A novel algorithm for flattening virtual subsystems in Simulink models

Recently embedded systems are often modeled using Simulink^® to simulate their behavior. In order to perform the simulation, the modeling tool has to process the model. An important processing step is to determine the execution order of the elements in a model. This execution order is based on a sorted list of all semantically relevant model elements. Therefore, before simulation, Simulink^® removes all model elements that only have a syntactic implication. In Simulink, the virtual subsystems are composite elements with no semantic bearing. Thus, Simulink performs a flattening model transformation that eliminates virtual subsystems. The work presented in this paper provides a novel algorithm for flattening composite elements in hierarchical models. Moreover, an optimized algorithm is also presented for Simulink models. With the implementation of these algorithms the level of abstraction of the model transformation can be raised. In this manner, a reusable, platform independent solution can be achieved for flattening Simulink subsystems.