Slides © Robert W. Heath Jr. (2016)
Vehicle-to-X communication using millimeter wavesProfessor Robert W. Heath Jr., PhD, PE
Wireless Networking and Communications GroupDepartment of Electrical and Computer EngineeringThe University of Texas at Austin
www.profheath.org
Thanks to sponsors including the U.S. Department of Transportation through the Data-Supported Transportation Operations and Planning (D-STOP) Tier 1 University Transportation Center, the Texas Department of Transportation under Project 0-6877 entitled “Communications and Radar- Supported Transportation Operations and Planning (CAR-STOP)”, National Instruments, Huawei, and Toyota IDC
Slides © Robert W. Heath Jr. (2016)
Motivation
2
Slides © Robert W. Heath Jr. (2016)
3
Fifth generation (5G) cellular communication
Multidimensional objectives* New industry verticals**
* Recommendation ITU-R M.2083-0, “IMT Vision – Framework and overall objectives of the future development of IMT for 2020 and beyond,” September 2015** “5G empowering vertical industries,” 5GPPP White Paper, Feb. 2016
Aut
omot
ive
Med
ia &
Ent
erta
inm
ent
e-H
ealth
Higher rates
Lower latency
Fact
ory
of th
e Fu
ture
Ener
gy
Mobility
Spectrumefficiency
Userexp.
data rate
Peakdata rate
Energyefficiency
Areatraffic
capacity
LatencyConnectiondensity
Slides © Robert W. Heath Jr. (2016)
Trends in vehicle automation
4
INCREASING NUMBER OF SENSORS
ADVANCING COMMUNICATION CAPABILITY
* 5G-PPP White Paper on Automotive Vertical Sector, October 2015, https://5g-ppp.eu/white-papers/
Higher automation
levels
TRAFFICEFFICIENCY
AUTOMATEDDRIVING
SAFETY
Slides © Robert W. Heath Jr. (2016)
5
Automation levels
Fullyself-driving automation
LEVEL 4
Driver providesdestination
Driver not available for control
Self driving cars
Driver can cedecontrol over a primary function (eg. ACC)Responsible for safe operation
Full driver control at alltimes
Diver can cede full control of all safety-critical functions
Driver does not have to monitor the roadway at all times
LEVEL 3
Limitedself-driving automation
LEVEL 2
Combinedfunction
automation
LEVEL 1
Function specific
automation
LEVEL 0
No automation
Driver can cedecontrol on at least two primary functions
Driver responsible for monitoring the roadway
NHTSA, “Preliminary Statement of Policy Concerning Automated Vehicles”, 2013
Driver assist
Slides © Robert W. Heath Jr. (2016)
Communication for driver assist
6
Both communication and automotive sensors are useful
for collision avoidance
Sensors require line-of-sight
Communication can expand sensing range
More informed safety decisions
Limited automation based on sensing + communication to assist driver
Slides © Robert W. Heath Jr. (2016)
7
Communication for full automation
10
Sharing local sensors information ~ 100x Mbps
for safety app.
Higher data rates and lower latency will be required for full automation
V2X can be used for other functions, for example
more precise navigation
Exchanging raw sensor data provides information for fully automated safety-
critical functions
Downloading high-definition 3D map data (~Gbyte) for precise
navigation
Slides © Robert W. Heath Jr. (2016)
8
Communication for traffic efficiency
V2X helps coordinate traffic through intersections,
eliminating lights
V2X leads to higher levels of traffic coordination like
platooning
Low latency but low rate connectivity may be suffiicent
Download HD 3D map
Upload local sensor data for dynamic map
Slides © Robert W. Heath Jr. (2016)
9
Infotainmentapplications
Download multimedia data(movie, music, apps)100x Mbps - Gbps
Mobile base station for passengers
Communication for infotainment
High rate and low latency Internet access required to keep passengers happy
Slides © Robert W. Heath Jr. (2016)
Expand the sensing range of the vehicle
Allows interactions between vehicles with
different automation levels
More informed safety decisions
Communication enhances automated vehicles
Higher levels of traffic coordination like platooning
autonomous
Slides © Robert W. Heath Jr. (2016)
11
Ex: Passing on rural roads
What if the bus or oncoming car do not have communication capability?
