Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
- 1 -
- 1 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Positioning using Signals-of Opportunity based on
OFDM
Olivier Julien, P. Thevenon (now CNES), D. Serant
Ecole Nationale de l’Aviation Civile, Toulouse, France, www.enac.fr
Funding from:
Colloquium on Satellite Navigation
TU München, May 23rd, 2011
- 2 -
- 2 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
ENAC Field
Short Presentation of ENAC
- 3 -
- 3 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
ENAC Key Figures
Initial Training: 1 700 students enrolled in 15 different cursus
(engineer/Master, technicians, air traffic controllers, pilots,
dispatchers)
6 "Mastère Specialisé" degrees (1 year) + 3 in China
2 Master of Science degrees (2 years)
Lifelong training: 5 000 attendants
Research and Innovation
International
– 6 000 foreign students trained
– 35 exchange agreements with universities
Short Presentation of ENAC
- 4 -
- 4 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
ENAC Research and Innovation
Short Presentation of ENAC
- 5 -
- 5 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
LTST (Signal Processing and Telecom Laboratory):
Human Resources (as of May 2011)
6 permanent staffs and 7 Ph.D. students
Fields of Research
– GNSS for Civil Aviation (integrity, augmentation, signal processing)
– Urban Navigation (high sensitivity, hybridization, integrity, SoO, precise
positioning)
– Precise positioning (RTK, PPP)
– GNSS Signal Design (satellite payload, modulation, navigation message)
– GNSS Software Receiver
Key elements
15 Ph.D. thesis defended, 3 post-docs
50+ expertise contracts, 100+ papers published
4 collaborative patents
ENAC Research in GNSS
Short Presentation of ENAC
- 6 -
- 6 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Context
Urban & indoor positioning represents a huge market:
– Regulatory incentives (E911)
– Convergence of telecommunication and localization services
– Military application
It is known that urban and indoor environment is challenging
for GNSS because of interference, signal blockage, multipath,
etc…
To deal with this challenge, specific GNSS developments have
been done (high-sensitivity, Aided-GNSS, system-level GNSS
upgrades). However, they only provide limited position
availability, accuracy, continuity in challenging environments
- 7 -
- 7 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Context
Alternative solutions are linked to the use of systems/signals
that are complementary to GNSS:
– Other navigation sensors: Inertial sensors, magnetometers, Wheel
Speed Sensors, laser, video cameras, …
– Dedicated radio-location systems: pseudolites, RFID, UWB
– Systems of signals of opportunity (SoO) that are not meant for
positioning a priori: Mobile telephony (2G or 3G), TV, Radio, WiFi signals
SoO have several advantages, even if they are not meant
primarily for navigation:
– Availability in urban centers
– Plurality of potential systems
– Integration with telecommunication services
The presentation will look at a subset of SoOs that are based
on OFDM modulation
- 8 -
- 8 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Presentation Outline
1. Introduction to OFDM and test signals
2. Propagation channel-related issues for positioning
3. Positioning principle
4. Proposed pseudo-range estimation algorithms
5. Results on pseudo-range estimation
6. Results on positioning
7. Conclusion and future work
8. Publications on this topic
- 9 -
- 9 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Frequency selectivity of multipath channel causes distortions
that degrade a wideband transmission performance
OFDM solution:– Transmit symbols on narrow-band orthogonal sub-carriers,
where channel distortion can be easily corrected.
