Study of the immunity of the GSM-R against electromagnetic … · Characterisation and Analysis of...
Transcript of Study of the immunity of the GSM-R against electromagnetic … · Characterisation and Analysis of...
Railcom final conference, UIC, Paris, 21 april 2009
Virginie Deniau, INRETS, France virginie [email protected]
R. Adriano, S. Dudoyer, N. Ben Slimen, J. Rioult, P. Massy, B. Meyniel, M. Berbineau, A. Raux and E. Smulders
Study of the immunity of the GSM-R against electromagnetic disturbances present on moving
trains
Railcom final conference, UIC, Paris, 21 april 2009
Outline
•
The GSM-R system•
Characterisation and Analysis of the disturbances received by GSM-R antennas
•
Impact of the supply voltage on the disturbances•
Estimation of the impact of the transient disturbances on the BER of GSM-R communications
•
Immunity testing on GSM-R communications in laboratory
•
Detection of transients disturbances for diagnostic in-situ
•
Conclusion, future works
Railcom final conference, UIC, Paris, 21 april 2009
ERTMS and GSM-R
ERTMS: European Rail Traffic Management SystemHarmonised signaling standards throughout Europe
GSM-R
GSM-R
Data and voicetransmissions
GSM-R : Global System for Mobiles-RAILWAYSradio system for providing voice and data communication between the track and the train
Railcom final conference, UIC, Paris, 21 april 2009
GSM-R PCRD 6STREP
The GSM-R is based on the standard GSM Phase 2+the modulation type is GMSK (Gaussian Minimum Shift Keying).
Specific frequencies:876-880 MHz for the up-link (trains → base stations) 921-925 MHz for the down-link (base stations → trains)200 kHz frequency spacing between each channel
+ Advanced functions specifically developed for rail.group calls, ads
or calls broadcast,
location-based
connections, call pre-emption
in case of emergency…
Base Transceiver
Station (BTS) close to the tracksThe distance between
base stations ≈
3-4 km
GSM-R antennas on the roof of the trains
Railcom final conference, UIC, Paris, 21 april 2009
GSM-R PCRD 6STREP
The GSM-R is a “Time Division Multiple Access” (TDMA) system
For each carrier frequency (physical channel)
-
data are organized per periodic TDMA frame, with a period of 4.615
ms.
-
each TDMA frame is divided into 8 time intervals of 577
μs
long called "Time Slots"
Each “Time Slot” (logical channel)
-
includes 156 bits which 148 bits of information. The one bit transmission duration is about 3.7 µs.
Railcom final conference, UIC, Paris, 21 april 2009
burst
3.7 µsbit time
GSM-R PCRD 6STREP
Time
user
2us
er 1
user
4
user
3
user
7us
er 8
user
6
user
5
TDMA frame= 4.6
ms
user
2us
er 1
user
4
user
3
user
7us
er 8
user
6
user
5
8 logical channels
924.8 MHz
924.6 MHz
921.2 MHz
921.4 MHz
200 kHz
Carrier frequenciesphysical channels
Freq
uenc
y
Time slot577
μs
Railcom final conference, UIC, Paris, 21 april 2009
Characterisation and Analysis of the disturbances received by GSM-R antennas
Railcom final conference, UIC, Paris, 21 april 2009
The disturbances
received
by GSM-R antennas
Main potential EM disturbances for the GSM-R:•
Public GSM, UMTS 900 on frequency channels adjacent to the GSM-R frequency bands⇔ «
permanent
»
disturbances• Transients coming from the catenary -
pantograph sliding contact
On-board
measurements:
1.
To characterise the coverage levels of the permanent disturbances2.
To characterise the level of noise produced by the transients in
the GSM-R frequency bands
3.
