Intercomparison Study of Time and Frequency Transfer between VLBI and Other Techniques (GPS, ETS8(TCE), TW(DPN) and DMTD)
Hiroshi Takiguchi, Moritaka Kimura, Tetsuro Kondo, Atsutoshi Ishii, Hobiger Thomas, Ryuichi Ichikawa, Yasuhiro Koyama, Yasuhiro Takahashi, Fumimaru Nakagawa, Maho Nakamura, Ryo Tabuchi, Shigeru Tsutshiya, Shinichi Hama, Tadahiro Gotoh, Miho Fujieda, Masanori Aida, Tingyu Li, Jun Amagai
aa r X i v : . [ a s t r o - ph . I M ] D ec Intercomparison Study of Timeand Frequency Transfer be-tween VLBI and Other Tech-niques (GPS, ETS8(TCE), TW(DPN) andDMTD)
Hiroshi Takiguchi ( hiroshi.takiguchi @ aut.ac.nz ),Moritaka Kimura , Tetsuro Kondo ,Atsutoshi Ishii , Hobiger Thomas , RyuichiIchikawa , Yasuhiro Koyama , YasuhiroTakahashi , Fumimaru Nakagawa , MahoNakamura , Ryo Tabuchi , ShigeruTsutshiya , Shinichi Hama , TadahiroGotoh , Miho Fujieda , Masanori Aida ,Tingyu Li , and Jun Amagai Institute for Radio Astronomy and SpaceResearch, Auckland University of Technology Space-Time Standards Laboratory, NationalInstitute of Information and CommunicationsTechnology Advance Engineering Services Co., LtdAbstract : We carried out the intercomparison ex-periments between VLBI and other techniques toshow the capability of VLBI time and frequencytransfer by using the current geodetic VLBI tech-nique and facilities as the summary of the exper-iments that we carried out since 2007. The re-sults from the two different types of experimentsshow that the VLBI is more stable than GPS butis slightly noisier than two new two-way techniques(TW(DPN), ETS8(TCE)), and VLBI can measurethe correct time difference as same as ETS8(TCE).
1. Introduction
As one of the new time and frequency trans-fer (hereafter T&F transfer) technique to comparethe next highly stable frequency standards, we pro-posed the geodetic VLBI technique [1]. Since 2007,to evaluate the capability of geodetic VLBI for pre-cise T&F transfer, we carried out intercomparisonexperiments between VLBI and GPS Carrier Phase(hereafter GPS) on the Kashima 11m and Koganei11m baseline several times. These intercompar-isons showed that the geodetic VLBI technique hasthe potential for precise frequency transfer [2], [3].Also, these results showed that the geodetic VLBIcan measure the correct time difference [4].Space-Time Standards Group of National Insti-tute of Information and Communications Technol- ogy (NICT) which we belong to, is conducting re-search and developments for precise T&F transfertechniques other than VLBI such as using GPS andtwo-way satellite time and frequency transfer (TW-STFT) at NICT Koganei Headquaters. In 2010,we carried out the intercomparison experiment be-tween VLBI and other techniques to show the capa-bility of VLBI time and frequency transfer by usingthe current geodetic VLBI technique and facilitiesas the summary of the experiments.In this paper, we describe the two intercompar-ison experiments from a viewpoint of the VLBImainly. Therefore, we leave the details of the resultof other techniques to different papers.
2. Two new TWSTFT techniques devel-oped by NICT
NICT developed the two new TWSTFT tech-niques. One is the method using a pair ofPseudo Random Noises (dual PRN, DPN) (here-after TW(DPN)). The other one is the methodusing Time Comparison Equipment (TCE) on theEngineering Test Satellite VIII (ETS8) (hereafterETS8(TCE)). To carry out the intercomparisonexperiment, we installed the TW(DPN) antennaand the ETS8(TCE) ground station at KashimaSpace Technology Center (KSTC, former KashimaSpace Research Center) next to the VLBI antenna(Kashima 11m). In this section, we describe thebrief overview of two techniques. Please see thereference papers for more details.
