HARDWARE EXPERIMENTS OF RESSOX WITH GROUND STATION EQUIPMENT

Transcription

1 HARDWARE EXPERIMENTS OF RESSOX WITH GROUND STATION EQUIPMENT Masato Fukui, Akira Iwasaki Department o Aeronautics and Astronautics, University o Tokyo, Japan Toshiaki Iwata and Takashi Matsuzawa National Institute o Advanced Industrial Science and Technology (AIST), Japan Abstract We have developed ground station equipment and an onboard control program or the remote synchronization system or the onboard crystal oscillator (RESSOX) o the quasi-zenith satellite system (QZSS). First, we introduce the ground station equipment. Second, we explain RESSOX control algorithm using the L1/L/L5-band navigation signals o the QZSS and related experiments using the ground equipment. We ocus on the selection priority o navigation signals or eedback control based on the signal quality estimated rom the standard deviation o the pseudorange. I ionospheric delay is not considered, the method that uses only L5Q navigation signal is the best. However, i ionospheric delay is considered, the combination o L1C/A, LCL, and L5Q navigation signals is the best. Third, we discuss a control method to be implemented during communication interruption that takes place or approximately 3 minutes twice a day. We examined an average o 100 control voltages (150 s) immediately beore the communication interruption, and were able to achieve a synchronization error o less than 5 ns. I. INTRODUCTION The quasi-zenith satellite system (QZSS) has been under development as a Japanese space project since 003, and its main mission is navigation [1]. Its constellation consists o three satellites orbiting on inclined orbital planes with a geosynchronous period. The QZSS utilizes a high inclined orbit because o high visibility over high-latitude regions. In the case o the QZSS, at least one satellite is highly visible at the zenith at any time. Thereore, users can always receive navigation signals rom at least one o the QZSs near the zenith. In general, global navigation satellite systems (GNSSs), such as the GPS o the US, GLONASS o Russia, and GALILEO o Europe, are equipped with onboard atomic clocks that are used as time reerence. This is because: (1) atomic clocks have good long-term stability, () the orbit o satellites makes monitoring rom one ground station impossible, (3) these satellite systems are used or military missions and are thereore expected to operate even i ground stations are destroyed, and (4) these systems consist o many satellites, making the control o each satellite with many antennae diicult. However, onboard atomic clocks have the ollowing disadvantages: they are bulky, expensive to manuacture and launch, power-demanding, and sensitive to temperature or magnetic ield. Moreover, they are one o the main actors contributing to the reduction o satellite lietime.

2 The ollowing have been taken into consideration in the design o the QZSS as a civilian navigation system: (1) some crystal oscillators have better short-term stability than atomic clocks [], () 4-hour control with one station is possible i the location o the control station is appropriate, or example, Okinawa, Japan, and (3) the number o satellites is assumed to be only three. Given these considerations, the remote synchronization system or the onboard crystal oscillator (RESSOX), which does not require onboard atomic clocks, was developed. In the case o RESSOX, modiication o the control algorithm ater launch is easy because it is basically a ground technology. The target synchronization accuracy o RESSOX is set at 10 ns and the target stability is or more than 100,000 s. These targets were determined on the basis o the synchronization perormance between GPS time and UTC (USNO) [3] and the long-term stability perormance o onboard cesium atomic clocks [4]. RESSOX ground experiments and computer simulations have been conducted since 003. Details o primary experimental results have been introduced in our previous papers [5-8]. We have developed a eedback method that uses the L1/L/L5-band navigation signals o the QZSS, and ound that we do not need precise orbit inormation or estimation o delays, such as those caused by the ionosphere and troposphere, to realize RESSOX technology. We also ound that we can estimate separately the delay o L1/L/L5/Ku-band signals caused by the ionosphere and other sources, during error adjustment [8]. Strictly speaking, open navigation signals o the QZSS are L1C/A, L1CP, L1CD, LCM, LCL, L5I, and L5Q. When RESSOX eedback algorithm is considered, the navigation signal selection method is arguable. We will discuss this later. The QZSS experiences communication interruption or approximately 3 minutes twice a day as a result o avoiding intererence with geostationary satellites (GEO). The control method during interruption communication is also discussed later. II. RESSOX OVERVIEW Figure 1 shows the schematic o RESSOX. In order to realize RESSOX, it is indispensable to identiy the error actors and the eedback mechanism by measuring the delay at the ground station. The ormer is related to the estimation o error and delay using models, and is considered to be a eed-orward loop. The latter is an error adjustment system that uses pseudoranges measured with the navigation signals o the QZSS and estimated pseudoranges, and is considered to be a eedback control. The error and delay models in the eed-orward loop are delays in the ground station and in the satellite, tropospheric delay, ionospheric delay, delay due to distance (orbit estimation), delay due to relativity eects, and errors caused by Earth s motion, such as daily rotation, nutation, and precession. These problems were discussed in our previous paper [5]. However, as is described later, i L1/L/L5-band navigation signals are used or eedback, use o the delay models becomes unnecessary [7,8]. GPS Block-IIR satellites adopt a timekeeping system (TKS) in which the crystal oscillator is used as the system clock and atomic clocks are used as the reerence clock, because the short-term stability o the crystal oscillator is superior to that o the atomic clock [9]. Figure shows a comparison o RESSOX with the TKS o GPS Block-IIR satellites. The GPS Block-IIR TKS utilizes the short-term requency stability o the crystal oscillator and the long-term stability o the rubidium atomic requency standard (RAFS); the crystal oscillator contributes to the whole system, whereas the RAFS is considered to be a reerence clock. In contrast, in the case o RESSOX, the QZSS is controlled by the crystal oscillator and the crystal oscillator is adjusted based on time inormation called RESSOX control signal rom the ground station where the standard time is kept, similar to the RAFS o the GPS Block-IIR TKS. We assume that the RESSOX control signal is uplinked with a pseudo-noise (PN) code using the Ku-band.

