PaperlayerView exact revision on arXiv

Source-linked AI summary

Millimeter Wave Vehicular Communication to Support Massive Automotive Sensing

Junil Choi, Vutha Va, Nuria Gonzalez-Prelcic, Robert Daniels, Chandra R. Bhat, Robert W. Heath

arXiv:1602.06456v2cs.IT

TL;DR

Future vehicles generate sensor data at rates that exceed the practical capacity of DSRC and 4G for raw-data exchange. This paper motivates mmWave V2X, examines its architectures and challenges, and proposes using DSRC or sensor-derived side information to reduce beam-alignment overhead. Examples and simulations show substantial overhead reduction with limited or marginal performance degradation in the evaluated cases.

  • Problem

    DSRC and 4G do not provide sufficient data rates for exchanging the terabyte-per-hour raw sensor data generated by next-generation vehicles.

  • Method

    The paper proposes mmWave V2X architectures and uses DSRC or automotive-sensor information as side information for beam alignment.

  • Results

    The proposed side-information approach reduces beam-alignment overhead, with average training requiring around 10 beam pairs versus exhaustive-search possibilities of 552 and 3180 for 8 × 8 and 16 × 16 UPAs.

  • Takeaways & Limitations

    Position information from DSRC or sensors can restrict candidate beams and reduce alignment overhead without performance loss in the restricted search-space formulation.

  • Takeaways & Limitations

    With a 16 × 16 UPA, the proposed approach has around 8 dB average receive-power loss, although more than half of cases show negligible degradation.

Abstract

from arXiv · show

As driving becomes more automated, vehicles are being equipped with more sensors generating even higher data rates. Radars (RAdio Detection and Ranging) are used for object detection, visual cameras as virtual mirrors, and LIDARs (LIght Detection and Ranging) for generating high resolution depth associated range maps, all to enhance the safety and efficiency of driving. Connected vehicles can use wireless communication to exchange sensor data, allowing them to enlarge their sensing range and improve automated driving functions. Unfortunately, conventional technologies, such as dedicated short-range communication (DSRC) and 4G cellular communication, do not support the gigabit-per-second data rates that would be required for raw sensor data exchange between vehicles. This paper makes the case that millimeter wave (mmWave) communication is the only viable approach for high bandwidth connected vehicles. The motivations and challenges associated with using mmWave for vehicle-to-vehicle and vehicle-to-infrastructure applications are highlighted. A high-level solution to one key challenge - the overhead of mmWave beam training - is proposed. The critical feature of this solution is to leverage information derived from the sensors or DSRC as side information for the mmWave communication link configuration. Examples and simulation results show that the beam alignment overhead can be reduced by using position information obtained from DSRC.

I. INTRODUCTION

Increasing sensor counts and data rates make raw sensor-data exchange valuable for connected vehicles, but DSRC and 4G cannot provide the required throughput. The paper motivates mmWave V2X and proposes using DSRC or sensor information to reduce beam-alignment overhead.

  • Motivation: Vehicles are generating more sensor data as automation advances, increasing the data handled by active-safety algorithms.The average number of sensors was around 100 and was expected to double by 2020; LIDAR data rates further increase vehicle data volumes.
  • Motivation: Wireless connectivity can extend sensing beyond line-of-sight and enable vehicles to exchange higher-rate raw sensor data.The paper calls vehicles exchanging minimally preprocessed raw sensor data fully connected vehicles.
  • Existing technologies: DSRC provides only 2-6 Mbps in practice, while 4G reaches at most 100 Mbps for high mobility, insufficient for terabyte-per-hour vehicle data rates.These rates are contrasted with the data volumes expected from next-generation vehicles.
  • Proposed approach: The paper motivates mmWave as a complement to DSRC or 4G because it can provide gigabit-per-second channels for raw sensor-data exchange.Candidate implementations include 5G cellular, modified IEEE 802.11ad, and a dedicated new standard.
  • Proposed approach: The proposed beam-alignment approach exploits DSRC or sensor information as side information to reduce alignment and tracking overhead.The paper focuses on position information obtained from DSRC as a concrete example.

II. SENSING FOR NEXT GENERATION VEHICLES

The paper reviews radar, camera, and LIDAR sensing for next-generation vehicles, emphasizing their complementary capabilities, drawbacks, and substantial data-generation requirements.

  • Automotive radar: Radar supports active-safety functions by estimating neighboring objects’ existence, location, and velocity from reflected waveforms.Radar serves long-, medium-, and short-range applications including adaptive cruise control, blind-spot detection, and parking assistance.
  • Automotive radar: Radar data are heavily post-processed into relevant-target point maps and may not identify whether detected objects are trucks, sedans, or motorcycles.Radar-generated rates range from kbps for point maps to hundreds of Mbps, according to the passage’s truncated discussion.
  • Automotive cameras: Cameras provide visible or infrared views for safety applications, but their generated data volume is large.Examples include speed-sign detection, blind-spot monitoring, lane-departure assistance, and driver-monitoring applications.
  • LIDAR: LIDAR scans with narrow laser beams to create high-resolution 3D images whose pixels include depth.Its 360-degree field of view supports 3D mapping and detection of cars, bicycles, and pedestrians.
  • Comparison: Table I summarizes the purposes, drawbacks, and data rates of automotive radar, camera, and LIDAR systems.The listed sensor data rates come from commercial-product specifications and conversations with industrial partners.

