Assessment of a sampling method for measuring safety-critical software reliability

The paper addresses the issue of using sampling to assess the reliability of safety-critical software. Several factors are considered, such as the number of test cases required, modifications in the program, determining the operational profile, generating test cases, and checking the correctness of the output. The problems are classified into two broad categories: those that can be overcome via advances in technology and those that are more fundamental in nature. The general conclusion is that, in principle, a combination of testing and verification will be successful in measuring ultrahigh software reliability