Session A4: PNT Solutions for Space Applications

Predicted Orbit Determination Performance of a Lunar Navigation System: Utilizing Inter-Satellite Measurements with Batch and EKF Estimation
Sungik Kim, Jisung Oh, Byungwoon Park, Sejong University
Location: Beacon A

The Moon is being considered an intermediate step toward Mars and future deep space missions, making it one of the most important destinations for space exploration. The number of missions targeting the Moon is expected to increase to about 400 over the next decade. For the success of these planned space missions, reliable Position, Navigation, and Timing (PNT) solutions are essential. Traditionally, the Deep Space Network (DSN), which relies on ground-based radiometric measurements from Earth stations, has been used to calculate position and velocity. However, the growing demand for lunar missions is expected to exceed the DSN's capacity, and its real-time solution accuracy may be inaccurate, with an error range of a few kilometers.
Given this situation, researches have been focused on using Earth Global Navigation Satellite System (GNSS) signals on the Moon. Employing GNSS signals in space offers several advantages, including reduced system complexity and operational costs, as it leverages existing infrastructure and equipment, minimizing the need for new developments. Also, several space agencies are developing Lunar PNT systems designed specifically for use on and around the Moon, in similar ways to Earth GNSS. Both NASA and ESA are actively working on initiatives such as the development of the LunaNet Interoperability Specification (LNIS), to establish standards for a future lunar communication and navigation framework called LunaNet. These efforts aim to ensure system interoperability between various future Lunar PNT service providers, allowing users to receive a unified service, regardless of the provider.
From this background, the Future Space Navigation & Satellite Research Center in Korea has been conducting simulation-based research on orbit determination for lunar navigation satellites as part of a broader Lunar PNT initiative. Through our internally developed simulation, we confirmed that the orbit determination error using an Extended Kalman Filter (EKF) based on GNSS measurements was less 100 meters in Near Rectilinear Halo Orbit (NRHO) and less 60 meters in Elliptical Lunar Frozen Orbit (ELFO). Based on these results, we further analyzed the orbit determination performance of lunar orbital vehicles when combining Lunar PNT measurements with inter-satellite measurements.
In this paper, we propose a reliable and practical two-step approach for orbit determination by utilizing Batch estimation to provide initial values for an Extended Kalman Filter (EKF)-based recursive estimation. This method allows for more accurate and stable orbit determination by combining the strengths of both techniques. First, Batch estimation provides a robust initial estimate, and then EKF refines this estimate through continuous error modeling by recursive measurement updates in dynamic environments. We present simulation results demonstrating the effectiveness of this approach, particularly in improving orbit determination accuracy for lunar orbital users.
The simulation consists of two parts: first, the actual trajectory of the lunar orbital user is generated using the General Mission Analysis Tool (GMAT), and then this trajectory is input into a Matlab-based measurement generation program to obtain observation data of lunar orbital user. The observation data is then input into the Matlab-based orbit determination simulation program, where the orbital filter used for orbit determination is set to have a lower accuracy than the orbit propagator used when generating the true trajectory in GMAT.