Scalable and reusable models for HLA-Ptolemy cosimulation framework

The Ptolemy-HLA framework aims to ease the creation of distributed simulation by providing an easy an user friendly interface for creating federate.One of HLA's end-goal is to have reusable simulators which can be used from one simulation to another. This requires to make no assumptions on the number of objects a simulator will have to handle. While this is not an issue with classical programming languages such as C++ or Java which can use dynamic memory allocation, but there is no such thing within Ptolemy models.Thus, there is a mandatory scaling phase : the model's designer will have to manually change the model in order to make it fit to the targeted simulation for handling the right number of objects. This phase is dull, error-prone and can be automated, which is this work's aim. To do so, our solution is to dynamically create actors in the model from a class upon an object's discovery. That way, the end user needs only to design a minimal model and it will automatically scales up to the right scale. We describe the associated abstract semantic and its implementation.