Section: New Results
Keywords : Integrated Modular Avionics, metamodeling, Generic Modeling Environment.
A modeling paradigm for Integrated Modular Avionics design
Participants : Christian Brunette, Thierry Gautier, Jean-Pierre Talpin.
We previously addressed the design of applications based on the Integrated Modular Avionics (IMA) architecture [14] , [13] , which relies on the avionic standard APEX-ARINC [26] , [27] . This leads to the implementation of a library of components in Signal, providing real-time executive services defined by the APEX-ARINC standard.
Now, we carry out this library in the General Modeling Environment (GME) [18] . The primary purpose is to increase the usability of the library by proposing the same concepts within a non domain-specific tool such as GME. Therefore, without being an expert of synchronous technologies, a user could still be able to design applications based on the IMA modeling approach proposed in the Polychrony environment. Today, we observe that the attention of the industry tends to shift to frameworks based on general-purpose modeling formalisms (e.g. UML), in response to a growing industry demand for higher abstraction-levels in the system design process.
GME [38] is a configurable object-oriented toolkit, which supports the creation of domain-specific modeling and program synthesis environments. Metamodels are proposed in the environment to describe modeling paradigms for specific domains: basic concepts required for model representation from a syntactical viewpoint to a semantical one.
Our modeling paradigm for IMA design in GME, called MIMAD, is represented by the layer on the top in Figure 18 . The layers on the bottom are dedicated to domain-specific technologies. Here, we consider Polychrony, which is associated with Signal. However, one can observe that the idea is extensible to further technologies that offer specific useful functionalities to the MIMAD layer (e.g., the integrated environment Uppaal , which enables validation and verification of real-time systems using timed automata). As GME enables to import and export XML files, information exchange between layers can rely on this intermediate format. This favors a high flexibility and interoperability.
The MIMAD layer aims at providing a user with a graphical framework allowing to model applications using a component-based approach. Application architectures can be easily described by just selecting these components via drag and drop. Component parameters (e.g. period or deadline of an IMA process model) can be specified. The resulting GME model is transformed in Signal (referred to as Mimad2Sig in Figure 18 ) based on the XML intermediate format.
In the synchronous data-flow layer, the XML description obtained from the upper layer is used to generate a corresponding Signal model of the initial application description. This is achieved by using the IMA-based components already defined in PolychronyPolychrony [6] . Thereon, the formal analysis and transformation techniques available in the platform can be applied to the generated Signal specification. Finally, a feedback is sent to the MIMAD layer to notify the user with possible incoherences in initial descriptions.