– 1 symbol per sub-carrier
– Implemented by iFFT / FFT in DSP
– A guard interval called Cyclic Prefix (CP) is
introduced to avoid Inter-Symbol Interferences and
allows demodulation in loose synchronization
conditions. The CP is the replica of the end of the
OFDM symbol
OFDM Principle
0 200 400 600 800 1000 12000
0.2
0.4
0.6
0.8
1
1.2
1.4
Spectrum of orthogonal subcarriers
OFDM symbol k OFDM symbol k+1 CP CP …
CPN samples CPN samples
0 200 400 600 800 1000 12000
1
2
3
4
5
Channel effect on the subcarriers
Multipath channel
spectrum
1- Introduction to OFDM and Test
Signals
- 10 -
- 10 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Modulator
Demodulator
OFDM Modulator and Demodulator
ns
pd
ic
ic : Tx symbolsns
)(nciFFTs in
: Tx samples
Serial
to
Parallel
IFFT
Parallel
to
Serial
Modulator
Add CP
Parallel
to
Serial
FFT
Serial
to
Parallel
Demodulator
Remove
CP
nr
nr : Rx samplespd
)( prFFTd np
: Demodulated symbols
1- Introduction to OFDM and Test
Signals
- 11 -
- 11 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Synchronization chain:
OFDM Transmission Model
ModulatorDAC
Ts
Multipath
channel ADC
Ts’
fc fc’
Synchronization
& demodulator
Timing
Synchronization
Frequency
Synchronization
Sampling Clock
Synchronization
FFT windows
positioning
Frequency
CorrectionResampling Demodulator
Rx
samples
Emitter Receiver
1- Introduction to OFDM and Test
Signals
- 12 -
- 12 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
OFDM pilot sub-carriers In order to be able to correct the distortion brought by the
channel, pilot symbols are introduced among the transmitted
data.
– Known symbols that can be compared to the corresponding
demodulated symbols for channel estimation.
Possibility to obtain the channel impulse response (CIR) for
equalization.
Ex: Pilot distribution in the DVB-SH/DVB-T standard:
Frequency (subcarriers index)Tim
e (O
FD
M s
ym
bo
ls)
Continuous pilots
Scattered pilotsData & TPS
1- Introduction to OFDM and Test
Signals
- 13 -
- 13 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
OFDM and Single Frequency Network
A SFN is a network where all emitters emit the same signal, on
the same frequency.
They are used to
– reach users in zones that could be shaded if only one powerful emitter
was used,
– to ensure smooth transition between emitters.
OFDM is well adapted for SFN thanks to the use of circular
FFTs, of the CP and of narrow sub-carriers. However, to be
useful, the delay spread of all received signals should be within
the CP duration very stringent emitters’ synchronization
Emitter synchronization is very interesting for positioning when
using multilateration: there is no need for station monitoring the
emitter clock drift + the emitter synchronization is usually done
with respect to the GPS time
1- Introduction to OFDM and Test
Signals
- 14 -
- 14 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
European Digital terrestrial TV standard, worldwide adopted
Base for the European mobile TV standards DVB-H (Handheld) and DVB-SH (Satellite-to-Handheld)
Multi- or single frequency networks (MFN/SFN) are possible
Interest for urban/indoor positioning:– High power signals → potentially high availability (even indoor) of the
positioning service
– Deployed in VHF and UHF bands, signal is less attenuated by walls than in GNSS band
– Fixed emitters
– Large signal bandwidth: good for
precise synchronization
– SFN
It is already available
The DVB-T standard
1- Introduction to OFDM and Test
Signals
- 15 -
- 15 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
European digital mobile TV standard
Emitters network composed of
– A network of terrestrial emitters
– One or several GEO satellites
Possibility of SFN
Interest for urban/indoor positioning:
– Dense coverage in urban centers
– Wideband signals (up to 8 MHz is foreseen)
– SFN
– Convergence with other telecommunication systems
The DVB-SH band is at 2.2 GHz, close to other telecommunication bands
(eg 3G band at 2.1 GHz, WiFi band at 2.45 GHz).
The DVB-SH standard
Frequency sharing between satellite and terrestrial repeaters
1- Introduction to OFDM and Test
Signals
- 16 -
- 16 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Terrestrial network propagation issues
The presence of multipaths
– Fast and strong fading of the
received signal power for short
multipath delays,
– Multiple replicas of the signal for
large multipath delays,
Large average power decay vs
distance created by the overall
environment.
Masking / blocking of the line-
of-sight (LOS)
– NLOS multipath may be received
with a stronger power.