3 The short-term requency stability o the QZSS using RESSOX is expected to be at least equivalent to that o the onboard atomic requency standard system. Gravity distribution, Sun, Moon, Planets, Solar radiation pressure, tide Error o orbit estimation, maneuvering and unloading GPS Satellite Delay in the QZS QZS Relativity eects Delay in the ionosphere Delay caused by distance Crystal Oscillator Ionosphere Delay caused by troposphere RESSOX control signal Time Signal QZS Signal Signal QZS Signal Signal GPS Satellite Delay in the ground station Atomic Clock Time Management Station Error adjustment based on the measurement User GPS Satellite Transormation o daily rotation, precession and nutation Fig. 1. RESSOX schematic. Radio Frequency GPS Block-IIR RESSOX PN Code Generator Atomic Clock Fig.. Assurance o long-term stability by Rb atomic clock Rb Atomic Clock 13.4 MHz Reerence Epoch Generator 1.5 s Phase Meter Phase Meter Assurance o long-term stability by ground station standard time Time Time Intermediate Comparison Comparator Frequency Unit Voltage Voltage 1.5 s Crystal Oscillator 10.3 MHz System Epoch Generator Assurance o short-term stability by Crystal Oscillator Crystal Oscillator 10.3 MHz Assurance o short-term stability by Crystal Oscillator Comparison between RESSOX and the TKS o GPS Block-IIR.

4 III. GROUND STATION EQUIPMENT FOR RESSOX In this section, the hardware or realizing RESSOX technology is described. Figure 3 shows the system diagram o RESSOX. The time o the time management station (TMS) is used as the standard time (called QZSS-Time). The RESSOX control signal transmitter (RCST) advances the time to compensate the delay during transmission between the QZSS-Time and the crystal oscillator onboard the QZS and the time inormation is modulated using PN codes. The RESSOX control signal is up-converted to GHz (Ku band) by the up-converter and is transmitted rom the antenna o the TMS to the QZS. At the QZS, the RESSOX control signal is received by the antenna, down-converted, demodulated, and compared with that o the onboard oscillator by the time comparison unit (TCU). On the basis o the comparison, the time-dierence inormation (PN-code phase dierence) is transerred to the navigation onboard computer (NOC) that then generates the control command (voltage to be applied) or the crystal oscillator inside the timekeeping circuit through some control algorithm. On the QZS, the navigation signals (QZS signals) o L1/L/L5 bands are generated using the onboard crystal oscillator as reerence clock. At the TMS, the QZS signals are received by the antenna and transmitted to the QZSS/GPS receiver or RESSOX. This receiver compares the time inormation in the QZS signal with the QZSS- Time, and outputs it as the pseudorange. The pseudorange is used as eedback inormation o RESSOX. The RESSOX controller at the TMS controls the RCST using the delay models and pseudorange inormation. Navigation Onboard Computer (NOC) RESSOX Control Sotware Phase Dierence Positioning Command JAXA Assets NICT Assets AIST Assets NICT signal A B QZS Site BPSK, MHz Down-Converter Time Comparison Unit (TCU) Standard Time Generator Voltage-Controlled Crystal Oscillator (VCXO) Voltage Command 10.3 MHz 1.5s pulse Ku Band Antenna NICT Signal Telemetry Up-Converter L1C/A, L1CD, L1CP, LCM, LCL, L5I, L5Q Navigation Signals (QZS Signals) L Band Antenna L Band Antenna Ku Band Antenna GHz GHz NICT signal A B Hybrid Combiner Up-Converter Down-Converter Time Management Station (TMS) Ground Site GPS Synchronizer RESSOX Control Signal RESSOX Control Signal Positioning Command RESSOX Control Signal NICT Signal Telemetry Time Comparison Results Telemetry Positioning Command L Band Signal Transmission Subsystem (LTS) Tracking and Control Station Master Control Station (MCS) RESSOX Control Signal Time Adjustment File RESSOX QZSS/GPS Transmitter (RCST) Time Adjustment Controller Receiver 10.3 MHz Command Pseudorange 1.5 s Pulse Adjustment 1.5 s Pulse 10 MHz 1 s Pulse Command Frequency Transormer 10 MHz 1 s Pulse Temperature AIST Terminal Humidity Pressure Fig. 3. System diagram o RESSOX. Meteorological Observation Equipment