82% of head-on fatal collisions take place in
rural areas
Radar requires line-of-sight
Both communication and radar are useful for collision avoidance
Alice Chu, Michael Motro, Junil Choi, Abdul Rawoof Pinjari, Chandra R. Bhat, Joydeep Ghosh, and R. W. Heath, Jr., ``Vehicular Ad-Hoc Network (VANET) Simulations Of Overtaking Maneuvers On Two-Lane Rural Highways,'' submitted February 2016.
Slides © Robert W. Heath Jr. (2016)
10
Infrastructure has value for automated vehiclesCan be used for other functions, for example
more precise navigation
Supports sensing of the environment, does not require
all cars to have complete sensing equipment
Helps coordinate traffic through intersections,
eliminating lightsEffective with non-connected cars, bicycles, and pedestrians
Slides © Robert W. Heath Jr. (2016)
13
Ex: Passing on rural roads
Infrastructure has enhanced sensing, better communication range
Infrastructure w/ sensing can broadcast position, velocity, and acceleration of vehicles
With enough infrastructure, can correctly avoid more
than 95% of collisions
Slides © Robert W. Heath Jr. (2016)
14
Key questions
Where is thecommunicationtheory and signalprocessing research?
What are the datarate requirements for
sensors?
What are the capabilities of current automotivecommunication solutions?
Slides © Robert W. Heath Jr. (2016)
State-of-the-art invehicular sensing
15
Slides © Robert W. Heath Jr. (2016)
16
Current technologies for vehicular sensing
Powerful sensing technologies with limited range in traffic conditions
Sensor Range(ideal)
Radar 200m
Camera 100m
LIDAR 100m
Sensor Range(traffic)
Radar 3-5m
Camera 3-5m
LIDAR 3-5m
Slides © Robert W. Heath Jr. (2016)
Sensor applications and data rates
17
Automotive sensors generate a huge amount of data
Purpose Drawback Datarate Update rateRadar Targetdetection,
velocityestimation
Hardtodistinguishtargets
Lessthan1Mbps
50-100ms
Camera Virtualmirrorsfordrivers
Needcomputervisiontechniques
100-700Mbpsforrawimages,10-90Mbpsforcompressedimages
60-100 ms
LIDAR Targetdetectionandrecognition,velocityestimation
Highcost 10-100Mbps 67-200ms
Slides © Robert W. Heath Jr. (2016)
18
State-of-the-art inconnected cars
Slides © Robert W. Heath Jr. (2016)
19
Connectivity for automated vehicles
Connectivity gives access to a
richer set of sensor data
Connectivity solves key challenges of
automated driving in congested urban areas
Connectivity motivates 5G and the
application of millimeter wave
Automated cars may have limited connectivity
Is it possible to exchange raw sensor data between cars with current technology?
Automated carsshould exploit connectivity
Slides © Robert W. Heath Jr. (2016)
DSRC: current technology for vehicular communications
20
!
Forward collision warning, do not pass warning, blind intersection warning, etc.
Based on IEEE 802.11p, IEEE 1609.x, SAE standards
* NHTSA, “Vehicle-to-Vehicle Communications: Readiness of V2V Technology for Application,” Aug. 2014** John B. Kenney, “DSRC: Deployment and Beyond,” WINLAB presentation, May 2015.
Supports very low data rates (27 Mbps max,
much lower in practice)
Non safety apps also possible
Slides © Robert W. Heath Jr. (2016)
21
DSRC: how it works for safety applications
1. Each transmitter broadcasts Basic Safety Messages
periodically (typically 10Hz)
2. Each receiver assesses collision threats
3. When threats are detected, the system warns the driver or
takes control of car
Safety messages sent on dedicated safety channel
Position, speed, etc.