0 0.2 0.4 0.6 0.8 1-25
-20
-15
-10
-5
0
5
Time (s)
Ta
p a
mp
litu
de (
dB
)M
ultip
ath
am
plit
ud
e (
dB
)
Time (s)
Multip
ath
dela
y (
s)
0 0.2 0.4 0.6 0.8
-0.5
0
0.5
1
1.5
2
2.5
3
x 10-6
2- Propagation channel-related issues
for positioning
- 17 -
- 17 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
SFN propagation channel issues Delay overlap
– At certain locations of the coverage called iso-delay
zones, signals from different emitters may arrive
simultaneously.
Emitter identification issue.
Near-Far Effect (commonplace in all terrestrial networks)
– A signal received from a closer emitter will have a
significantly stronger power than from remote emitters
and may prevent their detection.
Emitter detection issue (>3 required for 2D positioning).
LO
S
2- Propagation channel-related issues
for positioning
- 18 -
- 18 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Example of SFN CIR measurements No model taking into account all these characteristics was
found, in particular for the fine multipath delay modeling
Use of Channel Impulse Response (CIR) measurements.
CNES conducted a channel sounding campaign to
characterize the SFN CIR in urban environment (DVB-SH).
– 2 terrestrial emitters +1 helicopter acting as a GEO sat emitter
– Urban + dynamic
2- Propagation channel-related issues
for positioning
- 19 -
- 19 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
The multi-lateration principle is used (as in GNSS)
– Emitters’ locations are known.
– Tight synchronization of the emitters is required.
– The Line-of-Sight transit time from an emitter to the receiver is
measured, thanks to receiver synchronization processes.
– A minimum of 3 timing measurements are required for 2D positioning.
Can be done:
– with Time of Arrival
– with Time Difference of Arrival
Positioning principle
3- Positioning Principle
- 20 -
- 20 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Positioning principle using SFN Hypotheses
– Emitters’ locations are known: Terrestrial emitters are fixed (for DVB-SH,
the geo-stationary satellite should provide precise orbit information)
– Tight emitter synchronization:
Current SFN deployments: ±1 µs / 300 m
Current state-of-the-art: ±50 ns / 15 m
Assumption here: perfect sync or clock correction (currently under analysis
on DVB-T emitters)
Challenges
– Emitter identification in SFN: Same signal emitted synchronously from
every emitter.
– Fine pseudo-range estimation using OFDM signals in urban
environment:
OFDM does not need fine synchronization for telecommunication.
Urban environment creates NLOS bias and heavy multipaths.
3- Positioning Principle
- 21 -
- 21 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Proposition for solving the SFN issues Introduction of artificial delay
– One delay per emitter
– Known by the receiver
– Does not degrade data demodulation
if the overall delay is inferior to the CP length
Avoids the delay overlap issue;
Mitigates the near far effect;
Enable emitter identification.
Artificial
delay
𝑟(𝑡) = 𝑠(𝑡 −𝑑𝑘
𝑐− 𝜏𝑘). 𝑙 𝑑𝑘
𝑁𝑇𝑥
𝑘=1
received signal pathloss to emitter k
transit time
to emitter k
artificial delay
of emitter k
3- Positioning Principle
- 22 -
- 22 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Emitter identification in SFN
Emitter identification by reverse positioning
– Thanks to the introduction of artificial delays, only one combination of
pseudo-range measurements should correspond to a given location.
– With sufficient a priori knowledge, it is possible to find the association
between a set of anonymous pseudo-ranges and their emitter of origin.
– The required a priori information is an approximate position and the
characteristics of the present emitters
Mathematical formulation
– A cost function is defined
– All combinations of association between measurements and emitters
are tested. The one giving the smallest cost function corresponds to the
estimated emitter identities.
