A TESTING METHODOLOGY FOR HIGH RELIABILITY SOFTWARE

SINCE SOFTWARE IS A CRUCIAL COMPONENT OF CURRENT ENGINEERINGSYSTEMS, HIGH-RELIABILITY SOFTWARE IS REQUIRED FOR NASA PROGRAMS. THE USE OF FORMAL SOFTWARE ENGINEERING PRINCIPLESAND OF CASE TOOLS HAVE YIELDED ONLY MODEST PROGRESS IN THIS AREA. THE USE OF REDUNDANT SOFTWARE COMPONENTS IS NOT SUFFICIENTLY DEVELOPED TO PROVIDE AN EFFICIENT ALTERNATIVE. A TESTING METHODOLOGY WILL BE DEVELOPED THAT WILL PRODUCE HIGH RELIABILITY SOFTWARE BY USING INTERACTIVE, PROOF-BASED CORRECTNESS TESTING, ENFORCED TESTING DIVERSITY, AND KNOWLEDGE-BASED TESTING. THIS METHODOLOGY IS APPLICABLE TO PARALLEL SOFTWARE AS WELL. RECENT REDUNDANT SOFTWARE EXPERIMENTS INDICATE THAT NO SINGLE TESTING STRATEGY PRODUCES SUFFICIENT TEST CASES TO ENSURE HIGH RELIABILITY. HENCE, A COMBINATION OF TESTING STRATEGIES IS REQUIRED FOR HIGH RELIABILITY. EXPERIMENTAL RESEARCH ALSO INDICATES THAT DYNAMIC TESTING IS INEFFICIENT IN DETECTING FAILURES THAT OCCUR INFREQUENTLY. KNOWLEDGE-BASED STATIC TESTING IS MORE EFFICIENT. ORACLES FOR ASSERTING CORRECTNESS OF HIGH-RELIABILITY SOFTWARE ARE DIFFICULT TO BUILD SINCE THEIRRELIABILITY MUST EXCEED THE SOFTWARE. THE FIRM HAS DEVELOPED A GENERALIZATION OF INTERACTIVE PROOFS WHICH SOLVE THIS PROBLEM. IN THE FUTURE, PARALLEL ARCHITECTURES WILL BE USED FOR CRITICAL SOFTWARE APPLICATIONS. SINCE SOFTWARE IS A CRUCIAL COMPONENT OF CURRENT ENGINEERINGSYSTEMS, HIGH-RELIABILITY SOFTWARE IS REQUIRED FOR NASA PROGRAMS. THE USE OF FORMAL SOFTWARE ENGINEERING PRINCIPLESAND OF CASE TOOLS HAVE YIELDED ONLY MODEST PROGRESS IN THIS AREA. THE USE OF REDUNDANT SOFTWARE COMPONENTS IS NOT SUFFICIENTLY DEVELOPED TO PROVIDE AN EFFICIENT ALTERNATIVE. A TESTING METHODOLOGY WILL BE DEVELOPED THAT WILL PRODUCE HIGH RELIABILITY SOFTWARE BY USING INTERACTIVE, PROOF-BASED CORRECTNESS TESTING, ENFORCED TESTING DIVERSITY, AND KNOWLEDGE-BASED TESTING. THIS METHODOLOGY IS APPLICABLE TO PARALLEL SOFTWARE AS WELL. RECENT REDUNDANT SOFTWARE EXPERIMENTS INDICATE THAT NO SINGLE TESTING STRATEGY PRODUCES SUFFICIENT TEST CASES TO ENSURE HIGH RELIABILITY. HENCE, A COMBINATION OF TESTING STRATEGIES IS REQUIRED FOR HIGH RELIABILITY. EXPERIMENTAL RESEARCH ALSO INDICATES THAT DYNAMIC TESTING IS INEFFICIENT IN DETECTING FAILURES THAT OCCUR INFREQUENTLY. KNOWLEDGE-BASED STATIC TESTING IS MORE EFFICIENT. ORACLES FOR ASSERTING CORRECTNESS OF HIGH-RELIABILITY SOFTWARE ARE DIFFICULT TO BUILD SINCE THEIRRELIABILITY MUST EXCEED THE SOFTWARE. THE FIRM HAS DEVELOPED A GENERALIZATION OF INTERACTIVE PROOFS WHICH SOLVE THIS PROBLEM. IN THE FUTURE, PARALLEL ARCHITECTURES WILL BE USED FOR CRITICAL SOFTWARE APPLICATIONS.