DSRC is not designed for the exchange of raw sensor data
Slides © Robert W. Heath Jr. (2016)
22
Upcoming 3GPP standards for V2X: LTE-V2X (Rel. 14)
*3GPP Study on LTE-based V2X Services. TR 36.885, 3rd Generation Partnership Project (3GPP), July 2016.**M. Rumney et al. LTE and the evolution to 4G wireless: Design and measurement challenges. John Wiley & Sons, 2013
D2D
D2D
Standardization started in Apr. 2015
D2D-based is available both inside and outside the coverage
LTE-V2X = D2D-based + cellular-based
Higher data rates than DSRC (up to 44 Mbps for D2D and 75 Mbps for cellular)
Cellular-based is available only inside the coverage
D2D
Slides © Robert W. Heath Jr. (2016)
DSRC versus LTE-A for V2XFeatures DSRC D2D LTE-V2X Cellular LTE-V2X
Channel width 10 MHz Up to 20 MHz Up to 20 MHz
Frequency Band 5.9 GHz 5.9 GHz 450 MHz-3.8 GHz
Bit Rate 3–27 Mb/s Up to 44 Mb/s Up to 75 Mb/s
Range ~ 100s m ~ 100s m Up to a few km
Spectral efficiency 0.6 bps/Hz 0.6 bps/Hz (typical) 0.6 bps/Hz (typical)
Coverage Ubiquitous Ubiquitous Inside cell only
Mobility support High speed High speed High speed
Comm. fee Free ? ?
Latency x ms x10-x100 ms X10 ms
23*Giuseppe Araniti et al., “LTE for Vehicular Networking: A Survey”, IEEE Commun. Mag., May 2013
Gbps data rates are not supported
Slides © Robert W. Heath Jr. (2016)
24
Massive data rates from sensors vs DSRC/4G
New communication solution is needed for connected cars*http://low-powerdesign.com/sleibson/2011/05/01/future-cars-the-word-from-gm-at-idc’s-smart-technology-world-conference/**Cisco, “The Internet of Cars: A Catalyst to Unlock Societal Benefits of Transportation,” Mar. 2013***http://www.sas.com/en_us/insights/articles/big-data/the-internet-of-things-and-connected-cars.html
Automated vehicles can generate up 1 TB per
hour of driving
Current connected vehicles are expected to
drive 1.5GB monthly data in 2017**
4G and DSRC can not supportthese data rates
Handled with a combination of 4G and DSRC
Slides © Robert W. Heath Jr. (2016)
Millimeter wave for connected cars
25
5G
Slides © Robert W. Heath Jr. (2016)
High data rates with millimeter wave (mmWave)
26Millimeter wave offers the means to achieve high rates and low latency
Arrays needed for gain and aperture
Low freq. antenna mmWave antenna
Shrinking antenna aperture 802.11n 802.11ac 802.11ad
Band
wid
th
20 MHz 160 MHz
2 GHz
With channel bonding
Large bandwidth at mmWave
Adaptive beam steering
Slides © Robert W. Heath Jr. (2016)
27
mmWave for automated cars
MmWave is the only viable approach for high bandwidth connected vehicles*
V2V communication beams
Vehicle driving cloud
directionalbeamforming
blockageV2I communicationbeam
Joint communicationand radar
*Junil Choi, Nuria González-Prelcic, Robert Daniels, Chandra R. Bhat, and Robert W. Heath Jr, “Millimeter Wave Vehicular Communication to Support Massive Sensing”, to appear in IEEE Communications Magazine.
Sensing technologies can be used to help establish mmWave links
Exchanging raw sensor data is possibe
Enables high data rateinfotainment applications
Slides © Robert W. Heath Jr. (2016)
28
How will mmWave be realized?
