Broadband Wireless Channel Measurements for High Speed Trains

Transcription

Broadband Wireless Channel Measurements for High Speed Trains
Broadband Wireless Channel
Measurements for High Speed Trains
Florian Kaltenberger, Auguste Byiringiro,
George Arvanitakis, Riadh Ghaddab,
Dominique Nussbaum, Raymond Knopp,
Marion Bernineau, Yann Cocheril,
Henri Philippe
Eric Simon
C O R R I D O R F i n a l W o r k s h o p – I f s t t a r J u n e , 2 5 th 2 0 1 5
Motivation
 Achieving broadband communications at high speeds
 LTE has been designed for up to 150Mbps and up to 500km/h, but
realistic throughput is much lower (~20Mbps)
 LTE-Advanced increases throughput through
 Higher order MIMO (8 antennas)
 Carrier Aggregation (CA)
 However, fundamental problems remain
 High Doppler causes inter-carrier interference (ICI)
 High channel variation makes feedback difficult
 Different propagation conditions at different carriers
 This work presents
 High-speed (300km/h) channel measurement campaign
 Combining MIMO and CA
 Carried out in the French CORRIDOR project
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
2
Channel Sounding Campaign
5MHz
f1 = 771.5MHz
10MHz
20MHz
f2 = 2.590GHz
f3 = 2.605GHz
 3 LTE carriers
 4 transmit antennas each
 Sounding signal
 LTE-like (OFDM)
 Pilots only
Sampling Rate (MSPS)
Symbol duration
Prefix length
Frame duration
OFDM size
Useful carriers
5MHz
7.68
10MHz
20MHz
15.36
30.72
66 μs
16 μs
10 ms (120 OFDM symbols)
512
1024
2048
300
600
1200
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
3
Measurement Equipment
 TX and RX built upon Eurecom’s
ExpressMIMO2 cards
 Each card provides either 4 x 5MHz,
2 x 10 MHz, or 1 x 20MHz
 Multiple cards can be synchronized
 TX uses additional power amplifiers
 40dBm for 800MHz channel
 36dBm for 2.6GHz channel
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
4
Antennas
 Base station
 2 HUBER+SUHNER (2.6GHz)
 2 Kathrein (800MHz)
 All sectorized & dual polarized
 Train
 Sencity Rail Antennas from
HUBER+SUHNER
 2 double-directional antennas
at 2.6GHz (only one direction
used)
 3 omni-directional antennas at
800MHz
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
5
Scenarios
 Train: IRIS 320 TGV
 SNCF INFRA
 Rail line: LGV Atlantique
 Southwest of Paris
 Scenarios
1. Operator type: eNB 1.5km
away from rail line
2. LTE-R type A: eNB next to
railway line, antennas point
both sides
3. LTE-R type B: eNB next to
railway line, antennas point
in same direction
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
6
Data acquisition and Post processing
 Recording raw IQ data
 Continuously for 5MHz channel
 1/2 seconds for (10+20MHz) channel
 Timing synchronization and tracking
 Channel Estimation
 Delay Doppler Power Spectrum Estimation
 One estimate per second  1Hz Doppler resolution
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
7
Path loss results
Path loss
coefficient:
 3.2 - 4.5
Mean
 800MHz: 3.6
 2.6GHz: 4
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
8
Delay Doppler Spectrum 800MHz Scenario 1
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
9
Doppler Profile 800MHz Trial 1 Run 1
Zoom on Bock 30
 Strong frequency offset
 Calibrated mostly in trial 2
 Possible Model: (offset) Jakes + specular component
 Specular component corresponds well to geometry and speed of train
 Hypothesis yet to be tested
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
10
Delay Doppler Spectrum 2.6GHz (Carrier 1) Scenario 2A
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
11
Doppler Profile 2.6GHz Scenario 2A
Run 1 (Paris  Le Mans)
Run 2 (Le Mans  Paris)
 Second specular component when train is
southwest of base station!
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
12
Model of scattering environment
 Strong local reflection from gantries along
track
 Antenna main lobe pointing away from TX 
LOS strongly attenuated
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
13
Conclusions and Future Work
 First measurement campaign on a high speed
train combining MIMO and carrier aggregation
 Time-frequency analysis shows highly nonstationary behavior
 Antenna (port) selection very important to
reduce Doppler effects
 Channel could be modeled as geometry based
stochastic model
 Future work: spatial analysis (MIMO)
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
14
BACKUP
Motivation and Context
 CORRIDOR project
 Cognitive Radio for Railway Through
Dynamic and Opportunistic Spectrum Reuse
 Nov 2011 - Apr 2015, funded by ANR
PSD
f (Hz)
LTE-Advanced
Carrier Aggregation
PSD
f (Hz)
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
16
Huber&Suhner Sencity Rail Antenna 2.6GHz
Port 1
Port 2
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
17
Delay Doppler Spectrum 2.6GHz (Carrier 2) Scenario 2A
Final Workshop
Villeneuve d’Ascq – June, 25
th
2015
18