Navigation Accuracy and Interference Rejection for …...Navigation Accuracy and Interference...
Transcript of Navigation Accuracy and Interference Rejection for …...Navigation Accuracy and Interference...
Navigation Accuracy and Interference Rejectionfor an Adaptive GPS Antenna Array
David S. De Lorenzo, Jason Rife, Per EngeStanford University GPS Lab
Dennis AkosUniversity of Colorado
05 June 2006
The authors gratefully acknowledge the support of the JPALS Program Office, andthe Naval Air Warfare Center Aircraft Division through contract N00421-01-C-0022.
6/13/2006 Stanford University JPALS Research 2
Constrained Adaptive Processing
• JPALS has stringent limits on pseudorange & carrier-phase errors– Required to meet carrier-landing accuracy limits.
• GPS antennas introduce distortion on received signals– Apparent as deterministic pseudorange and carrier phase biases– Dependent on the incoming signal line-of-sight.
• JPALS will likely require multi-element STAP algorithms to improve C/No under jamming & RFI conditions.
• STAP algorithms can introduce additionalpseudorange and carrier phase biases.
Goal: Through temporal & spatialconstraints on a STAP algorithm,
move down & left in the trade space,to a point where deterministic
pseudorange & carrier phase correctionsbased on signal LOS can be applied.
6/13/2006 Stanford University JPALS Research 3
Outline
• Software tools for signal simulation & tracking– Signal simulator
• C/A or P-code signal generator that includes antenna distortion effects– Software receiver
• Tracks a wide range of GNSS signals – GPS C/A & P-code, Galileo L1 & E6
• Includes multi-antenna and space-time adaptive signal processing
• Verification of software receiver & multi-signal tracking performance
• Research methodology:– Characterize biases vs. RFI rejection– Evaluate different compensation schemes
• Preliminary results– FRPA vs. CRPA biases and interference rejection
6/13/2006 Stanford University JPALS Research 4
Stanford HW & SW Development
Analog front-end4-channel GP2015 with10 MHz common clock
Data collection computer(2) ICS-650 cards
sampling at 5-65 MHzand 12-bit A/D resolution
Software Receiverall-in view, multi-signal
GNSS receiverwith STAP array
processing
7-element arrayrectangular patch
antennas manufacturedat Stanford University
2m high-gain dish* Possibility of including data from OSU’s
multi-element antenna array.
HP VectorSignal
Analyzer
6/13/2006 Stanford University JPALS Research 5
Software-Based Signal Simulator
Bandlimited WGN& CW interference
C/A & “P” code1.023 & 10.23 Mchip/sec
Front-endA/D
Front-endA/D
…
…
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~19MHz
1450 1500 1550 1600 1650 1700-50
0
50
100
150
200
250
Frequency (MHz)
Pha
se
Antenna Phase Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~19MHz
• Gain/phase data for each look direction (10 S/V constellation) and antenna (7 rectangular patches)
– Data courtesy Ung-Suok Kim
Gain/phase response for PRN #1[Azimuth = 0°, Elevation = 40°]
6/13/2006 Stanford University JPALS Research 6
Signal Generation at Intermediate Freq.
