A Bootstrap Software Reliability Assessment Method to Squeeze Out Remaining Faults

摘要

This paper develops a bootstrap software reliability assessment method which can evaluate the number of remaining software faults at the final stage of the software testing process. The bootstrap method for reliability assessment problems has been already developed in the literature. However the method has a weak point which affects the applicability to the data set to be analyzed. We propose a new calculation formula in order to overcome this weak point. After showing the reliability assessment method by the traditional NHPP (nonhomogeneous Poisson process) models, we compare the performance of software reliability prediction with the bootstrap-based method by using a real software fault data set.