abstraction in strongmodelstrong checking multi-agent strongsystemsstrong.pdfVIP

Abstraction in model checking multi-agent systems Mika Cohen Mads Dam Alessio Lomuscio Department of Computing Access Linnaeus Center Department of Computing Imperial College London Royal Institute of Technology Imperial College London London, UK Stockholm, Sweden London, UK Francesco Russo Department of Computing Imperial College London London, UK ABSTRACT A major obstacle to model checking is the so called state We present an abstraction technique for multi-agent sys- explosion problem. Since model checking involves a search tems preserving temporal-epistemic specifications. We ab- through the state space of the system to be analyzed, model stract a multi-agent system, defined in the interpreted sys- checking becomes intractable for systems with sufficiently tems framework, by collapsing the local states and actions large state spaces. In fact, relatively small system descrip- of each agent in the system. We show that the resulting tions (“programs”) can induce intractable state spaces. abstract system simulates the concrete system, from which Often, however, some of the details of a large, intractable we obtain a preservation theorem: If a temporal-epistemic system are irrelevant to the property we aim to verify.