ABSTRACT: The quasi-zenith satellite (QZS) system (QZSS) is a Japanese project that was started in 2003 ... modem at the GS is called the transmitter modem.
Simulation and Ground Experiments of Remote Synchronization System for Onboard Crystal Oscillator of Quasi-Zenith Satellite TOSHIAKI IWATA, MICHITO IMAE, TOMONARI SUZUYAMA, HIROSHI MURAKAMI, and YOSHIKATSU KAWASAKI National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8568, Japan NAOTO TAKASAKI and AKIRA IWASAKI Department of Aeronautics and Astronautics, University of Tokyo, Bunkyo-ku, Tokyo 113-8656, Japan FABRIZIO TAPPERO and ANDREW DEMPSTER School of Surveying and Spatial Information Systems, University of New South Wales, Sydney, Australia Received February 2006; Revised December 2006
ABSTRACT: The quasi-zenith satellite (QZS) system (QZSS) is a Japanese project that was started in 2003 with the aim of positioning services. A simulation and experiment on a new timekeeping method for the QZSS was conducted on the ground. The remote synchronization system for the onboard crystal oscillator (RESSOX) of the QZS functions by maintaining an uplink to receive time information from a ground station (GS) equipped with accurate atomic clocks. The required apparatuses for the RESSOX are described. Some simulation and experimental results obtained on the ground, which indicate the feasibility of synchronization to within 10 ns between the GS and the QZS, are demonstrated and compared.
matic of the test bed for the preliminary ground experiments and Figure 2 shows the photograph. The details of some apparatuses used in the ground experiments are described below.
INTRODUCTION The remote synchronization system for the onboard crystal oscillator (RESSOX) of the Japanese quasi-zenith satellite (QZS), which does not require onboard atomic clocks, has been developed by the National Institute of Advanced Industrial Science and Technology (AIST) [1]. The target synchronization accuracy of the RESSOX is tentatively set at 10 ns, with the target stability being 1013 for more than 100,000 s, between the onboard voltage-controlled crystal oscillator (VCXO) and the ground-station (GS) atomic clock. These targets are determined on the basis of the synchronization performance between GPS time and UTC (USNO) [2] and the long-term stability performance of onboard cesium atomic clocks [3].
Transmitting Time Adjuster (TTA) and Timing Controller The TTA is an apparatus that generates advanced time used to compensate the delay of the uplink time signal between the GS and the QZS. The time adjustment files and commands for the TTA are generated and transferred by the timing controller at the GS with TCP/IP. The time adjustment files are prepared as feed-forward information by calculating the delay models, and the time adjustment commands are provided as feedback information from the QZS receiver in real time. The requirements for the TTA were reported in our previous paper [1]. To realize the requirements, the direct digital synthesizer (DDS) is used. To operate the TTA, the time delay file that describes the delay at a specific time in Coordinated Universal Time (UTC) is prepared by orbit calculation and other delay simulations based on knowledge from the
APPARATUSES FOR REALIZING RESSOX In this section, the hardware tools for realizing RESSOX are described. Figure 1 shows the scheNAVIGATION: Journal of The Institute of Navigation Vol. 53, No. 4, Winter 2006 Printed in the U.S.A.
231
Modems (Pseudo-Noise (PN) Code Generator and Time Comparator)
Fig. 1–Test bed for ground experiments. The goal of the experiment is synchronization between atomic clock and VCXO.
The principle of operation of the modem at the GS is PN code generation and that of the modem onboard is the time comparator. For simplicity, the modem at the GS is called the transmitter modem (Modem (T)) and the one onboard is called the receiver modem (Modem (R)). The same modem used in two-way satellite time and frequency transfer (TWSTFT) to compare two or more time standards all over the world is prepared for both the transmitter at the GS and the onboard receiver in this experiment [4]. The time signal, called the RESSOX control signal, is modulated and demodulated by the PN code and transmitted because the pure 10 MHz signal cannot transmit such time information as 1 pps.
Uplink Delay Simulator (UDS) The UDS is a hardware simulator that is used in the ground test bed. The UDS is based on the DDS technology of the TTA and the first-in first-out (FIFO) memory control techniques. The UDS assigns offset values to FIFO memory according to the time adjustment files for the UDS, which have the same format as those for the TTA. The UDS receives the intermediate frequency (IF) signal that includes time information (central frequency of 70 MHz, bandwidth of 2.5 MHz) from Modem (T), down-converts the frequency to 2 MHz, downloads the waveform of time information into FIFO memory in real time, and reads it out in delaycorresponded timing. Finally, the UDS up-converts the frequency to 70 MHz.
Fig. 2–Photograph of apparatuses.
