A Statistical Model Checker for Nondeterminism and Rare Events

Carlos E Budde,Arnd Hartmanns,Sean Sedwards,Pedro R D’Argenio

doi:10.1007/978-3-319-89963-3_20

Abstract

Statistical model checking avoids the state space explosion problem in verification and naturally supports complex non-Markovian formalisms. Yet as a simulation-based approach, its runtime becomes excessive in the presence of rare events, and it cannot soundly analyse nondeterministic models. In this tool paper, we present modes: a statistical model checker that combines fully automated importance splitting to efficiently estimate the probabilities of rare events with smart lightweight scheduler sampling to approximate optimal schedulers in nondeterministic models. As part of the Modest Toolset, it supports a variety of input formalisms natively and via the Jani exchange format. A modular software architecture allows its various features to be flexibly combined. We highlight its capabilities with an experimental evaluation across multi-core and distributed setups on three exemplary case studies.

Highlights

Statistical model checking (SMC [30,49]) is a formal verification technique for stochastic systems
We present a statistical model checker, modes, that addresses both of the above challenges: It implements importance splitting [45] to efficiently estimate the probabilities of rare events and lightweight scheduler sampling [39] to statistically approximate optimal schedulers
We describe the various methods implemented to make modes a correct and scalable statistical model checker that supports classes of models ranging from continuoustime Markov chain (CTMC) to stochastic hybrid automata in Sect

Summary

Introduction

Statistical model checking (SMC [30,49]) is a formal verification technique for stochastic systems. We present a statistical model checker, modes, that addresses both of the above challenges: It implements importance splitting [45] to efficiently estimate the probabilities of rare events and lightweight scheduler sampling [39] to statistically approximate optimal schedulers. Both methods can be combined to perform rare event simulation for nondeterministic models. The partial order and confluence-based methods have been replaced by LSS, enabling the simulation of non-spurious nondeterminism; automated importance splitting has been implemented for rare event simulation; support for MA and a subset of stochastic hybrid automata (SHA [22]) has been added; and the statistical evaluation methods have been extended and improved. Advances in the shared infrastructure of the Modest Toolset, at version 3, provide access to new modelling features and formalisms as well as support for the Jani specification

Ingredients of a Statistical Model Checker

Automated Rare Event Simulation

Scheduler Sampling for Nondeterminism

Experiments

Conclusion