Previous Abstract Return to Session A4b Next Abstract

Session A4b: LEO PNT 2: Precise Orbit Determination and Interference

Real-Time Kinematic Orbit Determination Performance Analysis of LEO Satellite Equipped with GNSS Receivers
Hyung-Seok Lee, Department of Geoinformatic Engineering, Inha University; Kwan-Dong Park, Department of Geoinformatic Engineering, Inha University, PP-Solution Inc.
Location: Prince Edward Room
Date/Time: Wednesday, Apr. 15, 3:17 p.m.

Recent positioning technologies utilizing low-Earth orbit satellites are gaining attention as a crucial alternative capable of overcoming the limitations of the Global Navigation Satellite System (GNSS). While GNSS is a widely used positioning system globally, it suffers from significantly reduced positioning accuracy in environments such as between tall buildings in cities or in heavily forested areas, due to signal blockage or the effects of multipath errors. Additionally, various factors such as changes in the ionosphere and troposphere, and satellite clock errors, affect GNSS positioning accuracy. In such environments, Low Earth Orbit (LEO)-based positioning technology can provide more accurate location information by leveraging its high speed and low altitude characteristics. Consequently, related research and technological development are actively progressing. Specifically, LEO-based PNT (Positioning, Navigation, and Timing) systems offer advantages over standalone GNSS, such as reduced initial convergence time and improved signal availability, enabling stable positioning even in environments where GNSS is vulnerable.
However, since the position and velocity information of LEO satellites directly impacts positioning accuracy, Precise Orbit Determination (POD) for these satellites is essential. POD can be performed using kinematic Precise Orbit Determination (K-POD), which utilizes GNSS receiver observation data onboard the satellite; dynamic Precise Orbit Determination (D-POD), which accounts for various perturbations like gravity models and solar radiation pressure; or reduced-dynamic Precise Orbit Determination (RD-POD), which combines observation data with physical models. While various LEO satellites for different purposes are currently in operation, there are significant differences in the scope and format of data provided by each satellite. For example, large-scale communication satellites like Starlink and OneWeb primarily provide Two-Line Element (TLE) data, which contains only simple orbital elements and thus has limitations for precise orbit determination. Conversely, Earth observation satellites like CHAMP, Swarm, and Sentinel release data observed by their onboard GNSS receivers. Utilizing this data enables Real-Time Kinematic Orbit Determination (RTK-OD) based on GNSS observation data.
This study utilizes the publicly available data from Earth observation LEO satellites to conduct a detailed analysis of the characteristics of the GNSS receiver observation data files carried by each satellite and compares the differences in GNSS system configurations contained within the observation data. Furthermore, it systematically analyzes potential error factors that may arise during the orbit determination process based on the characteristics of the observation data and evaluates the real-time kinematic orbit determination performance for each satellite.
Data collection utilized the Crustal Dynamics Data Information System (CDDIS) and the European Space Agency (ESA) data archives. GNSS observation Receiver Independent Exchange Format (RINEX) data and Standard Product 3 (SP3) data were acquired from the CHAMP, Swarm-A, Swarm-B, Swarm-C, and Sentinel-6A satellites, respectively. The data used in this study for the CHAMP satellite were collected on June 18, 2008; the Swarm satellites on April 16, 2024; and the Sentinel-6A satellite on April 20, 2025. Data for each satellite were collected at 1-second intervals over a 12-hour period starting at midnight.
The CHAMP satellite's RINEX files are in version 2.0 and have a relatively simple header structure. They contain only L1, L2, P1, P2, and C1 observation data from the GPS system and are jointly provided by GFZ and JPL. This data is collected from the zenith-pointing GPS antenna aboard the CHAMP satellite. In contrast, data from the Swarm-A, B, and C satellites is in RINEX version 3.0, following the standard format provided by the ESA. It includes various observation types such as L1C, L1W, and L2W from the GPS system, and the data provided is corrected for the receiver's internal delay (Hardware Delay). Data from the Sentinel-6A satellite is provided in RINEX version 3.03 and includes not only GPS but also Galileo observation data. The GNSS receiver type is PODRIX, which applies observation settings optimized for real-time satellite orbit determination. Phase correction (Phase Shift) information is included for each frequency, and a constant phase bias of 0.325 cycles is added to the L2L signal. These various data formats and correction information directly influence the differences in kinematic orbit determination accuracy between satellites.
This study performed LEO kinematic-OD based on Standard Point Positioning (SPP) using code pseudorange observations. The position, velocity, and clock errors of LEO satellites were estimated using Least Squares Estimation (LSE). The positions from the SP3 file were set as the true values and compared with the estimated satellite positions at one-second intervals. Low Earth Orbit satellites are minimally affected by the troposphere and experience less ionospheric influence than ground receivers; therefore, no additional corrections were applied.
The CHAMP satellite was operated in a near-polar orbit at an altitude of approximately 454 km and an inclination of about 87°. The CHAMP satellite’s kinematic orbit determination results indicate RMSEs of 0.78 m in the radial direction, 0.89 m along-track, 0.76 m cross-track, and 1.41 m in 3D, with an average of 8 GNSS satellites used.
For the Swarm satellite constellation, Swarm-A and Swarm-C operate in a polar orbit formation, flying in formation at an altitude of approximately 460 km and an inclination of approximately 88°, with a longitude difference of about 1.4°. Swarm-B satellite also uses a polar orbit but operates alone, at an altitude of approximately 550 km and an inclination of approximately 87°. The kinematic orbit determination results indicate RMSEs for the Swarm-A satellite of 0.71 m in the radial direction, 0.72 m along-track, 0.92 m cross-track, and 1.37 m in 3D. Similarly, the Swarm-C satellite shows RMSEs of 0.68 m in the radial direction, 0.52 m along-track, 0.90 m cross-track, and 1.24 m in 3D, while the Swarm-B satellite exhibits 0.78 m, 0.65 m, 0.87 m, and 1.34 m, respectively. The error levels among the three satellites were similar, with an average of 8 satellites used.
The Sentinel-6A satellite operates in a non-sun synchronous Orbit at an altitude of approximately 1336 km, minimizing ionospheric effects, with an inclination of about 66°. Kinematic orbit determination results show RMSEs of 0.81 m in the radial direction, 0.89 m along-track, 1.01 m cross-track, and 1.57 m in 3D. This satellite utilized both GPS and Galileo data, resulting in a total of 15 satellites being used. CHAMP and Swarm satellites, which use polar orbit, pass over the polar regions and experience varying relative position changes with GNSS satellites. However, since they operate at low altitudes, they are more affected by ionospheric delays, resulting in somewhat lower orbit determination accuracy. Conversely, satellites currently in operation that can utilize multiple satellite signals, such as Sentinel-6A, have high-quality observation data and a large number of visible satellites, leading to relatively higher orbit determination accuracy.
This study compared and analyzed the real-time kinematic orbit determination performance of various satellites using publicly available GNSS observation data from LEO satellites. This demonstrates the feasibility of real-time orbit determination using On-Board GNSS receivers on LEO satellites. It is anticipated that applying precise positioning techniques in the future will further enhance the accuracy of real-time kinematic precise orbit determination. Furthermore, this study can serve as foundational data for LEO-based positioning research using public data and provide crucial evidence for designing LEO-GNSS multi-layer PNT systems and developing satellite selection strategies. Specifically, the correlation analysis between the observational data characteristics of each satellite and its orbit determination performance is expected to be utilized in establishing optimized orbital configurations, inter-satellite communication, and observational data utilization strategies during future LEO constellation designs.

Previous Abstract Return to Session A4b Next Abstract