III. MMWAVE V2X COMMUNICATIONS

The paper outlines mmWave V2X architectures for high-rate sensing exchange and identifies antenna, blockage, mobility, and standardization considerations. It also proposes using lower-frequency side information to assist mmWave link setup.

  • MmWave fundamentals: MmWave uses large spectral channels, higher-order modulation, and MIMO to support higher rates than sub-6-GHz systems.IEEE 802.11ad uses 2.16 GHz at 60 GHz and supports up to 7 Gbps.
  • MmWave fundamentals: MmWave links require many antennas to form sharp transmit and receive beams, motivating analog or hybrid beamforming.Hybrid beamforming can achieve performance similar to full digital beamforming while using fewer RF chains than antennas.
  • V2X architectures: Vehicles may deploy multiple mmWave transceivers on bumpers, sides, and rooftops to support V2V and V2I links while mitigating blockage.The proposed conceptual architecture places rooftop transceivers for infrastructure links and other positions for vehicle links.
  • Link setup: A low-frequency control plane using DSRC or 4G can help establish mmWave links in 5G-based V2X systems.This approach leverages existing control connectivity while reserving mmWave for high-rate communication.

IV. CHALLENGES FOR MMWAVE IN V2X

The paper identifies three major challenges for mmWave V2X: accurate vehicular channel models, sufficient deployment penetration, and simple, fast beam alignment.

  • Challenges: MmWave V2X lacks accurate vehicular channel models that capture propagation, antenna placement, and blockage effects.Existing sub-6-GHz models must be modified and extended, and further mmWave measurements are needed.
  • Challenges: Early mmWave V2X deployments may have too few capable vehicles to realize the full benefits of especially V2V communication.High penetration could also create excessive interference in heavily loaded traffic conditions.
  • Challenges: Vehicular mobility creates a need for simple and fast mmWave beam-alignment algorithms.Beam alignment is listed as a central challenge alongside channel modeling and deployment penetration.

A. MmWave vehicular channel modeling

MmWave vehicular channel models remain incomplete because existing sub-6 GHz models require extension and additional measurements for realistic vehicle-to-vehicle propagation.

  • A. MmWave vehicular channel modeling: Existing vehicular channel models below 6 GHz must be modified and extended for mmWave bands.New parameters need to be inferred from measurement data.
  • A. MmWave vehicular channel modeling: Further mmWave measurements must characterize impulse responses between antenna arrays in vehicular scenarios.Measurements are also needed to capture antenna locations and blockage from vehicles, buildings, and pedestrians.
  • A. MmWave vehicular channel modeling: A practical channel model must incorporate antenna placement and blockage effects from nearby vehicles, buildings, and pedestrians.

B. Penetration rate of mmWave V2X-capable vehicles

Early mmWave V2X deployment may not deliver the full benefits of vehicle connectivity because too few vehicles may support V2X, especially for V2V communication.

  • B. Penetration rate of mmWave V2X-capable vehicles: Insufficient penetration of V2X-capable vehicles is an early deployment issue for realizing the full benefits of mmWave V2X.The concern is especially relevant to V2V communications.
  • B. Penetration rate of mmWave V2X-capable vehicles: High penetration rates may create excessive interference in highly loaded traffic conditions.Narrow mmWave beams and advanced MAC protocols are identified as possible mitigations.
  • B. Penetration rate of mmWave V2X-capable vehicles: Joint mmWave radar and communication systems may increase the penetration of mmWave V2X-capable vehicles during early deployment.The proposed rationale includes shared vehicle space, cost, and power savings.
  • B. Penetration rate of mmWave V2X-capable vehicles: A preliminary IEEE 802.11ad-based study achieved 0.1m range-estimation accuracy and 0.1m/s velocity-estimation accuracy using the preamble as a radar signal.

C. MmWave beam alignment overhead

MmWave’s narrow beams make alignment essential but costly. The paper proposes using relative-position information from automotive sensors or DSRC as side information to reduce alignment and tracking overhead, while retaining alignment to handle position uncertainty.

