Chinmay Gaikwad, Embry- Riddle Aeronautical University; Filipe Senra, Embry- Riddle Aeronautical University; Hao Peng, Embry- Riddle Aeronautical University
Keywords: Space Situational Awareness, LEO , Mega-constellations, OD, Space Surveillance and Tracking
Abstract:
With a rapid increase in unmanned and manned space activities, there is a growing demand for accurate and prompt space situational awareness (SSA) capabilities. Since the current SSA mainly relies on ground-based radars and telescopes, some intrinsic limitations can hardly be overcome, such as the atmosphere interference and the propagation attenuation due to distance. The authors have proposed a paradigm shift to enhance the SSA capability using spaceborne sensors deployed on a large scale.
In the new paradigm, mega-constellations are repurposed as platforms for spaceborne sensors to detect and track space objects. Mega-constellations consisting of tens of thousands of satellites in lower earth orbits (LEOs) have become a reality thanks to recent advances in technologies such as onboard processing power, phased array antenna, inter-satellite links, and so on. At present, the most massive mega-constellation is Starlink of SpaceX, with over 6,000 satellites in orbit and over 20,000 satellites planned for launch. More importantly, the mega-constellation is currently planned at an altitude as low as around 540 km and, therefore, below the majority of space objects. With the observation that powerful ground-based SSA radars are constantly illuminating space objects passing by, it is possible for passive radars deployed on a mega-constellation to detect the backward scattering and reflective nonmagnetic waves from space objects. Current spaceborne optical sensors or active radars cannot be adapted for massive deployment because of bulky sizes, high power demand, and high cost. Passive radars can be manufactured in a light-weight form factor since there is no need for an active feed, which makes them suitable for large-scale deployment.
There are several advantages of the new paradigm. First, the spaceborne sensors are above the atmosphere and closer to targets, so better resolutions and higher tracking abilities can be expected. Second, the large number of sensors will not only provide a broader coverage of the sky, but also generate an unprecedented amount of observation data that is crucial to enable machine learning and artificial intelligence (ML/AI) applications for space object detection and tracking. Third, the spaceborne sensors can be implemented as fully automated, and the ground segment only needs to track mega-constellation satellites accurately.
In this paper, the authors will report preliminary simulation results of onboard determination in a simulated environment. The authors will implement a mega-constellation inspired by that of Starlink and the entire space catalog with their relevant physical parameters. For the orbit, the number of satellites will be kept as an input variable to look at different configurations. The base configuration for the study will be thousands of satellites divided amongst different orbital planes in a near circular orbit with an eccentricity of 0.005. The distribution of the satellites amongst various orbital planes ensures uniform global distribution. A near circular orbit would help maintain nearly constant altitude and speeds ensuring coverage patterns remain stable over time. The open-source astrodynamics package Orekit is chosen to simulate the environment as it provides a complete control over the simulation environment. The authors will develop two models with different fidelities. For the lower fidelity model, the analytical SGP4 model will be utilized and refactored to run in parallel such that the thousands of satellite and space objects can be processed simultaneously. For the higher fidelity model, a numerical model with various force models representing key perturbations will be numerically propagated, including the gravitational field using the EGM96 model, the atmospheric drag modeled with the NRLMSISE00 model, the solar radiation pressure using the standard SRP model, and solar activity which depends on the object geometry and surface characteristics. Besides the computational burden, one great challenge is that all space objects need to be modeled with realistic cross sections and surface features to reflect radar waves emitted from ground-based radar transmitters.
The passive radar sensors attached to each satellite in the mega-constellation will measure the reflective waves from objects in higher orbits. This assumed configuration is because mega-constellation satellites are designed for ground communications with many antennas at the downside, but the upside is largely vacant. In this paper, the measurement model will be developed using insights from the current operational systems focusing on parameters such as frequency bands, transmitter power, radar waveform characteristics, antenna sizes, and system architecture. The measurement model will focus on angle-only tracking while incorporating a realistic field-of-view (FOV), relative speeds, observation frequencies, transmitter’s radio frequencies, radar signal strengths, signal-to-noise ratio, and any other hardware constraints. The ground-based illuminator will be modeled as a radio wave transmitter using real radar stations. The illuminated volume will be modeled as a cone or fan in the Earth-centered Earth-fixed (ECEF) frame. The FOV angle of each passive radar is a design parameter that will be surveyed. The detection criteria will be set such that when the target object enters the illuminated region and also falls inside the FOV of any satellite in the constellation, simulated measurements will be generated continuously.
For the proposed model, onboard orbit determination is preferred so that only processed orbit estimates will be beamed. Measurements consisting of azimuth and elevation angles in the local satellite frame are a typical angle-only OD problem. A Kalman Filter would be used after the Initial Orbit determination algorithm to filter and process the incoming measurement data. Kalman filter is chosen due to its ability to handle non-linearity in complex dynamic systems while efficiently managing high-dimensional state spaces. By utilizing a prediction-update framework, the filter would refine the estimation of states and reduce the errors associated with process and measurement noise. The ability of the filter to parallelize the computation helps ensure the operation of it in real time, making it ideal for on-board orbit determination and thereby significantly enhancing overall tracking performance.
The final paper will demonstrate the simulation environment with thousands of mega-constellation satellites equipped with passive radars and thousands of space objects with different physical characteristics. The paper will also present a preliminary analysis of the implemented IOD methods, focusing on the space objects detection rate, tracking accuracy, and the catalog coverage.
Date of Conference: September 16-19, 2025
Track: Space-Based Assets