CMU-ISR-22-103
Institute for Software Research
School of Computer Science, Carnegie Mellon University



CMU-ISR-22-103

Cyber-Forces, Interactions, Terrain:
An agent-based framework for simulating cyber team performance

Geoffrey B. Dobson

August 2022

Ph.D. Thesis
Societal Computing

CMU-ISR-22-103.pdf


Keywords: Agent-based modeling, simulation, military, cyber warfare, Cyber-FIT

Over the past decade, militaries all over the world have declared cyberspace a domain of war. This has led to the professionalization of cyber teams in combat arms. These cyber teams must understand how they control, defend, and maneuver within this new domain, cyberspace. Moreover, commanders must train and equip the cyber teams to be successful with clear expectations of tasking and success criteria. The definition of success, for cyber team performance, continues to be a very evasive concept. Military and private organizations are spending a large amount of resources on training cyber teams.The training comes in many forms, ranging from individual skill development to advanced tooling, to team-based exercises. Due to the lack of understanding of cyber team performance measures, determining which training is most beneficial is nearly impossible.

This research seeks to bridge the gap by 1) computationally defining the performance measures of cyber teams and 2) creating a software tool that can simulate the deployment of cyber forces into conflict. This would allow researchers to experiment with a multitude of variables such as team makeup, training status, adversary type, organizational interaction, behavioral theory, and cyber terrain factors. To accomplish this, I created the Cyber-FIT agent-based simulation framework. I describe all versions of the software, built in spiral development methodology to arrive at an architecture that can realistically simulate a cyber team deployment. I present the definitions and real-world applicability of the performance measures of cyber teams. A realistically scaled cyber conflict is simulated in order to collect synthetic data and analyze all of the measures. Several virtual experiments are conducted using the model to show its usefulness, along with a sensitivity analysis of the most important control variables. Finally, a validation of the Cyber-FIT model is presented.

202 pages

Thesis Committee:
Kathleen M. Carley (Chair)
Christian Lebiere
Greh Shannon
Leslie Blaha (Air Force Research Laboratory)

James D. Herbsleb, Director, Institute for Software Research
Martial Hebert, Dean, School of Computer Science


Return to: SCS Technical Report Collection
School of Computer Science

This page maintained by reports@cs.cmu.edu