Conference PaperPDF Available

Development of a network real-time kinematic processing platform

Authors:

Abstract and Figures

Single baseline "real-time kinematic" (RTK) GNSS positioning is a carrier-phase-based relative positioning technique that delivers centimetre-level accuracy in real-time. In general, this technique satisfies the GPS receiver manufacturers' accuracy specification (e.g. 5-10mm +/-1ppm) for baseline lengths of up to approximately 20km due to the distance dependent errors. The Network-RTK (NRTK) concept was introduced to overcome the limitation of the baseline distance while improving the positioning accuracy and repeatability. Rapid growth and development of information and communication technologies has enabled GPS service operators to broadcast the network correction via the Internet. In accordance with this trend, the development of a research-oriented real-time data processing platform for NRTK positioning was initiated by the School of Surveying and Spatial Information Systems at the University of New South Wales (UNSW), Sydney, Australia. This platform is being used to investigate different algorithms as well as issues such as network latency, data synchronisation, positioning quality, and others. This NRTK system is known as 'SNAPper'. SNAPper receives GPS data streams from Continuously Operating Reference Stations (CORS) and generates real-time network corrections using International GNSS Services (IGS) ultra-rapid orbits, and provides users with Virtual Reference Station (VRS) measurements. A robust ambiguity resolution algorithm was implemented for high precision positioning using corrections computed from a CORS network. In order to evaluate the performance of the SNAPper software, a range of numeral tests have been carried out. The Vicmap Position-GPSnet™, operated by the Victorian Department of Sustainability and Environment, Melbourne, was chosen as a test network, however, actual data processing was conducted at UNSW to demonstrate the capability of remote processing. The test results show that the implemented RTK algorithm can deliver precise positioning with centimetre-level accuracy in real-time.
Content may be subject to copyright.
Development of a Network Real-Time Kinematic
Processing Platform
Heo, Y., Li, B., Lim, S., Rizos, C., The School of Surveying and Spatial Information Systems, UNSW, Australia
BIOGRAPHIES
Yong Heo is a research assistant in the School of
Surveying & Spatial Information Systems, the University
of New South Wales (UNSW), Sydney, Australia. Yong
received his B.Sc. from the School of Computer Science
and Engineering, UNSW, in 2007.
Binghao Li is a research associate in the School of
Surveying & Spatial Information Systems, UNSW.
Binghao obtained B.Sc. in Electrical & Mechanical Eng.
from Northern Jiaotong University, P.R. China, in 1994,
and M.Sc. in Civil Eng., Tsinghua University, P.R. China,
in 2001. He received his Ph.D. from UNSW in 2006. His
research area is pedestrian navigation and network-RTK
algorithm development.
Samsung Lim is a senior lecturer in the School of
Surveying & Spatial Information Systems, UNSW. For
the past fifteen years his research has focused on the areas
of GNSS and GIS. Samsung's research interests are in
theoretical problems related to RTK-GPS and applying
geospatial information technologies to real-world
problems. Samsung received his B.A. and M.A. in
Mathematics from Seoul National University, South
Korea, and his Ph.D. in Aerospace Engineering and
Engineering Mechanics from the University of Texas at
Austin, USA.
Chris Rizos is currently Professor and Head of the
School of Surveying & Spatial Information Systems,
UNSW. Chris has been researching the technology and
high precision applications of GPS since 1985, and has
published over 350 journal and conference papers. He is a
Fellow of the Australian Institute of Navigation and a
Fellow of the International Association of Geodesy
(IAG). He is currently the Vice President of the IAG and
a member of the Governing Board of the International
GNSS Service.
ABSTRACT
Single baseline “real-time kinematic” (RTK) GNSS
positioning is a carrier-phase-based relative positioning
technique that delivers centimetre-level accuracy in real-
time. In general, this technique satisfies the GPS receiver
manufacturers‟ accuracy specification (e.g. 5-10mm +/-
1ppm) for baseline lengths of up to approximately 20km
due to the distance dependent errors. The Network-RTK
(NRTK) concept was introduced to overcome the
limitation of the baseline distance while improving the
positioning accuracy and repeatability.
Rapid growth and development of information and
communication technologies has enabled GPS service
operators to broadcast the network correction via the
Internet. In accordance with this trend, the development
of a research-oriented real-time data processing platform
for NRTK positioning was initiated by the School of
Surveying and Spatial Information Systems at the
University of New South Wales (UNSW), Sydney,
Australia. This platform is being used to investigate
different algorithms as well as issues such as network
latency, data synchronisation, positioning quality, and
others. This NRTK system is known as SNAPper.
SNAPper receives GPS data streams from Continuously
Operating Reference Stations (CORS) and generates real-
time network corrections using International GNSS
Services (IGS) ultra-rapid orbits, and provides users with
Virtual Reference Station (VRS) measurements. A robust
ambiguity resolution algorithm was implemented for high
precision positioning using corrections computed from a
CORS network.
In order to evaluate the performance of the SNAPper
software, a range of numeral tests have been carried out.
The Vicmap Position-GPSnet™, operated by the
