Red Cell Analysis for Mobile Networked Control Systems

In the near future, networked unmanned autonomous systems will increasingly be employed to support ground force operations. Approaches to collaborative control can find near-optimal position recommendations that optimize over system parameters such as sensing and communication to increase mission effectiveness. However, over time these recommendations can create predictable paths that may provide leading indications of the forces operational intent. Using time series forecasting methods and deep neural networks, this thesis conducts an adversarial assessment of unmanned mobile networked control systems. In the first scenario, the path of the teams ground motion predicted by the model follows the initially planned but not executed path. In a second scenario, the model achieves a maximum path error rate of only 75 meters. In both cases, this methodology correctly identifies the direction and distance the team would travel and even identified points where the team changed direction, allowing the autonomous red cell analysis to discern the ground forces intent. These results indicate that automated red cell analysis is a potentially valuable component in planning and executing unmanned mobile networked control systems supporting expeditionary ground teams. It provides near real-time feedback on the unmanned agents paths to determine if course adjustments can reduce operational intent predictability.