mm-Wave on Wheels: Practical 60 GHz Vehicular Communication Without Beam Training

Yung-Sheng Lu

OCT 31, 2017

@NCTU-CS

Communication Systems and Networks (COMSNETS) '17

Adrian Loch, Arash Asadi, Gek Hong Sim, Joerg Widmer, Matthias Hollick

Outline

  • Abstract

  • Introduction

  • Related Work

  • System Model

  • Experiments

  • Evaluations

  • Conclusions

  • References

Abstract

Abstract

  • Scenario​

    • ​The contact time during which a mobile node is in range of a fixed road side unit (RSU) is short.
       

    • When transmitting a large amount of information, short contact times become problematic.

Abstract (cont.)

  • ​Current research
    • Millimeter-wave (mm-Wave) communication provides larger throughput than the existing technologies.
    • The mm-Wave band has extremely high attenuation.
       

  • Problem
    Can the high throughput of mm-Wave make up for the reduction in the contact time?

Abstract (cont.)

  • Solution
    Analyze the trade-off and design a first-of-its-kind practical mm-Wave vehicular testbed to evaluate the resulting performance.
     

  • Experiments
    • Consider alternative locations for the RSU other than at the side of the road.
    • Use fixed beam-steering at both the car and the RSU, thus avoiding costly beam-training.

Introduction

Introduction

  • Millimeter-wave (mm-Wave)

    • Achieve multi-gigabit-per-second performance

    • Successfully combined with other vehicular systems

    • Extremely high attentuation

      • Uses directional antennas to overcome​

Introduction (cont.)

  • Highly mobile scenarios

    • ​How can transceivers perform efficient beam-steering at vehicular speeds such that they always reach each other?

    • Where shall mm-Wave antennas be located on cars?

      • Place the antenna on a pole on top of the car but stongly disagree with aethetic considerations

    • How wide shall the bandwidth of the mm-Wave antennas on a car be?

    • How much data can transceivers enchange in a mobile scenario?

Introduction (cont.)

  • First-of-its-kind mm-Wave vehicular testbed

    • ​Operating in the 60 GHz band

    • Consider the road side units (RSUs) to transmit data to vehicles

    • The data includes content of large volumes. 

  • Propose an approach that eliminates the use of beem-steering by leveraging the characteristics of vehicular communications.

Introduction (cont.)

Related Work

Related Work

  • [1][2] Performs channel measurements in the 60 GHz band for links within a car and from the car to a unit located outside
    • Not consider the effects of actual data transmission involving
       
  • [3] Focuses on the design of a frame structure for cellular mm-Wave networks operating at 28 GHz
    • Not consider actual traffic
       
  • ​[4] Performs an in-depth indoor measurement characterizing the coverage, bit-rate, beam steering impact, blockage, and spatial reuse of 60 GHz links using a softare-radio platform

System Model

Definitions

  • Contact time

    • The time during which the RSU and the in-car-unit (ICU) can exchange data.

    • Directly related to the speed of the car and the distance that the car travels within the beam of the transmitter

      • Contact distance

Definitions (cont.)

  • Infrastructure

    • ​Transmitter: RSU

    • Receiver: ICU
       

  • ​Symbols

    • Transmission direction:

    • Transmit beamwidth at the RSU: 

\theta_t
θt\theta_t
\theta_{\mathrm{rsu}}
θrsu\theta_{\mathrm{rsu}}

Definitions (cont.)

  • Three RSU Cases

    • ​At the road side

    • On top of a bridge

    • Within a roundabout

Fig.: At the road side.

Fig.: On top of a bridge.

Fig.: Within a roundabout.

RSU at the Road Side

  • The contact distance      between the RSU and the ICU

    • ​For


       

    • For

       

d_c(\theta_t, \theta_{ \mathrm{rsu} }) = l \left( \tan { \left( \theta_t + \displaystyle{ \frac{ \theta_{\mathrm{rsu}} }{2} } \right) }-\tan { \left( \theta_t - \displaystyle{ \frac{ \theta_{\mathrm{rsu}} }{2} } \right) } \right)
dc(θt,θrsu)=l(tan(θt+θrsu2)−tan(θt−θrsu2))d_c(\theta_t, \theta_{ \mathrm{rsu} }) = l \left( \tan { \left( \theta_t + \displaystyle{ \frac{ \theta_{\mathrm{rsu}} }{2} } \right) }-\tan { \left( \theta_t - \displaystyle{ \frac{ \theta_{\mathrm{rsu}} }{2} } \right) } \right)

Fig.: RSU at        .