Victorian Department of Sustainability and Environment,
Melbourne, was chosen as a test network, however, actual
data processing was conducted at UNSW to demonstrate
the capability of remote processing. The test results show
that the implemented RTK algorithm can deliver precise
positioning with centimetre-level accuracy in real-time.
1. INTRODUCTION
Real-Time Kinematic (RTK) positioning is a carrier-
phase-based GNSS relative positioning technique for
delivering centimetre-level accuracy in real-time. Since
RTK was introduced in the early 1990s it has been widely
used for high accuracy applications such as surveying,
mapping and machine guidance. A relative positioning
technique requires data communication links between a
reference receiver and a rover receiver. Traditionally,
radio frequency (RF) data communication channels such
as MF, HF, VHF and UHF have been used to transmit the
GPS corrections. However, such radio transmissions
suffer from distance limitations because RF links become
impractical for transmitting corrections for baselines more
than about 10km due to the signal power and licence
constraints. Furthermore, the positioning accuracy and
reliability of single-base RTK degrades as the baseline
length increases because of the residual distance
dependent biases, such as ionospheric and tropospheric
refraction, that remain after measurement double-
differencing. Thus single-base RTK satisfies the GPS
receiver manufacturers‟ accuracy specification (e.g. 5-
10mm +/- 1ppm) for baseline lengths up to about 20km
(depending on the location in the world, and what part of
the 11 year Solar Cycle it is used).
Rapidly developing information and communication
technologies now enable GNSS operators to broadcast
RTK correction via the Internet using wireless
communication links such as Global System for Mobile
Communications (GSM), General Packet Radio Service
(GPRS), Universal Mobile Telecommunications System
(UMTS) and so on (Wegener & Wanninger, 2005). The
easy access to high-speed Internet and the various options
for wireless connection are reducing the communication
costs significantly for accessing real-time GNSS data
products in the field, and has made Transmission Control
Protocol (TCP) and the Internet Protocol (IP) ubiquitous
(Yan, 2005). As a result, the real-time GNSS data service
providers such as government agencies and private
network operators have adopted the Internet as an
alternative correction service delivery channel (Rizos,
2007). Nevertheless, Internet-based delivery of single-
base RTK corrections still cannot resolve the baseline
distance problem.
The Network-RTK (NRTK) concept has been introduced
to overcome the distance constraint while maintaining the
high positioning accuracy. Many government agencies
and positioning service providers are beginning to express
their interests in the economics of NRTK. As a result,
Continuously Operating Reference Stations (CORS)
networks are being established in many countries in the
form of “positioning infrastructure” (Rizos, 2007).
Currently, there exist several variants of NRTK, such as
Virtual Reference Station (VRS), Individualised Master-
Auxiliary Corrections (iMAX), and Area Correction
Parameter (or Flächen Korrektur Parameter FKP in
German). In NRTK, network corrections are modelled
and interpolated for the position of rover. Noting this
trend, the development of a research-oriented real-time
data processing platform for NRTK positioning was
motivated by the need to investigate different algorithms
and address issues such as network latency, data
synchronisation, positioning quality, and so on. This
paper describes the UNSW in-house NRTK software
known as „SNAPper‟, that features receiver independence
(and therefore significant interoperability), network-
scalability, user flexibility, and research-oriented data
processing capabilities.
In order to transmit modelled NRTK corrections/data via
the Internet, a network protocol and a data format are
required. The Radio Technical Commission for Maritime
Services (RTCM) standards have been widely used as a
GNSS data format. In the United States, the Federal
Communications Commission and the U.S. Coast Guard
use RTCM standards to specify radar systems, Emergency
Position Indicating Radio Beacons, etc. Currently, many
GNSS receiver manufacturers incorporate RTCM SC-104
standard RTK message types into their real-time precise
positioning products. Therefore this facilitates
interoperability between different manufacturers‟
receivers. A new protocol Networked Transport of
RTCM via Internet Protocol (NTRIP) was introduced as
a transmission protocol for real-time GNSS data. NTRIP
has evolved from Hypertext Transfer Protocol (HTTP)
based on TCP/IP. It was introduced by the Federal
Agency for Cartography and Geodesy (BKG) in Germany.
NTRIP comprise of three components: NTRIP servers,
NTRIP casters and NTRIP clients.
Figure 1. NTRIP System.
Figure 1 illustrates the interaction between each NTRIP
component. An NTRIP server receives data from NTRIP
sources such as GNSS receivers and forwards them to an
NTRIP caster. Then the NTRIP caster multiply incoming
data and broadcast to clients based on request. An NTRIP
client is a component that sends requests to an NTRIP
caster and receives data streams in return (Georg &
Denise, 2005). NTRIP provides a capability of seamless
access to CORS anywhere in the world, in real-time. For
example, the RTCM messages are encapsulated in NTRIP
packets in the NTRIP server and transmitted to an NTRIP
caster on request.
SNAPper adopts the international standards of NTRIP
and RTCM as the transmission protocol and data format,
respectively. Originally the software was developed as a
joint project of the GPS Center of the Nanyang
