2009年度AOBセミナー
⻑基線RTK-GPS/GNSSの課題と展望
Overview of Long-Baseline RTK-GPS/GNSS
東京海洋大学 高須 知二
Long-Baseline RTK
2
RTK-GPS/GNSS
• Relative Positioning based on Carrier-Phase
– Real-time Position of A Moving Receiver
– Typical Accuracy: 1 cm + 1ppm x BL RMS (Horizontal)
– Transmit Reference Station Data to Rover via Wireless Link
– OTF (On-the-Fly) Integer Ambiguity Resolution
• Network-based RTK
– Single Receiver with Mobile Communication
– Commercial Services with Many Reference Stations
Reference Data Link Rover
Station Receiver
Applications of RTK-GPS/GNSS
Geodetic Survey Construction Precision Agriculture
Machine Control
ITS (Intelligent Mobile Mapping Sports
Transport System) System
http://www.trimble.com, http://www.leica-geosystems.com, http://www.gpsworld.com 4
Effect of Baseline Length
BL=0.3 km BL=13.3 km
RMS Error: RMS Error:
E: 0.2cm E: 2.2cm
N: 0.6cm N: 2.4cm
U: 1.0cm U: 10.6cm
Fix Ratio: Fix Ratio:
99.9% 94.2%
BL=32.2 km BL=60.9 km
RMS Error: RMS Error:
E: 10.0cm E: 14.0cm
N: 12.0cm N: 14.8cm
U: 30.2cm U: 26.7cm
Fix Ratio: Fix Ratio:
64.3% 44.4%
(24 hr Kinematic : Fixed Solution : Float Solution)
5
Demands of Long-baseline RTK
• No Reference Station in the vicinity of user
– Land of Vast Country (Russia, Canada, ...)
– Sea (Ships, Buy, Plants, Construction, …)
– Air (Flight), Space (Satellites)
• Need Separation of Movements between user
and reference
– Nation/World-wide Crustal Deformation Monitor
– Observation of Seismic Event by Large Earthquake
6
Long-BL RTK vs. Real-time PPP
Long-Baseline RTK Real-time PPP
Receiver Need Ref Station Single Receiver
Result Baseline Vector Position in Ref Frame
Baseline Length 100 km - 3,000 km Anywhere
Ephemeris Broadcast/Precise (IGU) Precise (IGU)
SV Clock Broadcast Precise (?)
Ionosphere Elimination/Estimation Iono-Free LC
Troposphere Model/Estimation Estimation
Antenna PCV Model Model
Earth Tides Ignored/Model Model/Estimated
Integer Ambiguity Fixed Float
7
Issues of Long-baseline RTK
8
Issues for Long-BL RTK
• Error Sources = Residuals of DD
– Satellite Ephemeris (BL=100 km- )
– Ionosphere (BL=10 km- )
– Troposphere (BL=50 km- )
– Antenna PCV, Earth Tides, Phase Windup ...
– Multipath + Receiver Noise (= Short BL)
• Strategy of Ambiguity Resolution
– Slow Convergence of Float Ambiguity Estimation
– Must Keep Integer Nature of Ambiguities
9
Broadcast Ephemeris
0.8 m
20,000
km
D.Boyd, GPS Constellation Status and Performance, CGSIC 49th, 2009
ΔrSD
ΔrSD = 0.8 m x 100 km/20,000 km = 4 mm 100 km
10
Precise Ephemeris
IGS Product Table
Final Rapid Ultra-Rapid (IGU)
Broadcast
(IGS) (IGR) Observed Predicted
Orbit ~2.5cm ~2.5cm ~3cm ~5cm ~100cm
Accuracy ~150ps
~75ps RMS ~75ps RMS ~3ns RMS ~5ns RMS
Clock RMS
~20ps STD ~25ps STD ~1.5ns STD ~2.5ns STD
~50ps STD
17-41
Latency 12-18 days 3-9 hours realtime realtime
hours
every at 17 UTC at 03, 09, at 03, 09,
Updates -
Thursday daily 15, 21 UTC 15, 21 UTC
Orbit 15min 15min 15min 15min daily
Sample
Interval Sat: 30s
Clock 5min 15min 15min daily
Stn: 5min
(2009/8, http://igscb.jpl.nasa.gov/) 11
Improvement of IGU Orbit
(J. Ray et. al., Status of IGS Ultra-Rapid Products for Real-Time Applications, 2008 AGU Fall Meeting)
12
Accuracy of IGU-Predicted
PRN2
RMSE: R:0.8, A:3.5, C:2.8cm
PRN32
2009/5/27-31 RMSE: R:2.9, A:12.7, C:2.6cm
13
Real-time Precise Ephemeris
• IGS RTWG+EUREF
• Orbit/Clock
