QZS-2 signal analysis, QZS-3 launched

This month we bring you a guest column by Steffen Thoelert, André Hauschild, Peter Steigenberger and Oliver Montenbruck of the German Aerospace Center (DLR) and Richard B. Langley of the University of New Brunswick.

UPDATE: Since Sept. 10, continuously operating DLR receivers in Sydney, Australia, and Chofu, Japan, have been reporting measurements from QZSS satellite J07, which, according to the QZSS Interface Control Document, is the geostationary satellite QZS-3.

The second satellite of Japan’s Quasi-Zenith Satellite System (QZSS) has started transmitting navigation signals. QZS-2, or Michibiki-2, was launched on June 1, 2017, and joins its predecessor QZS-1 (Michibiki-1), which has been in orbit since September 2010.

Both satellites have been placed into inclined geosynchronous, elliptical orbits, which enable extended satellite visibility periods over Japan and are characteristic features for this regional navigation system.

The third satellite, QZS-3, was launched on Aug. 19, 2017, into a geostationary orbit. If all goes according to plan, a fourth satellite in an eccentric orbit will follow by the end of this year and complete the constellation.

QZS-2 Signal Tracking

It is not straightforward to tell when QZS-2 started signal transmission exactly. About four weeks after launch, on June 27 between 10:17 and 12:37 UTC, several Septentrio PolaRx GNSS receivers in the Asia-Pacific region recorded continuous L5 observations. About one week later, on July 4 shortly after 03:02 UTC, Javad and Trimble receivers picked up L1 C/A and L5 signals from QZS-2 for a few seconds. Then again, between 23:03 UTC on July 6, and 01:36 UTC on July 7, several receivers intermittently tracked the L1 C/A, L2C and L5 signals. Finally, on July 10, starting at approximately 01:03 UTC, these three signals were continuously tracked until approximately 04:00 UTC on July 12. Up until Aug. 1, signal tracking had remained intermittent, but has been stable since. This was presumably the result of interruptions in the signal transmission due to test activities.

Figure 1. QZS-2 signals tracked by GNSS receivers in Chofu, Japan, (top plot) and Sydney, Australia, (bottom plot). The plots depict the measured C/N0 for L1 C/A (black), L2C (red) and L5 (green) together with the observed pseudorange (grey). The frequent discontinuities in the pseudorange are due to the receiver clock adjustments. Both receivers exhibited a short tracking outage at approximately 06:00 UTC. The interruption in tracking at Chofu around 08:00 UTC is due to the low elevation angle of the satellite.

The plots in FIGURE 1 show QZS-2 signals as tracked by GNSS receivers in Japan and Australia on July 10. The two first sets of broadcast messages were transmitted on July 16 at 6:00 and 7:00 UTC. Regular transmission of broadcast ephemerides started on July 27 at 22:00 UTC, but deviations from the hourly update rate still occur from time to time.

Identical or Fraternal Twins?

At first glance, QZS-2 seems like a look-alike of QZS-1, but there are many differences between the two spacecraft. Most apparent is the presence of an additional auxiliary antenna. Like QZS-1, QZS-2 transmits its navigation signals on the L1, L2, L5 and L-band Experiment (LEX) frequencies through the main antenna, while the augmentation signal L1S (formally known as Submeter-class Augmentation with Integrity Function or SAIF) is transmitted from a separate antenna. However, the new L5S signal, which is introduced with QZS-2, is transmitted with yet another antenna.

The new satellite also has a shorter “wingspan” of only 19 meters, since it is equipped with two solar panel segments on each side, compared to three segments for QZS-1 with a width of 25.3 meters. The second QZSS satellite also follows a different attitude model: Unlike QZS-1, which switches between yaw-steering mode and orbit-normal mode depending on the sun’s elevation angle with respect to the orbit plane, QZS-2 always remains yaw-steering except for short periods of time when orbit maneuvers are performed. Further differences will become apparent in the analysis of the signal spectra in the subsequent sections.

The Cabinet Office of the Government of Japan, which oversees QZSS as a national undertaking, has published QZSS satellite metadata information on its official website. At the time of writing, only one document for QZS-2 is available, which contains information about the satellite’s properties such as mass, dimension, attitude law and reference frame, but also antenna and laser retroreflector positions, antenna phase-center offsets and variations as well as signal group delays.

Additional documents containing metadata for QZS-1, -3 and -4 and further information about QZS-2 are in preparation.

Rubidium Clock

FIGURE 2 illustrates the stability of the QZS-2 rubidium atomic frequency standard (RAFS) by means of the Allan deviation (ADEV). Data from a global network of 150 GNSS stations was processed to estimate GPS and QZSS satellite orbit and clock parameters.

Figure 2. Allan deviation of the rubidium atomic frequency standards of GPS Block IIF satellite G32, QZS-1 (J01) and QZS-2 (J02).