Technological University (Singapore) and the Satellite
Navigation and Positioning Group at the University of
New South Wales (Australia), and the Singapore Land
Authority (Rizos, 2002). The software was tested based
on the Singapore Integrated Multiple Reference Station
Network (SIMRSN) (Chen et al., 2000; Hu et. al., 2002a;
Hu et. al., 2002b; Hu et. al., 2003). Test results
demonstrated centimetre-level accuracy over a 35km
baseline. After the joint project was completed, the
software was developed further by UNSW researchers.
The research was focused mainly on longer baseline
NRTK positioning.
2. UNSW NRTK PROCESSING PLATFORM
SNAPper generates network corrections by incorporating
ultra-rapid orbits from the IGS. Correction messages are
transmitted to rover receivers by wireless data links such
as GPRS. Real-time corrections derived from multiple
reference stations provide the data redundancy. Dual-
frequency GPS data processing allows reliable integer
ambiguity resolution and therefore enables accurate
medium-range (50km or more) baseline positioning. VRS
is implemented in order to estimate the spatially
correlated observation errors across a CORS network and
to interpolate the errors at a user‟s location within that
network. The VRS concept was first discussed by Lynn
and Anil (1995) for Differential GPS, and first introduced
in the SAPOS® network system in Germany (Retscher,
2002).
Figure 2. Data flow for SNAPper: the direct
communication model.
SNAPper consists of two components: Multi-Reference
Station Processor (MRS Processor) and Virtual Reference
Station Generator (VRS Generator). Figure 2 shows the
overall system architecture and data flow of SNAPper. A
rover receiver can be connected to SNAPper via a GPRS
wireless communication link. If a GPRS modem is
unavailable, an alternative way is to use a laptop
computer with wireless Internet connection and pass
corrections to the rover. Figure 3 illustrates this option.
A unique feature of SNAPper is that the MRS Processor
and VRS Generator can be separately operating. MRS
Processor transmits corrections to VRS Generator via
TCP/IP. The main advantage of this design is that the
rover receiver software or the VRS Generator will not be
affected even if algorithms for creating VRS corrections
are modified. Furthermore, the implementation can be
extended so that one MRS Generator can support multiple
VRS Generators, i.e. the proposed VRS can be operated
in a similar way FKP is. This concept of clustering
multiple VRS was introduced by Lim and Rizos (2007).
Figure 3. Data flow for SNAPper: the indirect
communication model.
2.1 Protocol and Data Format
SNAPper utilises NTRIP to deliver multiple GNSS data
streams to the NRTK server where the software is
running, and to broadcast VRS information to multiple
clients via an NTRIP caster. MRS Processor supports
standard formats such as RTCM version 2.3 and RTCM
version 3.0 for input data. In addition, MRS Prpocessor
has the capability of decoding some GNSS receiver
manufacturers‟ proprietary formats, such as Ashtech
MBEN, Leica LB2 and Trimble RT17. VRS Generator
generates VRS corrections with either a Compact
Measurement Record (CMR) format or a RTCM version
3.0 format. As a result, SNAPper has interoperability,
receiver independence and user flexibility. Ultra-rapid
ephemeris product from the IGS FTP server is
automatically downloaded and used by the software.
2.2 MRS Processor
The MRS Processor collects GNSS data streams from one
or more CORS via the Internet through the pre-allocated
ports, synchronises data streams, resolves network integer
ambiguities, computes residual vectors and network
corrections, and transmits the network corrections to the
VRS Generator via TCP/IP. PostSNAPper is a post-
processing module of SNAPper that is useful for testing
NRTK algorithms with Receiver INdependent EXchange
(RINEX) files.
2.3 VRS Generator
The VRS Generator is used to produce VRS
“observations”. It receives corrections from the MRS
Processor and generates synthetic VRS measurements for
a particular user‟s location. The user can be either client
software or a GNSS receiver. VRS Generator is also
capable of producing RINEX files for post-processing.
3. AMBIGUITY RESOLUTION
Centimetre-level accuracy can be achieved by using
carrier-phase-based differential GPS techniques and dual-
frequency data processing. Real-time ambiguity
resolution within the CORS network is the key to
generating reliable network corrections (Rizos et al.,
2003). Since dual-frequency GPS receivers are used,
linear combinations can be constructed from the original
carrier-phase observations.
Figure 4. Flow chart of the algorithm to fix the integer
ambiguities.
There are advantages in certain linear GPS measurement
combinations, such as assisting ambiguity resolution,
cycle slip detection and repair, multipath analyses and
smoothing the code pseudorange. But the main advantage
from linear combinations is the estimation, and
subsequent elimination (at least to the first order) of the
effect of ionospheric delay on GPS measurements. The
double-differenced (DD) linear combinations are widely
used for carrier-phase-based processing (Hoffmann-
Wellenhof et al., 2001). Figure 4 shows the flow chart of
the algorithm for fixing ambiguities within the CORS
network. First, the DD observables are generated. Once
the wide-lane (WL) ambiguity is fixed, a Kalman filter is
utilised to estimate the DD L1 float ambiguity and the
relative tropospheric zenith delay (RTZD). Finally, the
DD L1 integer ambiguity space is searched and the L1
ambiguities fixed. Then the DD L2 ambiguities are fixed.
3.1 Wide-Lane Ambiguity Resolution for Multiple
Reference Stations
The DD carrier-phase observable can be expressed as

