Matthew T. K. Koehler

Simulation in context: using data farming for decision support

Data farming leverages high performance computing to run simple models many times. This process allows for the exploration of massive parameter spaces relatively quickly. This paper explores a methodology to use data farming as a decision support tool. Data farming can be a highly effective in this role because it allows one to present to a decision-maker not only what may be the most likely outcome but what are possible outcomes, especially outliers that might have far reaching impact. The terrorist attacks of September 2001 are a good example of an outlier with very high impact. A case study is presented using a simple terrorist attack simulation and decision-maker utility model.

Leveraging agent based simulation for rapid course of action development

In the spring of 2005 a limited objective experiment was carried out to assess the feasibility of using agent based simulations to enhance co-evolutionary course of action development. In particular, relatively low fidelity simulations were employed to visualize the results of particular courses of action. Over four days multiple courses of action were developed by two opposing teams with similar force structures and then run against one another in an agent based modeling environment to test their ability to achieve the given mission. The results of the experiment indicate that there is significant potential for low fidelity simulations to stimulate objective thinking in course of action development

Using Generative Analysis for Homeland Security: Modeling the Possibilities and the Probabilities

Frequently problems of homeland security require systemic solutions. Tactics, techniques and procedures must be innovatively combined with the latest technological advances to meet an emerging and ever changing threat. This paper provides a simulation based systems engineering approach to evaluate the wide variety of combinations that complex solutions require. A briefcase study for the defense of a large venue is provided to illustrate the methodology described herein.

Cyber Defense as a Complex Adaptive System: A model-based approach to strategic policy design

In a world of ever-increasing systems interdependence, effective cybersecurity policy design seems to be one of the most critically understudied elements of our national security strategy. Enterprise cyber technologies are often implemented without much regard to the interactions that occur between humans and the new technology. Furthermore, the interactions that occur between individuals can often have an impact on the newly employed technology as well. Without a rigorous, evidence-based approach to ground an employment strategy and elucidate the emergent organizational needs that will come with the fielding of new cyber capabilities, one is left to speculate on the impact that novel technologies will have on the aggregate functioning of the enterprise. In this paper, we will explore a scenario in which a hypothetical government agency applies a complexity science perspective, supported by agent-based modeling, to more fully understand the impacts of strategic policy decisions. We present a model to explore the socio-technical dynamics of these systems, discuss lessons using this platform, and suggest further research and development.

Multiparty engineering is a contact sport

In this paper, the authors suggest that essential coordination aspects multiparty engineering can be formally modeled with utility and game theoretic approaches. The computational complexity of evolving utility theories which change the payoff structures and consequently the “game” that the boundedly rational participants are playing along with the stochastic nature of the decisions are investigated using an agent-based model.

Systems Engineering Solution for Technology Deployment to Combat Complex Threat Scenarios in the Homeland

Systems engineering has been applied to the problems faced by DHS, including defending the homeland against a complex threat. An initiative currently underway by DHS S&T will use systems engineering to perform analysis against a scenario prior to the field experiment. The systems engineering analysis will help define the specific configuration, location, and Concept of Operations (CONOPs) of the system. Results from the experiment will be fed back into the systems engineering framework architecture in order to refine and tune the analysis. This will give the systems engineers a higher confidence in the analysis and recommendations in future experiments as the system is improved, new sensors are added, and the threat becomes more complex.

Visualization of Systems of Systems Simulations: Density Plots and Trajectory Storyboarding

In the many cases where there are no solvable closed-form solutions for the dynamics of Systems of Systems (SOS), stochastic simulation may be the only practical solution to gaining insight into the systems interactions, processes and outcomes. Gaining this insight often requires producing many instantiations of the simulation, which, depending of the variability of the simulation, may produce huge amounts of data. Statistical techniques exist for summarizing the range of outcomes and identifying and analyzing various model states of interest. Often however, visualization techniques provide a more effective tool in gaining insight in the scope and result of the dynamic processes and interactions. The authors describe and present the use of Density Plotting (DP) and Trajectory Storyboarding (TS), visualization techniques for potentially extracting insights from SOS models.

Applying Complexity Science with Machine Learning, Agent-Based Models, and Game Engines: Towards Embodied Complex Systems Engineering

The application of Complexity Science, an undertaking referred to here as Complex Systems Engineering, often presents challenges in the form of agent-based policy development for bottom-up complex adaptive system design and simulation. Determining the policies that agents must follow in order to participate in an emergent property or function that is not pathological in nature is often an intensive, manual process. Here we will examine a novel path to agent policy development in which we do not manually craft the policies, but allow them to emerge through the application of machine learning within a game engine environment. The utilization of a game engine as an agent-based modeling platform provides a novel mechanism to develop and study intelligent agent-based systems that can be experienced and interacted with from multiple perspectives by a learning agent. In this paper we present results from an example use-case and discuss next steps for research in this area.

Epistemological Constraints When Evaluating Ontological Emergence with Computational Complex Adaptive Systems

Natural complex adaptive systems are of particular scientific interest in many domains, as they may produce something new, like structures, patterns, or properties, that arise from the rules of self-organization. These novelties are emergent if they cannot be understood as any property of the components, but as a new property of the system. One of the leading methods to better understand complex adaptive systems is the use of their computational representation. In this paper, we make the case that emergence in computational complex adaptive systems can only be epistemological, as the constraints of computer functions do not allow for the creation of something new, as required for ontological emergence. As such, computer representations of complex adaptive systems are limited in producing emergence, but nonetheless useful to better understand the relationship between emergence and complex adaptive systems.

Optimization of Heterogeneous Simulations of Complex Enterprise Systems

Attempting to optimize the design of a complex enterprise system is extremely difficult at best. The interconnections and human elements inherent in these systems make determining the impact of design elements and changes to these elements challenging. We suggest that a combination of sophisticated modeling and simulation techniques, including the use of agent- based models, and biologically inspired optimization techniques can be used to address the aforementioned difficulties. Finally, we discuss visualization techniques for these types of systems that enhance the ability of subject matter experts and decision-makers to understand the results.