7/27/2019 J-Tiempo 2
http://slidepdf.com/reader/full/j-tiempo-2 1/4
PRECISE TIME TRANSFER USING GPS CARRIER PHASE-BASED TECHNIQUES
Jan Johansson and Kenneth Jaldehag
SP Swedish National Testing and Research Institute, Box 857, S-501 15, Borås, Sweden
ABSTRACT
We have used the carrier phase signal from the
satellites in the Global Positioning System (GPS) in an
attempt to perform precise time and frequency transfer.
The method is commonly used in the geodetic
community to obtain mm-level precision over global
distances. When estimating the station position other
parameters must be solved for. Such parameters include,
e.g., the atmospheric propagation delay as well as
determination of clocks in the GPS-receivers. In this
study we have focused on the evaluation of these clock
solutions for two Swedish GPS stations with particular
interest for the timing community. One station is
collocated with the Swedish national laboratory for time
and frequency and equipped with several cesium
frequency standards. The other station is located at the
Onsala Space Observatory where all receivers are
connected to a hydrogen maser.
1. INTRODUCTION
The GPS carrier phase data have been in extensive use
in the geodetic and geophysical community for more
than a decade. It is possible to reach mm-level precision
even over global distances using GPS carrier phase- based relative positioning. Simultaneously, several ”by-
products” have come available because of the modeling
and the estimation technique used. For example, it is
necessary to calculate or estimate the signal propagation
path delay in the atmosphere in order to obtain accurate
coordinates of a station. Products such as the amount of
water vapor in the troposphere and the total electron
content in the ionosphere are standard deliverables from
several GPS data analysis centers.
Another parameter, which has to be solved for in the
estimation process, is the receiver clock. If the clock of
the GPS receiver is based on the signal from an atomic
frequency standard the parameter may be preciselydetermined. We might actually be able to study the
relative performance of an external frequency standard
connected to a high-quality GPS receiver.
Recent developments in atomic frequency standards
such as the cesium fountain and frequency standards
based on linear ion traps calls for new methods in time
and frequency transfer techniques. GPS carrier phase-
based time transfer is considered to have the potential
required [1]. In order to take full advantage of this
technique, the delays of various parts of the receiving
equipment must be stabilized and measured (includes
e.g., cables, receiver, and antennas). Especially, serioushardware delay instabilities may result from temperature
variations in the vicinity of the receiver system. For timelaboratories it is essential to minimize this effect.
2. GPS NETWORKS AND PRODUCTS
2.1 The Swedish permanent GPS network.
The Swedish permanent GPS network SWEPOS® has
been in continuos operation since August 1993. The
network consists of 21 stations (see figure 1). The
average station separation is 200 km covering a region
from latitude 55° to 69° north [2]. The communication
between the SWEPOS® station and the network control
center, hosted at the National Land Survey of Sweden, is
managed via 64 kbit/s TCP/IP lines. With this real time
data flow, SWEPOS® is a multipurpose network with
applications stretching from navigation to geophysics.
Figure 1: The Swedish network of continuously
operational GPS stations, SWEPOS®.
Data from the permanent GPS stations are analyzed
daily within several different projects such as studies of
present-day glacial isostatic adjustment and sea-level
rise, monitoring of water vapor in Earth’ troposphere,
and for addressing reference frame issues
(IGS/EUREF). In order accomplish this, the design of
each site and the choice of hardware are importantissues [2].
7/27/2019 J-Tiempo 2
http://slidepdf.com/reader/full/j-tiempo-2 2/4
2.2 The International GPS Service (IGS).
The IGS was founded to improve geodetic
applications of GPS by providing an infrastructure for a
global network of high quality and continuously
operational GPS stations, data collection and archiving,
and by making available products based on post- processing of GPS data. The basic products include
precise determination of satellite orbit parameters and
clocks. In high-precision carrier phase-based relative
positioning we need to obtain accurate information
about the satellite orbital parameters and the satellite
clocks. This can be obtained from the IGS with a lag-
time of 24 hours up to 11 days depending on the level of
accuracy needed. One may also determine these
parameters simultaneously with other parameters but
this requires a fairly big network of GPS tracking
stations.
3. EXPERIMENTAL SETUP
The two SWEPOS® stations, Borås and Onsala, are of
special interest to the time keeping community since
they contribute to the realization of the atomic time
scale and to the global reference frame. Data from
several receivers hosted at these two stations have been
used in order to investigate different aspects of precise
frequency transfer. In this section we will briefly
describe the hardware at each site and the strategy used
for GPS data processing.
3.1 The GPS station at SP in Borås
The station in Borås, located at the SP Swedish
National Testing and Research Institute, is a member of
SWEPOS® and collocated with the national laboratory
for time and frequency. Three cesium frequency
standards are available at SP and one is used for the
realization of UTC(SP). Data from two HP5071A high-
performance and one Oscilloquartz 3200 cesium
standards are contributing to TAI. The entire time and
frequency laboratory is hosted in a temperature-
controlled environment where the temperature is
continuously monitored and known to be 24 ± 0.5 °C.