i
i
iNiTIi
i
(1)
where the subscript i refers to a carrier frequency i.e.
either L1 or L2, λ is the wavelength of the carrier,

is the
DD carrier-phase measurement expressed in units of full
cycles, ρ is the DD satellite-to-receiver distance, N is the
DD integer ambiguity, T is the DD tropospheric delay, I is
the DD iononospheric delay and ε is the DD observation
noise. The wide-lane (WL) combination of L1 and L2
measurements is a difference between the two:

WL
1
2
(2)
The WL wavelength and the WL integer ambiguity can be
defined as:

1
WL
1
1
1
2
(3)
NWL = N1 N2 (4)
Since the WL wavelength is relatively long (~86.2cm)
and the error propagation theory indicates the standard
deviation decreases down to less than a half cycle, the
WL ambiguities among the reference stations can be
easily resolved to their likely integer values. Because the
VRS algorithm is basically an inverse distance weight
interpolator, the residuals of the CORS contain distance
dependent errors. For this reason, the ionospheric delay is
considered an error rather than a parameter that can be
modelled or determined when the WL integer ambiguity
is computed:

NWL
WL (
T)
WL
(5)
where the tropospheric delay T is estimated by a Hopfield
model (Hoffmann-Wellenhof et al., 2001) or other models.
Then a simple criterion is used to fix the WL ambiguities.
Firstly, the data gaps and cycle slips are checked from the
reference satellite. Then, the ambiguities are rounded up
to the closest integer values, and the differences of the
fixed and float values for all satellite-satellite pairs are
computed. Lastly, the minimum and the maximum values
of the differences are computed. If the difference is
greater than a specified threshold, it is assumed that there
has been a failure to fix ambiguities, and more data is
accumulated. Otherwise the ambiguities are fixed. An
alternative method to fixing the integer ambiguities is to
use the Least-squares Ambiguity Decorrelation
Adjustment (LAMBDA) algorithm (Joosten and Tiberius,
2002).
3.2 L1 and L2 Ambiguity Resolution
The DD ionosphere-free (IF) combination can be obtained
by:
(6)
The IF wavelength and the IF ambiguity are:

1
IF
77
1
60
2
(7)
NIF = 77N1 60N2 (8)
The L1 integer ambiguity N1 and the L2 integer ambiguity
N2 can be fixed by combining the WL integer ambiguity
and the IF integer ambiguity (Goad, 1992) or by using
LAMBDA.
The residual tropospheric delay can be approximately
expressed as (Zhang and Lachapelle, 2001):
/sin
z
T T z
(9)
where
z
T
is the relative tropospheric zenith delay (RTZD)
and
z
is the elevation angle. Instead of using LAMBDA,
RTZD and
1
N
are estimated in real-time with a Kalman
filter based on the model (Hu et al., 2003):
),0(~,
),0(~,
11, kkkkkkk
kkkkkk QNvvxx
RNuuxHy
(10)
where
),(
N
represents that the error follows the
normal distribution function with its mean µ and standard
deviation

. The observation vector
k
y
is given by:
nn
IFIF
IFIF
IFIF
k
y
,1,1
3,13,1
2,12,1
(11)
The superscript 1 denotes the reference satellite, and 2 to
n denote the other satellites. The state vector and the
design matrix are:
n
WLIF
n
IF
WLIFIF
Z
k
NN
NN
T
x
,1,1
1
2,12,1
1
6017
6017
(12)
100)sin(/1)sin(/1
010)sin(/1)sin(/1
001)sin(/1)sin(/1
1
31
21
n
k
zz
zz
zz
H
(13)
RTZD is assumed to be a first-order Gauss-Markov
process. The state transition matrix is:
100
010
00
/
1,
t
kk
e
(14)
and the corresponding covariance matrix is defined as:
16
16
2/
100
010
00)1(
2
e
e
qe
Q
t
k
(15)
where q is the variance of the RTZD process noise for the
correlation time τ; and Δt is the sampling rate.
IFIFIF
IFIF
IF
k
R
422
42
4
(16)
vIF = λIF 2 (77ν12 + 60ν22)
where ν1 is the L1 carrier-phase residual (cycles) and ν2is
L2 carrier-phase residual (cycles).
A residual-based adaptive procedure can be used to
improve the reliability of the Kalman filter:
 
T
kkkkkk
T
kkkkkk
kkkkkkkkkk
k
T
kkkk
T
kkkk
k
Tkkkkkkk
kkkkkk
HKIPHKIKRKPP
XHZKXXX
RHPHHPK
QPP
XX
)(
)(
)(
1,,
1,1,,
1
11,1,
11,11,1,
1,11,1,
(17)
where Xk,k-1 and Pk,k-1 are the predicted state vector and its
covariance matrix respectively, Xk,k and Pk,k are the
estimated ones, and Kk is the gain matrix. The float L1
ambiguity and RTZD can be estimated using the Kalman
filter recursive formula. For example:

