DEVS Distributed Modeling Framework - A Parallel Implementation for the Cloud
Robert Kewley, Neil Kester, Joseph McDonnell · 2016
This paper introduces DEVS Distributed Modeling Framework (DEVS-DMF), a publicly available implementation of DEVS for integrating simulation models as parallel and distributed microservices suitable for containerization and automated deployment. Based on the Parallel DEVS framework, DEVS-DMF explicitly enables asynchronous execution of distributed models via the actor model of computation. It is optimized for interoperability of models and flexibility of deployment, as opposed to performance. The DEVS-DMF framework provides two simplifying benefits to the simulation developer. It encapsulates internal state and synchronizes simulation execution to free the developer to focus on designing the flow of messages and encoding the state transitions and outputs generated by their flow. It also provides a natural model for partitioning the simulation into local processes, because each DEVS model, as an actor, performs a unit of execution on a separate thread each time it handles a message. An additional advantage of this framework is location transparency, its flexibility to deploy actors in a single Java Virtual Machine to scale up, or to deploy actors to different containers in the cloud to scale out. Two simulation implementations of DEVS-DMF include a parallel simulation test case and a combat weapons simulator that assesses the performance of alternative small arms weapons designs.