Page 1
© 2014, Frank van Diggelen
Consumer GNSS Receiver Design & comparison with ionospheric scintillation studies
Reference: Chapters 2,3 of: “A-GPS; Assisted GPS, GNSS & SBAS”, van Diggelen. Chapters 11,12 of: “Global Positioning System”, Misra & Enge
African School on Space Science
Page 2
© 2014, Frank van Diggelen
GPS (Civilian) Signal at the Satellite
~ 1575.42 MHz
C/A Code 1 Mbps
Data 50 bps
PRN Code
Data includes: Almanac, Ephemeris, HOW
BPSK signal
Page 3
© 2014, Frank van Diggelen
Standard GPS receiver architecture
fIF+fD Correlate
fIF+fD
+ noise
fIF+fD - fe
Σ
integrate received PRN code
locally generated copies of PRN code
corr
elat
ion
peak
NCO
+ noise
RF Front End
IF
~ local oscillator
Page 4
© 2014, Frank van Diggelen
fIF+fD Correlate
fIF+fD
+ noise
code delay
fIF+fD - fe
Σintegrate
received PRN code
locally generated copy of PRN code
corr
elat
ion
peak
NCO fIF+fD Correlate
fIF+fD
+ noise
code delay
fIF+fD - fe
Σ integrate
received PRN code
locally generated copy of PRN code
corr
elat
ion
peak
NCO fIF+fD Correlate
fIF+fD
+ noise
code delay
fIF+fD - fe
Σ integrate
received PRN code
locally generated copy of PRN code
corr
elat
ion
peak
NCO
+ noise
RF Front End
IF
~ local oscillator
BASEBAND BLOCK REPEATED ONCE PER CHANNEL
Standard GPS receiver architecture
Page 5
© 2014, Frank van Diggelen
Tri-band front end
5
1561.098 1575.42 1602 A/D
A/D
A/D Ban
d S
epar
atio
n Fi
lters
To B
aseb
and
Pro
cess
ing Gain Control
and Filters
Gain Control and Filters
Gain Control and Filters
Page 6
© 2014, Frank van Diggelen
Search space
Page 7
© 2014, Frank van Diggelen
Acquisition space review:
Real-time animation of standard GPS search of freq/code space. Click picture to play
Background
Page 8
© 2014, Frank van Diggelen
Search Engine Evolution (1) Correlator 1
Correlator 2
Correlator 3
Correlator n
Results
Samples
Gen 1
Correlators 1-4
Results
Samples
Gen 2
Correlators 5-8
Correlators 9-12
Correlators n-m
Correlators 1-2046
Results
Samples
Gen 3
Correlators 2047-4092
Correlators 4092-6138
Correlators n-m
Results
Samples Sample Storage
FFTMultiply
IFFT
Gen 4
Matched Filter Processing FFT Processing
Processing ca. 1993
Broadcom
Page 9
© 2014, Frank van Diggelen
Search Engine Evolution (2)
One search bin
Coarse-Time Acquisition Sensitivity (@ fixed TTFA of 10s) vs. number of code-epoch bins
* With A-GPS assistance data: ± 100 ppb frequency, ± 2 s time, ± 3km position, ephemeris
10-3 10-2 10-1 1 10 100
-130 dBm
-140 dBm
-150 dBm
-160 dBm
Number of full code-epoch bins that can be searched in parallel
these are actual receivers built over the last 20 years
Magellan
SiRF
Global Locate
Broadcom
1992
1999
2002
2012
Broadcom
Page 10
© 2014, Frank van Diggelen
Broadcom GPS
1) Search all code delays simultaneously 2) Search all over 100 bins in parallel
One “bin”
Broadcom
Page 11
© 2014, Frank van Diggelen
8.4 kHz (satellite motion)
0.15 kHz / 100 km/h (receiver speed)
1.5 kHz/ppm (oscillator)
0.1 kHz / 100km (init. position)
0.0008 kHz / s (init. Time)
Contributors to frequency offset
Dependence on oscillator stability Does this mean the consumer market will lead to better oscillators?
