evaluation and calibration of receiver inter-channel ... gnss 20… · these biases are known as an...
TRANSCRIPT
Evaluation and Calibration of
Receiver Inter-channel Biases for
RTK-GPS/GLONASSHideki Yamada,
Tomoji Takasu, Nobuaki Kubo and Akio Yasuda
Tokyo University of Marine Science and Technology
Outline
• Background and Objective
• Inter-channel Bias Calibration
• RTK-GPS/GLONASS with Calibrated Biases
• Use of GLONASS Float Ambiguity
• Summary and Conclusions
Mobile RTK PositioningInstantaneous RTK (Real Time Kinematic) is precise positioning
with epoch by epoch carrier-phase ambiguity resolution (AR).
Results of Mobile RTK Positioning
Fix solution : 43 %
Float solution : 57 %
NovAtel - NovAtel
Fix solution : 1 %
Float solution : 99 %
NovAtel - JAVAD
We aim to improve GLONASS positioning performance
in RTK-GPS/GLONASS between different types of receivers.
GPS + GLONASS L1 + L2
0 1 2 3 4 5 6 7 8 x 104-0.2-0.15-0.1-0.0500.050.10.150.2
Time (sec)Position Error (m)
RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=29.5% RMSE : E: 0.0032m N: 0.0057m U: 0.0157m East (m)North (m)Up (m)Results of Static RTK Positioning
GPS + GLONASS L1 + L2 instantaneous AR
Base length = 300 m
0 1 2 3 4 5 6 7x 104-0.2-0.15-0.1-0.0500.050.10.150.2
Time (sec)Position Error (m)
RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=99.0% RMSE : E: 0.0025m N: 0.0044m U: 0.0157m East (m)North (m)Up (m)P
osi
tio
nin
g E
rro
rs (
m)
Fix solution : 99.0 %
Float solution : 1.0 %
NovAtel -NovAtel
Time (s)P
osi
tio
nin
g E
rro
rs (
m)
NovAtel - JAVAD
Fix solution : 29.5 %
Float solution : 70.5 %
Time (s)
Background and Objective
Inter-channel bias
- The RF hardware design of a receiver will introduce frequency-
dependent biases. These biases are known as an inter-channel bias.
- These biases can not be canceled out with double differencing
between different types of receivers in GLONASS, except for GPS.
What is main source of observation errors to prevent
GLONASS ambiguity from being fixed ?
Background and Objective
• Estimates of GLONASS inter-channel bias for several
receivers are obtained in short baseline [Wanninger, L., et al,
ION GNSS 2007].
• Narrow band loop filter smoothes SD residual to get on line
code bias [Kozlov, D., et al, ION GNSS 2000].
• In this study, we try to improve the positioning performance
of instantaneous RTK with off line tables of GLONASS
inter-channel bias at various receivers calibrated in only
zero baseline measurements.
Inter-channel Bias Calibration
Antenna
Splitter
TopconNovAtel
GPS/GLONASS Antenna
TrimbleJAVAD
Zero baseline test
In the zero baseline, double-difference (DD) of code measurements can cancel
out terms of orbital errors, clock errors, atmospheric delays and multipath,
and remain DD of receiver noises and inter-channel biases.
Inter-channel Bias Calibration
Roof of our university (Open sky)Site
2009/12/23
24 hours/30 s24 hours/30 s20 hours/30 s24 hours/30 s
Date
Total/interval
Trimble (Zephyr 2 )Antenna
R7GNSSNET-G3OEMVLegacyReceiver type
TrimbleTopconNovAtelJAVAD
Experimental Environment
Calibrated L1 Inter-channel Code Biases
Tables of bias with slot 21
-6 -4 -2 0 2 4 6-3
-2
-1
0
1
2
3
GLONASS frequency number
NovAtel-NovAtel
Javad-Javad
DD
of L
1 in
ter-chan
nel co
de b
iases (m)
1 mD
D o
f L1
inter-ch
ann
el cod
e biases (m
)
-6 -4 -2 0 2 4 6-3
-2
-1
0
1
2
3
GLONASS frequency number
NovAtel-Trimble
NovAtel-Topcon
NovAtel-Javad1 m
Calibrated L2 Inter-channel Code Biases
Tables of bias with slot 21
-6 -4 -2 0 2 4 6-1
-0.5
0
0.5
1
GLONASS frequency number
NovAtel-NovAtel
Javad-Javad
DD
of L
2 in
ter-chan
nel co
de b
iases (m)
50 cm
-6 -4 -2 0 2 4 6-1
-0.5
0
0.5
1
GLONASS frequency number
NovAtel-Trimble
NovAtel-Topcon
NovAtel-JavadD
D o
f L2
inter-ch
ann
el cod
e biases (m
)50 cm
RTK-GPS/GLONASS with Calibrated Biases
( ) ( )
( ) ( )ij
rb
ij
rb
ij
rb
ij
rb
j
b
j
r
i
b
i
r
ij
rb
ij
rb
ij
rb
ij
rb
ij
rb
ij
rb
j
b
j
rj
i
b
i
ri
ij
rb
PPmpPd
ppppP
mpdN
)()()(
)()()(
ερ
φεφφρ
φφλφφλ
+++=
−−−≡
++++=
−−−≡Φ
1.GLONASS double difference equation
2.Compensation by tables of biases
sj
rb
si
rb
ij
rb ddd +−=
d(P) : inter-channel code bias, dφ: inter-channel carrier bias
Assumption on d(P) = d(φ)
RTK Positioning
RTK Processing Options
We evaluate positioning accuracy and fix rate of RTK-GPS/GLONASS
in order to improve the GLONASS positioning performance by tables of bias.
