Incorporating Burr Type XII Testing-efforts into Software Reliability Growth Modeling and Actual Data Analysis with Applications

Software reliability is the probability that the given software functions correctly under a given environment, during the specified period of time. During the software-testing phase, software reliability is highly related to the amount of development resources spent on detecting and correcting latent software errors, i.e. the amount of testing effort expenditures. This paper develops software reliability growth models (SRGM) based on non homogeneous Poisson process (NHPP) which incorporates the Burr Type XII testing-effort functions (TEF). Numerous testing-effort functions for modeling software reliability growth based on NHPP have been proposed in the past decade. This paper shows that the Burr Type XII testing-effort function can be expressed as the actual testing-effort consumption during software development process. Its fault-prediction capability is evaluated through the numerical experiments. SRGM parameters are estimated by  least square estimation (LSE) and maximum likelihood estimation (MLE) methods and computational experiments performed on actual software failure data set from various software projects. The results show that the proposed testing-efforts functions predicts fault better than the other existing models. Thus, the proposed models evaluate software reliability more realistically. In addition, the optimal release policy based on reliability and cost criteria for software system are proposed.