To characterise the time characteristics and the repetition rate
of the transients
Railcom final conference, UIC, Paris, 21 april 2009
Pow
er (d
Bm
)300 MHz
Pow
er (d
Bm
)F.F.T
S11-antenna1 GHz
≈
-35 dBm
≈
-35 dBm
GSM-R frequencies are systematically
covered
•
Characterisation of the noise levels produced by the transients in the GSM-R frequency bands:
Analysis
of transient
disturbances
received
by GSM-R antennas
Railcom final conference, UIC, Paris, 21 april 2009
• Characterisation
of the time characteristics
of the transients
:
90%
10%
AA
100%
50%
time timeDuration Rise time
Time caracteristics
of the transient disturbances
Railcom final conference, UIC, Paris, 21 april 2009
2 4 6 8 10x 10
-10
0
0.5
1
1.5
2
2.5
3x 109
Rise time (s)
Pro
babi
lity
Den
sity
Fun
ctio
n
Rise Time
experimental distributionempirical distribution
0 0.5 1 1.5 2x 10
-8
0
5
10
15
x 107
Duration (s)
Pro
babi
lity
Den
sity
Fun
ctio
n
Duration
experimental distributionempirical distribution
• Statistical study with about 20000 collected transients on trains:
Duration
(s) Rise time (s)
20 ns 1 ns
Typical
duration
= 5 ns Typical
rise
time = 0.4 ns
Time caracteristics
of the transient disturbances
Railcom final conference, UIC, Paris, 21 april 2009
• Characterisation
of the repetition
rate «
Rr »
of the transients
:
Analysis
of transient
disturbances received
by GSM-R antennas
Measurements
performed
in 400 µs time windowsObjective: to establish distributions of the time delays between the transients according to the operating conditions
0 1 2 3 4x 10
-4
-0.4
-0.2
0
0.2
0.4
0.6
Time (s)
Am
plitu
de (V
) Time delay
400 µs
Very
variable Rr according
to the trains operating conditions -0 1
-0.05
0
0.05
0.1
0.15
400 µs
1 transient/ 5µs
0 1 2 3 4-1
-0.5
0
0.5
1
Am
plitu
de (V
)
1 transient/ 10 µs
400 µs
Railcom final conference, UIC, Paris, 21 april 2009
• Number
of transients
in each
400 µs time window
:
Analysis
of transient
disturbances received
by GSM-R antennas
1568 recorded files
Number of transient
disturbances
0 200 400 600 800 1000 1200 1400 16000
10
20
30
40
50
60
70
80
90
file order
9080706050403020100
≈
4.5 µs medium time delay betweentwo successive transients
400 µs Time Window Loading
Time
Over a 1500 s
Measurements duration :1568
recorded files ==> 0,95 s
Time duration of the measurements process+ = ~ 1 s
Measurement equipment was continuously detecting transients
0 200 400 600 800 1000 1200 1400 1600Successive time windows
Railcom final conference, UIC, Paris, 21 april 2009
Impact of the supply
voltages on the time characteristics
of the transients
1500 Vcc
25000 Vcc
About 300 transient
disturbances
collected
under
1500 Vcc
and 25000 Vac
Comparison
of the rise
times
0 50 100 150 200 250 3000
0.5
1
1.5
2x 10-9
0 50 100 150 200 250 3000
0.5
1
1.5
2x 10-9 Rise Time
Time (ns)
Typicalrise
time
= 0.4 ns
= 0.4 ns
Railcom final conference, UIC, Paris, 21 april 2009
Impact of the supply
voltages on the noise levels
over the GSM-R channels
About 300 transient
disturbances
collected
under
1500 Vcc
and 25000 Vac
0 50 100 150 200 250 300-80
-70
-60
-50
-40
0 50 100 150 200 250 300-80
-70
-60
-50
-40Maximal Amplitude 921 MHz - 925 MHz
FFT of the 300 transients
and post pocess
to extract
the maximal noise level
over the GSM-R frequency
bands
1500 Vcc
25000 Vcc
921 MHz-925 MHz : Down-link
GSM-R
Railcom final conference, UIC, Paris, 21 april 2009
Impact of the supply
voltages on the peak values
About 300 transient
disturbances
collected
under
1500 Vcc
and 25000 Vac
1500 Vcc
25000 Vcc
0 50 100 150 200 250 3000
0.5
1
1.5
2
0 50 100 150 200 250 3000
0.5
1
1.5
2Peak Amplitude
0 50 100 150 200 250 300-80
-70
-60
-50
-40
0 50 100 150 200 250 300-80
-70
-60
-50
-40Maximal Amplitude 921 MHz - 925 MHz
Peak
valueF.F.T
921 MHz-925 MHz
Pow
er (d
Bm
)
Railcom final conference, UIC, Paris, 21 april 2009
Estimation of the impact of the transient disturbances on the BER of GSM-R communications
Railcom final conference, UIC, Paris, 21 april 2009
Transient disturbances and BER on GSM-R transmissions
Hypothesis :
• Transients duration <<
duration
of one bit→ approximated by a punctual event,•
Transients produce high levels of interference in the GSM-R band, •
Transients produce an arbitrary decision in a bit inside the burst (worst case?).