45 \degree
45°45 \degree
d_c(\theta_t = 90 \degree, \theta_{ \mathrm{rsu} }) = 2l \tan{ \displaystyle{ \left( \frac{ \theta_{ \mathrm{rsu} } }{2} \right) } }
dc(θt=90°,θrsu)=2ltan(θrsu2)d_c(\theta_t = 90 \degree, \theta_{ \mathrm{rsu} }) = 2l \tan{ \displaystyle{ \left( \frac{ \theta_{ \mathrm{rsu} } }{2} \right) } }
\theta_t = 90 \degree
θt=90°\theta_t = 90 \degree
0 \degree < \theta_t < 90 \degree
0°<θt<90°0 \degree < \theta_t < 90 \degree
d_c
dcd_c

contact distance

  • ​Transmission direction:

  • Transmit beam-width at the RSU:

\theta_t
θt\theta_t
\theta_{\mathrm{rsu}}
θrsu\theta_{\mathrm{rsu}}

RSU at the Road Side (cont.)

  • The contact time      at which the vehicle is driving       



     

  • We do not condiser angles              .

  • Let

  • Let 

t_c (\theta_t = 90 \degree, \theta_{ \mathrm{rsu} }) = \displaystyle{ \frac{ 2l \tan{ \left( \frac{ \theta_{\mathrm{rsu}} }{2} \right) } }{ v_{ \mathrm{icu} } } }
tc(θt=90°,θrsu)=2ltan(θrsu2)vicut_c (\theta_t = 90 \degree, \theta_{ \mathrm{rsu} }) = \displaystyle{ \frac{ 2l \tan{ \left( \frac{ \theta_{\mathrm{rsu}} }{2} \right) } }{ v_{ \mathrm{icu} } } }
\theta_t > 90 \degree
θt>90°\theta_t > 90 \degree
t_c (\theta_t, \theta_{ \mathrm{rsu} }) = \displaystyle{ \frac{ d_c(\theta_t, \theta_{ \mathrm{rsu} }) }{ v_{ \mathrm{icu} } } }
tc(θt,θrsu)=dc(θt,θrsu)vicut_c (\theta_t, \theta_{ \mathrm{rsu} }) = \displaystyle{ \frac{ d_c(\theta_t, \theta_{ \mathrm{rsu} }) }{ v_{ \mathrm{icu} } } }
v_{ \mathrm{icu} }
vicuv_{ \mathrm{icu} }

Fig.: RSU at       .

90 \degree
90°90 \degree
\theta_{\mathrm{rsu}} = 20 \degree
θrsu=20°\theta_{\mathrm{rsu}} = 20 \degree
t_c (45 \degree, \theta_{ \mathrm{ rsu } }) > 2 \times t_c (90 \degree, \theta_{ \mathrm{ rsu } })
tc(45°,θrsu)>2×tc(90°,θrsu)t_c (45 \degree, \theta_{ \mathrm{ rsu } }) > 2 \times t_c (90 \degree, \theta_{ \mathrm{ rsu } })
\theta_{\mathrm{rsu}} = 80 \degree
θrsu=80°\theta_{\mathrm{rsu}} = 80 \degree
t_c (45 \degree, \theta_{ \mathrm{ rsu } }) > 6 \times t_c (90 \degree, \theta_{ \mathrm{ rsu } })
tc(45°,θrsu)>6×tc(90°,θrsu)t_c (45 \degree, \theta_{ \mathrm{ rsu } }) > 6 \times t_c (90 \degree, \theta_{ \mathrm{ rsu } })

  • ​Transmission direction:

  • Transmit beam-width at the RSU:

\theta_t
θt\theta_t
\theta_{\mathrm{rsu}}
θrsu\theta_{\mathrm{rsu}}
t_c
tct_c

RSU at the Road Side (cont.)

