A Gaming Environment for Resilient Network Design and Adversarial Co-Evolution Modeling

Abstract

Abstract — The design and evaluation of resilient communication and computation infrastructures still relies heavily on empirical studies, and are often based on advanced simulation or emulation techniques that include external attacks or specific failure modes. While generally useful to provide an initial baseline, such static approaches fail to take into account the adaptive nature of the attacker. Recognizing the co-evolutionary nature of the attacker, we argue that, especially in the case of adversarial environments, an intelligent and adaptive attacker model must be considered for the test and evaluation of defense infrastructures. This paper describes the implementation of an adversarial gaming environment for the development, test, and evaluation of Resilience Network and Electronic Warfare procedures. The motivation for the gaming approach is based on the concept of attack co-evolution, and red teaming. In this work, we propose that serious games can be used to engage humans in high-fidelity adversarial simulations that will allow us to capture, and possibly model the co-evolutionary behavior of adversaries, and will provide the basis for the design of online electronic defense mechanisms that will take such effects into account.