Randomization for testing systems of systems

摘要

This paper applies randomization theory to the problem of selecting software test cases for software systems and applications in order to overcome the hurdle of high cost in testing componentized systems of systems (SoS). We have used a corner point semantics, which can approximate a proof of correctness - termed a pseudo-proof of correctness. Test cases for each component are designed to be mutually orthogonal, or randomized. Integration testing is performed through a composition of the test cases for components with some value-added test cases to cover integration aspects of the system. Integration testing is also designed in such a way that the testing algorithm is written in randomized form. The advantages offered by such randomization are ever present in the algorithm, programming language, integration, and workflow design.