N1
1,2 (Xk
1,2 /
IF 60NWL
1,2 )17
(18)
4. NETWORK CORRECTION GENERATION
There are several algorithms that can be used to generate
the network corrections (Wübbena et al., 1996; Gao et al.
1997; Han and Rizos, 1996). Dai et al. (2001) showed that
these algorithms have similar performance. A linear
combination method proposed by Han (1997) is adopted
by SNAPper.
The DD observation can be rewritten as:
VN
(19)
where V is the residual which is used to generate the
correction. For each pair of satellites, and each master-
reference station, one V can be calculated. After obtaining
all residuals from the master-reference station
combinations, the VRS Generator is waiting for a rover‟s
request. A rover receiver sends its approximate position to
the VRS Generator via a wireless communication link in
the NMEA format. The approximate rover‟s position
becames a VRS position.
A VRS position is used to compute the satellite to VRS
distance ρv. The range difference between VRS and the
master station will be:

v,m
v
m
(20)
where ν indicates VRS and m indicates the master station.
VRS corrections are applied to the master station‟s
observations to generate VRS observations on a satellite-
by-satellite and epoch-by-epoch basis. Firstly, VRS
observations for the primary satellite i are constructed:

Pk,v
iPk,m
i 
v,m
iPk,m
i(
v
i
m
i)
Pk,m
i
v
i
m
i
kLk,v
i
kLk,m
i 
v,m
i
(21)
Secondly, VRS observations for non-primary satellites j
(j: 1, 2 n) are constructed:

Pk,v
jPk,m
j 
v,m
jVk,v,m
i,j
kLk,v
j
kLk,m
j 
v,m
jVk,v,m
i,j
(22)
Finally, corrections for the tropospheric delay are added
because T has been subtracted from the residual
previously by the MRS generator:

