PaperlayerView exact revision on arXiv

Source-linked AI summary

RadChat: Spectrum Sharing for Automotive Radar Interference Mitigation

Canan Aydogdu, Musa Furkan Keskin, Nil Garcia, Henk Wymeersch, Daniel W. Bliss

arXiv:1908.08280v2cs.ITeess.SPmath.IT

TL;DR

Automotive radar and communication systems face mutual interference, including R2R and cross-system interference in shared spectrum. RadChat coordinates automotive radar sensors through radar communications using a PHY/MAC protocol. In dense VANETs, it reduces radar interference probability by about one order of magnitude every radar frame time, with protocol performance reported within 80 ms.

  • Problem

    Automotive R2R interference degrades radar detections, while C2R and R2C interference in shared bands can impede reliable communication and radar sensing.

    Source: §I · p. 1

  • Method

    RadChat uses wireless communications to coordinate automotive radar sensors, with FDM for radar and communication, rTDMA radar slots, and cCSMA communication access.

    Source: §I · p. 1, §V · p. 8

  • Results

    About one order of magnitude every radar frame time in dense VANETs, RadChat reduces automotive radar interference probability; its protocol is reported to meet sensing requirements within 80 ms.

    Source: §I · p. 2, §VII · p. 12

  • Takeaways & Limitations

    For similar-power radar and communication signals, avoid sharing time-frequency resources; starting FMCW chirps in different time slots can make R2R interference negligible.

    Source: §I · p. 1

Abstract

from arXiv · show

In the automotive sector, both radars and wireless communication are susceptible to interference. However, combining the radar and communication systems, i.e., radio frequency (RF) communications and sensing convergence, has the potential to mitigate interference in both systems. This article analyses the mutual interference of spectrally coexistent frequency modulated continuous wave (FMCW) radar and communication systems in terms of occurrence probability and impact, and introduces RadChat, a distributed networking protocol for mitigation of interference among FMCW based automotive radars, including self-interference, using radar communications. The results show that RadChat can significantly reduce radar mutual interference in single-hop vehicular networks in less than 80 ms.

I. INTRODUCTION

Automotive radar and vehicular communications face interference within and across systems, motivating RadChat, which uses wireless communication to coordinate automotive radar sensors. The introduction distinguishes coexistence, cooperation, and co-design approaches and presents interference analysis and a PHY/MAC protocol for mitigation.

I. INTRODUCTION

  • RadChat’s PHY/MAC protocol targets R2R interference while meeting radar sensing requirements across vehicles and among radars on the same vehicle, within 80 ms.The introduction also identifies an in-depth comparison with standard FMCW in a single-hop dense VANET.Source: §I · pp. 1–2

A. Classification of Joint Radar and Communications Systems

  • Joint radar and communication systems are classified as coexistence without information exchange, cooperation with explicit information sharing, or co-design of both systems together.Co-design generally uses the same waveform, though it does not require shared hardware.Source: A. Classification of Joint Radar and… · p. 2

1) Coexistence:

  • Earlier coexistence work primarily aimed to improve spectrum efficiency, whereas RadChat targets R2R interference mitigation using communications and shared hardware with different waveforms.The paper notes that radar and communication coexistence must account for R2R, C2C, C2R, and R2C interference.Source: §I · p. 1, 1) Coexistence · p. 2

2) Cooperation:

  • Prior cooperation work used shared hardware and different radar and communication waveforms, including a 79 GHz system with centrally controlled TDMA.That system sought to improve pulsed-radar sensing accuracy through communications.Source: 2) Cooperation · p. 2

3) Co-design:

  • Co-design examples include using an 802.11ad preamble as a radar signal, while OFDM flexibility is constrained in automotive radar by low-rate ADCs that cannot fully occupy the radar band.The cited limitation reduces distance-resolution capability.Source: 3) Co-design · p. 2

B. Medium Access Control for Cooperation

Prior radar-communication cooperation MAC approaches differ in how they schedule radar and communication access. The RadChat unit switches conventional FMCW radar hardware between radar and communication paths, with separate radar-chirp and communication-signal descriptions.

