Section: New Software and Platforms


Keywords: Energy - Statistics - Security - Runtime Analysis - Model Checker - Statistical - Model Checking - Aeronautics - Distributed systems

Scientific Description: Statistical model checking (SMC) is a fast emerging technology for industrial scale verification and optimisation problems. SMC only requires an executable semantics and is not constrained by decidability. Therefore we can easily apply it to different modelling languages and logics. We have implemented in PLASMA Lab several advanced SMC algorithms that combine formal methods with statistical tests, which include techniques for rare events estimation and non-deterministic models. PLASMA Lab comes with a simulator plugin that allows to verify LLVM code.

Functional Description: PLASMA Lab is a compact, efficient and flexible platform for statistical model checking of stochastic models. PLASMA Lab includes simulators for PRISM models (Reactives Modules Language-RML) and Biological models. It also provides plugins that interface external simulators in order to support Matlab/Simulink, SytemC and LLVM . PLASMA Lab can be extended with new plugins to support other external simulators, and PLASMA Lab API can be used to embed the tool in other softwares. PLASMA Lab provide fast SMC algorithms, including advanced techniques for rare events simulation and nondeterministic models. These algorithms are designed in a distributed architecture to run large number of simulations on several computers, either on a local area network or grid. PLASMA Lab is implemented in Java with efficient data structures and low memory consumption.

News Of The Year: In 2018 Tania Richmond and Louis-Marie Traonouez have extended PLASMA Lab to propose statistical model checking analysis of discrete time Markov chains with unknown values (qDTMC). We have defined a new logic, called qBLTL, that extends the semantics of BLTL properties to take care of the unknown information in the path of the qDTMC. We have also adapted the model checking algorithm of probabilistic model checking of incomplete models to perform a three hypotheses test and provide bounds on the probability of errors of this test.

  • Participants: Jean Quilbeuf, Louis-Marie Traonouez, Tania Richmond, Sean Sedwards, Benoît Boyer, Kevin Corre, Matthieu Simonin and Axel Legay

  • Contact: Tania Richmond

  • URL: https://project.inria.fr/plasma-lab/