S
r
RRBER ⋅=
21
Rs is the symbol rate of the communication systemRr represents the repetition rate of the transients
GSM burst
Railcom final conference, UIC, Paris, 21 april 2009
Immunity testing on GSM-R communications in laboratory
Railcom final conference, UIC, Paris, 21 april 2009
Immunity testing on GSM-R communications
CMU 200924.8 MHz
Combiner
GSM-RMobile
Spectrum analyzerCalibration of the power levels
-40dB
Loop back
Signal generator
Directional
combiner
Amplifier
SMIQGSM Public925.2 MHz
CM
U 2
00
SM
IQ
50 Ω
load
combiner
Erroneous
bits * 100%total number
of bitsBER =
Over 1200 speech frames
• Test set-up
:
Railcom final conference, UIC, Paris, 21 april 2009
GSM-R Mobile
Spectrum analyzer
CMU 200
SMIQ
Arbitrary signal
generator
Oscilloscope
combiners
Railcom final conference, UIC, Paris, 21 april 2009
Immunity testing on GSM-R communications
Combiner
SMIQGSM Public925.2 MHz
Signal generator
GSM-RMobile
load
CMU 200924.8 MHz
• Power calibration based
on the preliminary
on-board
measurements:
?Transients:analysis
by applying
FFT
→ Maximum level
≈
-35 dBm
?
GSM-R :measurement
campaign
for
coverage
levels
and specifications→ -90 dBm < GSM-R < -25 dBm
?
GSM public:measurement
campaign
for
measurements
of coverage
levels→ Maximum level
≈
-25 dBm
Railcom final conference, UIC, Paris, 21 april 2009
Immunity testing on GSM-R communications
3 -
Normalization
to 1V peak
to peak
4 –
Variation of the repetition
rate
• Wave
form
of the transients
based
on the statistical
distributions:
1 –
“Double exponential model”• duration = 5 ns• rise time = 0.4 ns
Combiner
SMIQGSM Public925.2 MHz
Signal generator
GSM-RMobile
load
CMU 200924.8 MHz
2 –
Application of“Bandpass
numerical filter”
or modulation with a sinus
Railcom final conference, UIC, Paris, 21 april 2009
-100
-90
-80
-70
-60
-50
-40
920 922 924 926 928 930Frequency (MHz)
Pow
er (d
Bm) without transients
GS
M-R
GS
M
CMU 200924.8 MHz
Combiner
Transient
GSM-RMobile
50 Ω
GSM Public925.2 MHz
Spectrum analyzerMaxhold
combiner
Immunity testing on GSM-R communications
Railcom final conference, UIC, Paris, 21 april 2009
Time interval
-100
-90
-80
-70
-60
-50
-40
920 922 924 926 928 930Frequency (MHz)
Pow
er (d
Bm)
4 µs time interval10 µs time interval20 µs time interval1.7 ms time intervalwithout transients
CMU 200924.8 MHz
Combiner
Transient
GSM-RMobile
50 Ω
GSM Public925.2 MHz
Spectrum analyzerMaxhold
combiner
Immunity testing on GSM-R communications
Time interval
≥
10 µsfor the immunity
tests
Railcom final conference, UIC, Paris, 21 april 2009
BER and RXQUAL
Quality level i
Rxqual: parameter
employed
to control the quality
of the service in situ
Range of valuesTypical values
of BER
Railcom final conference, UIC, Paris, 21 april 2009
- 17/23 -
Impact of public GSM signals
1 2 3 4 5 6
GSM-R FREQUENCY
GSM-R POWER (dBm)
GSM public FREQUENCY
GSM public POWER (dBm)
Measured
BER (%)
RXQUAL
924.8 MHz -70 925.2 MHz -70 0.004 0
-52 0.011 0
-30 0.01 0
-17 0.086 0
-15 0.137 1
-12 0.44 2
+ 55 dBm
GSM-R924.8 MHz
GSM-RMobile
GSM Public925.2 MHz
Combiner
Railcom final conference, UIC, Paris, 21 april 2009
0.0
0.4
0.8
1.2
1.6
2.0
-60 -50 -40 -30 -20 -10 0GSM public power (dBm)
Transients with TI = 90 µsTransients with TI = 150 µsTransients with TI = 550 µsBER without Transient
Rxqual
= 1
Rxqual
= 3
Impact of the transient disturbances in presence of public GSM
BER (%)
Railcom final conference, UIC, Paris, 21 april 2009
BER induced
by two
different
collected transients
GSM-R924.8 MHz
GSM-RMobile
Combiner
Signal generator
Transients
collectedon board
0.00.20.40.60.81.01.21.41.61.82.0
0 400 800 1200 1600Transients time interval (µs)
BER
(%)
Recorded transient D=6.1 ns and RT= 0.35 nsRecorded transient D=6.75 ns and RT= 0.4 ns
