Model-based design of IoT systems with the BIP component framework
Résumé
The design of software for networked systems with nodes running an Internet of Things (IoT) operating system faces important challenges, due to the heterogeneity of interacting things and the constraints stemming from the often limited amount of available resources. In this context, it is hard to build confidence that a design solution fulfills the application's requirements. This paper introduces a design flow for web service applications of the Representational State Transfer (REST) style that is based on a formal modeling language, the BIP (Behaviour, Interaction, Priority) component framework. The proposed flow applies the principles of separation of concerns in a component-based design process that supports the modular design and reuse of model artifacts. The BIP tools for state space exploration allow verifying qualitative properties for service responsiveness, i.e. the timely handling of events. Moreover, essential quantitative properties are validated through statistical model checking of a stochastic BIP model. All properties are preserved in actual implementation by ensuring that the deployed code is consistent with the validated model. We illustrate the design of a REST sense-compute-control application for a Wireless Personal Area Network architecture with nodes running the Contiki OS. The results validate qualitative and quantitative properties for the system and include the study of error behaviours.
Domaines
Systèmes embarqués
Origine : Fichiers produits par l'(les) auteur(s)