PaperlayerView exact revision on arXiv

Source-linked AI summary

Joint Radar-Communications Strategies for Autonomous Vehicles

Dingyou Ma, Nir Shlezinger, Tianyao Huang, Yimin Liu, Yonina C. Eldar

arXiv:1909.01729v2cs.IT

TL;DR

Joint radar-communications design for autonomous vehicles raises a relatively fresh resource-allocation problem. This survey reviews automotive radar basics, maps DFRC strategies, and reports that OFDM improves communications over IM-FAR while radar performance remains relatively similar, with no single method suitable for all scenarios.

  • Problem

    Allocating resources to optimize radar and communications jointly remains a relatively fresh research area.

  • Method

    The article reviews automotive radar fundamentals and maps existing DFRC strategies into four main categories.

  • Results

    OFDM achieves improved communications performance over IM-FAR while their radar performance is relatively similar; no single DFRC method suits all scenarios and requirements.

  • Takeaways & Limitations

    DFRC systems provide potential gains in performance, size, cost, power consumption, and robustness for autonomous vehicles.

  • Takeaways & Limitations

    A major limitation of the discussed approach is its use of a single directed beam.

Abstract

from arXiv · show

Self-driving cars constantly asses their environment in order to choose routes, comply with traffic regulations, and avoid hazards. To that aim, such vehicles are equipped with wireless communications transceivers as well as multiple sensors, including automotive radars. The fact that autonomous vehicles implement both radar and communications motivates designing these functionalities in a joint manner. Such dual function radar-communications (DFRC) designs are the focus of a large body of recent works. These approaches can lead to substantial gains in size, cost, power consumption, robustness, and performance, especially when both radar and communications operate in the same range, which is the case in vehicular applications. This article surveys the broad range of DFRC strategies and their relevance to autonomous vehicles. We identify the unique characteristics of automotive radar technologies and their combination with wireless communications requirements of self-driving cars. Then, we map the existing DFRC methods along with their pros and cons in the context of autonomous vehicles, and discuss the main challenges and possible research directions for realizing their full potential.

I. INTRODUCTION

Autonomous vehicles combine radar sensing with wireless communications to navigate safely and interact with surrounding entities. The article surveys DFRC strategies to match these shared functions with different vehicular scenarios and requirements.

  • I. INTRODUCTION: Autonomous vehicles use radar and communications to sense and interact with complex surroundings in real time.Communications support information exchange with vehicles, infrastructure, pedestrians, networks, and cloud applications.
  • I. INTRODUCTION: Automotive radar must detect many nearby scatterers in dense urban environments while operating under tighter size, power, and cost constraints than conventional radar.Radar is also valued for detecting distant objects and robustness to poor visibility.
  • I. INTRODUCTION: DFRC jointly designs radar and communications instead of deploying separate systems that transmit and process electromagnetic signals independently.The two functions share system resources, motivating integrated designs for vehicular applications.
  • I. INTRODUCTION: Joint radar-communications designs can reduce antennas, system size, weight, and power consumption while alleviating electromagnetic compatibility and spectrum-congestion concerns.These potential gains make DFRC attractive for autonomous vehicles.
  • I. INTRODUCTION: The proliferation of DFRC strategies makes it difficult to determine which scheme suits each autonomous-driving scenario.Communications-waveform approaches favor throughput but offer limited sensing, whereas radar-waveform approaches support lower-rate additional communications channels.
  • I. INTRODUCTION: The article reviews automotive radar, surveys four DFRC categories, and maps representative methods by radar capability, information rate, and complexity.The categories are coordinated separated signals, communications waveform-based, radar waveform-based, and dedicated dual-function waveform designs.

II. BASICS OF AUTOMOTIVE RADAR

Automotive radar has requirements distinct from conventional radar, including short-range detection of many objects, low implementation cost, and interference robustness. The section introduces FMCW processing and the four broad DFRC strategy categories.