• Received signal at master element:
• Mixing signal:
• Signal after LPF:
• Received signal at subsidiary element:
• Signal after mixing & LPF:
( ) ( ) ( )[ ]θπττ ++⋅−⋅−⋅ tfftxtDP DLS 12cos2
( ) ( ) ( )( )[ ]θπττ +Δ++⋅−Δ+⋅−Δ+⋅ ttffttxttDP DLS 12cos2
( )[ ]tff IFL −⋅ 12cos2 π
( ) ( ) ( )[ ]θπττ ++⋅−⋅−⋅ tfftxtDP DIFS 2cos
( ) ( ) ( ) ( )[ ]tfftffttxttDP DLDIFS Δ++++⋅−Δ+⋅−Δ+⋅ 122cos πθπττ
6/13/2006 Stanford University JPALS Research 7
Signal Generation w/ Antenna Distortion
s(t) and gain/phase data are antenna & S/V specific
Gain/phase distortion data available from HFSS simulation
– Verified by comparison to anechoic chamber testing
– Data courtesy Ung-Suok Kim
( )ts
( )tsout
fft ( )fS
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
0 5 10 15 20 25 30 35 40Frequency (MHz)
Pow
er/fr
eque
ncy
(dB
/Hz)
Power Spectral Density Estimate via Welch
⊗ 1−fft
-1.5 -1 -0.5 0 0.5 1 1.50
0.5
1
1.5
2
2.5
3
3.5
4 x 104 P-code Correlation Peak Distortion - PRN #1
Code Offset (chips
Cor
rela
tor O
utpu
t
-1.5 -1 -0.5 0 0.5 1 1.50
0.5
1
1.5
2
2.5
3
3.5
4x 104 P-code Correlation Peak w/o Distortion - PRN #1
Code Phase (chips)
Cor
rela
tion
Out
put
Autocorrelation of s(t)
Autocorrelation of sout(t)
1450 1500 1550 1600 1650 1700-50
0
50
100
150
200
250
Frequency (MHz)
Pha
se
Antenna Phase Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
6/13/2006 Stanford University JPALS Research 8
Standard Satellite Constellation
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #10 [Az = 300 deg, El = 80 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)G
ain
Antenna Gain Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #2 [Az = 30 deg, El = 30 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #3 [Az = 60 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #4 [Az = 90 deg, El = 50 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #5 [Az = 120 deg, El = 20 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #6 [Az = 150 deg, El = 60 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #7 [Az = 210 deg, El = 70 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #8 [Az = 240 deg, El = 20 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #9 [Az = 270 deg, El = 30 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
PRN #1
PRN #2
PRN #3
PRN #4PRN #5
PRN #6
PRN #7
PRN #8PRN #9 PRN #10
6/13/2006 Stanford University JPALS Research 9
0 5 10 15 20 25 30 35 40Frequency (MHz)
Pow
er/fr
eque
ncy
(dB
/Hz)
Power Spectral Density Estimate via Welch
Wideband Jammer Signal Generation at IF
( )tj
( )tjout
fft ( )fJ
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
⊗ 1−fft
( ) ( ) ( )[ ]∑ Δ++Δ+′++⋅i
JLJJIFJ tffttftffPii 1
2 22cos2 ππ
Gain/phase distortion data are currently not available ( ) ( )tjtj out≡
• For each jammer – six swept sinusoids cover frequency band from ~0 Hz to ~fs/2
• Amplitude scaled as needed to achieve desired J/S ratio
6/13/2006 Stanford University JPALS Research 10
Satellite & RFI Constellation
8030010
302709
202408
702107
601506
201205
50904
40603
30302
4001
ElevationAzimuthPRN
102706
02505
02254
01203
0452
001
ElevationAzimuthRFI
Bandlimited WGNand/or
CW interference
6/13/2006 Stanford University JPALS Research 11
Multi-signal Software GNSS Receiver
• Software Receiver tracks a wide range of GNSS signals, runs in MATLAB– GPS C/A and pseudo-P-codes– Galileo L1-B/C and E6-B/C codes
• Includes multi-antenna and space-time adaptive signal processing
• Supports arbitrary sampling frequency, intermediate frequency, data format, etc.– The ideal s/w rcvr is fast, flexible, & uncomplicated
You can have 2 out of 3!!
6/13/2006 Stanford University JPALS Research 12
Software Receiver Block Diagram
WeightControl
Algorithm
Correlatorsintegrate& dump
TrackingLoops
CarrierNCO
CodeNCO
cos
C/No Estimator