5G is promising for mmWave connected cars
High data rates
Modificationof IEEE
802.11ad
5G mmWave cellular
DedicatedmmWave
V2XRequires special infrastructure
*Federal Communications Commission (FCC), “FCC 15-138,” 2015
Uses cellular infrastructureAccess is highly coordinatedLeverages (coming*) mmWave spectrum
Less efficient accessUse of unlicensed band
Use new dedicated spectrum
Slides © Robert W. Heath Jr. (2016)
© Robert W. Heath Jr. 29
Potential bandwidths and data rates at mmWave
10x to 100x gains in bandwidth going to mmWave
* IEEE 802.11ad is commercially available
Totalspectrum
Typicalbandwidth
Peakrates
IEEE802.11ad*in60GHz
7GHz 2GHz 6Gbps
IEEE802.11ayin60GHz
7GHz 4 GHz 100Gbps
28GHz5G 0.85GHz 200MHz 1.5Gbps39GHz5G 3GHz 400MHz 3GbpsEband5G 10GHz 2GHz 24Gbps
Slides © Robert W. Heath Jr. (2016)
© Robert W. Heath Jr. 30
mmWave enabled infrastructure for transportation
Sensing at the infrastructure
Combination of sensing, learning and
communication
mmWave relay
mmWavesensing-BS
multiband BSMultiband-connectivity
supporting V2X
radarbeam
Vehicles exchanging sensor data
Slides © Robert W. Heath Jr. (2016)
31
mmWave spectrum challenges forV2X
Cognitive radio for shared spectrum with satellite or radar**
Regulations not
harmonizedWays to reduce license cost but allow carriers to share
spectrum * New communication technology needed
*A.K.Gupta;J.G.Andrews;R.W.Heath,"OntheFeasibilityofSharingSpectrumLicensesinmmWaveCellularSystems,"in IEEETransactionsonCommunications,toappear**A.K.Gupta,A.Alkhateeb,J.G.Andrews,RobertW.Heath,Jr, “Gains ofRestricted Secondary Licensing inMillimeter WaveCellular Systems,arxiv 2016
Slides © Robert W. Heath Jr. (2016)
Designing a mmWave V2X system
32
Slides © Robert W. Heath Jr. (2016)
Overview of the mmWave V2X channel
33
Low antenna elevation
Prone to blockage
Tx & Rx moving
Fast changing topology
Large penetration and diffraction loss
Severe blockage
Shrinking antenna aperture
Directionality
V2X channels MmWave channels
MmWaveV2X channelsCombined
challenges from both sides
There are several measurements but still limited
Slides © Robert W. Heath Jr. (2016)
Channel coherence time and directional reception
34
Mathematical expression relating coherence time
and beamwidthMathematical expression
relating coherence time and beamwidth
Long term beamformingcan be used
*V. Va, J. Choi, and R. W. Heath Jr. The impact of beamwidth on temporal channel variation in vehicular channels and its implications. Submitted to IEEE Trans VT
Overheads of beam training are much less significant than expected
Beams should be narrow but not too “pointy”
Optimum beamwidthis a tradeoff between
pointing error and Doppler
Slides © Robert W. Heath Jr. (2016)
Efficient beam alignment leveraging position info
35Junil Choi, Vutha Va, Nuria González-Prelcic, Robert Daniels, Chandra R. Bhat, and Robert W. Heath Jr, “Millimeter Wave Vehicular Communication to Support Massive Sensing”, to appear IEEE Commun. Mag., 2016.
