Deciphering Doppler: How the Doppler effect affects GPS signals
2020-04-09 | 8 min read
Receivers for Global Positioning Systems (GPS), or more broadly speaking, Global Navigation Satellite Systems (GNSS), are used in a variety of applications today, with mapping on smartphones being one of the most widely-used cases. Cellular phone networks have always benefited from GPS’s timing signals, and the new 5G systems will be even more dependent on GPS.
In this blog, we will examine how we used Keysight’s FieldFox handheld RF and microwave analyzer to capture and display GPS signals over a period of time and how the Doppler effect affected those signals. GPS signals transmit in the L-band — L1 at 1,575.42 MHz, L2 at 1,227.60 MHz, and L5 at 1,176 MHz.
Figure 1 shows the hardware setup for the measurement. The setup included a horn antenna, a low-noise amplifier (LNA), and a spectrum analyzer (FieldFox). The antenna, which was pointed towards the southwest sky (setup was in the northern hemisphere), was connected to the LNA. The output of the LNA was connected to the spectrum analyzer input port of the FieldFox RF analyzer. FieldFox’s built-in DC voltage source powered the LNA. We used a directional horn antenna (with gain) and an external LNA because GPS signals have very low power levels at the receiver.
With the FieldFox in spectrum analyzer mode, we configured it to capture the GPS L1 signal, with the center frequency tuned to 1,575.42 MHz (or 1.575420 GHz) and the span set to 5 kHz. Figure 2 shows the measurement.
To increase the frequency accuracy of the measurement, we enabled FieldFox’s internal GPS receiver, and locked the internal frequency reference to GPS. The typical frequency accuracy of FieldFox’s receiver is ±0.5 ppm, translating to ±787 Hz for this measurement. By locking to GPS, the frequency accuracy was increased ±0.01 ppm, translating to ±16 Hz. This frequency locking does not have the Doppler error due to internal correction methods used in the GPS chip.
Since GPS signals are very low power, we configured the spectrum analyzer for maximum sensitivity which is the same as trying to obtain the lowest noise floor. To achieve this objective, we set FieldFox’s internal attenuation to zero, and enabled the internal preamplifier (adding ~15 dB of gain). Furthermore, we reduced the resolution bandwidth (RBW) and video bandwidth (VBW) to 3 Hz and 1 Hz respectively.
We used FieldFox’s spectrum analyzer trace-recording capability to capture the data over a four-hour period. After recording the data, we played them back and displayed them on the FieldFox in a regular spectrum analyzer display and also a spectrogram display. Spectrograms are a three-dimensional data display, with frequency on the x-axis, time on the y-axis, and amplitude indicated using colors. Spectrograms are a useful tool for displaying signals that vary over time.
The spectrogram display in Figure 3 shows the captured data over the four-hour period. The colors represent the strength or power of the signal — blue represents the lowest power (-170 dBm) and red represents the highest (-160 dBm).
The oldest signals are towards the upper part of the graph and the newest signals are near the lower part. The GPS signals are the diagonally slanted (~arc) lines. If there was no Doppler effect or other secondary effects, the GPS L1 signal would be a straight vertical line at the center of the graph (1.57542 GHz). However, since GPS satellites are in motion relative to earth, the received frequency at earth changes as the satellite moves, resulting in the arcs/slanted lines seen in Figure 3. This graph shows a Doppler shift of approximately 2 kHz.
Note that we did not move the hardware (FieldFox + antenna) during the measurement, that is, there was no rotation. The results seen show transmissions from the GPS satellites that appeared in the field of view of the antenna during the four-hour measurement period.
Another observation is the presence of electromagnetic interference or EMI signals. You can see approximately five vertical lines about 1 kHz apart with stronger signals at the edges. There is also another signal, seen as a horizontal slanted line across the screen. This is a signal whose frequency is changing over time, albeit a lot faster than the GPS signals. It would take further investigation to determine the source of these other signals.
Additionally, if you are wondering how is it that our our cell phones and other myriad of GPS devices measure these low power GPS signals without large external antennas or LNAs, it’s because they are making different measurements. GPS receivers such as those used in our phones use digital signal processing techniques to decipher the GPS signal from the noise. GPS receivers, often a small chip, down-convert the L-band signal to an intermediate frequency (IF) or baseband signal. The baseband signal is then digitized by an analog to digital converter (ADC) and processed. An internal LNA is still part of the RF chain, but it’s the results of the digital signal processing, specifically signal correlation, that provide the GPS signal.
Science history enthusiasts may find it interesting that the Doppler effect played a crucial role in the early development of GPS. The year was 1957 and the Soviet Union had just launched Sputnik, the world’s first artificial satellite, starting the U.S.-U.S.S.R Space Race. Scientists at Johns Hopkins’ Applied Physics Lab recognized that by measuring the Doppler shift in the transmissions from Sputnik, they could estimate its position. And inversely, they figured out if you knew the orbit of the satellite, you could estimate your position on earth.
And the rest, as they say, is history.
To find out more about FieldFox and its capabilities beyond spectrum analysis, visit the FieldFox web page.
Thanks to Dan Slater for his assistance with these measurements.