Pseudorange & Carrier-phaseBias Estimator
carrierwipeoff
code wipeoff
Early
Prompt
LateFFT-basedCoherent
Acquisition
sin
Weight control processing supportsFRPA & CRPA antennas and
adaptive LMS & Applebaum arrays
6/13/2006 Stanford University JPALS Research 13
Software Receiver Flow Chart
For each tracking channel:1ms sample buffer alignedw/ start of spreading code
Code/carrier phase alignedw/ first sample of 1ms epoch
Discrete operation from stored data sample indexing & alignment
Noncoherent or coherentintegration depends onnav. data bit alignment
This loop runs ona 1ms schedule
The devil isin the details
6/13/2006 Stanford University JPALS Research 14
Epoch-based Tracking of GNSS Codes
Code start tracked ineach ~1ms data epoch
Problems:– If fs is not an exact multiple of
1000Hz, the extra fractional sample causes a code-phase ramp error
– Approaching/receding satellites will cause code wrap (phase skew), leading to dropped sample epochs
PRN1
PRN2
PRN3
Frame-based data epochs
Multi-epoch tracking window
Code start tracked in the~5ms active tracking window
Active window
Active window
PRN1
PRN2
PRN3
Solutions:– Sample index for each
tracked PRN aligns with 0 code phase
– Logic is used to control sample index over-run or under-run
6/13/2006 Stanford University JPALS Research 15
S/W Rcvr Verification:GPS C/A Code Tracking – Real Data
• All-in-view tracking of GPS C/A code signals• Software
receiver is acombination ofin-house codeand structuresfrom Kai Borreof AalborgUniversity
0 500 1000 1500 2000 2500-0.2
0
0.2
0.4
0.6
0.8
1
Navigation Data Message - PRN 3
Time (ms)
Data source: NordNav Rcvr & Dennis Akos
0 500 1000 1500 2000 2500
1700
1750
1800
Doppler Frequency
Time (ms)
Dop
pler
freq
uenc
y (H
z)
0 500 1000 1500 2000 2500211
211.5
212
212.5
213
213.5
214
214.5Position of the C/A Code Start
Time (ms)C
/A c
ode
star
t (ch
ips)
0 500 1000 1500 2000 250030
35
40
45
50
55
60Filtered C/No
Time (ms)
C/N
o (d
B-H
z)Tracking mode transitions
6/13/2006 Stanford University JPALS Research 16
S/W Rcvr Verification:C/A Code Acquisition – High-Gain Data
• Acquisition of GPS C/A code signal collected with SRI Dish– fS = 124.5 MHz– fIF = 43.08 MHz– Data courtesy
Grace Gao ofthe StanfordGPS Lab
Why is thecorrelation
peak rounded? 106.5 107 107.5 108 108.5 109 109.50
0.2
0.4
0.6
0.8
1
C/A Code Correlation - PRN #8 for 1ms
Code Phase (chips)
Cor
rela
tion
Func
tion
(nor
mal
ized
)
6/13/2006 Stanford University JPALS Research 17
106.5 107 107.5 108 108.5 109 109.50
0.2
0.4
0.6
0.8
1
C/A Code Correlation - PRN #8 for 1ms
Code Phase (chips)
Cor
rela
tion
Func
tion
(nor
mal
ized
)
S/W Rcvr Verification:C/A Code Acquisition – High-Gain Data
• Acquisition of GPS C/A code signal collected with SRI Dish– fS = 124.5 MHz– fIF = 43.08 MHz– Data courtesy
Grace Gao ofthe StanfordGPS Lab
A band-passfilter was appliedto the data before
processing.
6/13/2006 Stanford University JPALS Research 18
105.5 106 106.5 107 107.5 108 108.5 109 109.50
0.2
0.4
0.6
0.8
1
C/A Code Correlation - PRN #8 for 1ms
Code Phase (chips)
Cor
rela
tion
Func
tion
(nor
mal
ized
)
S/W Rcvr Verification:C/A Code Acquisition – High-Gain Data
• Acquisition of GPS C/A code signal collected with SRI Dish– fS = 124.5 MHz– fIF = 43.08 MHz– Data courtesy
Grace Gao ofthe StanfordGPS Lab
This is thecorrelation
to unfiltereddata.
6/13/2006 Stanford University JPALS Research 19
-400 -200 0 200 400 600 800 1000 1200 1400 16000
0.2
0.4
0.6
0.8
1
Code Correlation - Galileo L1/B for 1ms
Code Phase (chips)
Cor
rela
tion
Func
tion
(nor
mal
ized
)
Code CorrelationGalileo L1/B for 1msPeak in 1st acq. blockDoppler= 2200 HzCode Ph.= 610.4 chipsCPPR= 11.7CPPM= 332.3
– Galileo L1-B/C and E6-B/C codes determined by Grace Gao of the Stanford GPS Lab
S/W Rcvr Verification:Galileo L1/B Code Acquisition – Real Data
