Variational Bayesian Estimation of Low Earth Orbits for Satellite Communication 4 Amélia Struyf and Anders Malthe Westerkam are co-first authors.
Anders Malthe Westerkam 1 4 , Amélia Struyf 2 4 , Dimitri Lederer 3 , Troels Pedersen 1 , François Quitin 2 · 2025
Read Variational Bayesian Estimation of Low Earth Orbits for Satellite Communication 4 Amélia Struyf and Anders Malthe Westerkam are co-first authors. by Anders Malthe Westerkam 1 4 , Amélia Struyf 2 4 , Dimitri Lederer 3 , Troels Pedersen 1 , François Quitin 2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-27345.txt.
Introduction
Anders Westerkam, Amelia Struyf, Dimitri Lederer, Troels Pedersen, and Francois Quitin track low Earth orbit satellites for millimeter-wave communication from a ground station. Such satellites cross the sky in tens of minutes, and large hybrid analog-digital antenna arrays must steer beams rapidly to follow them. The team presents a variational message passing algorithm for joint localization and beam tracking.
The algorithm estimates parameters of a circular orbit model, from which pointing angles follow directly. Simulations show high resilience to missed detections, reliable tracking even near the horizon, and effective resolution of ambiguities inherent in hybrid architectures. Reliable orbit-based tracking matters because beam misalignment at millimeter-wave frequencies quickly breaks the link, and visibility windows leave little time to reacquire.
Catalog record
| CREATOR | Anders Malthe Westerkam 1 4 , Amélia Struyf 2 4 , Dimitri Lederer 3 , Troels Pedersen 1 , François Quitin 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-27345
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Variational Bayesian Estimation of Low Earth Orbits for Satellite Communication 4 Amélia Struyf and Anders Malthe Westerkam are co-first authors. Anders Malthe Westerkam 1 4 , Amélia Struyf 2 4 , Dimitri Lederer 3 , Troels Pedersen 1 , François Quitin 2 Abstract Low-earth-orbit (LEO) satellite communication systems that use millimeter-wave frequencies rely on large antenna arrays with hybrid analog-digital architectures for rapid beam steering. LEO satellites are only visible from the ground for short periods of times (a few tens of minutes) due to their high orbital speeds. This paper presents a variational message passing algorithm for joint localization and beam tracking of a LEO satellite from a ground station equipped with a hybrid transceiver architecture. The algorithm relies on estimating the parameters of the orbit, which is modelled as circular. Angles are then obtained from the orbit in a straightforward manner. Simulation results show that the proposed method is highly resilient to missed detections, enables reliable satellite tracking even near the horizon, and effectively alleviates the ambiguities inherent in hybrid architectures. I Introduction Satellite based communication systems hinges on the ability to accurately steer the beam of the ground stations (GS) antenna to the direction of the satellite. In general antennas with high directivity are employed in a quest to minimize transmission losses and at the same time avoid interference. Considering non-geostationary satellites, the ground station should be able to track the satellite movement close enough to avoid losing connection to the satellite. This is a particularly pressing concern in communication systems employing Low Earth Orbit (LEO) satellites. In such cases, the low orbit, in combination with high speed, result in rapidly changing direction and very brief visibility windows, typically on the order of ten minutes during each orbital pass. Hence, effective beam management is crucial to ensure performance of LEO communication systems. For LEO communication, mechanically steered high-gain parabolic antennas are depreciated due to their slow steering and unflexible antenna beams. In contrast, electronically steered antenna arrays offer much greater speed and flexibility. Such arrays enable fast steering of multiple beams toward multiple satellites as well as null steering, both of which are important features for LEO satellite constellations [ 1 ] . Typically, hybrid analog–digital architectures are employed to reduce cost compared to fully digital arrays while preserving significant beamforming capabilities [ 2 , 3 ] . Research on beam management for LEO systems has mainly focused on the satellite side, while less attention has been paid to user equipment and GS. Existing beam tracking solutions for millimeter wave (mmWave) systems typically rely on continuous beam tracking [ 4 ] or two-step methods combining angle-of-arrival (AoA) estimation with Kalman filtering [ 5 ] . Both suffer from limited robustness in the case of signal blockage or interruption. Moreover, traditional AoA estimation techniques applied to hybrid architectures suffer from phase ambiguity when the phase center the sub-arrays of the transceiver is greater than \lambda/2 [ 6 ] . Fig. 1: Considered scenario with a GS tracking a LEO satellite. It is well-known that the information passed