  • II. BASICS OF AUTOMOTIVE RADAR: Automotive radar detects many objects at ranges of a few tens of meters, unlike conventional radar’s smaller number of targets at tens or hundreds of kilometers.Automotive systems are deployed in mass-produced vehicles and therefore face stricter implementation constraints.
  • II. BASICS OF AUTOMOTIVE RADAR: Automotive radar requires low cost, size, power consumption, and spectral efficiency, while remaining robust to interference in dense urban deployments.No single radar scheme handles the complete set of requirements.
  • II. BASICS OF AUTOMOTIVE RADAR: Existing DFRC methods fall into coordinated separated signals, communications waveform-based, radar waveform-based, and joint dual-function waveform categories.The categories are evaluated through their advantages and disadvantages in autonomous-vehicle contexts.
  • II. BASICS OF AUTOMOTIVE RADAR: FMCW is a constant-modulus radar waveform with linearly modulated frequency that can be generated and detected using simplified hardware.The example uses one transmit antenna, a receive ULA, and periodically transmitted pulses.
  • II. BASICS OF AUTOMOTIVE RADAR: After de-chirp, the received waveform has a much lower frequency and can be sampled with low-speed ADCs.A three-dimensional DFT of the sampled signal recovers target range, velocity, and direction across fast-time, slow-time, and spatial domains.

A. Separate Coordinated Signals

Separate coordinated-signal DFRC designs transmit distinct radar and communications signals, reducing cross-interference through time, frequency, or spatial separation. Resource partitioning creates performance trade-offs, while spectral interleaving can preserve radar resolution close to full-spectrum operation.

  • Signal separation: Separate radar and communications signals mitigate cross-interference through time/frequency division or spatial beamforming.These strategies coordinate dedicated signals rather than relying on a single dual-function waveform.
  • Time/Frequency Division: Time/frequency division assigns radar and communications to different bands or time slots, inevitably trading radar against communications performance.The trade-off arises because spectrum and time resources are divided between the two subsystems.
  • Time/Frequency Division: OFDM permits optimized spectrum division through subcarrier-selection matrices that assign radar and communications symbols across N subbands.The selection matrix determines how the bandwidth is divided at each transmit element.
  • Time/Frequency Division: Spectral interleaving yields radar resolution comparable to using the complete spectrum.This result applies when the subcarrier-selection matrix contains multiple interleaved zero and one blocks.
  • Resource allocation: Resource allocation can be fixed, random, or optimized using statistical knowledge of target responses and communications channels.Optimization can combine radar target-echo mutual information with communications input-output mutual information.

2) Spatial Beamforming:

Spatial beamforming jointly designs radar and communications transmission to suppress mutual interference while meeting sensing and communications constraints. It can support simultaneous full-band operation, but depends on channel knowledge and may require substantial optimization.

  • 2) Spatial Beamforming:: Radar interference can be reduced by projecting the radar waveform into the communications channel's null space.The approach relies on spatial processing based on the channel response to the communications receiver.
  • 2) Spatial Beamforming:: Joint beamforming matrices for radar and communications mitigate cross-interference while satisfying performance requirements.One example maximizes communications-receiver SINR subject to a radar beampattern constraint.
  • 2) Spatial Beamforming:: Spatial beamforming can let both functionalities use the full bandwidth simultaneously across all time slots.This flexibility may improve the trade-offs created by time/frequency resource division.
  • 2) Spatial Beamforming:: Fast-moving vehicles may lack the a-priori channel knowledge required to design effective spatial beamformers.This availability issue limits the practicality of channel-dependent spatial coordination in automotive settings.
  • 2) Spatial Beamforming:: Time/frequency division is considered more attractive for automotive applications because spatial optimization can require considerable computation.Fixed sub-optimal allocations such as spectral interleaving may therefore be preferable.

B. Communications Waveform-Based Schemes

Communications waveform-based DFRC reuses spread-spectrum, OFDM, or structured vehicular signals for radar probing. OFDM is especially prominent because it serves both domains, but mobility and hardware constraints limit its automotive performance.

