pCSSL: A stochastic extension to MARTE/CCSL for modeling uncertainty in Cyber Physical Systems - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Article Dans Une Revue Science of Computer Programming Année : 2018

pCSSL: A stochastic extension to MARTE/CCSL for modeling uncertainty in Cyber Physical Systems

Résumé

Cyber-Physical Systems (CPSs) are networks of heterogeneous embedded systems immersed within a physical environment, thus combining discrete and continuous processes. As for any complex systems, the global system behavior is difficult to predict, in an analytical way, from the individual behaviors of its parts. A global analysis can only be done through a holistic process, via simulation for instance, requiring precise models of the parts and of their interactions. While the subsystems are usually expected to be fully deterministic, their interactions with the uncertain environment can be difficult to characterize precisely. We propose an approach to characterize the environment and its interactions through stochastic properties, while the discrete part remains fully determined. The novelty of our work is that we explore a more standard-based approach relying on SysML/MARTE. CCSL and logical clocks are used to identify synchronization points in the various heterogeneous UML diagrams. A CCSL specification expresses a set of possible behaviors. Refinement is performed by adding new constraints and thus reducing the set of possible behaviors. The classical MARTE/CCSL-based process explores the remaining solutions through simulation by applying a simulation policy. To help exploring the solution state-space, we propose a stochastic extension of CCSL, called pCCSL, to characterize the likelihood of different configurations to occur. Then, we use Statistical Model Checking to explore alternative solutions and drive the refinement process. We illustrate our proposition by modeling an energy-aware building, with different control strategies and occupant energy usage models. We explore the impact on the energy footprint of the different variants and control strategies.

Dates et versions

hal-01898202 , version 1 (18-10-2018)

Identifiants

Citer

Dehui Du, Ping Huang, Kaiqiang Jiang, Frédéric Mallet. pCSSL: A stochastic extension to MARTE/CCSL for modeling uncertainty in Cyber Physical Systems. Science of Computer Programming, 2018, 166, pp.71 - 88. ⟨10.1016/j.scico.2018.05.005⟩. ⟨hal-01898202⟩
129 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More