Master Control Station (MCS), which may include some errors. Using the time delay file, time adjustment files are prepared by a utility program. The time adjustment files include coefficients that are calculated by Lagrange interpolation or least mean squares approximation (the maximum order is 11) of the delay. In our ground test bed, the TTA receives 10 MHz and 1 pulse per second (pps) signals from the atomic clock that is used as the time standard of our experiment instead of the QZSS time. The time adjustment commands are also provided as the changes in the coefficients. The database of the L1 delay is prepared beforehand through orbit and delay calculations based on knowledge from the MCS, and is compared with the pseudo-range measured by the QZS signal receiver. The differences between them are accumulated, and using the first-order least squares filter, the error to be compensated is calculated and output as the change in the coefficients to be adjusted. 232
QZS Signal Generator and Downlink Delay Simulator (SimQZ) In the ground test bed, the QZS signal generator is required to simulate the QZS signal. Although the QZSS signal structure is known as L1C, L1 C/A, L2C and L5, a GPS signal generator (Spirent GSS4730) that generates four channels of L1 C/A and L2P positioning signals of the GPS is used. To simulate one QZS, only one channel of L1 C/A is used. Because the commercially available software with GSS4730 called SimPLEX30 supports only simple orbit models and ionospheric delay, we have developed a new delay simulation method called SimQZ that uses the simulation delay data file made by us. The simulation delay data files are read by SimQZ in real time, and SimQZ controls the signal generator hardware to output the simulated QZS signal.
Navigation
Winter 2006
QZS Signal Receiver and Controller
Table 1—Specifications of the VCXO
Because the QZS signal includes time information of the onboard VCXO, the time information coming from the QZS can be adopted as the pseudo-range. To calculate the pseudo-range, an exclusive QZS signal receiver is used. The QZS signal receiver calculates the pseudo-range using the atomic clock at the GS and outputs it to the timing controller. The timing controller calculates the time adjustment commands using the pseudo-range. In the experiment, only the L1 C/A signal is used.
Manufacturer Name Standard Frequency Range of Control Voltage Weight Size Adjustment
Oscilloquartz S.A. OCXO 8607 5.000 MHz 0–10 V 900 g 138 73 88 mm 0.03 Hz/V
VCXO The requirements for the VCXO were reported in our previous paper [1]. Table 1 in this current paper shows the specifications of the VCXO used in our test bed, and Figure 3 shows the Allan deviation of the VCXO. During the measurement, the applied voltage to the VCXO is constant (not controlled). The dashed line shows the results obtained against a hydrogen maser used as a reference, and the solid line shows those obtained against another VCXO used as a reference. As the VCXO has higher short-term stability than the hydrogen maser, the latter results are better than the former for the short term, i.e., less than 10 s. In both results, the VCXO satisfies the requirements.
Fig. 3–Allan deviation of VCXO (open loop). During the measurement, the applied voltage to the VCXO is constant. The dashed line shows the results obtained against a hydrogen maser used as a reference, and the solid line shows those obtained against another VCXO used as a reference. The stability requirement of RESSOX was satisfied.
the TIC directly and the TIC evaluates the synchronization accuracy.
Time Interval Counter (TIC) To measure and evaluate the time difference between the VCXO and the TMS atomic clock, a TIC is used. One pps signals from each is input to
SIMULATION RESULTS The performance of the RESSOX was simulated on a computer. Table 2 shows the conditions for
Table 2—Delay Calculation Conditions Items
Simulation period Semi-major axis, m Eccentricity, m Inclination, deg
Values
2000.1.1 00:00:00UTC2000.1.2 00:00:00UTC 42164170.0 0.099 45.0
Right ascension of the ascending node, deg Argument of perigee, deg Mean motion, deg
205.0
Geopotential model
Vol. 53, No. 4
Items
Satellite mass, kg Satellite cross section, m2 CODE Data of Ionosphere Meteorological condition
270.0
Radiation pressure coefficient (Cr) Position of the TMS
120.0
Solid Earth tide
EGM96, n, m ¼ 360
Other celestial bodies
Values
3000.0 30.0 COD10426.ION 15 8C, 1013.25 hPa, 70% (relative humidity) 4.56 106 N/m2 (McCarthy 1996), Cr ¼ 1.2 26.5N, 127.9E, Height ¼ 0.0 m (Okinawa) Moon and Sun are considered, k2 ¼ 0.3 (IAG 1999) Moon, Sun, Mercury, Venus, Earth, Mars, Jupiter, Saturn, Uranus, Neptune, Pluto (JPL-DE405)
Iwata et al.: Simulation and Ground Experiments of Remote Synchronization
233
Fig. 4–Block diagram of simulation. The goal of the simulation is synchronization between the atomic clock and VCXO.
Fig. 5–Delays or error of L1 for the simulation. Maximum range error, ionospheric delay and tropospheric delay (absolute value) are approximately 40 ns, 35 ns, and 50 ns, respectively.