5 Other time systems such as GPS-Time can be used instead o QZSS-Time. In such a case, the output o GPS synchronizer or the results o GQTO prepared by the National Institute o Inormation and Communication Technology (NICT) will be used as standard time. Figure 4 shows the schematic o the test bed or the preliminary ground experiments using these ground station apparatuses. Most o the components are the same as those illustrated in Fig. 3; however, some special apparatuses and sotware are required to simulate the delay between the TMS and the QZS. The details o some apparatuses used in the ground experiments are described below. Time Interval Counter I.5 s Pulse D/A Converter Sim. o NOC I.5 s Pulse Phase Dierence Sim. o TCU UDS RCST Frequency Transormer H-Maser MHz, PN modulation MHz, PN modulation Time Adjustment File Time Adjustment Command I PPS, 10 MHz Apparatuses o Ground Station Apparatuses or Validation Control Voltage MINI-OCXO Voltage Commands 10.3 MHz I.5 s Pulse Delay Adjustment File 1.5 s Pulse Generator RESSOX Controller I.5 s Pulse, 10.3 MHz 1.5 s Timing Adjustment Command I PPS, 10 MHz I PPS QZSS Signal Generator L1C/A, L1CD, L1CP, LCM, LCL, L5I, Pseudorange L5Q QZSS/GPS Receiver Sim. o NOC: Simulator o Navigation Onboard Computer Sim. o TCU: Simulator o onboard Time Comparison Unit UDS: Uplink Delay Simulator RCST: RESSOX Control Signal Transmitter Fig. 4. Schematic o the test bed or preliminary ground experiments. The equipment or the ground experiments on the rack is shown in Fig. 5. The apparatuses used in the ground experiments are introduced below. FREQUENCY TRANSFORMER Although the TMS uses only the standard time signal o 10 MHz and 1 s pulse, the RESSOX system uses both the time signal o 10.3 MHz and 1.5 s pulse and that o 10 MHz and 1 s pulse. Thereore, a requency transormer is required. The requency transormer has one input o each time signal o 10 MHz and 1 s pulse, ive outputs o each time signal o 10.3 MHz and 1.5 s pulse, and two outputs o each time signal o 10 MHz and 1 s pulse. Since there are three relationships between the 1 s pulse and the 1.5 s pulse (i.e., shits o ±0.5 s and 0 s), the RESSOX controller incorporates a shit o 0.5 s with the RS3C interace.

6 Meteorological Observation Equipment GPS Synchronizer QZSS/GPS Receiver RESSOX Control Signal Transmitter (RCST) Frequency Transormer MINI-OCXO 1.5 s Pulse Generator Simulator o onboard Time Comparison Unit (TCU) Uplink Delay Simulator (UDS) Fig. 5. Equipment or ground experiments. RESSOX CONTROL SIGNAL TRANSMITTER (RCST) RCST consists o two parts: transmitting time adjuster (TTA) and PN code generator. (1) TTA TTA generates the advanced time used to compensate the delay o the uplink time signal between the TMS and the QZS according to time adjustment iles and commands. The time adjustment iles and commands or RCST will be generated and transerred by the RESSOX controller at the TMS with TCP/IP connection. TTA changes the requency using the waveront clock principle [10]. TTA uses time adjustment iles (eed-orward control) and time adjustment commands (eedback control). To realize eed-orward control, the delay estimation will be conducted using delay models or inormation rom MCS. To obtain the time adjustment iles, the delay estimation results will be prepared as a delay ile, as shown in Fig. 6. This ile will include the date and time in UTC and the uplink signal delay in nanoseconds. The time adjustment ile shown in Fig. 7 will be generated on the basis o the delay ile. This ile will include the next ile name, data number, modiied Julian day (MJD), and UTC o the irst data; valid time o each coeicient; and coeicients o the polynomial estimated by Lagrange s interpolation o the 11th order or less. To realize eedback control, the dierences between estimated pseudoranges o navigation signals and measured pseudoranges are used. The dierences are extrapolated and expressed in the polynomial.

7 The coeicients o the polynomial will be used or the eedback control. The mechanism o eedorward and eedback controls is shown in Fig. 8. Feed-orward polynomial coeicients are adjusted with eedback coeicients. 000/1/1 0:00:00, /1/1 0:01:00, /1/1 0:0:00, Fig. 6. Example o delay ile. Next ile name Data number MJD, UTC Valid time o each coeicient C0, C1,, C11 #1. C0, C1,, C11 #N Fig. 7. Example o time adjustment ile. () PN Code Generator The PN code generator generates time inormation o the RESSOX control signal using the 5115 bit o an LC code or 0.5 ms. Time inormation advanced by TTA will be transmitted. The RESSOX control signal has a data overlay o 000 bit/s to resolve the ambiguity o 0.5 ms. Overlay data will be treated as dummy data. Orbit Calculation/ Delay calculation with model Feed-orward polynomial y = c + c t + c t + L+ c t c 0,, c 11 are read rom the time adjustment ile. Y is calculated every 0.1 ms Measured Pseudo Range Estimated Pseudo Range Feedback polynomial y b = d L d1t + dt + d11t d 0,, d 11 are given as a command Y b is calculated every 1 s + Set to DDS Fig. 8. Mechanism o eed-orward and eedback controls. QZSS/GPS RECEIVER The QZSS/GPS receiver has the ollowing capabilities: (1) it measures and outputs raw data o pseudoranges, navigation message, carrier phase, delta range, and signal intensity o tracking satellites; () it sends to the RESSOX controller the satellite PRN number and the pseudoranges (three decimal places, in meters) o L1C/A, L1CD, L1CP, LCM, LCL, L5I, and L5Q navigation signals every second; and (3) it receives the standard time signal o 10 MHz and 1 s pulse. The L-band antenna (choke coil antenna abricated by Trimble) is Japan Aerospace Exploration Agency s (JAXA) asset and JAXA has graciously shared the signals with us. There are two communication channels to the RESSOX controller: serial ports 1 and. Serial port 1 is

