Statistical Model Checking of Dynamic Networks of Stochastic Hybrid Automata

In this paper we present a modelling formalism for dynamic networksof stochastic hybrid automata. In particular, our formalism is based on primitivesfor the dynamic creation and termination of hybrid automata components duringthe execution of a system. In this way we allow for natural modelling of conceptssuch as multiple threads found in various programming paradigms, as well as thedynamic evolution of biological systems.We provide a natural stochastic semantics of the modelling formalism based on re-peated output races between the dynamic evolving components of a system. Asspecification language we present a quantified extension of the logic Metric Tempo-ral Logic (MTL). As a main contribution of this paper, the statistical model checkingengine of U PPAAL has been extended to the setting of dynamic networks of hybridsystems and quantified MTL. We demonstrate the usefulness of the extended for-malisms in an analysis of a dynamic version of the well-known Train Gate example,as well as in natural monitoring of a MTL formula, where observations may lead todynamic creation of monitors for sub-formulas