- AR mode : Instantaneous AR (epoch by epoch)
- AR method : LAMBDA, ratio-test (threshold : 3)
- Base length :300 m
(Rover: NovAtel, Javad. Base: NovAtel, Javad, Trimble,Topcon )
- Date : 2009/12/23 24 h (epoch intervals 30 s)
- Elevation mask angle : 15 deg
0 2 4 6 8
x 104
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=9.0% RMSE : E: 0.0037m N: 0.0032m U: 0.0145m
East (m)
North (m)
Up (m)
0 2 4 6 8
x 104
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=25.1% RMSE : E: 0.0035m N: 0.0047m U: 0.0166m
East (m)
North (m)
Up (m)
Positioning Results of RTK with
GLONASS ambiguity fix solution RTK-GPS/GLONASS L1 + L2
Baseline : NovAtel - Trimble
Fix rate 25.1 % Fix rate 9.0 %
No improvement
No correction
Po
siti
on
ing
Err
or
(m)
Time (s)
Correction with tables of biases
Po
siti
on
ing
Err
or
(m)
Time (s)
5 cm 5 cm
Considerations
( ) ( )
( ) ( )ij
rb
ij
rb
ij
rb
ij
rb
j
b
j
r
i
b
i
r
ij
rb
ij
rb
ij
rb
ij
rb
ij
rb
ij
rb
j
b
j
rj
i
b
i
ri
ij
rb
PPmpPd
ppppP
mpPdN
)()()(
)()()(
ερ
φεφρ
φφλφφλ
+++=
−−−≡
++++=
−−−≡Φ
From the next slide, we try to evaluate the positioning performance of RTK-GPS/GLONASS
with GLONASS ambiguity float solution.
1.Estimation of d(ρ) has poor accuracy.
Why can not we fix GLONASS ambiguity by tables of inter-channel code bias ?
2. Assumption on d(ρ) = d(φ) might not be true.
Inter-channel carrier bias can be obtained by subtracting the
GLONASS integer ambiguity fixed in static mode from
DD of carrier-phase measurement.
→ Estimation of dρ is producted by code
measurements.
The variation of inter-channel code and carrier biases
(slot 23 - slot 7)
7500 8000 8500 9000 9500 10000 105000.74
0.75
0.76
0.77
0.78
0.79
0.8
0.81
0.82
Time (s)In
ter-c
han
nel
bia
ses
(m)
Code
Carrier
0 2 4 6 8
x 104
-5
-4
-3
-2
-1
0
1
2
3
4
5RTK POSITION ERROR (FLOAT) : BASELINE=0.3km RATIO=100.0% RMSE : E: 0.2500m N: 0.2927m U: 0.6581m
East (m)
North (m)
Up (m)
0 2 4 6 8
x 104
-5
-4
-3
-2
-1
0
1
2
3
4
5RTK POSITION ERROR (FLOAT) : BASELINE=0.3km RATIO=100.0% RMSE : E: 0.2677m N: 0.4188m U: 1.0192m
East (m)
North (m)
Up (m)
Use of GLONASS Float Ambiguity
We use GLONASS float ambiguity as the aid of GPS ambiguity resolution.
Accuracy of float solution has an effect on the performance of ambiguity fixing.
