Model-driven Development of Discrete Event Simulations

Simulation is a common tool for modeling and evaluating real-life systems. In discrete event simulation (DES) a system is represented from the perspective of its events. Modern tools for developing simulations are bound to general purpose programming concepts that cannot represent the simulation domain effectively. In this paper, we propose a model-driven approach to DES domain. We provide a meta-model from scratch based on Ecore, an alternative one that utilizes UML 2.* profiling and a domain specific language grammar. Two example models demonstrate the use of our modeling language and provided model-to-text and model-to-model transformations to Java and DOT models, respectively, show that DES can effectively be expressed by a model-driven approach.