By Brady O’Hanlon, Mark L. Psiaki, Paul M. Kintner Jr., and Steven P. Powell
Anomalous behavior of the L1 C/A-code carrier phase has been detected on PRN07/SVN-48. The anomalies are sudden step-like changes of phase by about 10 degrees/5 millimeters. These steps are followed by negative steps of the same magnitude that restore the original phase time history. These anomalous square pulses have been observed with durations as short as 0.1 seconds and as long as 600 seconds. They can occur about once a minute or be absent for hours.
These anomalies could be of consequence for some GNSS applications. For precise monitoring of differential total electron content (TEC), the magnitude of this anomaly is the same order as the signals of interest. Precise point positioning (PPP) systems seek to achieve CDGPS accuracy without direct double-differencing. The lack of double-differencing would allow any L1 C/A carrier phase anomaly to directly affect the PPP solution.
This behavior was detected when testing a dual-frequency software receiver that processes the GPS civilian signals on L1 and L2. The anomaly was first noted when calculating carrier-phase-based TEC:
where bTEC is a bias term that occurs in the phase-based calculation. Figure 1 shows a plot of the resulting TEC, after removal of its mean value, with six square-edged pulses that range in duration from 0.1 to 590 seconds, with the first a short one at t = 48 seconds. The last pulse starts at 710 seconds and ends at 1300 seconds. In all cases, the anomaly consists of a positive step change in TEC followed some time later by a negative step change of identical magnitude. Step magnitudes in the range 0.04 to 0.07 TEC units have been observed.
Figure 1. Square pulses on phase-based TEC due to L1 C/A carrier phase anomalies.
Tests were performed to ascertain whether the anomalies were caused by the L1 signal, the L2 signal, or a combination of the two. Additional tests ruled out receiver malfunction as the cause of the anomalies.
Observation of detrended L1 and L2 carrier-phase time histories quickly revealed that the anomalies occur on the L1 carrier phase. The detrended L1 C/A carrier phase shows square-edged pulses corresponding to times, magnitudes, and signs of the TEC anomalies, but the detrended L2C carrier-phase plots show no such pulses. Figure 2 shows a typical detrended L1 C/A beat carrier-phase anomaly.
Figure 2. A typical detrended L1 C/A beat carrier-phase anomaly.
Extensive tests checked whether the anomalies may have been caused by the receiver. They were initially discovered using a digital storage receiver of raw RF front-end samples followed by off-line software receiver processing. Such carrier-phase anomalies could result from signal glitches in the RF front-end’s mixing chain, from data recording anomalies in the RF front-end samples, or from errors in the software receiver code. The former two possibilities were ruled out by two means. One was to process signals from other satellites for the same RF samples. Mixing problems or data sample problems would cause similar anomalies on all GPS signals, but other GPS signals were found to be free of anomalies. Additional tests used simultaneous data collection by two digital storage receivers spaced 700 meters apart and using different RF front-end hardware. Both receivers showed identical anomalies at identical times.
Software receiver code errors were ruled out by employing two independent sets of receiver processing code, one developed in MATLAB, the other in C. These two pieces of software were developed independently by different individuals and run independently by their developers. Both showed identical anomalies.
A final check used a different receiver, the NovAtel GSV4004B. Figure 2 plots its detrended L1 C/A carrier phase along with that of the C-based Cornell software receiver. Both show the same anomaly. Thus, the anomalies appear to be caused by the SVN-48 transmitter.
All observations were made from roof-mounted antennas in Ithaca, New York. The anomalies were first observed on March 24, 2010 and were observed again on April 1, 5, 7, and 29, and as late as May 13th. For one period of several hours on May 11, no anomalies occurred. Other Block IIR-M satellites have been monitored briefly, but without finding any similar anomalies to date: SVNs 58, 55, 57, 49, and 50.