Edwin G. W. Peters, University of New South Wales Canberra; Timothy Bateman, University of New South Wales Canberra; Rabbia Saleem, University of New South Wales Canberra; Melrose Brown, University of New South Wales Canberra; Andrew Lambert, University of New South Wales Canberra
Keywords: doppler estimation, software defined radio, passive RF
Abstract:
With the increased access to space and an increasing number of private launch and satellite providers, the space domain awareness (SDA) field has received increasing attention within the past decade. This is predicted to increase significantly over the next decade. Keeping pace with the rapidly growing population of active, manoeuvring, civilian satellites, requires the development of a new civilian SDA sensor network to support the next generation of space traffic management systems. One way to determine the orbit of satellites is using radio frequency (RF) measurements. Typically, ranging measurements are used by satellite operators, since these provide both frequency offset measurements due to Doppler and round-trip time. However, ranging measurements typically can only be used by the operator of the satellite. An alternative to ranging is mono- or bistatic radar. These, however, require a high power transmitter to illuminate the targets and are costly to build. An alternative, more cost-effective, method is to utilize passive RF measurements, where ground-based receiver antennas capture spacecraft RF emissions, such as communications and beacons.
In this work, we present a flexible software defined radio (SDR) based method that can perform high fidelity RF transmitter frequency offset measurements for those satellites that transmit as part of their normal operation. The proposed method relies on a matched filter approach, where templates of the transmitter waveform are known or can be estimated. This method allows for separation of overlapping waveforms, can operate in low signal to noise ratio (SNR) settings and can be used for identification or single and multiple transmitters as well. In its simplest form, the filters in the proposed method use wavelets that resemble the modulation scheme that the transmitter is assumed to use. The SDR performs a coarse frequency estimation using a grid search across the spectrum. Once a coarse frequency estimation has been performed, a fine frequency estimation is performed, that can resolve the frequency offset to a frequency resolution of single Hertz and a time resolution that currently is equal to the interpolation rate.
The proposed method utilizes graphics processing units (GPUs) to perform the coarse frequency estimation, presented in our earlier work. This GPU-based method allows the system to perform coarse frequency offset estimations in real-time at bandwidths sufficient for typical VHF, UHF and S-band communications on entry level desktop GPUs. When the signal bandwidth is sufficiently high compared to the frequency offset caused by the Doppler effect, the coarse frequency estimation can be omitted, and the fine frequency estimation can be performed on a GPU in real-time.
The proposed method is flexible, since a large number of templates for different modulation schemes can be supported simultaneously, and thereby receives excellent SNR. In low SNR settings, the templates can even be used to resemble markers or known sequences that are present in the signal to increase the processing gain significantly.
Compared to existing frequency estimation methods, the proposed method requires a-priori knowledge or assumptions on modulation scheme and symbol rates. However, the time/frequency resolution is significantly higher for similar SNR compared to existing methods that do not require a-priori information.
The proposed method has been applied during the launch and early operation (LEOP) stage of our LEO M2 Pathfinder and M2 satellite missions, and the frequency estimation used to derive improved ephemerides using simple orbit determination tools, such as STRF. This was crucial in the early mission stages where the SSN generated TLE was inaccurate or not yet available.
In this work, we present the algorithm including a simulation study showing its performance across different SNR. Finally, we present applications of frequency estimation of in-orbit spacecraft.
Date of Conference: September 27-20, 2022
Track: Astrodynamics