𝑉(𝑝, 𝑝 ,𝛔) = 𝜌𝑖 − 𝑑𝑖 − 𝑐. 𝜏𝑖 2
𝑁𝑇𝑥
𝑖=1
approximate position PR model taken at 𝑝
true position
tested permutation
PR measurement
3- Positioning Principle
- 23 -
- 23 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Emitter #1
Emitter #2
Proposed Estimation Technique
Conventional techniques(used for telecom)
Proposed techniques(specifically designed for positioning)
t
1. CIR estimation
2. Multipath delay acquisition:
– Extraction of the delay of the
main multipath components in the
CIR
3. Multipath delay tracking:
– parallel tracking of previously
acquired delays
4. Pseudo-range calculation:
– selection of the earliest estimated
delay for each emitter
Coarse syncOFDM
Demodulation
Multipath delay
acquisition
Multipath delay
tracking
Pseudo-range
calculationCIR estimation
4- Proposed Pseudo-Range
Estimation Technique
- 24 -
- 24 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
CIR estimation by correlating the received signal with a local
replica containing the pilot sub-carriers only
The correlation peak’s sidelobes can mask the correlation
peaks from the other emitters.
Coarse syncOFDM
Demodulation
Multipath delay
acquisition
Multipath delay
tracking
Pseudo-range
calculationCIR estimation
CIR Estimation
Emitter #2
Emitter #1
Sat Emitter
4- Proposed Pseudo-Range
Estimation Technique
Ex: CNES DVB-SH CIR
- 25 -
- 25 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
CIR Estimation using Windowing
Windowing techniques are
used to reduce the sidelobes’
amplitude of the correlation fct.
– Use of rectangular, Hamming and
Blackman-Harris
The use of windowing
significantly reduces the near-
far effect problem:
– The satellite is visible and emitter
#2 is not shadowed
Drawback: the main lobe is
wider, meaning that the
detection of the peak location
will likely be less accurate.
4- Proposed Pseudo-Range
Estimation Technique
Hamming
Blackman-HarrisEx: CNES DVB-SH CIR
Ex: CNES DVB-SH CIR
- 26 -
- 26 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
From a CIR vector estimate, estimate the delay of a set of
multipath (Matching Pursuit, ESPRIT, etc…):1. Find the multipath with the highest amplitude
2. Subtract the multipath from the CIR estimation
3. Loop (1-2) with corrected CIR estimate
-1 0 1 2 3
x 10-6
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6x 10
-3
Multipath delay (s)
Estim
ate
d M
ultip
ath
am
plit
ude
-1 0 1 2 3
x 10-6
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6x 10
-3
Multipath delay (s)
CIR
am
plit
ude
Coarse syncOFDM
Demodulation
Multipath delay
acquisition
Multipath delay
tracking
Pseudo-range
calculationCIR estimation
4- Proposed Pseudo-Range
Estimation Technique
Path Acquisition
- 27 -
- 27 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Multiple DLLs are launched from previously estimated delays
Discriminator: normalized |E|²-|L|²
Tracking loop: 2nd order, with 10Hz of noise loop equivalent
bandwidth
Sensitivity around 20 to 30 dB below
demodulation threshold
Tracking error in AWGN conditions
(no multipaths)
– Sub-meter accuracy for low SNR (> -10 dB)
– Tracking could be possible for remote
emitters or in indoor environment.
Coarse syncOFDM
Demodulation
Multipath delay
acquisition
Multipath delay
tracking
Pseudo-range
calculationCIR estimation
4- Proposed Pseudo-Range
Estimation Technique
Signal Tracking
- 28 -
- 28 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Delay selection for pseudo-range
estimation
– Group the estimated delays by emitters (via
clustering techniques).
– Take the earliest delay for each emitter.
– BUT, we could lose the LOS signal or even not
track the LOS at all
Pseudo-range estimation strategy
– New acquisitions are periodically launched
– DLLs are stopped if converging towards the
same delay, or tracking a weak part of the CIR.