The GPS station in Borås is hosted in the samelaboratory and all the receivers are consequently in the
same temperature-controlled environment. Similarly as
all SWEPOS® sites the Borås station is equipped with
two Ashtech Z12 receivers both connected to same
antenna through a power splitting device. These two
receivers are both hooked up to TCP/IP and delivers
real-time data to the SWEPOS® operational center. In
addition, two TurboRogue SNR-8000 geodetic receivers
and three timing receivers also receive GPS signals from
the same GPS antenna. The four ”geodetic” GPS-
receivers in Borås all utilizes external 5 MHz from the
cesium frequency standard from which UTC(SP) is
based.
Figure 2: The SWEPOS® station at SP in Borås.
The GPS station in Borås is unique in the sense that,
in addition to the GPS receivers, also the antenna cable
is in a temperature controlled environment. The antennacable is placed in a water pipe where the temperature of
the water is kept at 7 ± 1 °C. Thus, we do not expect any
significant fluctuation in the electrical delay of the cable
due to temperature variations. The three-meter high
concrete pillar, on top of which the GPS antenna is
mounted, is shown in figure 2. The entire pillar is
temperature-controlled by means of electrical heating
and cooling water.
As at all other SWEPOS® sites the Borås station has
a Dorne-Margolin T choke-ring antenna. A hemispheric
radome is used to protect the antenna. The GPS antenna
is not included in the temperature-controlled
environment and may contribute to the error budget. Atemperature sensor was therefore installed next to the
GPS antenna in order to model this effect.
3.2 The IGS station at Onsala Space Observatory
The Onsala station, located at the radio astronomy
facility Onsala Space Observatory, is a long-standing
member of the global GPS-satellite tracking network,
IGS and also a member of SWEPOS®. The station is
equipped with two TurboRogue SNR-8000, three
Ashtech Z12 GPS receiver, and one Ashtech Z18,
tracking both GPS and GLONASS satellites. All the
receivers are connected to the same Dorne-Margolin B
GPS antenna via a power splitting device. The antenna
is mounted on top a one meter concrete pillar and
covered by a hemispheric radome (see figure 3).
The Onsala GPS-receivers operates using an external
hydrogen-maser frequency standard which is equivalent
to the frequency standard used in the radio-astronomy
and geodetic observations with the Very Long Baseline
Interferometry (VLBI) technique. The GPS antenna,
antenna cable, and GPS receivers are not in a
temperature controlled environment. However, the
temperature are carefully logged every half hour near
the antenna, the antenna cable, and inside the cabinhosting the GPS-receivers.
7/27/2019 J-Tiempo 2
http://slidepdf.com/reader/full/j-tiempo-2 3/4
Figure 3: The GPS antenna and pillar system at the
Onsala station in front of the radome-enclosed 20 meter
radio telescope used for e.g., in geodesy VLBI.
3.3 The GPS data analysis
The analysis of GPS data basically adapted
traditional methods used in space geodesy and remote
sensing applications. Even though data from about 50
stations have been processed, this study includes only
the results obtained from the TurboRogue receivers at
Onsala and Borås. The entire data set was analyzed
using the GIPSY software [3], based on a Kalman filter
approach. Station coordinates were constrained to their a
priori value within a mm. The signal propagation path
delay in the troposphere is modeled as a random-walk
process with stochastic updates of one zenith delay parameter and two gradient parameters per stations
every five minutes. We used dual-frequency
measurements, which eliminate most of the delay caused
by the ionosphere. The receiver clock parameters were
updated every five minutes using a white-noise process
with very loose (1 s) a priori sigma. We have used the
most accurate products produced by the IGS community
as described above. The satellite orbits from the IGS
were held fixed. Finally, we should mention the
inclusion of accurate modeling of Earth tides and ocean
loading.
4. RESULTS
4.1 Zero baseline test
In order to eliminate all error sources and
demonstrate the capability of the method, we have used
a zero baseline test. Two GPS receivers were connected
to the same antenna and also utilizing external 5 MHz
from the same cesium frequency standard. Basically, all
other error sources were eliminated except those
associated with instrumental biases within the two GPS
receiver themselves, data registration, and data analysis
methods. Under these ideal conditions the GPS carrier phase-based technique are capable of frequency transfer
at the level of a few parts in 1016
after less than 12 h
averaging time. Figure 4 shows the Allan deviation
based on the zero-baseline test over 3 days. Apparently,
the slope of the Allan deviation is close to –1 and may
be restricted by the measurement phase noise.
Figure 4: The Allan deviation as a function of
averaging time between two receivers sharing the same
antenna, antenna cable, and external cesium standard.
4.2 Zero-baseline with different frequency standards
In order to test the capability of frequency transfer
the two TurboRogue receivers in Borås were connected
to different cesium standards (see fig. 5). However, both
receivers were sharing the same GPS antenna. Using a
Time Interval Counter (TIC) the 1 PPS from each GPS
receiver and cesium standard can be monitored. Thus,
we may compare the difference between the cesium 1
and the cesium 4 obtained from the readings from theTIC and from the GPS data analysis, respectively.