DSRC modules or automotive sensors can be used to reduce overhead
Even with poor accuracy of position information the beam alignment overhead is reduced
Slides © Robert W. Heath Jr. (2016)
Multipath fingerprint for V2I beam alignment
36
Reflection off building could be used in NLOS
Such paths via static objects can be learned
beforehand
Location Path RxPower AoA AoD
1 #1 -59.81 84 87
#2 -67.66 80 61
… … … …
2 #1 -60.9 82 87
… … … …
Example of multipath fingerprint
Junil Choi et al., “Millimeter Wave Vehicular Communication to Support Massive Automotive Sensing”, to appear in IEEE Commun. Mag., 2016
blocked
Infra collect database of multipath fingerprint(i.e. AoA/AoD) of paths indexed by location
1. Rx request link via DSRC and inform its position
2. RSU responses with list of beam indices for training
3. Perform beam training
Slides © Robert W. Heath Jr. (2016)
37
Radar-aided millimeter wave V2XRadar can be used to configure
communication link more efficiently
Radar can be used to design
multiuser beamforming
* N. González-Prelcic, Roi Mendez-Rial, and R. W. Heath Jr., ”Radar aided beamforming in mmWave V2I communications supporting antenna diversity," Proc. of ITA 2016 .
0.5
1
30
210
60
240
90
270
120
300
150
330
180 0
Azimut
Rela
tive P
ath
Gain
Com Signal at 65 GHz
Radar Signal at 76.5 Ghz
0.5
1
30
210
60
240
90
270
120
300
150
330
180 0
Elevation
Com Signal at 65 GHz
Radar Signal at 76.5 Ghz The dominant DoAs for the communication signal also appear at the radar echo in a different band
Algorithms for hybrid precoder& combiner design based on covarianceinformation of the radar signal
Slides © Robert W. Heath Jr. (2016)
38
Joint mmWave comm. and radar using IEEE 802.11ad
* P. Kumari, N. González Prelcic, and R. W. Heath, Jr., “ Investigating the IEEE 802.11ad Standard for Millimeter Wave Automotive Radar ,” Proc. of the Vehicular Technology Conference, Boston, USA, September 6-9, 2015
Joint system provides safety capabilities at lower cost
Special structure of preamble enables good ranging performance
Existing WLAN RX algorithms for
radar parameter estimation fine range estimation
achieves the desired accuracy of 0.01m
Slides © Robert W. Heath Jr. (2016)
39
Prototyping mmWave forV2X
National InstrumentsPXI chassis interfaces with
custom RF
communication transmitter radar receiver
automotive radar & DSRC
communication receiverautomotive radar &
DSRC
mmWaveV2X and joint mmWave / radar prototype
automotive radar test
mmWave tx/rx
baseband and IF
2x2 MIMOprototype
60 GHz arrays
mmWave 60 GHz phased array testbed
Slides © Robert W. Heath Jr. (2016)
40
Research challenges for PHY design
Effect of hardware impairments on mmWave V2X
Fast beam alignment and tracking
MIMO architectures for mmWave V2X: analog or hybrid?
Diversity solutions againstblockage
Slides © Robert W. Heath Jr. (2016)
41
What are we doing at UT to address these
challenges?
Slides © Robert W. Heath Jr. (2016)
Transportation and wireless @ UT Austin
WIRELESS COMMUNICATION
SIGNAL PROCESSING
MACHINE LEARNING & BIGDATAPOLICY AND PLANNING
TRAFFIC MODELING
SMART TRANSPORTATION
UT is well positioned to develop wireless networks for transportation systems
Deep industryconnections
Renowed wirelessexpertise
50 years of experienceon transportation
Strong conection with transportationadministrations
Slides © Robert W. Heath Jr. (2016)
43
Situation-aware vehicular engineering systems
UT-SAVES
Sensing
Data analytics and learning
Communication
Trafficcontrol center
An initiative in partnernshipwith Toyota IDC, Huawei, & National Instruments*
* Looking for more partners!
Slides © Robert W. Heath Jr. (2016)
Research challenges
44
Slides © Robert W. Heath Jr. (2016)
45
Open research challenges
o Fast beam training /channel estimation strategieso How and where to place the antennas in the carso Channel models are needed for the different scenarios
(rural road, highway, urban area, …)o Leveraging side information for beam trainingo More advanced MIMO communication techniques o Better integration with transportation simulators
Details on existing solutions and research challenges can be found here
General millimeter wave information here
Slides © Robert W. Heath Jr. (2016)
46
mmWaveV2X