Coarse syncOFDM
Demodulation
Multipath delay
acquisition
Multipath delay
tracking
Pseudo-range
calculationCIR estimation
4- Proposed Pseudo-Range
Estimation Technique
Pseudo-Range Calculation
- 29 -
- 29 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Semi-Simulated data
FFT size – NFFT 2048
CP length 1/4
Signal bandwidth B 5 MHz
Emitter EIRP 53.2 dBm [12]
Noise floor level -102.6 dBm [10]
SNR Between 9.2 - 49.2 dB
Noise equivalent loop bandwidth Bl 10 Hz
Time between 2 tracking updates Ti 448 µs
Time between 2 acquisition phases TACQ 14.336s
Acquisition threshold -120 dBm
Tracking threshold -140 dBm
Clustering threshold 2.5 µs
Max number of DLL launched after each acquisition 15
• The proposed pseudo-range
estimation technique was applied
to the CNES SFN measurements
including artificial delay.
– Moving van in urban environment
– 3 emitters
– 1000s of CIR measurements
• The simulation parameters were
chosen based on early
publications on DVB-SH system
deployments.
5 – Results on Pseudo-Range
Estimation
- 30 -
- 30 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Simulated data: Hamming window
• Result analysis:
– Clustering is working:
3 delay clusters, 1 for
each emitter
– Tracking of sidelobes
especially for emitter
#1 (improved
compared
to rectangular
window)
– Unavailability of
Emitter #3Availability
(%)
Mean error (m) Error standard deviation (m)
overall median overall median
Emitter #1 100 -150.2 -0.06 219.1 2.5
Emitter #2 100 26.7 0.41 61.9 3.9
Emitter #3 (sat) 88.6 29.7 0.02 169.0 1.3
5 - Results on Pseudo-Range
Estimation
- 31 -
- 31 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Simulated data: Blackman-Harris
Availability
(%)
Mean error (m) Error standard deviation (m)
overall median overall median
Emitter #1 98.6 35.0 8.3 80.5 32.4
Emitter #2 92.9 62.0 70.4 79.7 51.1
Emitter #3 (sat) 100 0.003 0.000 0.9 0.7
• Result analysis:
– Tracking of sidelobes
solved
– Availability of Emitter
#3
– Frequent leaps
between first and
second multipath for
Emitter #2
5 - Results on Pseudo-Range
Estimation
- 32 -
- 32 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Test on DVB-T (already available)
2 devices:
– A GPS receiver for time reference,
– An USRP2/WBX for TV signal down-conversion and digitization.
Recording a signal from
a French digital TV
emitter:– EIRP of 5kW,
– distance 80 km,
– frequency 554 MHz
No reference distance is
used (only PR variations
are investigated)
Real Signals: Test Bench Description
TV Antenna
Gain : 20 dB
NF : < 3.5 dB USRP2/WBXGain: 0 – 31.5 dB
LO: 50 MHz – 2.2 GHzFs : 100MHz/k k=[4..512]
GPS Active
Antenna L1/L2
Gain : 20 dB
NF : < 3.5 dB
GPS receiverNovatel OEM4
Channel I
Channel Q
To PC for recording
10MHz ref. locked on GPS time
5 - Results on Pseudo-Range
Estimation
- 33 -
- 33 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
The correlation image illustrates the evolution of the absolute
value of the correlation function over time.
In both cases: 2 strong peaks spaced by 4 km.
In the indoor case: higher noise floor and more multipaths
Uncertainty in knowing if this is the direct signal
Real Signal Test – Static Scenario
2 4 6 8 10 12
-2
0
2
4
6
Time (s)
Dela
y (
km
)
-30
-25
-20
-15
-10
-5
0
10 20 30 40 50 60
-2
0
2
4
6
Time (s)D
ela
y (
km
)
-30
-25
-20
-15
-10
-5
0
Outdoor case Indoor case
Colo
r s
cale
in d
B
Colo
r s
cale
in d
B
5 - Results on Pseudo-Range
Estimation
- 34 -
- 34 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Real Signal Test – Static Scenario
Outdoor case Indoor case
• A zoom on the main signal shows a very difficult environment
indoor. Still the strong transmitted signal ensures availability,
but with an unknown precision
10 20 30 40 50 60
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Time (s)
Dela
y (
km
)
4 5 6 7 8 9 10
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Time (s)
Dela
y (
km
)5 - Results on Pseudo-Range
Estimation
- 35 -
- 35 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Theory developped in [PLANS2010] : D. Serant et al.“Development and validation of an OFDM/DVB-T
sensor for positioning”, IEEE/ION PLANS, 4-6 May 2010
Tracking with a 1 Hz loop bandwidth:
The tracking error STD is quite low when considering the environment.
