Assessment of the minimal test effort in software reliability demonstration testing with software test effectiveness

Abstract

The reliability assessment of safety-related software systems and components is an issue of primary importance in the nuclear industry, the avionics sector, the military etc. In these areas, the required test effort calculated by the traditional software reliability demonstration testing (SRDT) schemes is too large or too costly to bear. We researched the effects of software test effectiveness on the efficiency of SRDT. Continuous-type software was discussed, and a quantity model of minimal sample size in SRDT with software test effectiveness was given. It was demonstrated in detail that the results given by the model were better than the results given by the traditional Bayesian method with no prior knowledge. Finally, a statistical fault injection method which was used to estimate the test effectiveness was introduced.