Checking sequences for distributed test architectures

Abstract

The objective of testing is to determine whether an implementation under test conforms to its specification. In distributed test architectures involving multiple testers, this objective can be complicated by the fact that testers may encounter problems relating to controllability and observability during the application of a checking sequence. A controllability problem, also known as a synchronization problem, exists when a tester cannot determine when to send a particular input to the IUT. An observability problem exists when a tester cannot determine whether a particular output has been received from the IUT in response to the related input in the specification. Solutions in the literature to these problems generally require either appending additional input-output sequences or the use of external coordination messages between testers. This thesis proposes two methods for constructing checking sequences with no potential controllability and observability problems. The first method assumes the presence of a reliable reset in the implementation and constructs digraphs to facilitate the construction of state and transition cover subsequences. A Rural Chinese Postman Path on the final digraph constructed by the method yields a synchronizable ordering of these subsequences. The second method does not require a reliable reset and constructs digraphs to generate subsequences that verify each state and transition. An Euler Tour of the final digraph constructed by the method yields a checking sequence. In both methods, by considering controllability and observability problems during the construction of the checking sequence, the use of coordination messages is either minimized or, if possible, avoided altogether.