608 608.5 609 609.5 610 610.5 611 611.5 612 612.5 6130
0.2
0.4
0.6
0.8
1
C/A Code Correlation - PRN #1 for 1ms
Code Phase (chips)
Cor
rela
tion
Func
tion
(nor
mal
ized
)
6/13/2006 Stanford University JPALS Research 20
S/W Rcvr Verification:Galileo L1/C Code Tracking – Real Data
• Tracking of Galileo L1/C BOC(1,1) pilot signal• No modification
to s/w rcvrexcept newcode generators
• GalileoL1-B/C andE6-B/C codesdetermined byGrace Gao ofthe StanfordGPS Lab
0 200 400 600 800 1000 1200 1400 1600 1800-0.2
0
0.2
0.4
0.6
0.8
1
Navigation Data Message - Galileo L1/C
Time (ms)
Data source: Galileo L1 w/ SRI Dish & VSA by Grace Gao
0 500 1000 1500
2140
2160
2180
2200
2220
2240
2260
Doppler Frequency
Time (ms)
Dop
pler
freq
uenc
y (H
z)
0 500 1000 1500604
605
606
607
608
609
610
611Position of the Galileo Code Start
Time (ms)
Gal
ileo
code
sta
rt (c
hips
)
200ms secondary code
6/13/2006 Stanford University JPALS Research 21
S/W Rcvr Verification:GPS P-Code Tracking – Simulated Data
• GPS pseudo-P-code signals, 10.23 Mchips/sec• Signals have
sufficient BWto elucidateeffects due todifferentialantenna gainand phaseresponses– fS = 80 MHz– fIF = 20 MHz
0 500 1000 1500 2000-0.2
0
0.2
0.4
0.6
0.8
1
Navigation Data Message - PRN 6
Time (ms)
Data source: Multi-Antenna Signal Simulator
0 500 1000 1500 2000
160
180
200
220
240
260
280
Doppler Frequency
Time (ms)
Dop
pler
freq
uenc
y (H
z)
0 500 1000 1500 20005444.35
5444.4
5444.45
5444.5
5444.55Position of the P-Code Start
Time (ms)
P-c
ode
star
t (ch
ips)
0 500 1000 1500 200030
35
40
45
50
55
60Filtered C/No
Time (ms)
C/N
o (d
B-H
z)
6/13/2006 Stanford University JPALS Research 22
S/W Rcvr Verification:CRPA Processing – Simulated Data
• 7-element CRPA with gradually decreasing C/No• No special
receiverprocessingfor low C/No
• Loss-of-lockbelow C/Noof ~25 dB-Hz
0 2000 4000 6000 8000-90
-45
0
45
90PLL Discriminator
Time (ms)
Pha
se o
ffset
(deg
rees
)
0 2000 4000 6000 8000-500
0
500DLL Discriminator
Time (ms)
Cod
e of
fset
(m)
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
10
20
30
40
50
60Filtered C/No
Time (ms)
C/N
o (d
B-H
z)
FRPACRPA
Increasing noise power
Loss-of-lock
6/13/2006 Stanford University JPALS Research 23
Navigation Errors due to Multi-Antenna Arrays
Common-mode: errors areshared by all signals on aparticular antenna channel
Non-common-mode: errorsare a function of satellite #
az/el-dependent(also present for single-antenna GPS)
Do errors introduce a pseudorange or carrier-phase bias? What RFI or noise
rejection performance is attainable?Algorithm: errors may beintroduced by the weightcontrol algorithm itself
LO
Σ TrackingLoops PVT
code replica
IMUattitude
ephemeris
WeightControl
Algorithm
tracking reference
carrier replica
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
LO
Σ TrackingLoops PVT
code replica
IMUattitude
ephemeris
WeightControl
Algorithm
tracking reference
Σ TrackingLoops PVT
code replica
IMUattitude
ephemeris
WeightControl
Algorithm
tracking reference
carrier replica
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&QHatke, 1998Fante & Vaccaro, 2000Gupta & Moore, 2001Moore, 2002Fante, et.al., 2004McGraw, et.al., 2004Hatke & Phuong, 2004Kim, et.al., 2004Kim, 2005
Pseudorange/Carrier-phase errors
6/13/2006 Stanford University JPALS Research 24
Methodology – Biases vs. RFI Rejection
• Develop adaptive algorithms that balance noise/RFI rejection against pseudorange & carrier-phase errors – through constraints on weight adjustment– End-to-end: from signal reception through the receiver
tracking loops to the pseudorange and carrier phase estimates
Goal: Through temporal & spatialconstraints on a STAP algorithm,
move down & left in the trade space,to a point where deterministic
pseudorange & carrier phase correctionsbased on signal LOS can be applied.