8 used or eedback control and serial port is used or maintenance. The measurement results o constant pseudorange (50.0 km, all signals do not have ambiguity) generated by a simulator o the QZSS signal generator are shown in Fig. 9. Since there are biases among signals, the results are shown or each signal (pseudorange scales are the same to acilitate comparison). The powers o the signals simulate the actual strength, that is, when L1 total signal power is assumed to be Pseudorange (m ) L1C/A Elapsed T im e (s) Pseudorange (m ) L1CD Elapsed T im e (s) Pseudorange (m ) Pseudorange (m ) Pseudorange (m ) L1CP Elapsed T im e (s) LCL Elapsed T im e (s) L5Q Elapsed T im e (s) Pseudorange (m ) Pseudorange (m ) LCM Elapsed T im e (s) L5I Elapsed T im e (s) Fig. 9. Measurement results o constant pseudorange (50.0 km).

9 110 dbm, the signal power o L1C/A is dbm, that o L1CD is dbm, that o L1CP is dbm, those o LCM and LCL are dbm, and those o L5I and L5Q are 113. dbm, respectively. Pseudorange deviations o the signals are as ollows: or L1C/A, it is ns; or L1CD,.5394 ns; or L1CP, ns; or LCM,.0749 ns; or LCL, ns; or L5I, ns; and or L5Q, ns. To calculate the deviations, data between 900 and 1100 s were used. UPLINK DELAY SIMULATOR (UDS) UDS is a hardware simulator used in the ground test bed to simulate the delay between the ground and the QZS. UDS assigns oset values to irst-in irst-out (FIFO) memory according to the delay adjustment iles or UDS, which have the same ormat as the time adjustment iles or RCST. Then, UDS adjusts the time inormation o the modiied 10.3 MHz generated by RCST. UDS is based on DDS technology o RCST and FIFO memory control techniques. UDS receives an intermediate requency (IF) signal that includes time inormation (central requency o MHz, bandwidth o.5 MHz), down-converts the requency, downloads the waveorm o time inormation into FIFO memory in real time, and reads out FIFO memory data using the modiied 10.3 MHz generated by TTA inside UDS. Herewith, UDS upconverts the requency to MHz and outputs the modiied IF signal using the time adjustment iles or UDS. RESSOX CONTROLLER The RESSOX controller is a Windows XP PC, as shown in Fig. 10. The RESSOX controller will conduct orbit calculation o the QZS and delay estimations o Ku/L1/L/L5- band signals, and output the total delay in nanoseconds with the designated UTC and time interval. The orbit calculation involves the ollowing procedures: 1) The initial values are given as six elements in the International Celestial Reerence Frame (ICRF), the International Terrestrial Reerence Frame (ITRF), the True o Date (TOD) or Keplerian, and the solution or a predeined period is given by solving the equation o motion. The integral interval can be set by the user. ) JGM-3 or EGM-96 is selected as the geopotential model, and orders o the model can be set by the user. 3) The eect o the sun, the moon, and other planets can be selected by the user. 4) The eect o solar pressure can be utilized. 5) The solid tide eect can be taken into consideration. 6) The eect o relativity can be included. 7) Impulse acceleration can be set by the user. For the delay estimation, 1) ionospheric delay, tropospheric delay, and geographical delay should be considered, and ) the delay in the TMS or the QZS should be taken into account. Every 30 s, the RESSOX controller will also receive rom MCS 1-s estimation o orbit inormation or a duration o 3 minutes, or receive L1C/A navigation message inormation rom MCS every hour. Using this inormation, the RESSOX controller will calculate the delay iles and the time adjustment iles every 30 s or every hour and make a chain o iles. For the results o total delay estimation o the Ku-band RESSOX control signal, the RESSOX controller will output the time adjustment iles or eed-orward control using Lagrange s interpolation o the 11th order or less. RCST will reer to these iles. The RESSOX controller will also output the delay adjustment iles or UDS, which are similar to the time adjustment iles or RCST. The pseudoranges o the L1/L/L5-band navigation signals will be received rom the QZSS/GPS receiver every second. Using the pseudoranges rom the QZSS/GPS receiver and the estimation results o the total delays o the L1/L/L5-band navigation signals, the RESSOX controller will calculate Lagrange s

10 extrapolating coeicients o the 11th order or less and generate the eedback command or RCST. The time interval o data or output will be chosen between 1 and 900 s. The RESSOX controller will generate the command or the requency transormer to adjust the 1.5 s pulse, and or RCST or UDS to start or stop the job or to set parameters. The RESSOX controller will also monitor the status o RCST and UDS. As reerence, the RESSOX controller will accumulate the results o the time dierence and orbit estimations o MCS obtained rom the TMS. SIMULATOR OF ONBOARD TIME COMPARISON UNIT (TCU SIMULATOR) TCU will be prepared by NICT, and to realize experiments on the ground, we prepared the simulator. Since the onboard system clock is 1.5 s epoch, the time dierence between RESSOX control signal and MINI-OCXO, which is the engineering model o the onboard crystal oscillator, is expressed between 0.75 s and s. SIMULATOR OF NAVIGATION ONBOARD COMPUTER (NOC SIMULATOR) NOC will be prepared by JAXA, and to realize experiments on the ground, we prepared the simulator. The Windows XP PC shown in Fig. 11 is used or this purpose. This simulator receives the phase dierence rom the onboard TCU simulator and drives the D/A converter that generates voltage between 0 V and 10 V. Fig. 10. RESSOX Controller. Fig. 11. Simulator o NOC. SIMULATOR OF QZSS SIGNAL GENERATOR The QZSS signal generator will be prepared by JAXA, and to realize experiments on the ground, we prepared the simulator (special equipment abricated by Spirent Communications). The speciications o the simulator are as ollows: (1) The types o signals are L1C/A (PRN: 193-0); L1CP and L1CD (BOC(1,1)); LCM and LCL; and L5I and L5Q o each channel with QZSS navigation message.