However, whereas about 60 of these stations provide QZS-1 observations, QZS-2 is only tracked by 13 stations. ADEV values for QZS-1, QZS-2 and a GPS Block IIF satellite were computed from a daily solution for Aug. 3 with 30-second clock sampling.

At an integration time of 100 seconds, the QZS RAFS reaches an ADEV of better than 3 × 10^-13.

At longer integration times, the QZS-2 clock almost reaches the stability of the GPS Block IIF RAFS.

Based on this preliminary analysis for only one day, the QZS-2 clock seems to perform as expected. The larger ADEV values compared to QZS-1 for integration times up to 1,000 seconds might be attributed to the significantly smaller number of tracking stations contributing to the QZS-2 clock solution. The quality of the clock solution will improve as soon as more stations are able to track QZS-2.

Signals with High-Gain Antenna

Complementary to the receiver measurements and analysis, the German Aerospace Center (DLR) has also recorded raw spectral and in-phase and quadrature (IQ) data of QZS-2 to get further insights into the transmitted signal structure and initial signal quality. FIGURE 3 shows a spectral measurement of the complete GNSS L-band frequency range, which shows the signal transmissions of QZS-2 in the L1, L2, L5 and L6 bands. The signal was captured with DLR’s 30-meter high-gain antenna at Weilheim, southwest of Munich, operated by DLR’s German Space Operations Center.

Figure 3. QSZ-2 L-band normalized power spectra recorded at Weilheim, Germany, on July 18, 2017 at 20:43 UTC.

This first view of the signal transmission shows a good spectral shape, appropriate band filtering and no out-of-band unwanted spurious emissions of the satellite. For further analysis, we looked closer at each signal-band spectrum and performed IQ-sample recording.

Comparing the QZS-2 spectra to that of QZS-1, we see differences in the signal structure for the L1 frequency band.

Figure 4. QZS-1 and QZS-2 L1 spectral flux density.

FIGURE 4 shows the L1 spectra of both satellites. The additional signal component can be seen at an offset of 6 x 1.023 MHz and 18 x 1.023 MHz from the L1 center frequency of 1575.42 MHz. This is the result of the new L1C-pilot modulation, which is based on the time-multiplexed binary offset carrier (TMBOC) modulation technique using a mixture of BOC(1,1) and BOC(6,1). See here for detailed information.

Another difference is present in the L6 band and can be seen within the signal time domain or the IQ domain. The new satellite transmits two components (one each for the I- and Q-channels) while QZS-1 transmits only one I-component. This observation is fully in line with the QZSS Interface Specification. On QSZ-2, an additional L6 signal component (Centimeter-Level Augmentation Message for Experiments, L6E) is implemented. FIGURE 5 shows the IQ constellation plots of QZS-1 and QZS-2 for the L6 band.

Furthermore, the L5 band IQ plot of QZS-2 exhibits significant differences compared to QZS-1. These differences, which are illustrated in the plots of FIGURE 6, are due to an additional L5S signal transmitted by QZS-2.

The QZS-2 L5 IQ diagram is fairly easy to understand as a coherent superposition of two distinct quadrature signals from two antennas. One signal is the GPS-like L5 signal transmitted from the main L-band antenna, while the other (L5S) signal originates from a new L5S antenna. This is illustrated in FIGURE 7.

Figure 7. QZS-2 L5 IQ constellation plot including demarcation of the L5 and L5S signals.

For illustration purposes, the dashed orange square in Figure 7 relates to the 10 MHz L5 signal, while the smaller red squares are the 10 MHz L5S signal.

A code generator has been setup according the QZSS L5 and L5S interface control document (ICD). An analysis of the correlations of possible pseudorandom noise (PRN) codes resulted in the detection of PRN 194 and PRN 196. Based on the information in the ICDs, PRN 194 is used for L5 and PRN 196 is used for L5S.

The performed code correlation analysis also yields the finding that the L5 signal is approximately 3.5 dB stronger than the L5S signal. Note, however, that both signals have a specified minimum receive power of -157 dBW. Due to the limited visibility of QZSS satellites from the Weilheim ground station, it is not possible to verify this value.

Conclusion

With the launch and activation of QZS–2, the deployment of Japan’s regional navigation system is moving forward again. The launch of a geostationary satellite, QZS-3, took place on Aug. 18. A fourth Japanese navigation satellite is scheduled to launch later this year. With this rapid  sequence, the target date of 2018 for the completion of an operational constellation with four satellites is quite realistic.

Steffen Thoelert, André Hauschild, Peter Steigenberger and Oliver Montenbruck are from the German Aerospace Center (DLR).

Richard B. Langley is from the University of New Brunswick and authors the monthly Innovation column for GPS World magazine.