Predicting Software Reliability and Availability in Operation

摘要

This paper presents methods and procedures to estimate software reliability and availability in operation. Software reliability analysis is conducted based on the non-homogenous Poisson process (NHPP). Software system test data is used to estimate the software the number of residual faults and post system test data provides the estimate for the software failure rate. Software system reliability and availability can be predicted from these estimates. Both execution times and calendar times are considered and method of transferring from one testing time to another is presented. A telecommunications application is used to illustrate how to calculate the failure rate and the testing time to meet the software availability requirement.