Orbit Determination for Space Surveillance

This thesis presents the assessment of candidates for orbit determination (OD) within the realtime simulation SpaceView & Analyst, developed at Fraunhofer FHR. The goal of this simulationis to investigate future scenarios for space surveillance, which implies the continuous monitoring of objects in space. This is necessary because most objects in space are non manoeuvrable,e.g. rocket bodies, and thus can pose a threat to active satellites. Five methods for obit determination were chosen from a group of candidates after a theoretical comparison concerning their expected accuracy and suitability to a real-time system. The chosen methods are: Least Squares (LS), Sequential-Batch Least Squares and the Kalman Filter (KF) in three variations, namely the Extendend KF (EKF), Linearised KF (LKF) and Controlled LKF. After the implementation, it was discovered that the EKF has the shortest execution time of the three Kalman Filters and it was also more accurate than the LKF. A focus of the OD analysis was on geographically distributed radar sensors at different locations. It was shown that is is possible to hand a track of a satellite over to a follow-up radar by the prediction of a future observation vector even from relatively short observation times. The success of this handover depends on the quality of the preceding measurement, which is also influenced by e.g. the observation geometry. The geographic spread between stations and thus a larger coverage along the orbit leads to an improved accuracy for an OD with measurements from multiple sites compared to a single site. Furthermore, an algorithm to correlate space object measurements with an orbit database was developed, which is exclusively based on orbit information at the current time and thus has a reduced computational burden compared to typical correlation methods in use. This new technique was successfully applied to solo flying satellites, but it was also able to distinguish between closely-spaced satellites within a constellation by the application of longer tracking times.