Braden Eichmeier, Ten One Aerospace; Ryan Goodman, Ten One Aerospace; Emily Gerber, Ten One Aerospace; Christopher Roscoe, Ten One Aerospace
https://doi.org/10.64861/SFDL3565
Keywords: Space Domain Awareness, Data Transformation Graphs, Graph Theory, Path Planning, Data Systems
Abstract:
Space Domain Awareness (SDA) data use cases are transitioning from linear flows to interconnected networks. In this approach, the resilience of the underlying data pipelines becomes critical for ecosystem health. Implementing these networks into existing real-world systems has faced difficulties due to a variety of challenges: 1) connecting data silos, 2) managing unreliable infrastructure, and 3) integrating new data schemas with existing analytics pipelines and platforms. These challenges are particularly prevalent in SDA applications where vendors provide data from sensors spread across the world to varying classification levels.
This paper applies data transformation graphs (DTG), a representation of complex data networks, to modern SDA data environments to mitigate these challenges. In the DTG paradigm, the data system is represented as a mathematical graph where nodes represent data types and edges represent functions or applications that convert data from one morphology to another. Edge weights can be dynamically parameterized based on operational constraints such as computation time, historical service reliability, functional performance, and API financial cost. This is in contrast to common architectural representations, where nodes represent platforms and systems, and data flows out from these nodes. The infrastructure-centric representation obfuscates the pathway from data generation to consumption and can hide weaknesses in the pipeline for specific operational use cases.
We describe how the general framework of DTGs applies across a variety of data architectures and design patterns, from embedded flight software to enterprise cloud-native environments. We demonstrate the utility of this approach through two illustrative examples modeled from operational use cases. The first example shows tactical level benefits where we implement a dynamic DTG-based data computation workflow in a Rendezvous and Proximity Operations (RPO) detection and classification pipeline. This pipeline receives requests from multiple schema, pulls data from two two-line element (TLE) providers, namely SpaceTrack and the Unified Data Library (UDL), and applies machine learning to detect and classify RPO maneuvers. We use an optimal path-finding algorithm to process the request from raw input to enriched output. By modifying the edge parameterization, we show multiple optimal paths to prioritize speed, cost, or accuracy in response to mission needs. Then, we show a self-healing benefit of the framework where the computation path-finding automatically adapts to API service failures. For example, if UDL is the optimal data provider and experiences an outage, the pathing algorithm responds by rerouting the computation path through SpaceTrack as a fallback mechanism without human intervention. Finally, we show how the DTG framework reduces the complexity of adapting to schema changes and integrating new data sources with existing systems.
The second example focuses on a strategic-level benefit where leadership can readily digest quantifiable metrics of system status. We model a notional large-scale and complex SDA data system. Using various metrics, including algebraic connectivity and bridge analysis, we quantify fragility and failure points within the system. Then, we perform simulation scenarios to measure the resilience of these points of failure. Finally, we quantify and show the expected benefits of potential additions to the system. Using this framework and process, leaders have quantitative metrics to guide investment decisions towards improving the robustness of data systems.
Date of Conference: September 15-18, 2026
Track: Space Domain Awareness