Mridul Songara, Digantara; Rithwik Neelakantan, Digantara; Latha Sridhar, Digantara; Sivalinga Raja Shanmugam, Digantara; Tanveer Ahmed, Digantara; Anirudh Sharma, Digantara
Keywords: Pattern of Life, Maneuver Pattern Violation, Residence Orbit Clustering, Maneuver Quantification
Abstract:
An approach for detecting anomalous behaviors of Resident Space Objects (RSOs) through the analysis of historical orbital patterns is presented. In the increasingly congested and contested space domain, the ability to characterize the deviations from established Pattern of Life (PoL) behavior has become critical for enhancing Space Domain Awareness (SDA). Identifying deviations from normal orbital behavior shall assist in analyzing the intent and denies the first-mover advantage to the adversary. The solution approaches, as suggested in the literature, ranging from traditional statistics-based methods to sophisticated Artificial Intelligence methods to identify anomalous behaviors face a variety of challenges including scalability, generalization, explain-ability, speed, and integration with multi-modal data types. A hybrid approach is proposed to address some of these challenges, accomplishing the following two goals –
- Identify patterns of orbital residence to establish a comprehensive behavioral baseline of the RSOs, using unsupervised learning
- Develop a technique to identify and quantify statistically significant deviations from the established maneuver PoL
The proposed framework utilizes the publicly available data sources including historical Two-Line Elements (TLE) sets from Space-Track.org. The orbital history is clustered using a density-based method to differentiate the orbital residence of the RSO. Features extracted from the historical data are further used to assign a distance-based metric to each maneuver conducted by the RSO. The method has been validated using known maneuver violations – including the flagged PoL violation of Cosmos 2576 (NORAD ID 59773) maneuvers in February 2025. Evaluation with the proposed methodology showed different orbital clusters of the satellite, as well as the maneuvers which violated the historical pattern – identified as those falling outside the region of 95% confidence in the distribution of the historical maneuvers. It was further revealed that the maneuvers conducted by the same object in December 2024 resulted in an even higher PoL violation score, indicating anomalous maneuvers of higher severity.
The resulting framework is designed to operate in varying levels of prior knowledge about space objects, from well-documented RSO behavior to minimally characterized RSO behavior, with a scope to evolve with the changing space environment, incorporating new observations to refine PoL baseline and adjusting sensitivity thresholds based on operational needs. The outcomes of this research will have direct applications to several critical aspects of SDA, including a high-fidelity intent assessment. As the space domain continues to evolve with new actors, technologies, and operational concepts, Pattern of Life analysis is expected to become an increasingly essential component of comprehensive SDA systems.
Date of Conference: September 16-19, 2025
Track: Satellite Characterization