  • While a wider beamwidth may provide a longer contact time, may also result in lower transmit rates since the antenna gain typically.
     

  • Optimal       and

     

    • Contact time: 

    • Transmission rate: 

\theta_t^*, \theta_{ \mathrm{rsu} }^* = \arg \mathop{ \max } \limits_{ \theta_t, \theta_{ \mathrm{ rsu } } } t_c(\theta_t, \theta_{ \mathrm{rsu} })R_m(\theta_t, \theta_{ \mathrm{ rsu }})
θt∗,θrsu∗=argmaxθt,θrsutc(θt,θrsu)Rm(θt,θrsu)\theta_t^*, \theta_{ \mathrm{rsu} }^* = \arg \mathop{ \max } \limits_{ \theta_t, \theta_{ \mathrm{ rsu } } } t_c(\theta_t, \theta_{ \mathrm{rsu} })R_m(\theta_t, \theta_{ \mathrm{ rsu }})
\theta_t^*
θt∗\theta_t^*
\theta_{ \mathrm{rsu} }^*
θrsu∗\theta_{ \mathrm{rsu} }^*

  • ​Transmission direction:

  • Transmit beam-width at the RSU:

\theta_t
θt\theta_t
\theta_{\mathrm{rsu}}
θrsu\theta_{\mathrm{rsu}}
R_m(\theta_t, \theta_{ \mathrm{ rsu }})
Rm(θt,θrsu)R_m(\theta_t, \theta_{ \mathrm{ rsu }})
t_c(\theta_t, \theta_{ \mathrm{ rsu }})
tc(θt,θrsu)t_c(\theta_t, \theta_{ \mathrm{ rsu }})

RSU at the Road Side (cont.)

  • Placing the RSU at the road side can result in reflections.

Fig.: Car reflection case. The row of cars are parked cars.

RSU on Top of a Bridge

  • The cars on the road cannot cause blockage.

  • The signal can receive all of the car from above.

    • May not hold when a car driving closely behind a large truck.

Large truck

Fig.: RSU located at the top of a bridge.

RSU within a Roundabout

  • The cars must slow down when driving in a roundabout.

    • Increases the contact time

  • ​Provides coverage in all directions

  • To compensate for the low gain of the omnidirectional antenna, the car can use a highly directional receive antenna.

  • The contact time for given a
    roundabout of radius 

t_c(\theta_{\mathrm{ rsu }}) = \displaystyle{ \frac { 2 \pi r \left(\frac{ \theta_{ \mathrm{ rsu } } }{ 360 \degree } \right) }{ v_{ \mathrm{ icu } } } }
tc(θrsu)=2πr(θrsu360°)vicut_c(\theta_{\mathrm{ rsu }}) = \displaystyle{ \frac { 2 \pi r \left(\frac{ \theta_{ \mathrm{ rsu } } }{ 360 \degree } \right) }{ v_{ \mathrm{ icu } } } }
r
rr

  • Contact time:

  • Transmit beam-width at the RSU:

t_c
tct_c
\theta_{\mathrm{rsu}}
θrsu\theta_{\mathrm{rsu}}

Experiments

Vehicular 60 GHz Testbed

  • Environment

    • RSU

      • Based on GNU Radio developed at RWTH Aachen

      • Along with a USRP X310 to generate a stream of 4-QAM modulated data

      • Sends this data to the external upconverter for transmission in the 60 GHz band

    • ICU

      • Carrier-frequency offset (CFO) compensation

      • Decoding

Vehicular 60 GHz Testbed (cont.)

  • Metrics

    • ​Receive time
      ​Period of time during which the receiver gets packets.

    • Contact time
      Period of time during which the receiver is able to receive correct packets without bit errors.

    • Others - throughput, BER, PER, SNR

Vehicular 60 GHz Testbed (cont.)

  • RSU Setup

    • Desktop PC

    • USRP X310

    • SiversIMA FC1005V/00 60 GHz upconverter



       

    • Horn antenna or omni-directional antenna

Vehicular 60 GHz Testbed (cont.)

  • ICU Setup - ​Same elements as the RSU

Vehicular 60 GHz Testbed (cont.)

  • Location

    • ​Temporarily installed the RSU at multiple locations in the city of Leganés, Spain.