6/13/2006 Stanford University JPALS Research 25
-1.5 -1 -0.5 0 0.5 1 1.50
1
2
3
4
5
6
7x 104
Code Offset (chips)
Cor
rela
tion
Out
put
PRN #1PRN #2PRN #3PRN #4PRN #5PRN #6PRN #7PRN #8PRN #9PRN #10
0 200 400 600 800 1000
-6
-4
-2
0
2
4
6
Time (ms)
Car
rier P
hase
Err
or (c
m)
Carrier Phase Estimates vs. Truth Values
PRN 1PRN 2PRN 3PRN 4PRN 5PRN 6PRN 7PRN 8PRN 9PRN 10PRN 11
Preliminary Results:GPS P-Code w/ Antenna Distortion
• Simulated data and complete visibility of tracking output allow comparison between predicted and true pseudorange and carrier phase values
Carrier phase biases Pseudorange bias – distortionof P-code correlation peak
Test satellite – zero bias
6/13/2006 Stanford University JPALS Research 2635 40 45 50 55 60
-6
-4
-2
0
2
4
6
C/No (dB-Hz)
Car
rier P
hase
Erro
r (cm
)C
arrie
r-ph
ase
Erro
rs (c
m)
35 40 45 50 55 60-5
-4
-3
-2
-1
0
1
C/No (dB-Hz)
Cod
e P
hase
Erro
r (m
)Ps
eudo
rang
e Er
rors
(m)
Preliminary Results:FRPA vs. CRPA Processing – no RFI
• Code/carrier errors are not affected by deterministic array processing
• C/No enjoys healthy boost under these conditions
0 200 400 600 800 1000 1200 1400 1600 1800 200035
40
45
50
55
60
Time (ms)
C/N
o (d
B-H
z)
Filtered C/No
PRN 1PRN 2PRN 3PRN 4PRN 5PRN 6PRN 7PRN 8
FRPA CRPA
6/13/2006 Stanford University JPALS Research 27
Future Work – Exploring the Trade Space
1. Simulate signals with desired characteristics (C/No, J/S ratio, antenna distortion, etc.)
2. Track with Software Receiver (FRPA, CRPA, LMS, Applebaum processing)
3. Characterize C/No vs. pseudorange and carrier phase biases for various scenarios
4. Apply compensation schemes
Spatial Constraints
Temporal Constraints
DistortionC/NoJ/S level
ApplebaumLMSCRPAFRPA
6/13/2006 Stanford University JPALS Research 28
Conclusions
• Signal simulator captures parameters of interest– Broadband RFI, antenna gain/phase variation based on
detailed characterization models
• Software receiver supports adaptivealgorithms and estimation ofpseudorange & carrier phase errors
• Methodology is yielding data abouttradeoffs between RFI rejection andcode/carrier estimation errors
• Tools will support implementationand testing of antenna compensation schemes
Backup
6/13/2006 Stanford University JPALS Research 30
0 5 10 15 20 25 30 35 40Frequency (MHz)
Pow
er/fr
eque
ncy
(dB
/Hz)
Power Spectral Density Estimate via Welch
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1450 1500 1550 1600 1650 1700-50
0
50
100
150
200
250
Frequency (MHz)
Pha
se
Antenna Phase Response for PRN #1 [Az = 0 deg, El = 40 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #8 [Az = 240 deg, El = 20 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1450 1500 1550 1600 1650 170080
100
120
140
160
180
200
220
240
260
280
Frequency (MHz)
Pha
se
Antenna Phase Response for PRN #8 [Az = 240 deg, El = 20 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1480 1500 1520 1540 1560 1580 1600 1620 1640 1660 16800
0.2
0.4
0.6
0.8
1
1.2
Frequency (MHz)
Gai
n
Antenna Gain Response for PRN #10 [Az = 300 deg, El = 80 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
1450 1500 1550 1600 1650 1700-60
-40
-20
0
20
40
60
80
100
120
140
Frequency (MHz)
Pha
se
Antenna Phase Response for PRN #10 [Az = 300 deg, El = 80 deg]
Antenna #1Antenna #2Antenna #3Antenna #4Antenna #5Antenna #6Antenna #7
Front-end BW ~40MHz
PRN #1
PRN #8
PRN #10
Antenna Distortion Effects (2)
• Fourier transform of complex signal multiplied by antenna response characteristics
– Gain/phase data re-centered at IF
• Inverse Fourier transform back to time domain
• Real-valued samples stored to disk
• Gain/phase distortion data available from HFSS simulation
– Verified by comparison to anechoic chamber testing
– Data courtesy Ung-Suok Kim
6/13/2006 Stanford University JPALS Research 31
Wideband Jammer Signal Generation at IF