TW(DPN) was developed to improve the mea-surement precision and decrease operational costof TWSTFT. The precision can be improved byincreasing the chip rate of the PRN. However, thismethod of enhancing the precision is not feasiblebecause the rental costs of the commercial commu-nication satellites used for signal transfer are high.TW(DPN) is composed of a waveform genera-tor and an A/D converter. By using this method,we can improve the delay measurement precisionby one order of magnitude, even though the occu-pied bandwidth is only 400kHz, which is less thanone-sixth the currently used bandwidth. Since thetransponder cost is proportional to the occupiedbandwidth, we can reduce the operational cost ofthe TWSTFT [5].
ETS8 is a Japanese Geostationary Satellite,which launched in 2006. ETS8 has missions for mo-bile communication experiments and for precision1iming experiments using Cesium atomic clocks inspace.At the time of T&F transfer, TCE transmit andreceive signals to and from the ground. As thetwo-way uplink and downlink transmission path-ways are approximately equivalent, the effects oftransmission delay in the atmosphere or those dueto the motion of the satellite will be cancelled out,enabling highly precise time transfer, with antic-ipated precision on the order of several nanosec-onds in code-phase operation and approximatelyless than 100 picoseconds in carrier-phase opera-tion [6], [7], [8].
3. Intercomparison experiments3.1 Outline of the experiments
Figure 1 is the layout map of KSTC and KoganeiHeadquaters. The baseline length of Kashima 11m- Koganei 11m is about 109 km. In 2010, we car-ried out intercomparison experiments two times(August and October). At August experiment, toevaluate long term stability of these techniques,we acquired the over 100 hours data. At Oc-tober experiment, we compared the precision ofthese techniques by stretching the Coaxial PhaseShifter (hereafter trombone) which was insertedin the path of the reference signal from Hydro-gen maser to the Kashima 11m antenna [4]. Thereference signal which is provided from hydrogenmaser is transmitted by coaxial cable (the dis-tance is about 300m) in KSTC. In 2009, we in-stalled the RF distribution system using opticalfibers at Koganei Headquaters to transmit the ref-erence signal to VLBI back end which is coherentwith UTC(NICT). To cancel the length fluctuationof optical fibers, we adopted the feedback systemusing the round-trip signal. Hence, this transferstability reached the 10 − level over 1000 seconds.Therefore, the reference signal (10MHz/1PPS) atKoganei station is coherent with UTC(NICT) [9]. NICT are currently developing the two typesof sampling system named as K5/VSSP32 (here-after VSSP) and K5/VSI such as ADS1000 andADS3000+ (hereafter VSI) [10]. Also, we are devel-oping the software correlator and data conversionutilities [13] corresponding to each system. VSSPand K5 software correlator are one of the sets, andthat is mainly used for the geodetic VLBI experi-ments in Japanese stations [11]. VSI and GICO3software correlator are another sets. That is mainlyused for astronomical purpose. The processingspeed of the GICO3 worthy of special mention isabout 10 times faster than that of the DiFX at 2k FFT points [12]. In order to use VSI system in thegeodetic VLBI experiment, we developed the dataconversion programs and carried out the experi-ments. Figure 2 show the flowchart that indicatefrom K5 sampling system to the baseline analysissoftware. The gray background indicate the newprograms which were developed in this time.The setting parameters of both experiments areshown in Table 1. At first, we supposed VSSPwas main. Therefore, the effective bandwidth (X-band) of VSI was narrower than VSSP in spiteof a wideband sampler. In addition, because theschedule was optimized for VSSP, the scan lengthwas longer, and the number of observation was less,than the schedule that was optimized for VSI. Asthe result, the estimated delay precision of VSI was70% with reference to VSSP.Table 2 shows the baseline length calculated fromthe two types of K5 system. In the two time ex-periments, these results show the good agreement.Thus it is concluded that using the K5/VSI andGICO3 software correlator for the geodetic VLBIexperiment is not a problem. However, the resultsof the clock offsets from VSSP have daily variationswhich were influenced from the problem of phasecalibration system. Therefore, the results of VSSPshown afterward were corrected using VSI data.