Pk,vPk,vTvTm
 
kLk,v
kLk,vTvTm
 
(23)
5. TESTS
In order to evaluate the performance of the NRTK
processing platform using the above algorithms, a number
of tests were carried out. Test results presented in this
paper were conducted at UNSW whereas the real-time
data were transferred from the Vicmap Position-
GPSnet™, operated by the Victorian Department of
Sustainability and Environment, Melbourne, Victoria. The
observation data from the reference stations were supplied
to the platform via the Internet using NTRIP in the RTCM
v3.0 format.
The test results were obtained by simulating one reference
station as a rover, surrounded by three CORS. The closest
CORS was chosen as the master station. BALL was
chosen as a master station; COLA and WHIT are the
other two CORS, while BACC was used as a rover
station. Figure 5 shows the layout of the test network. The
test started at 01h 55m UTC and the duration was about 3
hours 17 minutes. A VRS was generated close to BACC.
The distance between VRS and BACC was 110.503km.
The distances from BALL, WHIT and COLA to the VRS
were 53km, 55km, and 105km, respectively.
Figure 5. Layout of test network.
Tables 1 and 2 summarise the test results. Table 1
includes results for all epochs during the test, while Table
2 shows results when initialisation epochs are excluded.
Table 1. Results including initialisation epochs.
Easting
error (cm)
Northing
error (cm)
Height
error (cm)
Mean
-0.6
0.3
2.0
Stdev
4.0
4.2
13.3
Table 2. Results excluding initialisation epochs.
Easting
error (cm)
Northing
error (cm)
Height
error (cm)
Mean
-0.8
0.2
4.1
Stdev
3.3
3.9
5.6
Figures 6, 7 and 8 show the easting, northing and height
errors for the whole test span. The mean easting errors is -
0.6cm and for the northing errors it is 0.3cm. Standard
deviations of the easting and northing errors are 4.0cm
and 4.2cm respectively. The mean height error is 2.0cm
and the standard deviation is 13.3cm. Centimetre-level
accuracy was achieved for the northing and easting
components, while the accuracy of the height was
noticeably worse.
44000 46000 48000 50000 52000 54000 56000
-20
-15
-10
-5
0
5
10
15
20
25
Time of Day (sec)
Easting error (cm)
VRS (BALL 53 km, WHIT 55 km, COLA 105 km) - Rover (BACC)
Easting Error (Mean: -0.6 cm, Stdev: 3.9 cm)
Figure 6. Easting error.
44000 46000 48000 50000 52000 54000 56000
-20
-15
-10
-5
0
5
10
15
20
25
30
35
Time of Day (sec)
Northing error (cm)
VRS (BALL 53 km, WHIT 55 km, COLA 105 km) - Rover (BACC)
Northing Error (Mean: 0.3 cm, Stdev: 4.2 cm)
Figure 7. Northing error.
44000 46000 48000 50000 52000 54000 56000
-120
-100
-80
-60
-40
-20
0
20
40
60
Time of Day (sec)
Height error (cm)
VRS (BALL 53 km, WHIT 55 km, COLA 105 km) - Rover (BACC)
Height Error (Mean: 2.0 cm, Stdev: 13.3 cm)
Figure 8. Height error.
However there are positioning gaps, indicating where the
NRTK corrections were not available for the VRS. There
were 1,321 epochs unavailable out of a total of 10,504
epochs, approximately a 11% loss. There are two reasons
for this. Firstly, one or more integer L1 ambiguities were
not able to be fixed at that epoch. Secondly, there is
network latency for data delivery from each CORS to the
server computer, consequently the data synchronisation
becomes an issue among three parallel processes.
Further tests are needed in different environments,
including field tests using wireless communications such
as GPRS since the wireless network latency can affect the
positioning as well.
The number of available satellites to generate the
correction is also an important factor. Figure 9 shows the
number of available satellites at the VRS during the test.
There were at least 8 satellites available during the test.
44000 46000 48000 50000 52000 54000 56000
7
8
9
10
11
12
13
Time of Day (sec)
# of satellite
VRS (BALL 53 km, WHIT 55 km, COLA 105 km) - Rover (BACC)
Number of available satellites at VRS
Figure 9. Number of available satellites.
In order to compare the performance of NRTK, single-
base RTK tests were also carried out. BALL was the
nearest reference station, hence RINEX files for BALL
and BACC were processed. Table 3 shows the post-
processed results of single-base RTK. Compared to the
VRS solutions the easting and northing errors of single-
base RTK are similar, however its height errors are much
worse i.e. NRTK heights are more accurate. On the
contrary, single-base RTK heights are more precise
because its standard deviation is much smaller than that of
NRTK.
Table 3. Results for single-base RTK
(BALL-BACC).
Easting
error (cm)
Northing
error (cm)
Height
error (cm)
Mean
-0.45
0.20
-13.91
Stdev
0.76
1.07
2.36
6. CONCLUDING REMARKS
There are high demands for real-time GNSS data in
precise positioning applications. A robust kinematic
ambiguity resolution algorithm was described for high
precision positioning using corrections computed from a
CORS network. In order to validate the performance of
the algorithms a test was carried out with a local CORS
network. Easting, northing and height errors were
computed: 0.6cm accuracy in easting, 0.3cm in northing
when the baseline length extends to more than 50km. The
test confirmed that the implemented RTK processing
platform can deliver centimetre-level accuracy
positioning in real-time and, more importantly, function
as a research platform for further studies into NRTK.
ACKNOWLEDGEMENTS
The authors would like to acknowledge Victorian
Department of Sustainability and Environment,
Melbourne for providing data access. This work is
supported by Cooperative Research Centre for Spatial
Information (CRC-SI) project 1.04 “Delivering Precise
Positioning Services in Regional Areas”, 2007-2010.
REFERENCES
BKG: The Federal Agency for Cartography and Geodesy
[online]. http://igs.bkg.bund.de/ [Accessed on 14
September 2009].
Chen, X.M., Han, S.W., Rizos, C., Goh, P.C. (2000),
“Improving real-time positioning efficiency using
the singapore integrated multiple reference station
network (SIMRSN)”, Proc 13th Int Tech Meeting of
the Satellite Division of the U.S. Inst. of Navigation,
Salt Lake City, Utah, 19-22 September, pp. 9-16.
Dai, L., Han, S.W., Wang, J., Rizos, C. (2001), “A study
of GPS/GLONASS multiple reference station
techniques for precise real-time carrier phase-based
positioning”, 14th Int. Tech. Meeting of the Satellite
Division of the U.S. Inst. of Navigation, Salt Lake
City, Utah, 11-14 September, pp. 392-403.
Gao, Y., Li, Z., McLellan, J.F. (1997), “Carrier phase
based regional area differential GPS for decimeter-
level positioning and navigation”, 10th Int. Tech.
Meeting of the Satellite Division of the U.S. Inst. of
