Prediction of Safety Critical Software Operational Reliability from Test Reliability Using Testing Environment Factors

It has been a critical issue to predict the safety critical software reliability in nuclear engineering area. For many years, many researches have focused on the quantification of software reliability and there have been many models developed to quantify software reliability. Most software reliability models estimate the reliability with the failure data collected during the test assuming that the test environments well represent the operation profile. User's interest is however on the operational reliability rather than on the test reliability. The experiences show that the operational reliability is higher than the test reliability. With the assumption that the difference in reliability results from the change of environment, from testing to operation, testing environment factors comprising the aging factor and the coverage factor are developed in this paper and used to predict the ultimate operational reliability with the failure data in testing phase. It is by incorporating test environments applied beyond the operational profile into testing environment factors. The application results show that the proposed method can estimate the operational reliability accurately.