Comments on NTIA GPS/UWB Measurement Test Plan
NASA Glenn Research Center
Rod L. Spence
8/28/00

The following comments are offered with regard to the proposed NTIA GPS/UWB measurement test plan.

In response to the GPS receivers to be tested, the types of receivers listed in Table 1 are appropriate. Three of the receivers (i.e. Garmin GPS-155XL, Ashtech Z-Surveyor, and Allen Osborne SNR-8000) also are being tested in the Texas University GPS/UWB measurement effort so that (hopefully) the two data sets will complement one another. We also note that the 3 aviation receivers listed in the table are conventional L1 C/A code tracking receivers that perform position-velocity-time (PVT) determination. It might be worthwhile to replace one of these receivers with one that uses code/carrier tracking to perform both PVT and attitude determination (e.g. Trimble TANS Vector or Force-19). These types of aviation receivers are similar to ones that will be used onboard Shuttle, ISS, and the ISS emergency crew return vehicle (ECRV). The carrier tracking loops in such a receiver will be the weak link and most susceptible to UWB interference. NASA had considered a space GPS receiver for testing, but given the time and funding constraints on the NTIA effort along with uncertainty in being able to make a space receiver available, this idea was dropped. Since most space GPS receivers are modified versions of aviation receivers (usually modifications to the software in addition to radiation hardening), test results for the aviation receivers should be adequate for assessing the UWB interference threat to spaceborne GPS. One significant difference between airborne and space GPS receivers, however, is in the signal acquisition process. Space receivers must contend with about 10 times higher LOS velocity (and hence Doppler shift) than aeronautical receivers (i.e. ±40 kHz vs ±4 kHz) and also shorter GPS satellite rise and set times (i.e. 45 min vs 6 hrs). Thus, the frequency-code search space for acquisition is typically much larger for space GPS with higher required C/No needed to achieve acquisition in a reasonable time (e.g. < 5 min). The tests being planned are simulated static tests. If possible, one of the aviation receivers should be tested with the GPS simulator configured to model maximum expected aircraft dynamics, so that this additional stress on the acquisition/tracking performance can be taken into account. Otherwise, the test results may underestimate the interference impact compared to when the receiver is operating in the actual environment. Of course, if the static tests reveal serious interference degradation, then additional dynamic stress will only make things worse and such a test may be unnecessary.

With regard to the measurement methodology and making measurements on a single simulated GPS satellite channel, it is recognized that such a configuration is necessary to eliminate other GPS ranging error sources such as multipath interference and atmospheric delays and to isolate the error due to the UWB interference and receiver thermal noise alone. As mentioned above, simulator tests also allow the capability to model the receiver dynamics to include dynamic stress on the tracking loops which cannot be done in static outdoor tests.

Nevertheless, we believe it is important to carry out some limited outdoor "live-sky" testing with at least one of the receivers (perhaps one of the public safety receivers) in order to assess performance degradation due to UWB interference under more realistic conditions and also to provide results in terms of position error rather than just single channel pseudorange or carrier phase error. Such testing will allow scatter plots of the receiver estimated position to be generated under both UWB and no-UWB interference conditions. Interference results presented in this way may be more meaningful to GPS users than numerous plots of pseudorange error and carrier phase error vs C/No. Since a GPS user is primarily interested in the final navigation solution (PVT) and not the pseudorange error per se, it is important to let the UWB induced ranging errors propagate through the receiver positioning software algorithms and see the impact on the positioning accuracy.

A possible test configuration for a live-sky test is shown in the block diagram below. Rather than trying to eliminate other GPS error sources to isolate the error due to UWB interference alone, this "zero-baseline" test uses two identical receivers sharing the same antenna and LNA. Thus, they both simultaneously experience the same error mechanisms (i.e. satellite ephemeris/clock, dynamic stress, GPS intra-system interference, sky background, atmospheric, multipath, DOP, LNA filtering/noise) except for UWB interference – which is applied to only one of the receivers. The attenuator along the receiver A path is included to account for the additional device, connector, and cable losses along the receiver B path. With the UWB turned off, one should be able to adjust the attenuator so that the composite C/No into A is equal to the C/No into B. Since both receivers are tracking the same GPS satellites and experiencing the same errors, they both should then generate nearly identical measurements. (There may be some variance due to slight differences in the receivers themselves such as oscillators, inter-channel biases, etc. but this can be calibrated out.)

By applying UWB interference to receiver B at various power levels, one should then be able to easily observe the degradation in the output parameters of receiver B as compared to receiver A. Although the UWB signal is not radiated and does not experience the GPS antenna/LNA polarization discrimination and filtering effects in this setup, it is still possible to measure UWB interference power levels at the receiver B input itself (after subtracting the combiner loss). Knowing the receiver front end gains and losses, the corresponding interference power at the antenna output can then be calculated. Depending on the specific receivers which are used, a number of output measurements may be recorded. These are listed in the diagram.