Note that results show only the tracking error STD but a bias can be
present if a reflection is tracked instead of the direct signal.
Real Signal Test – Static Scenario
Outdoor case Estimated SNRPseudorange
Error Std Dev
Theoretically
predicted result*
First peak -3.5 dB ~8 cm ~3 cm
Second peak -9 dB ~8 cm ~6 cm
Indoor case Estimated SNRPseudorange
Error Std Dev
Theoretically
predicted result*
First peak -13 dB ~0.7 m 0.1 m
Second peak -25 dB ~2.3 m 0.5 m
5 - Results on Pseudo-Range
Estimation
- 36 -
- 36 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
20 40 60
-0.2
0
0.2
0.4
0.6
0.8
Time (s)
Del
ay (
km
)
-30
-20
-10
0
This test is a succession of static and dynamic phases as described in the following table:
Real Signal Test – Dynamic Scenario
Phase Static Dynamic Static Dynamic Static
Duration 15 s 15 s 15 s 12s 15s
0 20 40 60-60
-50
-40
-30
-20
-10
Time (in sec)
Est
imate
d S
NR
(in
dB
)
Correlation image Estimated SNR
During dynamic phases the correlation is disturbed and the estimated SNR drops.
5 - Results on Pseudo-Range
Estimation
- 37 -
- 37 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Estimated pseudorange variation (1 Hz loop bandwidth):
WARNING: It is not the pseudorange error that is plotted but only its variation
since no reference position was recorded during the test.
Real Signal Test – Dynamic Scenario
0 10 20 30 40 50 60 70-30
-20
-10
0
10
20
30
Time (in sec)
Est
imat
ed P
seudora
nge
var
iati
on(i
n m
)
• During the static phases: the
pseudorange variation is quite stable.
• But during dynamic phase the
pseudorange estimation is clearly
disturbed
-> the absence of reference position and
the limited traveled distance during the
test do not permit to give a quantitative
performance in dynamic scenarios.
5 - Results on Pseudo-Range
Estimation
- 38 -
- 38 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Conclusion on PR Estimation
Error on the pseudo-range estimation is affected by:
– A potential unavailability of PR due to the Near-Far effect. Reduced
when using windowing.
– Numerous steps of the pseudo-range errors due to the propagation
channel.
– A large error for the terrestrial emitters due to the tracking of strong
sidelobes or multipath. This can be due to signal blockage.
Good performance for slices of the time series
– Numerous jumps (several hundreds of meters)
– If detected, PR estimation performances could reach meter-level
To be published soon:
– Promising work in reducing measurement jumps
– Test set up has been significantly improved (multi-receiver + mobile test
bench) with availability of reference distance
5 - Results on Pseudo-Range
Estimation
- 39 -
- 39 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Position-domain simulation The results are shown only using the semi-simulated CNES
SFN measurements
The PR error calculated previously was done between the
simulated delay estimate and the delay provided by the
proposed tracking technique.
However, the CNES measurements may be affected by an
unknown NLOS bias.
In order to observe this phenomenon, the position was
computed using the estimated pseudo-ranges:
– Conventional Non-Linear Least Square algorithm;
– Position averaged over 1s.
Comparison between these DVB-SH tracks and the GPS track
recorded during the CNES measurement.
