Integrating semantic properties within a Petri net based scheduling tool

We consider real-time applications with interacting periodic tasks which may contain conditional statements. When the behaviours of the tasks are linked to the context, they may be interdependent, and some theoretical configurations may actually be impossible. Our concern is to reduce the schedulability analysis to the only coherent behaviours. We thus propose a model-driven approach. We first model the application and the semantic constraints coming from the tests in the conditional statements. Then we propose a Petri net based model, whose aim is to enumerate the possible behaviours, and which takes the structural, the temporal and the semantic constraints into account.