11 () Operation is accomplished with a modiied SimQZ ile and SimPlex30 sotware [11], with 10.3 MHz. At least seven signals described in (1) can be broadcast simultaneously. The modiied SimQZ ile includes inormation o time stamp in ms, the satellite identiier as an integer value (193-0), ive bits o L1C/A navigation message to be transmitted during the 100 ms time step, power level or L1 navigation signal, L1 navigation signal carrier pseudorange in meters, oset o L1 code pseudorange rom L1 navigation signal carrier pseudorange in meters, L1 navigation signal carrier pseudorange rate o change in meters per second, oset o L1 code pseudorange rate rom the L1 carrier pseudorange rate in meters per second, power level or L navigation signal, L navigation signal carrier pseudorange in meters, oset o L code pseudorange rom L navigation signal carrier pseudorange in meters, L navigation signal carrier pseudorange rate o change in meters per second, oset o L code pseudorange rate rom the L carrier pseudorange rate in meters per second, ten symbols o navigation message to be transmitted during the 100 ms time step or L1C, ive symbols o navigation message to be transmitted during the 100 ms time step or LC, ten symbols o navigation message to be transmitted during the 100 ms time step or L5, power level or the L5 navigation signal, L5 navigation signal carrier pseudorange in meters, oset o L5 code pseudorange rom L5 navigation signal carrier pseudorange in meters, L5 navigation signal carrier pseudorange rate o change in meters per second, and oset o L5 code pseudorange rate rom L5 navigation signal carrier pseudorange rate in meters per second. (3) The simulator is operated in a stand-alone coniguration or synchronized with the GPS simulator. (4) When the simulator is synchronized with the GPS simulator, the synchronization should be within 10 ns. (5) Sotware that synchronizes the operations o both GPS and QZS simulator should be prepared. (6) Clock change rate is 3.3 Hz/1.5 s in maximum. (7) Speciy power range, code, and navigation message by reerring to IS-QZSS [1]. 1.5 S PULSE GENERATORS To synchronize MINI-OCXO, the simulator o QZS signal generator, and the TCU simulator in the ground experiment, a 1.5 s pulse generator is used. MINI-OCXO To simulate the onboard crystal oscillator, an engineering model o the onboard crystal oscillator, called MINI-OCXO, is used. Speciications or requency stability are the same as those o the onboard crystal oscillator. The measured Allan deviation o MINI-OCXO is shown in Fig. 1. IV. CONTROL METHODS RESSOX ALGORITHM RESSOX control algorithm or ground hardware experiments or computer simulation is outlined as ollows.

12 1.0E-11 Allan D eviation 1.0E-1 1.0E Time (s) Fig. 1. Allan deviation o MINI-OCXO. Step 1. Four (L1/L/L5/Ku-band signals) estimated delay iles are prepared, which include model errors such as those due to the orbit, ionosphere, or troposphere, and we assume that they are used at the TMS as measurement results. Three (L1/L/L5-band navigation signals) delay iles include the times (date and UTC) that the L1/L/L5-band navigation signals are received and the estimated delays o the L1/L/L5- band navigation signals. Another delay that is contained in the Ku-band estimated time adjustment ile includes the time (date and UTC) that the Ku-band signal is transmitted rom the TMS and the estimated delay o the Ku-band signal. These estimated delays are converted into database o L1/L/L5-band delays and the time adjustment ile or TTA, respectively. Step. Four (L1/L/L5/Ku-band signals) authentic delays that do not contain any errors are prepared. Three (L1/L/L5-band navigation signals) delay iles include the times (date and UTC) that the L1/L/L5-band navigation signals are received at the TMS and the authentic delays o the L1/L/L5-band navigation signals. Based on the authentic delays, a simulation data ile in the CSV ormat or SimQZ is generated. Another delay ile that is used as the delay adjustment ile or UDS includes the time (date and UTC) that the Ku-band signal is transmitted rom the TMS. Step 3. The time adjustment ile or RCST is ed into the RCST as eed-orward control. Step 4. UDS delays the RESSOX control signal according to the delay adjustment ile or UDS. Step 5. The onboard crystal oscillator is controlled using the time dierence between the RESSOX control signal and the time o the crystal oscillator itsel. Step 6. The QZS signal generator generates L1/L/L5-band navigation signals according to the simulation data ile in the CSV ormat or SimQZ. Step 7. Pseudoranges o L1/L/L5-band navigation signals obtained by the QZSS/GPS receiver are compared with the database o L1/L/L5-band delays and the dierences between the pseudoranges and the database are designated as E 1 or L1 (requency L1 = Hz), E or L ( L = Hz), and E 3 or L5 ( L5 = Hz). Step 8. The system o equations (1), (), and (3), which includes E 1, E, and E 3 and delays due to the nonrequency-dependent term e and the coeicient o delay k due to the requency-dependent term (i.e., ionospheric delay) as unknowns, is prepared.