Navigation, Kansas City, Missouri, 16-19
September, pp. 1305-1313.
Goad, C.C. (1992), Robust techniques for determining
GPS phase ambiguities, 6th International Geodetic
Symposium on Satellite Positioning, Columbus,
Ohio, 17-20 March, pp.245-254
IGS: The International GNSS Service [online],
http://igscb.jpl.nasa.gov/ [Accessed on 14 September
2009].
Joosten, P., and Tiberius, C.C.J.M (2002), “LAMBDA:
FAQ., GPS Solution, 6(1-2): 109-114.
Lim, S., and Rizos, C. (2007), “A new framework for
server-based and thin-client GNSS operations for
high accuracy applications in surveying and
navigation”, 20th Int Tech Meeting of the Satellite
Division of the U.S. Inst. of Navigation, Fort Worth,
Texas, 25-28 September, pp. 2169-2177.
Lynn, W., Anil, T. (1995), “DGPS architecture based on
separating error components, virtual reference
stations and FM sub-carrier broadcast”, 51st Annual
Meeting, 50 Years of Navigation Progress From Art
to Utility, Colorado Springs, Colorado, 57 June, pp.
128139.
NTRIP: Networked Transport of RTCM via Internet
Protocol [online],
http://igs.bkg.bund.de/index_ntrip.htm [Accessed on
14 September 2009].
Retscher, G. (2002), “Accuracy performance of virtual
reference station (VRS) networks”, Journal of
Global Positioning Systems, 1(1): 40-47.
RTCM: The Radio Technical Commission for Maritime
Services [online] http://www.rtcm.org [Accessed on
14 September 2009].
Rizos, C., (2002), “Network RTK research and
implementation: A geodetic perspective”, Journal of
Global Positioning Systems, 1(2): 144-150.
Rizos, C, Han S.W. (2003) Reference station network
based RTK systems-concepts and progress”, Wuhan
University Journal of Natural Sciences, 8(2B): 566-
574.
Rizos, C. (2007), “Alternatives to current GPS-RTK
services and some implications for CORS
infrastructure and operations”, GPS Solutions, 11:
151158.
SAPOS®: [online] http://www.sapos.de. [Accessed on 14
September 2009].
Han, SW. (1997), “Carrier phase-based long-range GPS
kinematic positioning”, PhD dissertation, rept
UNISURV S-49, School of Geomatic Engineering,
The University of New South Wales, Sydney.
Han, S.W. and Rizos, C. (1996), “GPS network design
and error mitigation for real-time continuous array
monitoring system”, 9th Int Tech Meeting of the
Satellite Division of the U.S. Inst. of Navigation,
Kansas City, Missouri, 17-20 September, pp. 1827
1836.
Hoffmann-Wellenhof, B., Lichtenegger, H., Collins, J.
(2001), GPS Theory and Practice, Fifth Edition,
Springer-Verlag, Wien New York.
Hu, G.R., Khoo, H.S., Goh, P.C., Law, C.L. (2003),
Development and assessment of GPS virtual
reference stations for RTK positioning, Journal of
Geodesy, 77: 292302.
Hu, G.R., Khoo, H.S., Goh, P.C., Law, C.L. (2002a),
“Testing of Singapore Integrated Multiple Reference
Station Network (SIMRSN) for precise fast static
positioning”, European Navigation Conference -
GNSS2002, Copenhagen, Denmark, 2730 May,
CD-ROM proc.
Hu, G.R., Kho,o H.S., Goh, P.C., Law, C.L. (2002b),
“Internet-based GPS VRS RTK positioning with a
multiple reference station network”, Journal of
Global Positioning Systems, 1(2): 113-120.
Vicpos Position-GPSnet™ [online],
http://www.gpsnet.com.au [Accessed on 14
September 2009].
Wegener, V., Wanninger, L. (2005), Communication
options for Network RTK / SAPOS® realization,
Joint 2nd Workshop on Positioning, Navigation and
Communication 2005 (WPNC‘05) & 1st Ultra-
Wideband Expert Talk 2005 (UET‘05), Hannover,
Germany, 17th March.
Wübbena, G., Bagge, A., Seeber, G., Boder, V.,
Hankemeier, P. (1996), “Reducing distance
dependent errors for real-time DGPS applications by
establishing reference stations networks”, 9th Int
Tech Meeting of the Satellite Division of the U.S.
Inst. of Navigation, Kansas City, Missouri, 17-20
September, pp. 18451852.
Yan, T.S. (2005). “Benefits of telecommunications
technology to GPS users”, Journal of GPS, 4(1-2):
240-244.
Zhang, J., Lachapelle, G. (2001), “Precise estimation of
residual tropospheric delays using a regional GPS
network for RTK applications”, Journal of Geodesy,
75(5-6): 255-266.
... CORS stations from BIG can be used as a base reference for positioning using GNSS, and data from InaCORS can be used for positioning using NTRIP (Network Transport of RTCM via Internet Protocol) services. For positioning in real-time kinematic (RTK), you can use the network method or the nearest/single-base rtk method. 2 NRTK (Network RTK) is a differential positioning method that develops a single base RTK [4]. The principle of NRTK is that data recording is carried out by reference stations from the GNSS satellite continuously, which is then stored and sent to the NRTK server via the internet network. ...
... The principle of NRTK is that data recording is carried out by reference stations from the GNSS satellite continuously, which is then stored and sent to the NRTK server via the internet network. Currently, there are several NRTK methods, namely VRS, MAX, iMAX, Nearest, and FKP [4]. ...
Article
Full-text available
Badan Informasi Geospasial (BIG) is a government agency engaged in the field of Geospatial Information (IG). BIG provides several services that can be accessed by the general public and related to IG. One of the services provided by BIG is the Indonesia Continuously Operating Reference System (InaCORS). InaCORS is divided into various services, as Rinex data services, Online Post-Processing, RTK NTrip, and Mobile InaCORS. This study aims to determine the effectiveness and accuracy of InaCORS services for GNSS surveys using the Rapid Static method and Network RTK. The rapid static survey data is processed using online post-processing services and Network RTK (iMax, Max, and Nearest) will use InaCORS points as a base reference. This study also uses the results of the Total Station tool as comparison data. The results showed that the average value of the difference between the rapid static coordinates and the TS observations was dN = 0.353 m dE = 0.180 m and dZ = 0.233 m, while the Network RTK and TS coordinates were dN = 0.408 m dE = 0.184 m and dZ = 0.176 m.
... After collecting data, a post-processed operation must be carried out to adjust these observations. The second technique is real-time kinematic, which can provide real-time corrections with up to centimeter-level accuracy (Heo et al. 2009). In this technique, one receiver is stationed at the reference point as a base, while the other ones move as rovers along the unknown points. ...
Article
Full-text available