Cesium1
Cesium4
FDA
Switch
TR2
TR1
Digitalclock
µ-phasestepper
TIC
5 MHz
CS1
5 MHz CS4
1-pps
1-pps
5 MHz 5 MHz
5 MHz
1-pps
1-pps
UTC(SP)
Figure 5: The setup used in the zero-baseline frequency
transfer between cesium 1 and 4. The dashed ”5 MHz-
line” corresponds to the setup used in section 4.1.
The difference between these two methods over a
period of almost three weeks in March 1999 is plotted in
figure 6. The GPS solutions were obtained daily and
based on daily IGS orbits. Jumps of up to 150 ps in the
time series are clearly evident from day-to-day and
related to the GPS solutions. The entire period have a
standard deviation of 120 ps. However, if we just look
over a one-day periods (one GPS solution) we obtain 60
ps. The IGS orbits have an uncertainty of about 5 cm.
Since a jump of 100 ps is roughly equivalent to 3 cm,
orbit errors are significant. The corresponding Allan
deviation is shown in fig. 7. We have also included a
line describing white frequency noise reflecting thecesium standards.
7/27/2019 J-Tiempo 2
http://slidepdf.com/reader/full/j-tiempo-2 4/4
4.3 A 60 km baseline with different frequency standards
Using the same time period as above, frequency-transfer
between the stations in Borås and Onsala have been
performed. The two stations are separated by
approximately 60 km. In figure 7 the time difference
between the two receivers is shown. The frequencyoffset has been removed before plotting. In the same
figure we have included both the indoor and outdoor
temperature at the Onsala GPS station. A correlation
between, especially, the temperature in the GPS cabin
and the time difference is evident. Similar hardware
temperature dependence has been detected by other
investigations.
In figure 8, the Allan deviation is plotted. The results
closely follow the –0.5 slope and one may conclude that
this is the performance level of the cesium frequency
standard. Variations, most likely caused by changes in
the temperature at Onsala, are visible in fig. 8.
-0,5
-0,4
-0,3
-0,2
-0,1
0
0,1
0,2
0,3
0,4
0,5
69 71 73 75 77 79 81 83 85 87
Day of Year 1999
T I C - G P S [ n s ]
Figure 6: The difference between the time intervalcounter and the GPS estimates.
Figure 6: The Allan deviation plotted as a function of
averaging time for a zero-baseline experiment where the
two GPS receivers were setup according to fig. 5.
5. DISCUSSION AND CONCLUSIONS
The capability of using GPS carrier phase-based
techniques in time and frequency transfer applications
has been demonstrated. In a zero-baseline test we have
obtained an Allan deviation of about 4 x 10-16
when
averaged over less than 12 hours. Furthermore, a GPS
zero-baseline was used to compare two cesium
frequency standards hosted at the same site and
simultaneously we also compared these two cesium
standards with a hydrogen maser over a 60 km baseline.
The Allan deviation plots show that the performance
level of cesium frequency standards is reached.
However, there is a correlation between temperature
variation in the vicinity of the GPS hardware at Onsalaand the results obtained in the frequency comparison.
The Onsala IGS stations will now be upgraded. The
receivers, the antenna cable, and the antenna will be
temperature controlled similar to the Borås site. Finally,
we intend for Borås to become an IGS station within the
framework of the new IGS/BIPM pilot project.
-6
-4
-2
0
2
4
6
8
10
78 79 80 81 82 83 84 85 86 87Day of Year 1999
B o r å s - O
n s a l a
[ n s ]
-40
-30
-20
-10
0
10
20
30
T e m p e r
a t u r e [ C ]
GPS
Onsala outdoor temperature
Onsala indoor temperature
Figure 7: The time difference between Borås and
Onsala GPS receivers over 9 days (lower curve). Also
plotted are the outdoor (mid curve) and the indoor (top
curve) temperature at the Onsala station.
Figure 8: Frequency stability obtained from a 60 km
baseline between Onsala and Borås.
6. REFERENCES
[1] G. Petit, C. Thomas, Z. Jiang, P. Uhrich, and F. Taris. Use
of GPS Ashtech Z12T Receivers for Accurate Time and
Frequency Comparisons. IEEE International Frequency
Control Symposium, Pasadena, Ca, May 27-29, 1998.
[2] H.-G. Scherneck, J.M.Johansson, J.X.Mitrovica, and
J.L.Davis. The BIFROST Project: GPS determined 3-D
displacement rates in Fennoscandia from 800 days of
continuous observations in the SWEPOS network.
Tectonophysics, 294, pp. 305-321, 1998.
[3] F.H.Webb and J.F. Zumberge. An Introduction to
GIPSY/OASIS-II. Jet Propulsion Laboratory, 1993.
Top Related