Distributed Simulation of Complex and Scalable Systems: From Models to the Cloud

Simulation is a standard technique to understand or to analyze complex Discrete Event Systems (DES). Distributed simulation techniques try to improve the elapsed time of sequential simulations for large DES models by dividing a monolithic simulation application into communicating concurrent Logical Processes. The performance of the simulator is usually evaluated on the basis of the time needed and the involved resources to complete a simulation run. Additionally, cloud computing, under a pay-per-use model, introduces the costs of the resources that must be allocated to run the simulation. In this paper, a Petri Net based modeling methodology for complex systems is presented producing hierarchical and modular models. From this model, an elaboration process produces a heterarchical model for efficient execution of the simulation over cloud platforms using well known techniques. The required partitioning of the model may be subject to different criteria such as cost, elapsed time, and synchronization constraints, where the structural properties of the Petri Nets can aid in this task.