  • B. Communications Waveform-Based Schemes: Communications waveform-based DFRC uses standard signals for probing, with OFDM receiving particular attention in automotive applications.The surveyed approaches include spread spectrum, shared OFDM, and structured vehicular communications protocols.
  • 1) Spread Spectrum Waveforms:: Spread-spectrum DFRC suffers limited radar dynamic range, computationally complex velocity recovery, and costly wideband ADC requirements.Imperfect spreading-sequence autocorrelation contributes to the dynamic-range limitation, while de-chirp processing used in FMCW is unavailable.
  • 2) OFDM Waveforms:: OFDM is attractive for DFRC because it is spectrally efficient, flexible, adaptable, and widely used in both radar and communications.Unlike FMCW, OFDM does not suffer from range-Doppler coupling.
  • 2) OFDM Waveforms:: OFDM radar estimates target range and velocity with matched filtering followed by two-dimensional DFT processing across carrier and slow-time domains.Dividing received subcarriers by their corresponding transmitted symbols removes data dependency before estimation.
  • 2) OFDM Waveforms:: OFDM subcarriers can be assigned among transmit antennas using equidistant, non-equidistant, or random interleaving schemes.Weighted OFDM can also control the maximum peak-to-average power ratio.
  • 2) OFDM Waveforms:: Shared OFDM in moving vehicles suffers subcarrier misalignment, high-rate ADC requirements, and high peak-to-average power ratio.These effects degrade maximal radar unambiguous range, increase cost and power consumption, and can distort signals in nonlinear amplifiers.

MIMO Radar

MIMO radar uses multiple transmit and receive antennas with orthogonal waveforms to form a larger virtual aperture. In automotive DFRC, communications-based probing offers integration benefits but constrains radar coverage, duty cycle, and detection capability.

  • MIMO Radar: MIMO radar transmits orthogonal waveforms from multiple antennas and combines transmit and receive steering responses.The formulation uses LT transmit elements, LR receive elements, and antenna spacings dT and dR.
  • MIMO Radar: A virtual array formed from multiple transmit and receive antennas increases angular resolution without requiring additional hardware elements.The equivalent angle resolution matches a phased array with LT LR receive antennas, with angular resolution enhanced by a factor of LT.
  • Communications waveform-based probing: Protocol-based DFRC can use known IEEE 802.11ad preambles for radar probing while minimally affecting communications functionality.Radar-suitable preamble design has also been studied because the preamble influences radar performance.
  • Communications waveform-based probing: Directional mmWave communications restrict radar detection to targets in the assigned beam direction after the data link is established.Scanning-area extensions are possible at the cost of power reduction.
  • Communications waveform-based probing: Protocol-based radar has a relatively low duty cycle because only the communications preamble is used for probing.Its detection range is also limited under vehicular peak-power constraints.
  • Communications waveform-based probing: Shared OFDM and communications waveforms support high data rates using conventional digital communications schemes but require costly wideband hardware.These approaches are nevertheless considered promising for autonomous-vehicle DFRC.

C. Radar Waveform-Based Techniques

Radar waveform-based DFRC embeds communication data into conventional radar waveforms through digital modulation or index modulation. These approaches can improve throughput or preserve power efficiency, but tradeoffs include limited rates, higher complexity, channel-knowledge requirements, and power-efficiency concerns.

  • C. Radar Waveform-Based Techniques: Radar waveforms can carry digital messages through modulation or index modulation while retaining their radar role.Index modulation conveys bits through radar-parameter indices rather than conventional symbol modulation.
  • C. Radar Waveform-Based Techniques: Phase or frequency modulation modifies conventional FMCW pulses to encode information in phase or frequency-slope parameters.A positive frequency-modulation rate can represent one bit and a negative rate another.
  • C. Radar Waveform-Based Techniques: Power-efficient, low-complexity modulation schemes provide very limited communication rates.The limitation is explicitly associated with phase- and frequency-modulated radar waveforms.
  • C. Radar Waveform-Based Techniques: Multiple orthogonal waveforms can increase communication rates, but they also raise system complexity and require transmitter knowledge of the communications channel.The receiver uses matched filtering to recover the waveform vector, after which amplitude or phase can convey the data.
  • C. Radar Waveform-Based Techniques: Amplitude- or phase-based communication can improve rates, but constant-modulus transmission is difficult to guarantee and may reduce power efficiency.The cited design also requires a-priori channel knowledge at the transmitter.

