Christopher Ingram, ExoAnalytic Solutions, Inc.; Jaycie Bishop, ExoAnalytic Solutions, Inc.; Phillip Cunio, ExoAnalytic Solutions, Inc.; Doug Hendrix, ExoAnalytic Solutions, Inc.; Mark W. Jeffries, ExoAnalytic Solutions, Inc.
Keywords: STM, SSA, change detection, object characterization
Abstract:
Space activity has grown dramatically in the past decade. This is driven by a number of trends, including a lowering of the cost of access to space and the cost of space systems themselves. As a result of these trends, the operators of space systems today face an environment more challenging and congested, with a wider variety of space operations, leading to a need for Space Traffic Management (STM). The associated technical and organizational challenges to STM are manifold. There are an increasing number of governmental and commercial actors operating platforms in space, and these systems are an increasing part of critical infrastructure in many parts of the globe. A traditional model of Space Situational Awareness (SSA) data collection which focuses on catalog maintenance and conjunction assessment will not be sufficient for affirmative flight safety. Flight safety services which constitute traffic monitoring and control ultimately dictate dedicated monitoring frameworks and have little tolerance for data gaps. Data strategies which collect the minimum amount of measurements required to provide conjunction warnings should not be the basis for a modern approach to STM.
With over 300 telescopes located worldwide, ExoAnalytic Solutions provides persistent SSA for space objects above 8,000 kilometers altitude with minimal solar exclusion gaps using the ExoAnalytic Global Telescope Network (EGTN). This persistent observation capability, operational since 2014, provides a basis for commercial services that can enable true Space Traffic Management. Geosynchronous Earth Orbit (GEO) is a domain suitable for the development and testing of STM services because it can be addressed comprehensively today.
Starting from algorithms based on the empirical observation of orbital behaviors both routine and anomalous, an alert taxonomy has been developed using data collected on the entire GEO object population since 2014. This taxonomy was put into operational use starting in 2017 with the EGTN capture of the Telkom-1 anomaly. As a Space Traffic Management service, this system is referred to as ExoALERT (ExoAnalytic AI-assisted human-in-the-Loop Exploitation and Reporting Tool). The technical underpinnings of this alerting system, which starts from persistent observation data, and includes automated correlation and object characterization, will be discussed in this paper. The alert taxonomy includes single object alerts, both astrometric and photometric in nature, as well as multi-object alerts such as conjunctions, various types of orbit matching maneuvers, and proximity operations.
The integrated workflow between the automated alerting system, which autonomously processes on the order of 30TB of raw data each day, and human analysts will be described. The alerting system is designed to manage leaks and false alarms by introducing human interactions only where needed for reviewing real-time alerts to override the AI/ML reasoning and inference engine. Alerts are presented to human analysts at a manageable rate, and can be reviewed in analyst-readable and machine-readable form. Looking back over the past year of operations, the frequency of different types of automated alerts across categories will be presented and discussed. A comparison with the alerts curated by human analysts will be shown. Performance estimates for selected alert categories in terms of Type I and Type II errors will be discussed.
Date of Conference: September 27-20, 2022
Track: SSA/SDA