    • All experiments were carried out under real traffic conditions at the beginning of August 2016.

Evaluations

RSU at the Road Side

  • Points the antenna of the RSU at               and the ICU at 

  • The faster the car,

    • the less time the transmit and the receive beams are aligned

    • the less packets are received

    • the lower the probability that we observe bit errors

\theta_t = 45 \degree
θt=45°\theta_t = 45 \degree
\theta_t = 0 \degree
θt=0°\theta_t = 0 \degree

Related to the antenna gain

RSU at the Road Side (cont.)

  • The wider the beam width, the larger the variance of SNRs.

Fig.: SNR distribution.

RSU at the Road Side (cont.)

  • Compares with lagacy 802.11p-like WiFi

  • Assumes a circular range of 100 m and a fixed rate of 27 Mbps

  • Achieves higher throughput only holds for narrow beam width

Only holds for narrow beam

Fig.: 60 GHz vs. legacy WiFi.

RSU on Top of a Bridge

  • Achieves higher contact times in bridge scenario compared to the road side case

Fig.: Communication times.

RSU on Top of a Bridge (cont.)

  • Achieves higher contact times in bridge scenario compared to the road side case

Fig.: In the Bridge Scenario.

Fig.: In Road Side Scenario.

The increase in contact
time when switching angles.

RSU on Top of a Bridge (cont.)

  • Communication gaps occur when antennas are temporarily misaligned due to, e.g., bump.

Fig.: Communication gaps.

RSU on Top of a Bridge (cont.)

  • The contact progress reflects the normalized contact time.

  • The bridge case is more stable than the road with bumps.

Fig.: SNR fluctuations.

Shakiness has less influence.

RSU within a Roundabout

  • Considers an RSU placed at the center of a roundabout

  • The omnidirectional antenna has a peak of its radiation pattern when the car is driving through a roundabout.

30 km/h

Path loss

Path loss

Road Side Reflections

Fig.: Packet arriving at the ICU via a car reflection.

  • The length of the main burst increases with the beam width.

    • The reflected path becomes wider.

20 km/h

Isolated packets

Car Blockage

  • Higher heights results in lower PER.

  • Wider beam widths tend to yield better results.

  • The transmission through parallel cars is feasible.

    • Easily result in link loss

Fig.: PER for transimission through cars multiple heights.

Height

Conclusions

Conclusions

  • The road geometry and the arrival direction of cars are known in vehicular scenarios.

    • Fixed beam-steering and beam width angles
       

  • Builds a first-of-its-kind SDR-based practical vehicular testbed

    • Enables 60 GHz packet-level transmissions
       

  • Our approach is feasible and yields significant throughput gains compared to legacy 802.11p-based systems.

Contributions

  • A simple yet effective approach to select a fixed antenna steering for multiple RSU scenarios

    • at the road side, on a bridge, and in a roundabout
       

  • The impact of antenna beam width in the above RSU scenarios, and analytically optimize it
     

  • The impact of reflectors in vehicular scenarios

    • the walls of buildings or cars parked nearby

References

References

  • Related work

    1. J. Blumenstein, T. Mikulasek, A. Prokes, T. Zemen, and C. Mecklen-brauker, "INtra-Vehicular Path Loss Comparison of UWB Channel for 3-11 GHz and 55-65 GHz," IEEE IXUWB, 2015.

    2. E. Ben-Dor, T. S. Rappaport, Y. Qiao, and S. J. Lauffenburger, "Millimeter-Wave 60 GHz Outdoor and Vehicle AOA Propagation Measurements Using a Broadband Channel Sounder," IEEE GLOBECOM, 2011.

    3. Y. Kim, H.-Y. Lee, P. Hwang, R. K. Patro, J. Lee, W. Roh, and K. Cheun, "Feasibility of Mobile Cellular Communications at Millimeter Wave Frequency," IEEE Journal of Selected Topics in Signal Processing, 2016.

    4. S. Sur, V. Venkateswaran, X. Zhang, and P. Ramanathan, "60 GHz Indoor Networking Through Flexible Beams: A Link-Level Profiling," ACM SIGMETRICS Performance Evaluation Review, 2015.  

Made with Slides.com