Collaborative Space Surveillance

This paper presents a space-based, space-surveillance study wherein the goal is to demonstrate the feasibility and scalability of the modeling and simulation of a distributed multi-agent multiple satellites tracking and prediction system. A flexible and modular system architecture that enables collaborative and efficient teaming among distributed agents is delineated. Hierarchical objective methodology is deployed to align the mission objectives with the diverse agents' capabilities and resources. A set of satellite platform and sensor configuration/models is considered. Detailed mathematical models of the satellite orbits including the mutual visibility function are simulated for combinations of GEO and LEO orbits. An Unscented Kalman Filter (UKF)/Distributed Unscented Information Filter (DUIF) for high-accuracy orbital determination and tracking is demonstrated to show that the LEO orbit estimation from the GEO satellite with only angle measurements based on UKF is an excellent approach. Simulation studies show that the rate of filter convergence depends on sample time period, initial error, process error, measurement errors as well as the relative geometry of the LEO and GEO satellite orbits.