Interval Estimation for Non-Parametric NHPP-based Software Reliability Model via Simulation-based Bootstrap
Xiao Xiao, Tadashi Dohi · 2021 8th International Conference on Dependable Systems and Their Applications (DSA) · 2021
The model estimation of the widely used non-homogenous Poisson process (NHPP) based software reliability model (SRM) is often done by point estimation. A number of interval estimation methods have been developed for parametric NHPP-based SRMs to take the uncertainty of the corresponding estimator into consideration. In this paper, we propose non-Bayesian interval estimation without asymptotic approximations for non-parametric NHPP-based SRMs. The key idea is to utilize the simulation-based bootstrap. To investigate the effect of different NHPP simulation methods to the result of the confidence interval of software reliability measures, we use two kinds of NHPP simulation methods to generate bootstrap samples, based on which the confidence interval of the software reliability measures such as the software intensity function and the mean value function will be constructed.