2) IM-Based Techniques:

Index-modulation DFRC embeds data in radar transmission parameters such as antenna allocation, frequency, time slot, or waveform indices. MIMO-radar variants encode bits through waveform assignments, providing rates determined by the number of available arrangements.

  • 2) IM-Based Techniques:: Index modulation embeds data bits into transmission building-block indices that are also important radar waveform parameters.Examples include spatial allocation and frequency division.
  • 2) IM-Based Techniques:: Index-modulation DFRC can retain conventional radar schemes by encapsulating communications in transmission parameters.The indexed parameters include carrier frequency, time slot, antenna allocation, and orthogonal waveforms.
  • 2) IM-Based Techniques:: MIMO-radar index modulation assigns orthogonal waveforms across transmit antennas, yielding LT! arrangements and a maximal rate of log2 LT! bits per PRI.This construction directly maps each waveform assignment to communication bits.
  • 2) IM-Based Techniques:: Sparse-array MIMO index modulation activates only K of LT transmit elements in each PRI.The supplied passage identifies this as an extension of the basic MIMO-radar approach but does not state the resulting rate.

Frequency Agile Radar

Frequency-agile radar changes carrier frequencies to address congested radar environments, and index modulation can embed communications in frequency permutations and antenna allocations. These methods preserve radar functionality relatively well but offer limited communications throughput and may increase decoding complexity.

  • Frequency Agile Radar: FAR randomly changes carrier frequencies across pulses, providing wideband coverage with narrowband waveforms and helping mitigate interference.FAR is presented as a promising approach for mutual interference in congested environments.
  • Frequency Agile Radar: The IM-FAR transmission pattern count combines antenna-allocation patterns with carrier-selection combinations, while the message is embedded in their joint pattern.The supplied passages give the antenna-allocation factor LT!/(LK!)^K and describe the carrier-selection factor without a visible numeric expression.
  • Frequency Agile Radar: Random carrier-frequency changes can raise sidelobe levels and affect weak-target detection, motivating compressed-sensing range-Doppler processing.Recovery guarantees are described for sparse and block-sparse target scenes.
  • Frequency Agile Radar: Frequency-permutation index modulation provides N! possible carrier-frequency patterns for a carrier set with N frequencies.The permutations are used for information embedding.
  • Frequency Agile Radar: IM-FAR selects K carriers from a set F and allocates them across K antenna sub-arrays, embedding data in both carrier selection and antenna allocation.The scheme was illustrated with a hardware prototype equipped with 64 antenna elements.
  • Frequency Agile Radar: Radar detection in IM-based DFRC uses standard radar processing, while communications symbols can be detected by maximum likelihood or reduced-complexity IM detectors.FAR detection is described as matched filtering followed by compressed-sensing recovery.
  • Frequency Agile Radar: Radar performance of MIMO radar and FAR combined with IM is reported as roughly equivalent to the corresponding radar-only counterparts.This is identified as the main advantage of radar-waveform-based DFRC methods.
  • Frequency Agile Radar: Radar-waveform-based DFRC typically has limited communication throughput and increased decoding complexity, making it better suited as an additional channel than a replacement for cellular vehicular communications.The stated scope positions these methods alongside existing communications rather than as substitutes.

D. Joint Waveform Design

Joint waveform design directly optimizes radar and communications objectives rather than adapting traditional signalling. It can control the radar–communications tradeoff, but automotive deployment is constrained by optimization complexity and the need for timely channel knowledge.