• For each jammer – six swept sinusoids cover frequency band from ~0 Hz to ~fs/2– Pseudo-random
center frequencies– For fs = 80 MHz and
fIF = 20 MHz, eachwave covers 10 MHz
• Amplitude scaled asneeded to achievedesired J/S ratio
( ) ( ) ( )[ ]∑ Δ++Δ+′++⋅i
JLJJIFJ tffttftffPii 1
2 22cos2 ππ
0 5 10 15 20 25 30 35 40Frequency (MHz)
Pow
er/fr
eque
ncy
(dB
/Hz)
Power Spectral Density Estimate via Welch
6/13/2006 Stanford University JPALS Research 32
Spatial & Temporal D.O.F
• Spatial: geometry of gain pattern, width (selectivity) of beams/nulls
• Temporal: spectral response of array, nulling of wideband signals
LMS WeightAdjust Circuit
Δw11
Δw21
s(t)
r(t) - LMS referenceΣ+
–
Σ
w12
w22
Σ TrackingLoops
replica
ε
Spat
ial
Temporal
from Widrow & Stearns, Adaptive Signal Processing (1985)
6/13/2006 Stanford University JPALS Research 33
slave
master
SamplingClock
SynchronizationCables
PCI Bus
IDEHD
NovAtelSuperStar
Analog line
Host PC
Serial cable
NovAtelSuperStar
power/ground wiresand buffer circuitsleft off for clarity
InterferenceGenerator
4 Antennas4 SuperStars
2 ICS-650 cardsCommon Clock
NovAtelSuperStar
GPS Display
ICS-650
ICS-650
NovAtelSuperStar
10 MHzClock
Analog line
Analog line
Analog line
ICS-65012-bit A/D5-65 MHz2-channel
SuperStar IIZarlink GP20154.309 MHz IF
4-Channel Test Hardware – Schematic
6/13/2006 Stanford University JPALS Research 34
Proposed 1st-gen Hardware Implementation
LO
Σ TrackingLoops
PVT
code replica
AdaptiveWeight
Algorithm
LMS reference
carrier replica
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
S/V-specific signal processing
LMS-based AdaptiveBlind Beamforming
Multi-Element GPS ASIC
6/13/2006 Stanford University JPALS Research 35
Software Receiver Block Diagram
WeightControl
Algorithm
Correlatorsintegrate& dump
TrackingLoops
CarrierNCO
CodeNCO
cos
C/No Estimator
Pseudorange & Carrier-phaseBias Estimator
carrierwipeoff
code wipeoff
Early
Prompt
LateFFT-basedCoherent
Acquisition
sin
Let’s focus on thisblock for a moment.
6/13/2006 Stanford University JPALS Research 36
Generic Adaptive Antenna Array
from Compton, Adaptive Antennas (1988)
Σ
FeedbackControl
PerformanceIndex
ArrayOutput
Gain and PhaseAdjustment
Maximize SINR at the array output(e.g., Applebaum, 1976; Frost, 1972)
Minimize MSE between the actualarray output and the ideal array output
(e.g., Widrow, et. al., 1967)
Optimization Criteria
• Using feedbackto optimize someperformance index
6/13/2006 Stanford University JPALS Research 37
Σ TrackingLoops PVT
code replica
IMUattitude
ephemeris
WeightControl
Algorithm
replicated signal
carrier replica
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Adaptive Array Processing for GPS
covarianceestimate
6/13/2006 Stanford University JPALS Research 38
Σ TrackingLoops PVT
code replica
IMUattitude
ephemeris
WeightControl
Algorithm
replicated signal
carrier replica
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Adaptive Array Processing for GPS
Deterministic CRPA: signal arrivaland platform/array orientation
Adaptive w/ maximum SINR, e.g. Applebaum: signalarrival, platform/array orientation, and SINR metric
Adaptive w/ minimum MSE, e.g. Widrow LMS:ideal array output reference signal and SINR metric
covarianceestimate
e.g., Hatke, 1998; Nicholson, et al., 1998; Fante & Vaccaro, 2000;Gupta & Moore, 2001;Hatke & Phuong, 2004; Fante, et al., 2004
e.g., Gecan & Zoltowski, 1995
6/13/2006 Stanford University JPALS Research 39
Σ TrackingLoops PVT
code replica
IMUattitude
ephemeris
WeightControl
Algorithm
replicated signal
carrier replica
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Front-end A/D I&Q
Adaptive Array Processing for GPS
But what happens when there are
errors introduced by the hardware
that are not completely mitigated
by software or calibration?
Antennaphase-center
and groupdelay effects
Front-endfilter and
timingeffects