The Global Positioning System (GPS) is considered as one of the most powerful tools that is used in geodetical works. It is increasingly considered as an alternative tool for conventional surveying techniques. The goal of this research paper is to produce a methodology for generating topographic contour maps using the post-processed kinematic differential GPS technique (PPK-GPS). The performance of this technique is assessed through a comparison with the traditional spirit-leveling technique taking into consideration the accuracy, time, and cost. The study area is located along the Mediterranean coast in the Rosetta area of the middle Delta in Egypt with a total area of about 39 square kilometers. The field work includes 14 control points and about 50000 PPK observations. In addition, 20 check cross sections that extend along the coastal line from Rosetta to El Burullus cities were observed using both spirit-leveling and GPS-PPK techniques. The results of the comparison indicate that the PPK-GPS technique can be used instead of spirit-leveling in producing a topographic contour map with an accuracy of about 20 cm in orthometric height. We find that the PPK-GPS technique reduces the estimated time of leveling works in our study area by about 70%.
Chapter
Based on Beidou regional navigation satellite system in the phase of 4GEO + 5IGSO constellation, systematic research on Beidou high-precision relative positioning has been carried out by sufficient measured data from Beidou test network. A comparison scheme with Beidou/GPS dual-mode data is designed, and feasibility and attainable precision of Beidou relative positioning have been verified effectively. As shown in results: coordinate accuracy realized by Beidou relative positioning has achieved centimeter level, and baseline repeatability has reached millimeter level, which are equivalent to GPS high-precision relative positioning level. For short baseline, coordinate repeatability in the North, East and Up direction is respectively better than 0.5, 1.3 and 2.9 mm, and absolute error of coordinate is respectively better than 0.2, 0.6 and 1.7 mm by comparison with GPS; for medium/long baseline, the coordinate repeatability in the NEU direction is respectively better than 2.1, 5.9 and 7.3 mm, and absolute error of coordinate is respectively better than 1.6, 1.3 and 2.2 mm; for long baseline over 1000 m, NEU coordinate repeatability is respectively better than 8.4, 14.6 and 21.3 mm, and absolute error is respectively better than 1.5, 1.5 and 14.5 mm, and baseline repeatability is approximately 0.003 ppm. The conclusion from the article provides technical criteria for widespread use of Beidou in the high precision application fields, and makes sense for promotion of application prospect and enhancement of usage confidence with Beidou.
Article
Full-text available
Multiple reference station networks have been established for high precision applications in many countries worldwide. However, real-time application is still a difficult task in practice. Virtual reference station (VRS) concept is an efficient method of transmitting corrections to the network users for RTK positioning. Today's challenge for VRS RTK positioning lies in adapting advance wireless communication technologies for real time corrections. With the availability of GPRS technology, an Internet-based VRS RTK positioning infrastructure via GPRS has been developed and tested. This paper discusses the VRS data delivery mechanism, and gives an overview on VRS data generation for RTK positioning. Field test results are presented to evaluate the performance of the proposed system. The results demonstrate that Internet-based VRS RTK positioning can be achieved to better than 4 centimeters accuracy in horizontal position. Height accuracy is in the range of 1 to 6 centimeters.
Article
Full-text available
In this paper a near real-time (typically a baseline update once per hour, or once per day) continuous array system, with at least three dual-frequency GPS receivers connected to the IGS network, and many single frequency receivers, is proposed for monitoring applications covering a region of 50km radius. A weighted differential GPS method, appropriate for the suggested array design, can be implemented in order to eliminate or mitigate the orbit bias, including the effects of SA. A local area epoch-by-epoch and satellite-by-satellite ionospheric delay model, determined using dual-frequency observations, is used to correct single frequency observations. Tropospheric delay is taken into account by modelling it as a first order Gauss-Markov (or random walk) process with temporal correlations. Multipath elimination is achieved by using an Finite Impulse Response (FIR) lowpass filter for the local ionospheric delay model, and multipath extraction using an FIR bandpass filter to account for the multipath in real-time. The biases dependent on the GPS receiver and antenna are also discussed and associated optimal network designs for data processing are suggested. In addition, efficient and reliable ambiguity resolution, and automatic cycle slip detection and repair procedures will be incorporated. This system would be capable of millimeter horizontal accuracy and centimeter vertical accuracy, and is intended to address real-time applications for earthquake studies, and for volcano and engineering deformation monitoring.
Article
Recent developments in differential GPS (DGPS) services have concentrated mainly on the reduction of the number of permanent reference stations required to cover a certain area and the extension of the possible ranges between reference and rover stations. Starting from networked DGPS stations where all stations are linked to a central control station for data correction and modeling, the most advanced technique nowadays is based on the virtual reference station (VRS) network concept. In this case, observation data for a non-existing "virtual" station are generated at the control center and transmitted to the rover. This leads to a significant improvement in positioning accuracy over longer distances compared to conventional DGPS networks. This paper summarizes the various DGPS architectures and the corresponding accuracy. This is followed by a description of the models and algorithms used for the VRS station concept. Practical examples of correction data services in Europe are given to highlight the achievable positioning accuracy. The results of an analysis of test data in a virtual reference station network in southern Germany show that always a horizontal positioning accuracy in the order of ± 5 cm can be achieved for baselines with a length up to 35 km.