Neil Dhingra, Orbit Logic Incorporated; Cameron DeJac, Orbit Logic; Alex Herz, Orbit Logic; Trevor Wolf, University of Texas; Brandon Jones, University of Texas at Austin
Keywords: Sensor Resource Management, Sensor Tasking, Space Domain Awareness, Space Situational Awareness, Government-Commercial Partnerships, Non-Traditional Sensor Networks, Catalog Maintenance
Abstract:
The growing proliferation of space objects, and the increasingly contested nature of space as a warfighting domain, make the Space Domain Awareness (SDA) mission more difficult because more objects must be monitored and because effectively monitoring the emerging diverse types of space objects requires more sensor tasking with more complicated requirements. To alleviate the larger burden that these considerations impose on government sensors, there has been much recent activity centered around fostering collaboration between the government and non-traditional sensor networks operated by commercial or international partners. Such partnerships seek to allow exquisite government sensors to focus on high-priority or sensitive tasks to ensure coverage so the entire space catalog is sufficiently monitored and so all tasking requirements are satisfied and to limit redundancy so that time from high-value sensing assets is not used when time from cheaper assets could be used to complete the same mission.
However, more sensing hardware and data sharing agreements themselves are not enough; intelligent sensor tasking is required to unlock the potential that these partnerships hold. To reduce the burden on exquisite government sensors, their planned tasking must be conditioned on partner sensor plans. To improve key catalog health metrics, plans must be generated that target these metrics directly.
Orbit Logics Heimdall SDA tasking software performs intelligent sensor network tasking that empowers such government-commercial and international SDA partnerships to impact space catalog health. Heimdall cooperatively tasks ground- and space-based sensors to optimize an SDA-specific figure of merit (FOM) that reflects catalog and/or mission objectives. Heimdall runs several optimization algorithms in parallel to generate different sensor schedules, and chooses the schedule that scores best in terms of the FOM. Recently, Orbit Logic demonstrated Heimdalls upgraded capability to ingest commercial provider plans from the LeoLabs, Inc. and Numerica Corporation commercial sensor networks and to output them to the Unified Data Library (UDL) allowing the government to condition their sensor scheduling on partner plans. Orbit Logic has also partnered with the University of Texas at Austin to enhance Heimdall for sensor scheduling that targets specific catalog improvement metrics, including those related to the predicted error uncertainty associated with different objects. This would allow operators to task sensors around the estimated accuracy of object orbit estimates, e.g., to ensure that the maximum error covariance across a catalog is below a certain threshold. Moreover, these requirements can be heterogeneous, e.g., so that objects in crowded orbits have better estimates, and they can interact with other requirements, e.g., so that objects that frequently maneuver are still persistently monitored with a minimum cadence.
To quantify the effect of tasking on space catalogs, we simulated data collection by model sensor networks representative of government and partner networks in different scenarios with different sensor tasking strategies. These include static scenarios, in which space objects behave nominally, as well as dynamic scenarios, in which observed actions, e.g., satellite maneuvers, prompt revised tasking requirements and updated scheduling to satisfy those requirements, e.g., elevated collection frequency on maneuvering objects. We simulated data collection using the different sensor schedules with a non-zero probability of missed detection. To generate space catalogs from these different sets of simulated data, we updated the object orbit estimates with the simulated collected data assuming perfect data association; the effect of inaccurate data association may be studied in follow on work. By deriving and comparing metrics from these resulting space catalogs, we illustrate the effect of different tasking strategies on catalog health.
The different sensor tasking strategies include: when government sensor network tasking is and is not conditioned on partner sensor network tasking, when tasking does and does not target specific mission objectives, such as limiting the catalog-wide maximum error uncertainty below a threshold, and different optimization algorithms used for generating task plans. If government tasking is not conditioned on partner plans, there is a risk that partner-provided data is redundant with government-collected data and that not all tasks are fulfilled. If tasking does not target specific mission objectives, there is a risk that sensor time is poorly used on collection tasks that yield low value-of-information data. Finally, the sensor scheduling problem itself is very difficult even simplified versions of it are NP hard so it is unlikely that any one sensor tasking algorithm will always generate effective plans. We demonstrate how different algorithms excel for different mission scenarios and how running several algorithms in parallel and choosing the best generated sensor schedule (as measured by an SDA-specific FOM) enhances the quality of the collected data.
The studies we present are enabled by Orbit Logics Heimdall Sensor Tasking software for SDA. Heimdall performs intelligent sensor scheduling and facilitates effective partnerships between the government and non-traditional data providers.
Date of Conference: September 14-17, 2021
Track: Dynamic Tasking