Page 12
© 2014, Frank van Diggelen
Reminder of receiver design
12
fIF+fD Correlate
fIF+fD
+ noise
code delay
fIF+fD - fe
Σintegrate
received PRN code
locally generated copy of PRN code
corr
elat
ion
peak
NCO fIF+fD Correlate
fIF+fD
+ noise
code delay
fIF+fD - fe
Σ integrate
received PRN code
locally generated copy of PRN code
corr
elat
ion
peak
NCO fIF+fD Correlate
fIF+fD
+ noise
code delay
fIF+fD - fe
Σ integrate
received PRN code
locally generated copy of PRN code
corr
elat
ion
peak
NCO
+ noise
RF Front End
IF
~ local oscillator
BASEBAND BLOCK REPEATED ONCE PER CHANNEL
Page 13
© 2014, Frank van Diggelen
Assisted GNSS (1)
Satellite nav data from the internet
Page 14
© 2014, Frank van Diggelen
Assisted GNSS (2)
Location Server
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
frequency (kHz)
code-delay (chips)
Location Server
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
Location ServerLocation Server
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
frequency (kHz)
code-delay (chips)
Reduced search space ⇒ quicker acquisition ⇒ higher sensitivity
Page 15
© 2014, Frank van Diggelen
Long Term Orbits (LTO) (aka Extended Ephemeris)
15
Location Server
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
frequency (kHz)
code-delay (chips)
Location Server
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
Location ServerLocation Server
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
Assistance:Acquisition assistAlmanacEphemerisFrequencyTimePosition
reduce search space: frequency
code delay
frequency (kHz)
code-delay (chips)
Ephemeris is calculated for many days into the future
Page 16
© 2014, Frank van Diggelen
North Korea: 6
Broadcom, LTO Server, Unique Android Visitors, in 24 hours
Tuvalu: 1
USA: 37M China: 80M
South Korea: 1.6M
Brazil: 36M
UK: 27M
Vatican City: 12
Greenland: 388
South Africa: 0.5M
Russia: 1.1M
Spain: 0.9M
Page 17
© 2014, Frank van Diggelen South Africa: 484,000
Lesotho: 150
Swaziland: 300
Reunion: 14,000
Mauritius: 14,000
Madagascar: 2,000
Mayotte: 400 Comoros: 90
Seychelles: 700
Botswana: 2,000
Namibia: 17,000
Zimbabwe: 3,000
Above 1,000 rounded to nearest 1,000 Below 1,000 rounded to 50 Below 100 rounded to 10
Mozambique: 5,000
Malawi: 1,000 Zambia: 1,000 Angola: 17,000
Tanzania: 4,000
Kenya: 62,000
Somalia: 200
Djibouti: 150
Eritrea: 10
Egypt: 454,000
Ghana: 6,000 Togo: 150
DRC: 1,000 Congo: 300
Gabon: 1,000 Equatorial Guinea:
Sudan: 57,000
Libya: 21,000
Tunisia: 38,000 Morocco: 136,000
Algeria: 40,000
Mauritania: 1,000
Cape Verde: 450
Senegal: 14,000 Gambia: 650
Guinea Bissau: 100 Guinea : 350
Liberia: 200 Cote Divoire: 9,000
Sierra Leone: 300
Mali: 2,000 Niger: 750
Burkina: 2,000
CAR: 80
Chad: 20
Uganda: 1,000
Rwanda: 300
Burundi: 100
Ethiopia: 1,000
St Tome & Principe: 80
Cameroon: 2,000 Nigeria: 18,000
Benin: 500
Broadcom, LTO Server, Unique Android Visitors,
24 hours
Page 18
© 2014, Frank van Diggelen
Back to search space with A-GNSS
8.4 kHz (satellite motion)
0.15 kHz / 100 km/h (receiver speed)
1.5 kHz/ppm (oscillator)
0.1 kHz / 100km (init. position)
0.0008 kHz / s (init. Time)
Page 19
© 2014, Frank van Diggelen
Frequency assistance Cell towers have oscillators that are known to ±100ppb
A cell-phone communicating with a tower can calibrate its internal oscillator to ±100ppb
Page 20
© 2014, Frank van Diggelen
8.4 kHz (satellite motion)
0.15 kHz / 100 km/h (receiver speed)
1.5 kHz/ppm (oscillator)
0.1 kHz / 100km (init. position)
0.0008 kHz / s (init. Time)
± 0.15 kHz/100ppb
Result: remaining search space is a fraction of a kHz, easily within the capabilities of modern receivers. And so the trend is towards worse (= cheaper) oscillators in consumer products L.
Back to search space with A-GNSS
Page 21
© 2014, Frank van Diggelen
OSCILLATORS & IONO ...
Page 22
© 2014, Frank van Diggelen
Studying ionospheric scintillation
“Crystal Oscillator Noise Effects on the Measurement of Ionospheric Phase Scintillation Using GPS”, A.J. Van Dierendonck & Quyen Hua IEEE Frequency Control Symposium. May 1998
Measuring phase scintillation: must remove effects of receiver oscillator Frequency jumps are not tolerable:
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
175000 180000 185000 190000 195000 200000 205000
GPS Time-Of-Week - Seconds
Sigm
a Ph
i - R
adia
ns
-120
-115
-110
-105
-100
-95
-90
-85
-80
-75
-70
Spec
tral
Pow
er @
20.
473
MHz
- dB
ph 1ph3ph10ph30ph60Phase Density @ 1 Hz OffsetThermal Noise Contribution
Feb 10 1998
Frequency Jumps
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
373200 376800 380400 384000 387600 391200 394800
GPS Time Of Week - Seconds
Sigm
a Ph
i - R
adia
ns
-120
-115
-110
-105
-100
-95
-90
-85
-80
-75
-70
Spec
tral
Pow
er @
10
MH
z - d
B
ph 1ph3ph10ph30ph60Spectral Density @ 1 HzThermal Noise Contribution Loss of Lock
SV 1 - Feb 97
Strong Multipath
20 MHz OCXO (Bad) 10 MHz OCXO (Good)
Conclusion: higher frequency OCXO showed jumps of the order of 1 rad/s in measured phase ≈ 0.1 ppb
Page 23
© 2014, Frank van Diggelen
Oscillator summary
ppb 0.1 1
Typical frequency jumps in different types of oscillators
10
OCXO ~$100
TCXO ~$0.50
TSX ~$0.25 cost (USD)
Summary: for consumer products to measure iono scintillation effect on phase you would need (at least) to change the crystal oscillator.
Page 24
© 2014, Frank van Diggelen
Measuring scintillation using observed C/No
24
Page 25
© 2014, Frank van Diggelen
MEASURING TEC ...
Page 26
© 2014, Frank van Diggelen
State of the art, and trends
• Current consumer GNSS is multi-frequency, but across different systems (therefore different satellite clocks)
• However, the trend is towards L1 and L5 • In the next decade you may see consumer products
measuring multi GNSS systems on dual frequencies (L1, L5)
26
Page 27
© 2014, Frank van Diggelen
Summary
• You’ve seen consumer GNSS designs and trends – half for your general knowledge – half relevant to your work
• Consumer products have some (small) overlap with GNSS for space science today
• And may be quite useful in years to come
27