from one step to the next should be a sufficient statistic— a condition which is generally hard to ensure and is not met in traditional two-step tracking schemes such as direction estimation followed by Kalman filtering. This motivates the research interest in joint direction estimation and tracking in related applications such as drone tracking in radar systems [ 7 , 8 ] . In these works Bayesian Variational message passing (VMP) is employed to develop joint estimation and tracking methods which has been shown to improve robustness and accuracy compared to traditional multi-step methods. The present contribution develops a novel VMP algorithm for joint localization and beam tracking of a LEO satellite from a GS with a hybrid architecture at mmWave frequencies. The proposed method estimates the parameters of the satellite orbit, which is modelled as circular, and computes the AoA based on these estimates. We assume a scenario in which the GS has no access to prior information about the satellite position. Simulation results highlight the advantages of this joint approach compared to conventional methods. II System and orbit model II-A System model Consider the downlink transmission between a LEO satellite and a GS illustrated in Fig. 1 . As shown in Fig. 2 , the satellite orbit is specified with respect to a a right-handed coordinate system at the GS, such that the z-axis points to zenith. At time t , the position of the satellite is given by the 3D vector \bm{\Omega}(t;\bm{\Gamma}) specified by the orbit parameter vector \bm{\Gamma} . Thus the unit vector \tilde{\bm{\Omega}}(t)=\bm{\Omega}(t;\bm{\Gamma})/||\bm{\Omega}(t;\bm{\Gamma})|| is the the satellite direction at time t . We assume that the direction varies slow enough to be considered constant over a recording time interval n\Delta t\leq t<(n+1)\Delta t , i.e., \tilde{\bm{\Omega}}(t)=\tilde{\bm{\Omega}}_{n}. The GS is equipped with a Uniform Planar Array (UPA) with a hybrid analog-digital architecture with infinite resolution fixed-amplitude phase shifters and \lambda/2 -spaced antennas. We consider a partially-connected hybrid architecture, with M RF chains each connected to a different subset of N_{sub} antennas. Because of this independent sub-array architecture, the signals after analog combining do not meet the Nyquist spatial sampling criterion, leading to a 2\pi phase ambiguity [ 6 ] . The satellite is equipped with a UPA with an analog architecture with N_{t} \lambda/2 -spaced antennas, with its beam assumed to always perfectly point towards the GS with maximum gain. The signal received by the m^{th} subarray at time instant n is modeled as \mathbf{y}^{(m)}(\bm{\Omega}_{n})\hskip-2.5pt=\hskip-2.5pt(\bm{b}^{(m)}(\tilde{\bm{\Omega}}_{n}))^{H}\bm{a}^{(m)}(\tilde{\bm{\Omega}}_{n})f^{(m)}(\tilde{\bm{\Omega}}_{n})h(\bm{\Omega}_{n})\bm{s}+\bm{v}_{n}^{(m)}, (1) where \bm{b}^{(m)}(\tilde{\bm{\Omega}}_{n})\in\mathbb{C}^{N_{sub}\times 1} corresponds to the analog combining weights, \bm{a}^{(m)}(\tilde{\bm{\Omega}}_{n})\in\mathbb{C}^{N_{sub}\times 1} is the receive steering vector, f^{(m)}(\tilde{\bm{\Omega}}_{n})\in\mathbb{C}^{1\times 1} is the phase shift between the first and the m^{th} subarray, h(\bm{\Omega}_{n})\in\mathbb{C}^{1\times 1} represents the channel gain, \bm{s}\in\mathbb{C}^{N_{s}\times 1} corresponds to the transmitted signal and \bm{v}_{m}\in\mathbb{C}^{N_{s}\times 1} is complex circular symetric distributed noise \bm{v}_{m}\sim N^{C}(0,\bm{\Lambda}_{v}) , where N^{C}(\bm{\mu},\bm{\Lambda}) denotes the complex circular symmetric distribution with mean \bm{\mu} and precision matrix \bm{\Lambda} . For brevity, the explicit dependence on the AoA \bm{\Omega}_{n} is dropped, and the time dependence is noted by a subscript, for example, h(\bm{\Omega}_{n})=h_{n} . By stacking received signals of all M subarrays in a single vector, we obtain \mathbf{y}_{n}=h_{n}\,\mathbf{x}_{n}+\bm{v}_{n}. (2) Here, \mathbf{x}_{n} corresponds to received signals that include beamforming effects arising from a satellite in direction \tilde{\bm{\Omega}}_{n} without the channel effects. An Orthogonal Frequency Division Multiplexing (OFDM) system is considered, where each subcarrier is represented by a narrowband line-of-sight (LOS) channel. Multipath effects are neglected in this scenario and the Doppler shift due to satellite motion is assumed to be perfectly compensated for at the receiver. Under these assumptions, the channel model is h_{n}=\rho_{n}e^{j\chi_{n}} where \chi_{n}\sim U(0,2\pi) and \rho_{n} includes the attenuation due to the free space path loss, the atmospheric attenuation, and the gains of the transmit and receive antenna arrays. In this work, atmospheric attenuation is set according to models provided by the ITU recommendation [ 9 ] . II-B Orbit Model Satellite orbits can be complicated and are affected by factors such as atmospheric drag, non-uniform Earth gravity, and the gravitational pull of other celestial bodies [ 10 ] . However, considering a single revolution, we approximate the orbit as circular. Considering the orbit to have fixed and