The Purdue System of Systems Laboratory (SoSL) has provided significant value-add for many sponsors in the area of medium-fidelity modeling and simulation--as well as novel metric design--for system of systems (SoS) problems. These problems frequently touch on common concepts including how best to deploy new systems into an enterprise SoS, how to re-allocate required functions across existing systems, and how to distinguish "good" configurations from "bad" configurations across scenarios or operational conditions. As a direct result of these commonalities across projects in the last 10 years, the Purdue SoSL has iteratively developed and refined a simulation engine called the Discrete Agent Framework (DAF). It's current iteration, DAF 2.0, has been successfully applied in several projects across multiple disciplines. This framework facilitates rapid development of low- to medium-fidelity simulation and visualization tools that heavily leverage inheritance to 1) be extensible to new and related problems of interest, 2) separate controlled or proprietary work from public research, and 3) provide a well-tested library of functions to prevent dedicating limited project resources to reinventing existing logic.
The proposed paper under development seeks to provide an update to Purdue SoSL capabilities and also showcase the ways in which a flexible computing environment complements SoS research. The rapid generation of simulation and visualization tools to tackle problems of interest is independently useful, but is insufficient for providing decision support in the area of SoS analysis and design. To fully explore SoS problems it is necessary to capture their structure, behaviors, and system interactions; characterize the design space of interest; and conduct a representative exploration of that design space with respect to metrics of interest to expose the underlying problem structure. In a representative multi-target tracking example, this paper will highlight the capability of DAF 2.0 to flesh out the problem space, and also identify the changing computational costs for each phase of SoS analysis across the lifecycle of the example problem. At the end of this research effort it should be clear that DAF 2.0 is a well-tailored solution to tackling SoS problems, and that it is further enabled by integration with a flexible computing environment to provide conclusions and insights of interest at a reduced monetary cost.
Kris Ezra is a Research Scientist at Purdue University in the Center for Integrated Systems in Aerospace who specializes in modeling and simulation approaches for system-of-systems problems. He received his B.S. (2010), M.S. (2011), and Ph.D. (2015) degrees in Aeronautics and Astronautics from Purdue University under the direction of Dr. Dan DeLaurentis. His dissertation, entitled “An Integrated Optimization Approach to the Problem of Sensors, Weapons, and Targets” explored large-scale policy trades between solution paradigms in the context of missile defense problems by leveraging agent-based modeling techniques and combinatorial optimization. His current research interests remain in this area and focus on mathematical modeling, simulation, and optimization for system-of-systems problems for aerospace- and defense-related problems.
Dr. Ezra has been an active member of the System of Systems Laboratory at Purdue University since 2007 and has accumulated more than 10 years of experience in agent-based modeling in this context. As a result of his research, he has accumulated a growing expertise in simulation design, software architecting, and more recently high-performance computing and system administration. Based on this growing skillset, Dr. Ezra co-founded Integrated Modeling Solutions Incorporated in 2017 which provides modeling and simulation consulting services and also leverages the Discrete Agent Framework (DAF) 2.0 that he co-invented in 2016. His research experience has been applied to multiple sponsored projects and activities through Purdue University including work for the MIRE corporation, the Missile Defense Agency, Sandia National Laboratories, Torch Technologies, the Jet Propulsion Laboratories, Lockheed Martin Company, and many others.