Historically, GPS receiver designers have relied on a combination of simulation and drive testing to characterise a receiver’s performance in difficult environments. While GPS signal simulation provides a repeatable signal source, simulators cannot reproduce difficult multi-path signal anomalies often seen in the real world. Moreover, drive testing also contains inherent challenges. Not only is drive testing expensive, but it fundamentally introduces a receiver to signals that are difficult to repeat from one trial to the next.
As a result of these challenges, an increasingly common approach to receiver validation is to test receivers with recorded GPS waveforms. This approach uses an RF vector signal analyser (such as the NI PXI-5661) to record live GPS signals as a continuous IQ data file. Then, an RF vector signal generator (such as the NI PXIe-5672) is used to play the signal back to the receiver. Testing GPS receivers with recorded signals has several benefits over traditional simulation or drive testing approaches. Not only does this approach introduce real world impairments, but it does so in a manner that is repeatable. Thus, you can test many different receivers by observing how they will react to the same test stimulus.
Configuring an RF recording device
There are two main concerns to consider when configuring an RF record and playback system: using the full dynamic range of the RF vector signal analyser and ensuring that the recording device does not add additional noise to the signal. Concerning the first issue, an easy way to capture the GPS signal’s full dynamic range is to use a vector signal analyser with significant dynamic range. With typical vector signal analysers offering up to 80 dB of dynamic range, signals with a small signal-to-noise ratio (SNR) — usually less than 30 dB — such as GPS can easily be recorded without significantly affecting the SNR of the off-the-air signal. Thus, the only remaining task is to amplify the off-the-air signal while adding as little noise as possible.
In a typical environment, each GPS satellite will have an average power level (course acquisition (C/A) codes) ranging from -135 dBm to -125 dBm, depending on its position and environmental factors. A typical scenario will result in signals in the L1 (1.57542 GHz) band having a peak power that can range from -120 dBm to -110 dBm. In our testing, we observed a peak power of -116 dBm.
As might be expected, recording such low-power signals requires careful attention to both antenna and amplifier selection. In fact, to use the full dynamic range of the RF vector signal analyser, amplification is required. There are several ways to amplify an off-the-air GPS signal. However, you can achieve the best results by using an active GPS antenna in conjunction with an additional low noise amplifier. With two cascaded low-noise amplifiers (LNAs), each providing 30 dB of gain, the total gain applied is 60 dB. Thus, the resulting peak power observed by the vector signal analyser is increased from -116 dBm to -56 dBm. The required power at an RF vector signal analyser will vary from one instrument to the next, and this value is determined by the maximum gain applied by the vector signal analyser.
Powering an active antenna
To capture GPS signals while adding the smallest amount of noise possible, use an active GPS antenna with a noise figure that is less than 2 dB. This achieves the best results. Active antennas provide the best gain performance versus noise figure, but they introduce the inherent challenge of providing a DC bias signal of anywhere from 2.5 to 5 V.
One common method that can be used to power an active antenna is with a DC bias “T”. Using this component, a DC signal (3.3 V in this case) is applied to the DC port of the bias T, which applies the appropriate DC offset to the active antenna. Note that the precise DC voltage you should apply depends on the DC power requirements of the active antenna. Figure 1 shows a diagram illustrating the system setup.
Observe in Figure 1 that you can use any off-the-shelf DC supply to supply the DC bias signal. While we used the NI PXI-4110 in our experiments, any generic power supply will work. Also, it is important that the DC bias T is rated for operation up to 1.57542 GHz, the frequency range of L1 GPS signals. A DC bias T like the one used in this experiment was purchased from minicircuits.com.
Once the RF front end of the recording device is configured, you can test the system simply by performing a basic RF spectrum measurement in the L1 band. To do this, configure the RF signal analyser to a centre frequency of 1.57542 GHz (the L1 band) and a span of 4 MHz. Note that the antenna should be placed in an open-air environment where it has a clear view of the sky.
The GPS C/A code signal will occupy a bandwidth of about 1 MHz, so a slightly wider span is required to visualise this signal. In addition, because the power level is significantly low, a narrow resolution bandwidth (RBW) combined with a low RF reference level (-50 dBm) is also required. With a 10 Hz RBW configured with 20 averages, the GPS satellites should clearly be visible just above the noise floor. Figure 2 illustrates an example RF spectrum after 60 dB of gain.
Figure 2 shows a small “bump” right at 1.57542 GHz. This “bump” is the off-the-air GPS signals and it indicates that the RF front end is correctly configured. Now that the RF front end is configured, the next step is simply to perform a continuous IQ acquisition. Connected to a large storage volume, typical RF recording systems can capture up to 25 hours of continuous GPS waveform.
Experimenting with recorded GPS signals
One of the biggest benefits of RF record and playback systems is that you can use them to test receivers with real world data. In addition, you can observe how repeatable a receiver will react to the same environmental conditions. In the following experiment, we observed how a GPS receiver reacted to 10 playback trials of an identical recorded GPS waveform. The specific receiver used in these experiments was a SiRFstar III chipset. All receiver information was reported through a serial RS232 interface and decoded as NMEA-183 data.
An interesting experiment that can be performed with recorded GPS signals is to observe the relationship between signal strength and position repeatability. To observe this relationship, we can compute the standard deviation of latitude and longitude of each of the 10 trials. If the number of satellites in view does indeed affect the position accuracy and repeatability, we should see the standard deviation increase as satellites fade out of view. For example, suppose you take the average C/N ratio of the four highest satellites reported by the GPS receiver as a proxy for signal conditions. With access to NMEA-183 data, you could also track HDOP and the number of satellites in view. However, in Figure 3, there is a strong correlation between signal strength and position repeatability.
The peak deviation between each of the trials occurs at time = 120 seconds. At this time, the standard deviation is approximately 2 metres of deviation, while it is less than 1 metre during most other times. This jump in standard deviation occurs concurrently with the top four satellites dropping in signal strength from 45 dB-Hz to 41 dB-Hz. Moreover, the standard deviation of position is directly correlated with satellite C/N ratio and not with receiver velocity.
The experiment above highlights the effects of environmental factors on receiver performance. Most likely, the receiver was reacting to an obstructed view of the sky where the number of satellites in view also dropped at the same time. Either way, the experiment simply provides an example of analysis that you can perform with recorded GPS waveforms. Because the real world RF signal is stored on disk, you can perform experiments similar to the one above at a future time if more analysis is required.
While receiver drive tests in deployment environments such as urban canyons are still common, RF record and playback systems have emerged as a new solution to RF receiver validation. As this article illustrates, careful attention to the RF front end of the recording device can ensure signals are captured without adding a significant level of impairment. Finally, with recorded GPS waveforms, you can perform a wide variety of experiments. Because these recordings enable you to generate repeatable RF signals, it is easy to observe how different receivers will react to the same RF conditions.
Reference
Add a comment