B. Medium Access Control for Cooperation

  • Prior cooperation MAC approaches include dedicated communication control with OFDM and time-division multiplexing that treats radar as a CSMA packet, potentially leaving radar sensing with low duty cycle during congestion.The cited approaches emphasize data communication or give radar sensing no priority over communications.Source: B. Medium Access Control for Cooperation · p. 2

A. RadChat Unit

  • A RadChat unit switches a conventional FMCW radar’s transmitter input and receiver output between radar and communication modules, and its signals connect to the MAC layer.The assumed VANET uses identical units and radar and communication parameters.Source: A. RadChat Unit · pp. 2–3

B. RadChat Transmitter

  • The RadChat transmitter sends radar or communication signals, but not both simultaneously.The unit reuses hardware for the two functions.Source: B. RadChat Transmitter · p. 3

1) Radar Transmitter:

  • FMCW radar transmission is modeled as a sequence of chirps, with parameters chosen to meet range, velocity, and resolution requirements.The receiver processes reflections within its bandwidth of interest, which is bounded by the ADC bandwidth.Source: 1) Radar Transmitter · p. 3

2) Communication Transmitter:

  • The communication transmitter uses a pulse-shaped complex baseband signal of bandwidth Bc, formed from unit-energy pulses and transmit symbols.Source: 2) Communication Transmitter · p. 3

C. RadChat Receiver

The RadChat receiver switches between radar and communications reception rather than receiving both simultaneously. The section describes radar-return processing and the scenarios and assumptions used to analyze interference.

C. RadChat Receiver

  • The receiver handles radar and communication signals at different times, not simultaneously.The radar and communication receiver paths are selected by switching the receiver input.Source: C. RadChat Receiver · p. 3

1) Radar Receiver:

  • Radar returns pass through a mixer, ADC, and digital processor before range and Doppler peaks are detected using FFTs over fast and slow time.The signal model assumes a narrowband radar and target displacement during a chirp much smaller than the wavelength.Source: 1) Radar Receiver · pp. 3–4

2) Communication Receiver:

  • The communication receiver is modeled under free-space propagation, with Doppler shift fD,c = vfc/c and carrier wavelength λc.Source: 2) Communication Receiver · p. 4

IV. INTERFERENCE ANALYSIS

  • The analysis considers two-radar R2R interference and a three-vehicle case in which communication ACKs disturb Vehicle 1's radar while Vehicle 1's radar affects Vehicle 2's communication.Source: §IV · p. 4, Figure 3 · p. 4
  • The interference analysis assumes identical FMCW waveform parameters across radars and equal antenna gains from signals in the respective main beams.Source: §IV · p. 4

A. R2R Interference Analysis

R2R interference can create false radar detections, and its occurrence depends on chirp timing and the vulnerable period. The analysis also relates that period to interference-path distance and network density.

1) Impact and Power of R2R Interference:

  • A strong ghost target appears at half the vehicles’ relative speed and distance: 15 m/s and 50 m for the illustrated 30 m/s, 100 m case.With identical radar parameters, ghost targets dominate; differing chirp parameters instead produce an effective noise-floor increase.Source: 1) Impact and Power of R2R Interference · p. 4
  • The direct line-of-sight interferer can be much stronger than the desired backscattered radar signal when target and interferer distances are of the same order.Source: 1) Impact and Power of R2R Interference · p. 4

2) Probability of R2R Interference:

  • The analysis expresses the probability of R2R interference as the fraction of vulnerable time over a radar frame, and says it is minimized with reduced Bmax or longer chirps.The vulnerable duration is approximately 2(1 + αd)NTmax when N ≫ 1.Source: 2) Probability of R2R Interference · p. 5
  • The vulnerable period is the range of relative chirp-start times that cause interference, and its duration depends on the longest interference path.The paper considers direct line-of-sight and reflected interference, including bistatic returns.Source: A. R2R Interference Analysis · p. 4, 2) Probability of R2R Interference · p. 5
  • Sparse VANETs have a longer vulnerable period but fewer interferers, whereas dense VANETs have a shorter period but more potential interferers.Source: 2) Probability of R2R Interference · p. 5

3) Example:

The section verifies the R2R interference-probability analysis and examines communication-to-radar interference. C2R interference raises the radar noise floor, with its time ratio affected by signal bandwidth choices.

3) Example:

  • The simulated R2R interference probabilities exactly match the analysis for two radars within 2dmax across varying Bmax and U.The verification uses 10^6 Monte Carlo simulations with uniformly distributed radar start times.Source: 3) Example · p. 5, Figure 5 · p. 5

1) Impact and Power of C2R Interference:

2) C2R Interference Time Ratio:

  • The C2R interference time ratio can be minimized with smaller communication bandwidth Bc, smaller radar bandwidth of interest Bmax, or higher radar sweep bandwidth Br.The analysis assumes Vehicle 2 transmits continuously and finds periodic radar interference irrespective of the transmission-time delay.Source: 2) C2R Interference Time Ratio · p. 6

3) Example:

The examples examine how interference affects radar detection and communication symbol error. The supplied analysis frames the communication evaluation around a 16-QAM system under differing parameter settings.