the simulation. The simulation date was January 1, 2000, and the data of the ionosphere and other celestial bodies on that day were used for the orbit and signal delay calculations. Figure 4 shows the block diagram of the simulation. ProportionalIntegral (PI) control for the applied voltage of the onboard VCXO was used as follows: vk ¼ offset
k2
k k1 X ðtOCXO tRESSOX Þi l þ 1 i¼kl
k1 Z X i¼0
iþp
ðtOCXO tRESSOX Þ dt i
where vk is the k-th output voltage, offset ¼ 5.0 (V), k1 is a proportional gain set at 3.8 106, k2 is an integral gain set at k1/120, l is the number of past data used for proportional control set at 1, k is the data number from the beginning of the control, p is the integral interval, which means an overlapping integral number, and is set at 2, and tRESSOX is time information of the received RESSOX control signal. Onboard time comparison error was added as noise simulating the time comparator and the VCXO and was generated by Stable 32, which is software to simulate clock behavior. In the simulation, it was assumed that knowledge from MCS that is shown as ‘‘Orbit/Delay Calculation (with error)’’ has 5.0 m error in each International Celestial Reference Frame (ICRF) axis initially and the velocity has no error. To calculate the orbit precisely, the EGM96 geopotential model with the spherical harmonic coefficient of degree 360, gravity effects of the Sun, the Moon, and other planets taken from the Jet Propulsion Laboratory (JPL) ephemeris DE405, radiation pressure, and solid tide effects were considered. Ionospheric and tropospheric delays were also considered; however, relativistic effects were not adopted. To calculate ionospheric delay, data (COD10426.ION) from the 234
Fig. 6–Simulation results with feedback. Synchronization error within 1 ns was realized.
Center for Orbit Determination in Europe (CODE) were used. Tropospheric delay was calculated with the Saastamoinen model. The error and delays of L1 are shown in Figure 5. Maximum range error, ionospheric delay and tropospheric delay (absolute value) were approximately 40 ns, 35 ns, and 50 ns, respectively, during simulation. Gaussian noise with r ¼ 3 ns was added to the QZS signal and the RESSOX control signal. On the ground, the result of comparison five seconds before adjustment between the L1 C/A pseudo-range and Orbit/Delay Calculation (with error) was used as the time adjustment command. The VCXO was synchronized to within 1 ns (Figure 6). EXPERIMENTAL RESULTS Long-Term Stability of VCXO Figure 7 shows the Allan deviation for the longterm stability of the PI-controlled VCXO when the cesium atomic clock and the hydrogen maser were used as references. In this experiment, no delays or errors were considered; the VCXO was controlled
Navigation
Winter 2006
required to calculate the time adjustment command. Figure 8 shows the results. When the error and delays were compensated for, and the time difference converged to zero, the synchronization error was within 1 ns. CONCLUSIONS This study is summarized as follows.
Fig. 7–Allan deviation for the long-term stability of the PI-controlled VCXO when the cesium atomic clock and the hydrogen maser were used as references. Stability at 100,000 s was less than 1014.
Fig. 8–Experimental results with feedback. Synchronization error within 1 ns was realized.
only by transmitter/receiver modems and PI control. Regarding the stability of the Allan deviation over a short term of less than 100 s, the stability using the hydrogen maser as the reference was approximately one order of magnitude smaller than that using the cesium atomic clock, and the stability is expected to be better than 1 3 1014 at 100,000 s. This means that the GS clock used as the reference affects the properties of the RESSOX. Experiments of RESSOX Experiments of RESSOX were conducted for seven hours. The experimental conditions were the same as those in the former simulation. The TIC was used to evaluate synchronization. In this experiment, the first-order least squares filter for 100 sets of comparison results between L1 pseudorange and Orbit/Delay Calculation (with error) was used to extrapolate the time adjustment command in the timing controller. Six seconds were
Vol. 53, No. 4
1. Apparatuses for realizing the RESSOX were considered. As exclusive apparatuses for the RESSOX, the TTA, the UDS, the modems, the QZS signal generator, the QZS receiver, the VCXO and the TIC were introduced. 2. Feedback control including the GS was simulated. The synchronization error was within 1 ns even though the maximum orbit error, tropospheric and ionospheric delays were approximately 40 ns, 35 ns, and 50 ns, respectively. 3. In the VCXO experiments using PI control, the long-term stability is expected to be better than 1 1014 at 100,000 s if the appropriate reference clock is chosen. 4. Experiments of the RESSOX were conducted. The error and delays were compensated and the final synchronization error was within 1 ns.
ACKNOWLEDGMENTS This study was carried out as part of the ‘‘Basic Technology Development of Next-Generation Satellites’’ project promoted by the Ministry of Economics, Trade and Industry (METI) through the Institute for Unmanned Space Experiment Free Flyer (USEF). REFERENCES 1. Tappero, F., et al., Proposal for a Novel Remote Synchronization System for the On-board Crystal Oscillator of the Quasi-Zenith Satellite System, Navigation, Journal of The Institute of Navigation Vol. 53. No 4, 2007, p. 219. 2. Koppang, P. A., Matsakis, D., and Miranian, M., Alternate Algorithms for Steering to Make GPS Time, Proceedings of ION GPS 2000, pp. 933–936. 3. Allan, D. W., Ashby, N., and Hodge, C. C., The Science of Timekeeping, Application Note 1289, Hewlett Packard, 1997, p. 60. 4. Imae, M., et al., Two-Way Satellite Time and Frequency Transfer Networks in Pacific Rim Region, IEEE Trans, Instrumentation and Measurement, Vol. 50(2), 2001, pp. 559–562.
Iwata et al.: Simulation and Ground Experiments of Remote Synchronization
235