To evaluate long term stability of these tech-niques, we acquired the over 100 hours data at Au-gust experiment. Figure 3 show the time series oftime difference calculated from these techniques atKashima-Koganei baseline. The common trend ofthese time series was already removed up to 2ndorder. Table 3 show the data property (integrationtime, etc) of each techniques. Also, Table 4 showthe root-mean-square of time series variation cal-culated from with reference to ETS8. The resultof ETS8(TCE) is extremely stable than other tech-niques. TW(DPN) is also stable, but it has cleardaily variation. The cause of the daily variationdoes not yet clear, but we think that it is causedby interference from spread signal and/or sunlight.The results of the VLBI agree with GPS, butthese results vary than other results. Figure 4show the difference of the atmospheric delay cal-culated from VLBI and GPS between Kashimaand Koganei. Time delay and atmospheric delayvariations agree well. As already described, theinfluence of atmospheric delay was removed fromTW(DPN) and ETS8(TCE) because both tech-niques are TWSTFT. Usually in the analysis ofVLBI and GPS, the atmospheric delay and timedelay are estimated at the same time. These re-sults suggested the estimation of atmospheric de-2ay in the analysis of VLBI and GPS is not enough.To obtain the more precise result, it is necessary touse the more precise model and/or another methodsuch as KARAT [14]. & F transfer precision
At October experiment, we compared the preci-sion of these techniques by stretching the trombonewhich was inserted in the path of the reference sig-nal from Hydrogen maser to the Kashima 11m an-tenna. Figure 5 show the reference signal setupdiagram at Kashima station. This experiment isalmost same strategy in the case of [4]. In thistime, we stretched the trombone more slowly andmore constantly. In addition, we expanded scantime of VLBI according to the time of stretchingtrombone. Also, as the reference of correct changeof time difference, we introduced the new DMTDequipment (TSC511A, Phase Noise and Allan De-viation Test Set).Figure 6 show the time difference of each tech-niques. The large steps (A to E) were artificialdelay change parts by trombone. Black lines areDMTD. Gray thin lines are VLBI (VSSP and VSI)and ETS8(TCE). Variety lines are GPS. Also, weshow the summary of the amount of the steps ob-tained from these techniques at the artificial delaychange parts in Table 5. These results show thateach technique agree very well. The differences ofeach technique except GPS are only a few picosec-onds on the average. Anyway, the result of ourexperiment clearly show that the geodetic VLBItechnique can measure the correct time differenceas same as ETS8(TCE) and DMTD.
