Adaptive Weighting-Based Multi-Station Doppler Orbit Determination for Noncooperative LEO Satellites

Precise orbit information is generally unavailable to users of noncooperative low Earth orbit (LEO) communication satellites such as Iridium NEXT, making publicly available two-line element (TLE) data the primary source of orbital states. However, TLE-derived orbital-state errors degrade Doppler positioning performance. This paper proposes an adaptively weighted multi-station Doppler orbit determination method. Using TLE-derived states as the orbit prior, the method formulates an orbital-dynamics-constrained batch least-squares (BLS) model. A priori weights are constructed from satellite elevation angles, and multi-station, multi-epoch Doppler measurements are adaptively reweighted by minimizing the uncertainty of the estimated satellite velocity projected onto the line of sight (LOS) from the ground station network centroid to the satellite. A Levenberg-Marquardt (LM)-type damping mechanism with diagonal scaling constrains the state correction vector to improve iterative stability. Experiments using real measurements show that the estimated orbital states yield a user-station two-dimensional root-mean-square error (2D RMSE) of 41.30 m, which is 16.06% lower than that obtained with the a priori weighted solution, 24.64% lower than that obtained with the equally weighted solution, and 64.25% lower than that obtained with the TLE-derived orbit. These results indicate that the proposed method improves the utility of the estimated orbital states for positioning with Iridium NEXT signals of opportunity.