6 – Results on Position Estimation
- 40 -
- 40 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
With Ideal PR estimates Heavy NLOS bias present even with perfect PR estimates
Mean absolute error at 24.3 m (whole time series).
6 – Results on Position Estimation
- 41 -
- 41 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
With Best PR estimate combination
Best combination
– 2 PR estimates (emit #1 and #3) from Blackman-Harris simulation
– 1 PR estimate (emit #2) from the Hamming simulation
Mean absolute error at 76.8 m (whole time series)
41.9 m (first 800 s)
6 – Results on Position Estimation
- 42 -
- 42 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Conclusions The main contributors to the position error are
– The NLOS bias
– The pseudo-range estimation bias due to the tracking of multipath
The best achieved performance is a mean absolute error
around 40m
– Considering only 3 emitters
– Considering heavy multipaths
Performance is improved when combining different windowing
techniques.
To be published:
– Test with real DVB-T signals
– Hybridization with GNSS
6 – Results on Position Estimation
- 43 -
- 43 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Conclusions
The overall results have demonstrated the feasibility of
autonomous positioning using a dense terrestrial network of
emitters transmitting OFDM-based signals:
– System level modification to permit emitter identification in an SFN;
– Emitter identification technique working in SFN;
– Pseudo-range estimation technique working with OFDM signal, in urban
environment.
The methods were validated by simulation using realistic
propagation channel measurements provided by CNES.
– The best positioning performance is around 40 m (mean absolute error)
with the use of only 3 emitters and a moderate signal bandwidth.
The proposed methods were shown not impact the provision of
TV broadcast if the right signal parameters are chosen (not
shown in this presentation)
7 – Conclusions and Future Work
- 44 -
- 44 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
Future work The first results look promising. However, improvements can be
expected and are all underway:
– Improving the pseudo-range estimation method and avoiding the
measurement jumps (currently underway)
– Doing extensive real-world measurements to obtain a valid PR error
model
– Implementing an elaborate position calculation technique
Finally, most results are applicable to other OFDM, SFN-based
standards
– Terrestrial digital radio / TV broadcast: DAB, DVB-x, DMB-x, ISDB
– Mobile communication: 3GPP LTE, mobile WiMAX
7 – Conclusions and Future Work
- 45 -
- 45 -
MAI 2011
Laboratoire de Traitement du Signal et des Télécommunications
More information Thevenon P, Julien O, Macabiau C, Serant D, Ries L, Corazza S., Bousquet M.,
Positioning principles with a mobile TV system using DVB-SH signals and a Single Frequency Network. In: 2009 16th International Conference on Digital Signal Processing. IEEE; 2009. p. 1-8.
Thevenon P, Julien O, Macabiau C, Serant D, Corazza S, Bousquet M., Pseudo-Range Measurements Using OFDM Channel Estimation. In: ION GNSS 2009; 2009. p. 481-493.
Thevenon P., S-Band Air Interface for Navigation System: A focus on OFDM Signals, University of Toulouse, PhD Thesis, 2010
Thevenon P., S. Damien, O. Julien, C. Macabiau, M. Bousquet, L. Ries, S. Corazza, Positioning Using Mobile TV Based on the DVB-SH Standard, to be published in Navigation: the Journal of the ION.
Serant D., Julien O, Macabiau C, Thevenon P, Dervin M, Corazza S, M.-L. Boucheret, Development and Validation of an OFDM/DVB-T Sensor for Positioning. In: IEEE/ION PLANS 2010; 2010. p. 988-1001.
Serant D., Julien O, Macabiau C, Thevenon P, Dervin M, M.-L. Boucheret, Use of OFDM-based Digital TV for Ranging: Tests and Validation on Real Signals, NAViTEC conference, 2010.
Serant D., O. Julien, C. Macabiau, P. Thevenon, M. Dervin, L. Ries, M.-L. Boucheret, Positioning Using OFDM-based Digital TV: New Algorithms and Tests with Real Signal, accepted for ION GNSS 2011
8 – Publications on this Topic