An extensible modeling methodology for embedded and CPS design
(English)Manuscript (preprint) (Other academic)
Abstract models are important tools to manage the increasing complexity of system design. The choice of a modeling language for constructing models governs what types of systems can be modeled and which subsequent design activities can be performed. This is especially true for the area of embedded electronic and cyber-physical system design, which poses several challenging requirements on modeling and design methodologies. This article argues that the ForSyDe methodology with the necessary extensions can fulfill these requirements and thus qualifies for the design of tomorrow’s systems. Based on the theory of models of computation and the concept of process constructors, heterogeneous models are captured in ForSyDe with precise semantics. A refined layer of the formalism is introduced to make its denotational-style semantics easy to implement on top of the commonly used imperative languages and an open-source realization on top of the IEEE standard language SystemC is reported. The introspection mechanism is introduced to automatically export an intermediate representation of the constructed models for further analysis/synthesis by external tools. Flexibility and extensibility of ForSyDe is emphasized by integrating a new timed model of computation without central synchronization, and providing mechanisms for integrating foreign models, parallel and distributed simulation, modeling adaptive, data-parallel, and non-deterministic systems. A set of ForSyDe features are demonstrated in practice and compared to similar approaches using two relevant case studies.
Modeling methodologies, Cyber-physical systems, Electronic system-level, Embedded systems, Formal models, Heterogeneous systems, Models of computation, Simulation, SystemC
IdentifiersURN: urn:nbn:se:kth:diva-155937OAI: oai:DiVA.org:kth-155937DiVA: diva2:763525
QS 20142014-11-152014-11-152014-11-17Bibliographically approved