Cost Optimizations in Runtime Testing and Diagnosis of Systems of Systems

In practically all development processes tests are used to detect the presence of faults. This is not an exception for critical and high-availability systems. However, these systems cannot be taken offline or duplicated for testing in some cases. This makes runtime testing necessary. This paper presents work aimed at optimizing the three main sources of testing cost: preparation, execution and diagnosis. First, preparation cost is optimized by defining a metric of the runtime testability of the system, used to elaborate an implementation plan of preparative work for runtime testing. Second, the interrelated nature of test execution cost and diagnostic cost is highlighted and a new diagnostic test prioritization is introduced.