Dynamic Environment Generations for an ASTRAL Process

ASTRAL is a high-level formal speciication language for real-time systems. It includes structuring mechanisms that allow one to build modularized speciications of complex real-time systems with layering. Based upon the ASTRAL symbolic model checker reported in DK99b], an approximation technique to speed-up the ASTRAL symbolic model checker for debugging a speciication is presented. The technique uses dynamic environment generation coupled with partial image approximations. A sequence of concrete environments for an ASTRAL process instance is randomly generated along each execution path in the execution tree of the ASTRAL process. Doing this greatly reduces the time for nding an error in a speciication, as demonstrated by a number of mutation tests, while still ensuring reasonable coverage of the search procedure. The results of the tests show that the techniques presented in the paper are eeective.