13 k 9 e + = E1, 1 = [Hz] L L1 k 9 e + = E, = [Hz] L L k 9 e + = E3, 5 = [Hz] L L5 (1) () (3) Step 9. I the navigation signals o all the three requencies are available, the system o equations (1), (), and (3) is expressed as ollows: / 1/ 1/ L1 L L5 E1 e Ax = = E = E k E 3 (4) This means that three equations exist or two unknown parameters (e and k). Thereore, the solution o simultaneous equations (1), (), and (3) is given as ollows: T 1 T x = ( A A) A E (5) I the navigation signals o two requencies are available, then two equations exist with two unknown parameters. Using the solutions o the system o equations, we obtain the time to be adjusted o the RESSOX control signal using the Ku band ( Ku = Hz) or the TTA. e k, = Ku Ku 10 [Hz] (6) I the navigation signals o only one requency are available, the time to be adjusted is given by E i (i=1 or or 3). Step 10. By combining the time adjustment ile in Step 3 and the time adjustment command based on the time to be adjusted, TTA is controlled. We consider some ilters in this step, as described later. Then, we go back to Step 4. The calculation o the time to be adjusted and the time adjustment command is conducted every second. Figure 13 shows the block diagram o the RESSOX algorithm using the three-requency navigation signals in Step 8. CRYSTAL OSCILLATOR CONTROL METHOD To control MINI-OCXO using the dierence between uplinked time inormation and MINI-OCXO time, a kind o PI control or applied voltage was utilized. The ollowing ormula describing PI control was used:

14 v k = oset k1 l + 1 k i= k l ( t OCXO t RESSOX ) i k k 1 ( i= 0 i+ p i ( t OCXO t RESSOX ) dt, (4) where v k is the k-th output voltage, oset = 5.4 (V), k 1 is a proportional gain set at , k is an integral gain set at , l is the number o past data used or proportional control set at 1, k is data number rom the beginning, p is the integral interval, which means an overlapping integral number, set at, and t RESSOX is time inormation o the received RESSOX control signal. Crystal Oscillator Time dierence QZS Site QZS Signal Generator Voltage (5) Crystal Oscillator Controller (PI control) + L1/L/L5 authentic delay ile Time inormation o Crystal Oscillator Delay o QZS signal + + Noise () Orbit/Delay Calculation (without error) L1/L/L5 band Delay adjustment ile TMS Site Advanced time inormation o Atomic Clock + QZSS/GPS Receiver - Time adjustment ile or TTA Noise + Atomic Clock + Stable (4) Delay o Ku band RESSOX Control Signal Time + RESSOX Transmitter (RCST) Comparator - control signal + (10) Every 1.5 s Every 1 s (3) Serial Data Date, UTC, L1 Pseudorange L Pseudorange L5 Pseudorange Delay reerence ile Timing Controller Date, Database UTC, L1_delay, o L1/L/L5- L_delay band delays + (1) Orbit/Delay Calculation (with error) (8) Time adjustment command (9) Time to be adjusted e, k - (7) E 1, E, E 3 k 9 e + = E1, [Hz] L1 = L1 k 9 e + = E, = [Hz] L L k 9 e + = E3, 5 = [Hz] L L5 (6) k 10 e +, = [Hz] Ku Ku Least-squares ilter Fig. 13. Block diagram o the RESSOX algorithm. GROUND STATION CONTROL METHOD At the ground station, the control method was dependent on how the eedback was conducted, i.e., the iltering method in Fig. 13. The ilter was constructed with 100 data o time to be adjusted using the dierence between measured pseudoranges o L1/L/L5-band navigation signals and estimated pseudoranges o the same signals prepared as the database o L1/L/L5-band navigation signal-delays every second rom 6 s beore to 105 s beore the time to transmit the time adjustment command to RCST. The 100 data o time to be adjusted were used or the irst-order least-squares iltering, and the time to be adjusted was extrapolated to the current time, as shown in Fig. 14. To calculate the iltering result and send it to the TTA as the time adjustment command, 6 s is required.

15 V. HARDWARE EXPERIMENTS EXPERIMENTAL CONDITIONS The experimental conditions are shown in Table 1, Fig. 14, and Fig. 15. The simulated date was 1 January 000, and ionospheric conditions and conditions o other celestial bodies or that day were used as data. In Table 1, typical Keplerian elements o the QZSS are shown as the initial conditions (i.e., on 1 January 000, 00:00:00 in UTC). These conditions can be expressed as x = m, y = m, z = m, v x = m/s, v y = m/s, v z = m/s in the ICRF. The TMS was assumed to be located in Okinawa, and meteorological conditions used to calculate tropospheric delay were assumed to be constant. Based on the calculations o orbit and delay, the delay adjustment ile or UDS and the CSV ile or SimQZ were prepared. Time to be adjusted Extrapolated with irstorder least-square ilter -105 Elapsed time with current time set as 0 (s) -6 0 Fig. 14. Control method at TMS. Table 1. Experimental conditions. Items Simulation period Semimajor axis, m Eccentricity Inclination, deg Right ascension o the ascending node, deg Argument o perigee, deg Mean anomaly, deg Geopotential model Potential deg Satellite mass, kg Values :00:00 UTC :00:00 UTC EGM96 n, m= Items Satellite cross section, m CODE data o ionosphere Meteorological condition Radiation pressure coeicient (Cr) Position o ground station Solid Earth tide Other celestial bodies Values 37.0 COD1046.ION 15, hPa, 70% (relative humidity) N/m (McCarthy 1996), Cr = N, 17.9E, Height = 0.0m (Okinawa) Moon and Sun are considered, k = 0.3 (IAG 1999) Moon, Sun, Mercury, Venus, Mars, Jupiter, Saturn, Uranus, Neptune, Pluto (JPL-DE405)