  • D. Joint Waveform Design: Traditional signalling offers established performance and compatibility with existing hardware, but was not originally designed for DFRC scenarios.The paper contrasts this practicality with the potential improvements of dedicated dual-function waveforms.
  • D. Joint Waveform Design: Dedicated joint waveforms are designed around both radar and communications performance, including throughput, complexity, and hardware requirements.Their objective is explicitly dual-function rather than inherited from one conventional signalling scheme.
  • D. Joint Waveform Design: The joint signal X is designed for multiple single-antenna receivers using a joint beamformer, channel matrix, and additive-noise model.The formulation targets desired observations at communications receivers and at a radar target direction.
  • D. Joint Waveform Design: Unconstrained observations in other directions can produce high sidelobes outside the mainlobe, so signal-covariance constraints can restrict the radar beampattern.The cited designs also impose receiver SINR, interference, constant-modulus, or waveform-similarity constraints.
  • D. Joint Waveform Design: Dedicated dual-function waveforms can balance radar and communications in a controllable manner and potentially realize achievable performance tradeoffs through joint optimization.This benefit is presented as theoretical rather than as an automotive deployment result.
  • D. Joint Waveform Design: Automotive application remains limited because joint-waveform optimization is relatively complex, depends on prior channel knowledge, and must be frequently repeated for fast-moving vehicles.The repeated optimization induces increased computational burden when accurate instantaneous channel knowledge is difficult to obtain.

E. Discussion

The surveyed DFRC schemes differ in radar–communications trade-offs, complexity, hardware needs, and suitability for vehicular scenarios. No single method satisfies all autonomous-vehicle requirements, so selection must be scenario-dependent.

  • No unified joint measure currently enables rigorous evaluation of different DFRC schemes.
  • Numerical comparison: OFDM-based DFRC achieves improved communications performance over IM-FAR while maintaining relatively similar radar performance in interference-free conditions.The comparison targets a 10 m range target moving at 5 m/s over a Rayleigh flat-fading channel.
  • Numerical comparison: In dense urban-like scenarios, frequency-agile radar is expected to mitigate mutual interference more effectively through random spectral sparsity.
  • Communications waveform-based schemes: OFDM and other communications-waveform methods support high data rates, but directed-beam operation, costly hardware, and vehicle motion can degrade automotive suitability.OFDM uses conventional digital communications signals and exhibits subcarrier misalignment when used from moving vehicles.
  • Radar waveform-based schemes: Radar-waveform schemes integrate naturally with automotive radar, while IM-based methods offer robustness and reduced complexity but limited bit-rates and decoding complexity.IM-based schemes are therefore positioned as additional communications channels rather than replacements for cellular links or dedicated radar.
  • DFRC strategies: Joint waveform design can achieve a desired radar–communications performance trade-off by optimizing a dual-function waveform under combined constraints.
  • Conclusion: No single DFRC method suits all autonomous-vehicle scenarios, making understanding each approach’s advantages and disadvantages important for technology selection.

IV. CONCLUSIONS AND FUTURE CHALLENGES

This survey reviews DFRC designs for autonomous vehicles, organizes existing strategies into four categories, and evaluates their trade-offs against vehicular radar and communications requirements. It concludes that no single scheme fits every scenario while identifying theoretical, algorithmic, resource-allocation, and practical challenges.

  • Survey scope and taxonomy: The survey reviews automotive radar basics and maps DFRC strategies into four main categories for autonomous-vehicle applications.The radar discussion includes FMCW, frequency-agile, and MIMO waveforms; the strategy taxonomy covers coexistence, communications-waveform, radar-waveform, and joint-waveform approaches.
  • Supported conclusion: No single DFRC scheme is suitable for all self-driving scenarios and requirements.The conclusion is repeated as a central result of the survey’s analysis.
  • Future challenges: A unified performance measure is lacking, so comparing DFRC approaches requires heuristic arguments rather than a common theoretical metric.The article identifies analysis of fundamental limits and optimal gains over separate systems as an open theoretical direction.
  • Future challenges: Future work should develop recovery and decoding algorithms for non-standard joint waveforms and resource-allocation methods that optimize radar and communications together.Resource allocation for both functionalities is described as a relatively fresh research area, while conventional waveforms such as OFDM also require efficient approaches.
  • Future challenges: Vehicular implementations and real-road testing are needed to characterize DFRC benefits and limitations and translate theoretical potential into performance gains.The paper highlights implementation on vehicular platforms and evaluation in real road environments as practical research needs.
Loading 1909.01729v2…