osaki · 2011
This paper proposes software reliability growth models (SRGM) where the software failure time follows a normal distribution. The proposed model is mathematically tractable and has sufficient ability of fitting to the software failure data. In particular, we consider the parameter estimation algorithm for the SRGM with normal distribution. The developed algorithm is based on an EM (expectation-maximization) algorithm and is quite simple for implementation as software application. Numerical examples are devoted to investigating the fitting ability of normal distribution based SRGMs through software failure data collected in real software projects.