16 FEED-FORWARD CONTROL EXPERIMENTAL RESULTS Uplink experiments (eed-orward control) were conducted. In the experiments, the delay compensation was applied to the TTA, the delay due to transmission was applied to the UDS, and MINI-OCXO was controlled by PI control. As the estimated delays, three cases were prepared. The initial satellite positions have no error, +5 m error in each ICRF axis, and 5 m error in each ICRF axis. The synchronization results o the experiments and the expected synchronization errors are shown in Figs. 17 and 18, respectively. They are consistent with each other. Measured position Ionospheric delay, other celestial +5 m bodies: Data o January 1, m Z Estimated range Ground station X ICRF Tropospheric delay Ionospheric delay Authentic range Y +5 m Earth Authentic position Fig. 15. Experimental conditions (+5 m ICRF error). Ionospheric delay, other celestial bodies: Data o January 1, 000 Ground station X ICRF Z Earth Authentic position -5 m Authentic range Ionospheric delay Tropospheric delay Estimated range Y Measured position -5 m -5 m Fig. 16. Experimental conditions ( 5 m ICRF error). Synchronization Error (ns) m Error No Error +5 m Error Elapsed Time (h) Fig. 17. Feed-orward control results. Scynchronization Error (ns) m Error m Error /1/1 0:00 000/1/1 6:00 000/1/1 1:00 000/1/1 18:00 000/1/ 0:00 Time (UTC) Fig. 18. Expected synchronization error. CONTROL DURING COMMUNICATION INTERRUPTION Figure 19 shows the control results during communication interruption (approximately 190 s) when intererence with GEO satellites is avoided. To simulate the communication interruption, experiments were conducted once every hour. The average o the voltages in the last 150 s (100 samples) beore communication interruption was supplied to MINI-OCXO. In the experiments, the time dierence o μs corresponds to complete synchronization between the QZSS and the ground station (3.370 μs is the internal delay o RCST and UDS). The maximum error was approximately 5 ns in the experiments. However, since the errors are statistical, the probability o error exceeding 10 ns is possible.

17 VI. SIGNAL PRIORITY Based on the distribution o pseudoranges shown in Fig. 6, the combination o navigation signals or RESSOX was evaluated using RESSOX algorithm. In this evaluation, only transmission delay error was considered (the initial position o the QZS has +5 m error or each ICRF axis, and the error does not contain ionospheric or tropospheric error). The distribution o synchronization error o RESSOX was evaluated. The results are shown in Table. The results show that using L5Q navigation signal alone is the best or RESSOX. However, in the actual case, ionospheric and tropospheric delays exist. In particular, ionospheric delay can be estimated exactly by using two or more requency navigation signals. This means that i only one requency navigation signal is used or RESSOX, especially L5 that will cause the largest ionospheric delay due to the lowest requency, the ionospheric delay will not be cancelled. Thereore, the synchronization errors were investigated again in the case o ionospheric and tropospheric delays. The results are shown in Table 3. In such a condition, the combination o L1C/A, LCL, and L5Q navigation signals is the best synchronization result Time Dierence (μs) D ata sam pling Communication Interruption :45:00 13:45:00 14:45:00 15:45:00 16:45:00 17:45:00 Time (hh:mm:ss) Fig. 19. Control results during communication interruption. VII. CONCLUSIONS This study is summarized as ollows. (1) The RESSOX concept and the apparatuses used in this research were introduced. () The RESSOX algorithm using any combinations o L1/L/L5-band navigation signals was introduced. (3) Feed-orward control experiment was explained. When orbit calculation was correct, synchronization error luctuation was within ns. When errors o +5 m and -5 m in the x direction, y direction, and z direction in the ITRF coordinate were added to the epoch satellite position, the luctuation o synchronization error was 40 ns. The dierence o synchronization errors was almost as much as the dierence o expected delays.