GSM-R power = -70 dBm
Railcom final conference, UIC, Paris, 21 april 2009
GSM-R924.8 MHz
GSM-RMobile
Combiner
Signal generator
S
r
RRBER ⋅=
21
Comparisons
between
measured
and estimated
BER
Transient
collected
on boardRT= 0.35 ns, D= 6.1 ns
GSM-R power = -70 dBm
0.3
0.2
0.1
0
‐0.1
‐0.2
‐0.3
Temps (µs)0 0.1 0.2 0.3 0.4
(V)
Variable Time interval
Time
V
( ) ( )tRTt
FTtAtV γ⎟⎟
⎠
⎞⎜⎜⎝
⎛⎟⎠⎞
⎜⎝⎛−−⎟
⎠⎞
⎜⎝⎛−= expexp
Double exponential
ModelRT= 0.4 ns and D= 5 ns
Prediction
of the BER
Railcom final conference, UIC, Paris, 21 april 2009
Comparisons
between
measured
and estimated
BER
0.0
0.4
0.8
1.2
1.6
2.0
0 200 400 600 800 1000 1200 1400 1600Intervalle de temps entre les transitoires (µs)
BER (%)
estimation du BER (%)
transitoire réel
modèle de transitoire
0.3
0.2
0.1
0
‐0.1
‐0.2
‐0.3
Temps (µs)0 0.1 0.2 0.3 0.4
(V)
( ) ( )tRTt
FTtAtV γ⎟⎟
⎠
⎞⎜⎜⎝
⎛⎟⎠⎞
⎜⎝⎛−−⎟
⎠⎞
⎜⎝⎛−= expexp
RsRrBER ⋅=
21
Double exponential
ModelRT= 0.4 ns and D= 5 ns
Time interval
between
transients
GSM-R power = -70 dBmS/N=1 over the duration of the transient
Railcom final conference, UIC, Paris, 21 april 2009
Detection of transient disturbances for diagnostic in-situ
Railcom final conference, UIC, Paris, 21 april 2009
Characterisation
of the noise level
0 50 100 150 200 250 300-80
-70
-60
-50
-40
Maximal Amplitude 921 MHz - 925 MHz
Over 300 transients
Noise level
varies between-40 dBm and -70 dBm
GSM-R Reception
level
canVary
between
About -20 dBm and -92 dBm
The definition
of a maximum level
of noise is
not adapted to this
application
We
propose to control the recurrence
of the transient disturbances
and to compare it
with
the Rxlevel
of the GSM-
R signal
Railcom final conference, UIC, Paris, 21 april 2009
Minimum GSM-R power level
to keep
a BER inferior
to 1.13 %
GSM-R924.8 MHz
GSM-RMobile
Combiner
Signal generator
Variable GSM-R power
-80
-78
-76
-74
-72
-70
-68
-66
-64
0 400 800 1200 1600Transients time interval (µs)
Req
uire
d G
SM-R
Pow
er (d
Bm)
With Collected transient D=6.1 ns, RT = 0.35 ns
With Collected transient D=6.75 ns, RT = 0.4 ns
BER>1.13 %
BER<1.13 %
Railcom final conference, UIC, Paris, 21 april 2009
Detection
of the transient
disturbancesMethod
employed
during
the project
is
to «
count
»
the transient
is
too
«
heavy
»
for a diagnostic methodology
in situ
-50
-40
-30
-20
-10
0
300 400 500 600 700 800 900 1000Frequency (MHz)
|Sii|
(dB
)|S11|- straight antenna|S11|- oblique antenna
850 MHzFree channel
Reflexion
S-parameters
of the GSM-R antennas
EMI Test ReceiverIn zero spanMeasures 1point/3.7 µs
Approach
proposed
Railcom final conference, UIC, Paris, 21 april 2009
-80
-70
-60
-50
-40
-30
-20
0.0 0.5 1.0 1.5 2.0Time (ms)
Pow
er (d
Bm
)
Detection
of the transient
disturbances
Approach
proposed
-
Results
70 µs 200 µs
Model transient
Real transient
Test sequence
Arbitrary
waveform
generator EMI Test Receiver
70 µs 200 µs
Advantage: we
collect
1 point by transient
disturbance
Railcom final conference, UIC, Paris, 21 april 2009
Conclusion
•
Complete characterisation
of the time characteristics
of the transient disturbances
produced
by the catenary-pantograph
sliding
contact
•
Proposition of a laboratory
testing
method
to control the immunity
of the GSM-R communications against
EM disturbances
representative
of the in
situ conditions (+ permanent and transient
disturbances
simultaneously)
•
Proposition of a prediction
method
of the BER induced
by the transients observed
on board
as a function
of the repetition
rate of the transients
!Characterisation with a GSM-R antenna → the bandwidth of the antenna can impact the rise time distribution obtained
•
Proposition of a methodology to preliminary verify the conditions required to guarantee the quality of the communications