Section: Overall Objectives

Context and motivations

The design of embedded software from multiple views and with heterogeneous formalisms is an ubiquitous practice in the avionics and automotive domains. It is more than common to utilize different high-level modeling standards for specifying the structure, the hardware and the software components of an embedded system.

Providing a high-level view of the system (a system-level view) from its composite models is a necessary but difficult task, allowing to analyze and validate global design choices as early as possible in the system design flow. Using formal methods at this stage of design requires one to define the suited system-level view in a model of computation and communication (MoCC) which has the mathematical capability to cross (abstract or refine) the algebraic boundaries of the specific MoCCs used by each of its constituents: synchronous and asynchronous models of communication; discrete and continuous models of time.

We believe these requirements to be met with the polychronous model of computation. Historically related to the synchronous programming paradigm (Esterel, Lustre), the polychronous model of computation, implemented with the data-flow language Signal and its Eclipse environment Polychrony, stands apart by the capability to model multi-clocked systems. This feature has, in turn, been proved and developed as one ability to compositionally describe high-level abstractions of GALS architectures.

The research and development performed in the team aim at completely exploiting this singularity and to implement its practical implications in order to provide the community with all benefits gained from this property of compositionality.

Our main research results are, first and foremost, to consolidate the unique capability of the polychronous model of computation to provide a compositional design mathematical framework with formal analysis and modular code generation techniques implementing true compositionality (i.e., without a global synchronization artifact as with most synchronous modeling environments) [42] , [2] , [13] .

The most effective demonstrations of these features are found in our recent collaborative projects Spacify, Opees and Cesar to equip industrial toolsets with architecture/functions co-modeling services and provide flexible and modular code generation services.

Our research perspectives aim at pursuing the research, dissemination, collaboration and technology transfer results obtained by the team over the past years and, in doing so, further exploit the singularity and benefits of our model of computation and maximize its impact on the academic and industrial community.