Float Solution
E/N/U (m) 0. 27/0.42/1.02
Only GPS
Po
siti
on
ing
Err
or
(m)
Time (s)
E/N/U (m) 0.25/0.29/0.65
GPS + GLONASS
Po
siti
on
ing
Err
or
(m)
Time (s)
1 m 1 m
0 2 4 6 8
x 104
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=52.1% RMSE : E: 0.0524m N: 0.0733m U: 0.1120m
East (m)
North (m)
Up (m)
0 2 4 6 8
x 104
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=31.9% RMSE : E: 0.1156m N: 0.1650m U: 0.3184m
East (m)
North (m)
Up (m)
Positioning Results of RTK with GLONASS ambiguity float solution
RTK-GPS/GLONASS L1
Improved by 20 %
No correction
Fix rate : 31.9 %
E/N/U (m) : 0.12 /0.19/0.32
Time (s)
Po
siti
on
ing
Err
or
(m)
Correction with tables of bias
Fix rate : 52.1 %
E/N/U (m) : 0.05/0.07/0.11
Time (s)
Po
siti
on
ing
Err
or
(m)
Baseline : NovAtel - Trimble
5 cm 5 cm
0 2 4 6 8
x 104
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=99.0% RMSE : E: 0.0026m N: 0.0052m U: 0.0169m
East (m)
North (m)
Up (m)
0 2 4 6 8
x 104
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2RTK POSITION ERROR (FIX) : BASELINE=0.3km RATIO=96.7% RMSE : E: 0.0026m N: 0.0051m U: 0.0168m
East (m)
North (m)
Up (m)
Positioning Results of RTK with GLONASS ambiguity float solution
No correction
Fix rate : 96.7 %
E/N/U (m) : 0.01 /0.01/0.02
Correction with tables of bias
Fix rate : 99.0 %
E/N/U (m) : 0.01/0.01/0.02
RTK-GPS/GLONASS L1 + L2
Po
siti
on
ing
Err
or
(m)
Po
siti
on
ing
Err
or
(m)
Time (s) Time (s)
Baseline : NovAtel - Trimble
5 cm 5 cm
Accuracy of RTK positioning with L1 measurements
0.04/0.06/0.140.04/0.06/0.140.09/0.11/0.32
68.8 %68.7 %41.0 %JAVAD - Topcon
0.10/0.11/0.300.17/0.23/0.530.12/0.14/0.32
36.5 %21.9 %35.8 %JAVAD - Trimble
0.11/0.18/0.350.12/0.17/0.400.13/0.21/0.54
33.8 %31.9 %29.9 %NovAtel - JAVAD
0.07/0.09/0.150.10/0.16/0.270.09/0.17/0.39
47.0 %41.5 %45.5 %NovAtel - Topcon
0.05/0.07/0.110.11/0.17/0.310.07/0.09/0.25
52.1 %31.9 %49.5 %NovAtel -Trimble
E/N/U (m)E/N/U (m)E/N/U (m)
Fix rateFix rate Fix rate
GPS + GLONASS
(GLONASS float + bias)
GPS + GLONASS
(GLONASS float)Only GPS
Baseline
300 m
Accuracy of RTK positioning with L1 + L2 measurements
0.01/0.01/0.020.01/0.01/0.020.01/0.01/0.02
98.0%98.0%98.2%JAVAD -Topcon
0.01/0.01/0.020.01/0.01/0.020.01/0.01/0.02
97.0%95.3%98.1%JAVAD -Trimble
0.01/0.01/0.020.01/0.01/0.020.01/0.01/0.02
98.5%98.0%98.2%NovAtel - JAVAD
0.01/0.01/0.020.01/0.01/0.020.01/0.01/0.02
98.9%96.4%98.9%NovAtel - Topcon
0.01/0.01/0.020.01/0.01/0.020.01/0.01/0.02
99.0%96.7%98.9%NovAtel - Trimble
E/N/U (m)E/N/U (m)E/N/U (m)
Fix rateFix rateFix rate
GPS + GLONASS
(GLONASS float + bias)
GPS + GLONASS
(GLONASS float)Only GPS
Baseline
300 m
Summary and Conclusions
• We analysis and improve the performance
degradation on GLONASS AR epoch by epoch.
• We can obtain tables of inter-channel biases
with zero baseline test.
• Tables of inter-channel biases are effective in
GLONASS ambiguity float solution.
• More precise calibration method of the biases is
required for GLONASS Fixed ambiguity.
Future work
• Complete tables of inter-channel carrier bias as
well as more precise code bias calibration
• Evaluate the difference between carrier and
code bias
7500 8000 8500 9000 9500 10000 105000.74
0.75
0.76
0.77
0.78
0.79
0.8
0.81
0.82
0.83
0.84
Time (s)
Inte
r-chann
el b
iase
s (m
)
Code
Carrier