Changes in Satellite Ecosystem Health Following Space Weather Events

The objective of this proposal is to determine the feasibility of integrating cutting-edge data science, machine learning, and business intelligence technologies to enhance Space Domain Awareness and decision making. We fuse vast quantities of diverse space data from diverse sources including environmental, positional, and systems health and status, and analyze at the space ecosystem level to derive meaningful incites and predictions for decision making. This will help enable space operators and decision makers to close knowledge gaps, characterize space system activities, understand space environmental impacts to operations, predict future impacts to space services, and then rapidly make informed decisions. Our research addresses two current Air Force R&D focus areas – Focus Area 3 - exploration of AI, Machine Learning, and data science tools for situational awareness and decision making, and focus area 23 - AI and Machine Learning dual use technologies for Air Force Application. The problem addressed is that large volumes of space data from diverse sources is not being analyzed or utilized, and may never reach space-system operators, decision makers, and service customers who would benefit, in a meaningful way.  This routinely creates significant knowledge gaps in Space Domain Awareness (SDA) from satellite positioning and health relative to the larger ecosystem, to an understanding of all ecosystem effects, to threat monitoring. The combination of having too much data to analyze, a lack of understanding and characterization of our space systems within the context of a larger dynamic environment, and the fragmented nature of our networks, has set the conditions for a need to approach the protection of our space assets and understanding of the space environment in a new AI enabled way. We apply automated data fusion techniques to aggregate data from diverse sources. Then we apply Process Mining to map complex space ecosystems and generate pattern of life understanding. Next, we use Machine Learning to build data models that describe the cause and effect relationships between internal and external forces to predict outcomes. Our solution, while protected by Trade Secrets, is built using standing programming languages, an IEEE data format, widely adopted studies in Artificial Intelligence, and cloud-compatible data management tools.  Our development environment enables the rapid prototyping, testing and evaluation of process improvements and the broad dissemination of results across other network architectures.   Lastly, our tools can be developed/deployed on the Air Force Chief Data Office’s Visible Accessible Understandable Linked Trusted (VAULT) platform for dissemination to user.  This architecture not only ensures the broadest access to the data tools within the Air Force, but also helps expand the utility of the data tools to address other concerns within the Space and Air Domains.