3) Example:

  • For Pr = Pc = 5 mW, Br = 1 GHz, and Bc = 40 MHz, radar-communication coexistence without significant radar performance reduction is possible only for close interferers, such as those less than 50 m away.The analysis attributes greater degradation from more distant interfering cars to the d2 versus d4 power-attenuation scaling for communication and radar.Source: 3) Example · p. 6

C. R2C Interference Analysis

1) Impact and Power of R2C Interference:

R2C interference can degrade communication when radar and communication powers are comparable, while its duration depends on the relative bandwidths. The 16-QAM example finds substantial SER degradation at equal powers for both tested communication bandwidths.

1) Impact and Power of R2C Interference:

  • When radar and communication distances are similar and radar power generally exceeds communication power, R2C interference is strong enough to cause loss of some data.The example treats dI = d as the worst case for radar detection; radar performance is less affected when the interferer is farther than the target.Source: 1) Impact and Power of R2C Interference · p. 6

2) R2C Interference Time Ratio:

  • The R2C interference time ratio can be reduced by choosing a smaller communication bandwidth or a larger radar bandwidth.The radar chirp interferes while its instantaneous frequency falls within the communication receiver's passband.Source: 2) R2C Interference Time Ratio · p. 7

3) Example:

  • At Pr = Pc = 5 mW, R2C interference from a 1 GHz FMCW radar causes unacceptably high SER for 16-QAM at Bc = 20 MHz and 40 MHz.The interference-related SER degradation is larger at higher communication bandwidths.Source: 3) Example · p. 7, Figure 7 · p. 7

D. Interference Analysis Outcome

R2C and C2R interference can impair communication and radar sensing concurrently, while comparable signal powers should not share time-frequency resources. RadChat coordinates radar sensing and communication through distributed scheduling, with network, timing, and state mechanisms supporting that coordination.

D. Interference Analysis Outcome

  • FMCW radar and communication signals with similar powers should not share time-frequency resources, and R2R interference can be avoided with non-overlapping chirp vulnerable periods.Increasing communication power relative to radar can improve communication SER while degrading radar detection performance.Source: D. Interference Analysis Outcome · p. 7

V. RADCHAT: PROTOCOL DESCRIPTION

  • RadChat schedules radars into non-overlapping vulnerable periods and uses a separate communication control channel to avoid R2R and R2C/C2R interference.Its MAC combines frequency division multiplexing for radar and communication, rTDMA for radar, and cCSMA for communication.Source: §V · p. 8, Figure 10 · p. 8

1) Network:

  • Each connected RadChat unit receives or transmits communications to another unit over a symmetric link, and units on the same vehicle coordinate through a central processor.The system assumes equal radar and communication transmit powers and leaves dynamic topologies and independent power control for future work.Source: 1) Network · p. 8

2) Timing:

  • Timing slots are sized to prevent chirp overlap within a radar frame, with GPS synchronization and a bound on the number of vehicles that can be supported.The maximum is Mmax ≤ ⌊1/U′⌋⌊Br/((1 + αd)Bmax)⌋.Source: 2) Timing · p. 8

3) Data Structure:

  • Each RadChat unit broadcasts its time-reference ID, rTDMA slot index, radar start time, and priority strength to support distributed slot assignment.Units sharing a vehicle use common ID and strength values, while slot indices and radar start times are specific to each unit.Source: 3) Data Structure · p. 8

B. MAC Operation

RadChat uses communication-channel sensing and control packets to assign radar slots without overlap. Distributed units exchange slot and timing state, using carrier sensing and backoff to manage contention.

B. MAC Operation

  • Non-persistent CSMA with binary exponential backoff assigns non-overlapping rTDMA slots, with control packets sent after an idle channel sense or deferred when busy.Each vehicle preferably reserves a preceding slot with no radar transmissions for communication shared by its RadChat units.Source: B. MAC Operation · p. 9
  • Vehicles select a contention period, typically the slot before most radar starts, and broadcast control packets containing unit identity, time reference, slot assignments, strength, and base-unit timing.The packet transmission schedule is derived from the chosen base RadChat unit's communication start time.Source: B. MAC Operation · p. 9
  • A RadChat unit senses the communication channel outside its radar slot and uses the prior slot to contend for and transmit scheduled control packets.Transmission occurs when the channel is idle; otherwise the unit backs off.Source: B. MAC Operation · p. 9
  • On receiving a control packet, a unit updates its ID, strength, and slot state, selecting an unused rTDMA slot and recalculating timing when its assignment changes.The receiver tracks announced state in a local database to monitor unused slots and time-reference priority.Source: B. MAC Operation · p. 9

C. Properties of RadChat

Under specified network and duty-cycle conditions, RadChat can converge to non-interfering radar slots and mitigate R2R interference within its communication range. Its evaluation uses simulations of dense vehicular networks and measures interference probability and convergence behavior.