– Orbit - IGU-Predicted
– Clock - Estimated
• Protocol/Format
– NTRIP v.1.0
– RTCM3 MT1057-1068
(Draft Proposal)
http://igs.bkg.bund.de/ntrip/orbits
14
Solar Cycle
International Sunspot Number (ISN): 1700-2009
20 21 22 23
by SIDC (Solar Influences Data Analysis Center) in Belglum (http://sidc.oma.be)
Solar Cycle Prediction: Cycle 24
23 24 24
by NOAA SWPC (Space Weather Prediction Center) (http://www.swpc.noaa.gov/SolarCycle)
15
Ionosphere Delay
SD L1 Vertical Ionospheric Delay (1 Week)
BL=100 km
RMS: 12.4 cm
BL=500 km
RMS: 30.8 cm
16
Tropospheric Delay
SD ZTD 2009/1/1-1/7 SD ZTD 2009/7/1-7/7
BL=100 km
BL=500 km
17
Integer Ambiguity Resolution (1)
• Ambiguity Resolution for Long Baseline RTK
– Eliminate Ionos with Integer Nature of Ambiguity
• Search Strategy for Short Baseline RTK
– Solve Statistically Optimal ILS Problem
– Efficient Strategy like LAMBDA
– Issue: No Ionos Elimination
• WL/NL Strategy for Static Post Processing
– Fix WL by Rounding of MW LC Average
– Fix NL by Rounding Iono-Free LC + WL
– Issue: Long Convergence, Reliability with Dual-Freq
18
Integer Ambiguity Resolution (2)
• RTKLIB v.2.3.0 (T. Takasu, 2009)
– EKF-based States Estimation
Rover Position, SD Vertical Ionos Delay, ZTD
Residuals for Rover/Ref, SD Float Ambiguity L1/L2
– Ionos: Random-Walk + Single Layer Model
– Tropos: A Priori Model + Random-Walk + NMF
– ILS Search by LAMBDA/MLAMBDA
– Partial Fixing with AR Elevation Mask
– Tight Constraint to Fixed Ambiguity (Fix and Hold)
19
Integer Ambiguity Resolution (3)
Full Fix
No Hold
(43.1%)
Partial Fix
No Hold
(86.2%)
Partial Fix
Fix and Hold
(96.8%)
v.2.3.1b, BL=100 km, 24 hr, Eph=IGS 20
RTKLIB
21
History
• 2006/4 v.0.0.0First version for RTK+C prog. lecture
• 2007/1 v.1.0.0Simple post processing AP
• 2008/7 v.2.1.0Add APs, support medium-range
• 2009/1 v.2.2.0Add real-time AP, support NTRIP,
start to distribute it as Open Source
• 2009/5 v.2.2.1 Support RTCM, NRTK, many receivers
• 2009/12 v.2.3.0 Support GLONASS, INS/GPS, …
• 2010/3? v.2.3.1 …
• 2010/5? V.2.4.0 Support Real-time PPP with IGS
22
RTKLIB APs
RTKNAVI, RTKRCV : Real-time positioning
RTKPOST, RNX2RTKP : Post-processing baseline analysis
RTKPLOT : Plot raw observation data and solutions
RTKCONV : RINEX converter for raw receiver log
...
23
RTKLIB v.2.3.1
• New Features
– Support Precise Ephemeris for Real-time AP
– Support Stream Type of FTP Download
– Tuning for Long-Baseline RTK
– Support Output Swap at day/hour boundary
– Improvement of Troposphere Model
– Full Positioning Options for RNX2RTKP
• Tuning for Long-Baseline RTK
24
Preliminary Evaluation (1)
BL 2009/1/1 - 1/7 2009/7/1 - 7/7
Rov Ref
(km) SDE SDN SDU Fixing SDE SDN SDU Fixing
3023 50.6 0.5 0.6 1.6 99.8% 1.6 1.9 5.7 90.2%
0586 100.2 0.7 0.9 2.5 98.7% 4.0 2.7 9.3 78.1%
0562 150.8 0.6 0.8 2.4 99.0% 3.7 2.2 6.6 83.6%
0241 200.4 0.7 0.9 2.2 99.5% 3.4 3.4 15.0 85.2%
0601 250.3 0.7 1.0 2.7 96.8% 3.2 2.1 6.5 79.1%
2110 0174 300.0 0.7 1.0 2.8 98.7% 2.0 1.8 5.2 88.0%
0579 351.9 0.9 0.9 2.8 99.3% 3.4 3.9 11.2 81.3%
0324 400.6 1.0 0.9 2.9 96.7% 2.6 2.0 6.5 74.3%
0905 450.6 3.2 5.4 21.2 64.8% 7.2 5.7 19.7 63.2%
0369 500.4 1.0 0.9 3.3 98.7% 2.4 1.8 6.0 85.6%
0837 995.6 3.1 1.9 8.5 91.4% 7.0 7.3 19.4 66.0%
SDE/SDN/SDU: Standard Dev E/N/U (cm), v.2.3.1b, Eph=IGS
25
Preliminary Evaluation (2)
2009/1/1-1/7 BL=100 km 2009/7/1-7/7
BL=300 km
26