4. Summary and Outlook
We carried out the intercomparison experimentsbetween VLBI and other techniques (GPS CP,TW(DPN), ETS8(TCE), and DMTD) to show thecapability of VLBI time and frequency transfer byusing the current geodetic VLBI technique and fa-cilities as the summary of the experiments that wecarried out since 2007.The results from the August experiments showthat the VLBI is more stable than GPS but isslightly noisier than two new two-way techniques(TW(DPN), ETS8(TCE)). Also, these results showthat the estimation of atmospheric delay in theanalysis of VLBI and GPS is not enough.At October experiment, we produced artificialdelay changes by stretching the trombone whichwas inserted in the path of the reference signalfrom Hydrogen maser to Kashima 11m antenna.At the artificial changes, the results of VLBI, ETS8and DMTD hardly had a difference. Consequently, the geodetic VLBI technique can measure the cor-rect time difference as same as ETS8(TCE) andDMTD.Currently the T&F transfer experiment usingETS8 in NICT was finished. And the projectshifted to the next phase of the R&D of T&F trans-fer using Quasi-Zenith Satellite System (QZSS).Also, the VLBI project shifted to the next phasewhich is the R&D of the new facilities and strategysuitable for T&F transfer. In the near future, weare planning to carry out the following list. • apply KARAT for the VLBI and GPS analysis • using MARBLE [16] and ADS3000+ [15] forT&F transfer • international experiment MK3TOOLsCalc/Solve K5coutkombkomb K5/VSSP32GICO3K5/VSI gco gico2komb cout softwarecorrelator
Figure 2. The flowchart from the two K5 samplingsystems to the baseline analysis software.Figure 5. The reference signal setup diagram atKashima station.Acknowledgments : The authors would like to ac-knowledge the IVS and the IGS for the high qual-ity products. We are grateful that GSFC and3 oganei 11m
UTC(NICT)UTC(NICT)GPSGPS ETS8ETS8 TWTW
Koganei /Tokyo
Kashima
Kashima Space Research CenterHeadquarters
Kashima 11m Kashima34m
Optical fiber linkOptical fiber link
Figure 1. The layout map of KSTC and Koganei Headquaters.Table 1. The setting parameters of experiments
K5/VSSP32 K5/VSI (ADS1000)Band S/X S/XInput Freq. Width 16MHz/ch 512MHz/chSampling Rate 32Mbps, 1bit 1024Mbps/ch, 1bitNumber of Channels 16ch 2chEffective Bandwidth of X-band 364.8MHz 147.8MHz
Table 3. The data property of each techniques
VLBIVSSP VSI GPS CP TW(DPN) ETS8(TCE)Integration Time Scan Length 1 120 1Analysis Software K5 + Calc/Solve GICO3 + Calc/Solve NRCan PPP developed by NICTData Interval Slew time + (Sum of two scan /2) 30 120 1Average 100 Unit: second4 p s K5/VSSP32K5/VSIGPS CPTW(DPN)ETS8(TCE)
Time Difference Kashima11−Koganei11DOY
Figure 3. Time difference obtained from these techniques. -400-300-200-100 0 100 200 217.5 218 218.5 219 219.5 220 p s DOYAtmospheric Delay VLBIGPS
Figure 4. The difference of the atmospheric delay calculated from VLBI and GPS between KSTC andKoganei Headquarters
NR Canada provided the VLBI and GPS analysissoftware (CALC/SOLVE and NRCan’s PPP). TheVLBI experiments were supported by M. Sekidoand E. Kawai of the Kashima Space TechnologyCenter.
References [1] Takiguchi, H., T. Hobiger, A. Ishii, R. Ichikawa, and Y. Koyama,Comparison with GPS Time Transfer and VLBI Time Transfer,
IVS NICT-TDC News , No.28, 10-15, 2007.[2] Takiguchi, H., Y. Koyama, R. Ichikawa, T. Gotoh, A. Ishii, T.Hobiger and M. Hosokawa, Comparison Study of VLBI and GPS Carrier Phase Frequency Transfer using IVS and IGS data,
IVSNICT-TDC News , No.29, 23-27, 2008.[3] Takiguchi, H., Y. Koyama, R. Ichikawa, T. Gotoh, A. Ishii, andT. Hobiger, Comparison Study of VLBI and GPS Carrier PhaseFrequency Transfer - Part II -,
IVS NICT-TDC News , No.30,26-29, 2009.[4] Takiguchi, H., Y. Koyama, R. Ichikawa, T. Gotoh, A. Ishii, T.Hobiger, and M. Hosokawa, VLBI Measurements for FrequencyTransfer,
IVS NICT-TDC News , No. 31, 21-24, 2010.[5] Amagai, J., and T. Gotoh, Current Status of Two-way SatelliteTime and Frequency Transfer Using a Pair of Pseudo RandomNoises,
Proc. ATF 2008 , 2008.[6] Takahashi, Y., M. Imae, T. Gotoh, F. Nakagawa, H. Kiuchi, M.Hosokawa, M. Aida, Y. Takahashi, H. Noda, and S. Hama, Time igure 6. Time difference of October experiment. The large steps (A to E) were artificial delay changeparts by trombone. Black lines are DMTD. Gray thin lines are VLBI (VSSP and VSI) and ETS8(TCE).Variety lines are GPS.Table 2. The baseline length calculated fromthe data that was sampled with each system ofK5/VSSP32 and K5/VSI. Exp. date System Baseline Length 1 σ August VSSP 109099639.00 0.53VSI 109099639.00 0.57October VSSP 109099635.43 0.58VSI 109099635.58 0.66Unit: mm
Table 4. The root-mean-square of time series vari-ation calculated from with reference to ETS8.
