Making Modeling Assumptions an Explicit Part of Real-Time Systems Models
Résumé
Modeling is an intellectual process that consists in making abstractions. A model conveys a point of view of one system in order to assist a designer in his or her attempt to master the complexity of the system. The statement particularly applies to real-time systems that capture complex problems in terms of parallelism, synchronization, distribution and time constraints.
Making abstractions implies making assumptions. In other words, a model works for a precise set of assumptions. For instance, a pressure controller model may be valid under the assumption the pressure sensor it is connected to never fails. Such an information is important to share the model with other people and to make the model easy to understand.
Whether the importance of assumptions has regularly been acknowledged in the literature, the inclusion of modeling assumptions into models has not been formally proposed and discussed.
In the paper, we conversely support the idea of making assumptions an explicit part of the model and we insist on the importance of assumptions in incremental modeling processes.
We also consider that modeling assumptions play a key role in incremental modeling. Therefore, the approach of "versioning" discussed in the paper make it possible to take the way assumptions evolve into account (throughout the entire modeling process).
The first part of the paper gives general principles and proposes a meta-model that allow including assumptions in a diagrammatic shape that may accompany a broad variety of models. Then, we show how to add modeling assumptions to real-time systems models expressed in SysML and we implement the approach in the TTool tool. An existing UAV platform, aiming at autonomously navigating in buildings, serves as a case study.
Domaines
Systèmes embarqués
Origine : Fichiers produits par l'(les) auteur(s)
Loading...