  • C. MmWave beam alignment overhead: MmWave beam alignment is critical because sharp transmit and receive beams require proper alignment before communication can proceed.The brute-force approach tests all transmit and receive beam pairs sequentially, creating high overhead.
  • C. MmWave beam alignment overhead: Automotive sensors or DSRC can provide neighboring vehicles’ relative-position information to mitigate mmWave beam-alignment and tracking overhead.The approach exploits this information as side information for link configuration.
  • C. MmWave beam alignment overhead: Automotive radar can estimate detected objects’ distance, angle, and velocity, although it may not identify whether they are vehicles, obstacles, or infrastructure.Combining multiple sensors may improve position accuracy.
  • C. MmWave beam alignment overhead: DSRC basic safety messages provide absolute position, velocity, size, and heading information that vehicles can use to deduce neighboring vehicles’ relative positions.Infrastructure can likewise broadcast its absolute position for mmWave V2I beam alignment.
  • C. MmWave beam alignment overhead: Position side information does not eliminate beam alignment because GPS noise, limited satellite visibility, and unknown transceiver locations can cause position errors.Vehicles can nevertheless use relative position and size to adaptively select candidate beams.
  • C. MmWave beam alignment overhead: The proposed idea is illustrated through examples using DSRC or automotive-sensor position information to reduce mmWave beam-alignment overhead.

VI. EVALUATION ON MMWAVE V2X BEAM ALIGNMENT

The evaluation presents example scenarios where DSRC or automotive sensors can reduce mmWave beam-alignment overhead and uses numerical results to assess the idea in practical vehicular environments.

  • VI. EVALUATION ON MMWAVE V2X BEAM ALIGNMENT: Example scenarios demonstrate how DSRC or automotive sensors can reduce mmWave beam-alignment overhead.
  • VI. EVALUATION ON MMWAVE V2X BEAM ALIGNMENT: Numerical results evaluate the proposed idea under practical vehicular-environment conditions.
  • VI. EVALUATION ON MMWAVE V2X BEAM ALIGNMENT: The evaluation connects illustrative scenarios with numerical assessment of the proposed beam-alignment approach.

A. Illustrative examples

Position information from DSRC or automotive sensors can restrict mmWave beam searches, reducing alignment overhead without performance loss. The benefit is greater in V2I, where wider angular regions and 3D beamforming enlarge the search space.

  • V2V scenario: Position information from DSRC or automotive sensors restricts candidate beams and reduces mmWave beam alignment overhead without performance loss.Advanced alignment methods can further reduce overhead within the restricted search space.
  • V2I scenario: V2I benefits more from position side information because infrastructure placement creates wider angular search regions than V2V.Different vehicle and infrastructure heights also require 3D beamforming, increasing the beam search space.
  • V2I scenario: Infrastructure can exploit stationary-environment knowledge, DSRC, and real-time 3D maps to restrict beam searches for entering vehicles.Preferred beams for specific regions provide additional prior information.

B. Numerical evaluation

A ray-tracing evaluation of 60 GHz V2I links shows that position-assisted beam selection sharply reduces training attempts. The reduction has negligible average power loss for 8 × 8 arrays but incurs about 8 dB average loss for 16 × 16 arrays.

  • Evaluation setup: The evaluation uses ray tracing for realistic 60 GHz V2I environments with infrastructure and a communicating vehicle establishing links by beam alignment.Vehicle spacing follows an Erlang distribution, and multiple vehicle types are considered.
  • Evaluation setup: Training beams use Kronecker products of horizontal and vertical DFT vectors for N × N uniform planar arrays.The resulting beams support three-dimensional beam patterns.
  • Beam-training overhead: 552 and 3180 exhaustive-search attempts are reduced to around 10 average training beam pairs for both 8 × 8 and 16 × 16 UPAs.The simulation divides roads into 5 m grids and uses prior angle-of-arrival and angle-of-departure information for each grid.
  • Performance trade-off: The proposed method has negligible receive-power loss with the 8 × 8 UPA but around 8 dB average loss with the 16 × 16 UPA.More than half of the 16 × 16 cases have negligible performance degradation.
  • Performance trade-off: Overall, the proposed method significantly reduces mmWave beam-alignment overhead while maintaining marginal performance degradation.The authors expect more advanced beam-alignment algorithms to mitigate the observed loss.

VII. CONCLUSIONS

The paper argues that future connected vehicles need mmWave alongside existing V2X technologies to exchange raw sensor data. It proposes using DSRC or automotive sensors as side information to reduce beam-alignment overhead and discusses deployment options and multiple transceivers.

  • Conclusions: DSRC and 4G cellular systems are insufficient for large-scale raw sensor-data exchange in future connected vehicles.The paper focuses on data from automotive sensors such as cameras and LIDARs.
  • Conclusions: The paper proposes 5G cellular, modified IEEE 802.11ad, or a dedicated new standard as possible mmWave vehicular-network approaches.These options are presented for exchanging large volumes of automotive-sensor data.
  • Conclusions: Multiple mmWave transceivers are proposed to mitigate blockage and improve spatial packing in vehicular environments.The conclusion also identifies DSRC and automotive sensors as side information for reducing beam-alignment overhead.
  • Conclusions: The paper expects out-of-band sensor side information to help mmWave systems achieve sufficient link quality with reduced control overhead.This expectation extends beyond mmWave V2X communications to mmWave communication systems generally.
Loading 1602.06456v2…