August OctoberTW(DPN) 95 (20) -GPS CP 75 56VLBI(VSI) 60 36Unit: ps
Table 5. The amount of the steps obtained fromthese techniques at the artificial delay change parts.
A B C D E AverageDMTD 347 346 346 347 348 347GPS CP 352 340 385 353 345 355ETS8 348 340 343 349 347 345VSSP - 349 342 350 340 345VSI 347 340 - 351 348 346Unit: ps
Comparison Equipment for ETS-VIII Satellite -Part 1 Develop-ment of Flight Model-,
J. NICT , Vol.50, Nos.1/2, 4-4, 135-143,2003.[7] Fujieda, M., Y. Takahashi, T. Gotoh, F. Nakagawa, and M.Imae, Time Comparison Experiment, Earth Station of TimeComparison Experiment,
J. NICT , Vol.50, Nos.3/4, 4-8-1, 243-249, 2003.[8] Nakagawa, F., M. Imae, Y. Takahashi, H. Kiuchi, T. Gotoh,and M. Fujieda, Time Comparison Experiment, Time Compar-ison Equipment for ETS-VIII -Data Processing, Analysis andCapability of Time Comparison-,
J. NICT , Vol.50, Nos.3/4, 4-8-2, 251-259, 2003.[9] Fujieda, M., M. Kumagai, S. Nagano, and T. Gotoh,UTC(NICT) signal transfer system using optical fibers,
IVSNICT-TDC News , No. 31, 17-20, 2010.[10] Koyama, Y., T. Kondo, M. Sekido, and M. Kimura, Develop-ments of K5/VSI System for Geodetic VLBI Observations,
IVSNICT-TDC News , No. 29, 15-20, 2008.[11] Kondo, T., Y. Koyama, R. Ichikawa, M. Sekido, E. Kawai, andM. Kimura, Development of the K5/VSSP System,
J. GeodeticSoc. Japan , 54(4), 233-248, 2008.[12] Kimura, M., J. Nakajima, T. Kondo, Y. Koyama, M. Sekido, andT. Oyama, High Speed VLBI Software Correlator (now prepar-ing) .[13] Hobiger, T., R. Ichikawa, and Y. Koyama, MK3TOOLS - seam-less interfaces for the creation of VLBI databases from post-correlation output,
Proc. the Fifth IVS General Meeting , 153156,2008.[14] Hobiger, T., R. Ichikawa, Y. Koyama, and T. Kondo, Kashimaray-tracing service (KARATS) online provision of total tropo-sphere slant delay corrections for east asian sites,
Proc. Interna-tional Symposium on GPS/GNSS 2008 , 1, 4044, 2008.[15] Takefuji, K., M. Tsutsumi, H. Takeuchi, and Y. Koyama, Cur-rent Status of Next Generation A/D Sampler ADS3000+,
IVSNICT-TDC News , No. 31, 6-9, 2010.[16] Ishii, A., R. Ichikawa, H. Takiguchi, K. Takefuji, H. Ujihara, Y.Koyama, T. Kondo, S. Kurihara, Y. Miura, S. Matsuzaka andD. Tanimoto, Current status of development of a transportableand compact VLBI system by NICT and GSI,
IVS NICT-TDCNews , No. 31, 2-5, 2010., No. 31, 2-5, 2010.