C. Properties of RadChat

  • For a fixed connected topology with fewer than Mmax active radars, RadChat is guaranteed to converge to distinct vehicle slots and eliminate R2R interference when its radar duty cycle U′ ≤ 1/3.A higher duty cycle may split the network into disjoint interfering networks because RadChat units cannot receive control packets while radar is active.Source: C. Properties of RadChat · pp. 9–10
  • RadChat addresses R2R interference between vehicles and radars on the same vehicle, but cannot eliminate interferers beyond 2αddmax or beyond its maximum communication range.The range guarantee assumes Pc = Pr is sufficient to communicate with those interferers.Source: C. Properties of RadChat · p. 10

VI. RADCHAT PERFORMANCE EVALUATION AND RESULTS

  • The performance evaluation uses Matlab R2017b simulations of a network with one RadChat unit per vehicle and examines interference probability, convergence time, synchronization error, radar jitter, penetration rate, and communication parameters.With αd = 1 and Bc = 40 MHz, the setup supports a maximum of 70 vehicles; interference probabilities are based on 10,000 Monte Carlo runs.Source: §VI · p. 10, A. Simulation Parameters · p. 10

1) Time to minimize interference:

RadChat reduces R2R interference by coordinating radar transmissions, with the selected contention-window setting bringing interference below 10^-3 within 80 ms for a newly formed 70-vehicle VANET. Its performance depends on synchronization, deployment, and communication parameters, while the conclusion identifies larger heterogeneous scenarios as future work.

1) Time to minimize interference:

3) Radar Jitter:

  • Radar periodicity is distorted by at most one Tf during initialization, with no jitter afterward.The result is reported without being shown.Source: 3) Radar Jitter · p. 11

4) RadChat Deployment:

  • Almost total R2R interference elimination requires 100% RadChat deployment; benefits diminish quickly as deployment falls.With no RadChat-equipped vehicles, the reduced radar bandwidth increases R2R interference.Source: 4) RadChat Deployment · p. 11, Figure 14 · p. 11

5) Effect of Communication Parameters:

  • Larger control-channel bandwidth speeds convergence slightly, while a small bandwidth leaves an interference floor; increasing W0 from 6 to 64 lowers maximum tfinal from 10Tf to 4Tf.Allocating communication bandwidth slightly degrades radar range estimation, with reported reductions of 0.63 cm in range resolution and a 1.64 cm increase in range-estimation error.Source: 5) Effect of Communication Parameters · pp. 11–12, Figure 15 · p. 11, Figure 16 · p. 12
  • 80 ms: With W0 = 64, RadChat reduces R2R interference below 10^-3 for a newly formed 70-vehicle VANET.The authors report that this setting reduces interference by an order of magnitude almost within one radar frame time.Source: 5) Effect of Communication Parameters · p. 12

VII. CONCLUSION:

  • RadChat combines FDM, radar TDMA, and communication CSMA on shared hardware; future work will examine heterogeneous FMCW radars and inter-vehicle data communication.The conclusion also expects mitigation in sparse networks by adapting the vulnerable period alongside fewer interfering vehicles.Source: §VII · p. 12

APPENDIX A VULNERABLE PERIOD FOR R2R INTERFERENCE

The appendix characterizes when one FMCW radar’s chirp can interfere with another, accounting for perceived Doppler delay and target-detection range. It then gives an approximate vulnerable period under stated sampling and filtering assumptions.

APPENDIX A VULNERABLE PERIOD FOR R2R INTERFERENCE

  • A received chirp is treated as starting at t′ − τD, where τD is the perceived Doppler time delay.The delay is derived using the triangle proportionality theorem for one FMCW chirp.Source: Appendix A · p. 12
  • Interference occurs when t′ − τD falls within [0, Tmax], with Tmax corresponding to the maximum delay for detectable radar targets.Considering possible distances d/c ∈ [0, αdTmax] gives the vulnerable period.Source: Appendix A · p. 12
  • When Br ≫ 1/Tmax, the vulnerable period is approximated as [−αdTmax, Tmax], assuming I/Q sampling and perfect low-pass filtering.The maximum detectable relative velocity is given as vmax = c/(4frT).Source: Appendix A · p. 12

Related papers

Loading 1908.08280v2…