18 (4) Drit behaviors o MINI-OCXO during communication interruption were measured. In this experiment, the average voltage o the last 100 voltage commands immediately beore communication interruption was used as ixed control voltage. The maximum synchronization error in the experiments was 5 ns; however, since the errors are statistical, the probability o error exceeding 10 ns is possible. (5) For the combination o navigation signals, priority signals are investigated. I ionospheric and tropospheric delays are not considered, the method using only L5Q is the best. However, i the ionospheric and tropospheric delays are considered, the method that combines L1C/A, LCL, and L5Q is the best. Table. Signal combination priority in case that ionospheric and tropospheric delays are not considered. SignalC om bination Standard deviation A verage R ange M inimum Maximum L5Q E E E E E-10 L5I E E E E E-10 L1CA E E E E E-10 L1CP.08868E E E E E-10 L1C A/L5Q.1647E E E E E-10 L1C A/L5I.175E E-1 1.6E E E-10 L1CA/LCL/L5Q.85E E E E E-10 L1CA/LCL/L5I.877E E E E E-10 LCL.4014E E E E E-10 L1CA/LCM /L5Q.45457E E E E E-10 L1CA/LCM /L5I.45513E E E E E-10 L1CA/LCL E E-1.641E E E-09 L1C P/L5Q E E-1.618E E E-09 L1C P/L5I E E-1.618E E E-09 L1CP/LCL/L5Q E E-1.73E E E-09 L1CP/LCL/L5I E E-1.731E E E-09 LCM E E E E E-09 L1C P/LC M /L5Q E E-1.964E E E-09 L1C P/LC M /L5I E E-1.964E E E-09 L1CD E E E E E-09 L1C P/LC L E E E E E-09 L1CA/LCM E E E E E-09 L1C P/LC M E E E E E-09 L1C D /L5Q E E E E E-09 L1C D /L5I E E E E E-09 L5CD/LCL/L5Q E E E E E-09 L1CD/LCL/L5I E E E E E-09 L1C D /LC M /L5Q 8.95E E E E E-09 L1C D /LC M /L5I E E E E E-09 L1C D /LC L E E E E E-09 L1C D /LC M E E E-09-4.E E-09 LC L/L5Q.003E E E E E-09 LC L/L5I.00358E E E E E-09 LC M /L5I E E E E E-08 LC M /L5Q E E E E E-08 Table 3. Signal combination priority in case that ionospheric and tropospheric delays are considered. SignalC om bination Standard deviation Average Range Minimum Maximum L5Q E E E E E-08 L1CA E E E E E-08 L1CA/LCL/L5Q.787E E E E E-10 L1C A/LC M /L5I.45471E E E E E-10

19 VIII. ACKNOWLEDGMENT This study was carried out as part o the Basic Technology Development o Next-Generation Satellites project promoted by the Ministry o Economics, Trade, and Industry (METI) through the Institute or Unmanned Space Experiment Free Flyer (USEF). REFERENCES [1] M. Kishimoto, H. Hase, A. Matsumoto, T. Tsuruta, S. Kogure, N. Inaba, M. Sawabe, T. Kawanishi, S. Yoshitomi, and K. Terada, 007, QZSS System Design and its Perormance, in Proceedings o the ION National Technical Meeting, -4 January 007, San Diego, Caliornia, USA (Institute o Navigation, Alexandria, Virginia), pp [] J. J. Suter, L. J. Craword, B. G. Montgomery, and W. E. Swann, 000, Syntonics LLC APL- Developed Technology Makes Its Commercial Debut, Johns Hopkins APL Technical Digest,, [3] P. A. Koppang, D. Matsakis, and M. Miranian, 000, Alternate Algorithms or Steering to Make GPS Time, in Proceedings o the ION GPS Meeting, 19- September 000, Salt Lake City, Utah, USA (Institute o Navigation, Alexandria, Virginia), pp [4] D. W. Allan, N. Ashby, and C. C. Hodge, 1997, The Science o Timekeeping, Application Note 189 (Hewlett-Packard), p. 60. [5] F. Tappero, A. Dempster, T. Iwata, M. Imae, T. Ikegami, Y. Fukuyama, K. Hagimoto, and A. Iwasaki, 006, Proposal or a Novel Remote Synchronization System or the On-Board Crystal Oscillator o the Quasi-Zenith Satellite System, Navigation, 53, [6] T. Iwata, M. Imae, T. Suzuyama, H. Murakami, Y. Kawasaki, N. Takasaki, A. Iwasaki, F. Tappero, and A. Dempster, 006, Simulation and Ground Experiments o Remote Synchronization System or Onboard Crystal Oscillator o Quasi-Zenith Satellite, Navigation, 53, [7] T. Iwata, Y. Kawasaki, M. Imae, T. Suzuyama, T. Matsuzawa, S. Fukushima, Y. Hashibe, N. Takasaki, K. Kokubu, A. Iwasaki, F. Tappero, A. Dempster, and Y. Takahashi, 007, Remote Synchronization System o Quasi-Zenith Satellites Using Multiple Positioning Signals or Feedback Control, Navigation, 54, [8] T. Iwata, M. Imae, T. Suzuyama, Y. Hashibe, S. Fukushima, A. Iwasaki, K. Kokubu, F. Tappero, and A. G. Dempster, 008, Remote Synchronization Simulation o Onboard Crystal Oscillator or QZSS Using L1/L/L5 Signals or Error Adjustment, International Journal o Navigation and Observation, 008, Article ID [9] H. C. Rawicz, M. A. Epstein, and J. A. Rajan, 1993, The Time Keeping System or GPS Block IIR, in Proceedings o the 4th Annual Precise Time and Time Interval (PTTI) Applications and Planning Meeting, 1-3 December 199, McLean, Virginia, USA, NASA Conerence Publication 318, pp [10] T. Kiuchi and T. Kondo, 1996, The Waveront Clock Technique Applied to Current VLBI System, Publications o Astronomical Society o Japan, 48,

20 [11] T. Iwata, F. Tappero, M. Imae, Y. Fukuyama, T. Suzuyama, H. Murakami, Y. Ikemoto, A. Iwasaki, and T. Masubuchi, 005, Ground Experiments o Remote Synchronization System or Onboard Crystal Oscillators o Quasi-Zenith Satellites, in Proceedings o AIAA Guidance, Navigation, and Control Conerence and Exhibit, August 005, San Francisco, Caliornia, USA, Paper No. AIAA [1] Japan Aerospace Exploration Agency, 007, Quasi-Zenith Satellite System Navigation Service Interace Speciication or QZSS (IS-QZSS), Ver. 1.0.