PaperlayerView exact revision on arXiv

Source-linked AI summary

5G PRS-Based Sensing: A Sensing Reference Signal Approach for Joint Sensing and Communication System

Zhiqing Wei, Yuan Wang, Liang Ma, Shaoshi Yang, Zhiyong Feng, Chengkang Pan, Qixun Zhang, Yajuan Wang, Huici Wu, Ping Zhang

arXiv:2211.11488v1eess.SP

TL;DR

The paper addresses how to integrate radar sensing with 5G communication and positioning using an existing standardized signal. It applies PRS to sensing, derives estimation bounds, and reports improved range accuracy together with feasibility and compatibility for JSC.

  • Problem

    Existing pilot-based sensing approaches can be incompatible with 5G NR, while direct investigation of modern 5G signals for radar sensing is scarce.

  • Method

    The paper uses 5G PRS as a sensing reference signal, analyzes range and velocity estimation, derives CRLBs, and proposes multi-frame velocity estimation and frame-structure guidance.

  • Results

    1.17 m RMSE for range estimation was reduced to 0.33 m with the fractional factor, reaching centimeter-level ranging accuracy.

  • Takeaways & Limitations

    PRS can support joint sensing, communication, and positioning without additional changes to the 5G signal.

Abstract

from arXiv · show

The emerging joint sensing and communication (JSC) technology is expected to support new applications and services, such as autonomous driving and extended reality (XR), in the future wireless communication systems. Pilot (or reference) signals in wireless communications usually have good passive detection performance, strong anti-noise capability and good auto-correlation characteristics, hence they bear the potential for applying in radar sensing. In this paper, we investigate how to apply the positioning reference signal (PRS) of the 5th generation (5G) mobile communications in radar sensing. This approach has the unique benefit of compatibility with the most advanced mobile communication system available so far. Thus, the PRS can be regarded as a sensing reference signal to simultaneously realize the functions of radar sensing, communication and positioning in a convenient manner. Firstly, we propose a PRS based radar sensing scheme and analyze its range and velocity estimation performance, based on which we propose a method that improves the accuracy of velocity estimation by using multiple frames. Furthermore, the Cramer-Rao lower bound (CRLB) of the range and velocity estimation for PRS based radar sensing and the CRLB of the range estimation for PRS based positioning are derived. Our analysis and simulation results demonstrate the feasibility and superiority of PRS over other pilot signals in radar sensing. Finally, some suggestions for the future 5G-Advanced and 6th generation (6G) frame structure design containing the sensing reference signal are derived based on our study.

I. INTRODUCTION

The paper motivates using 5G PRS for joint sensing and communication because existing pilot-based sensing schemes can be incompatible with 5G NR. It proposes PRS-based sensing with estimation analysis, CRLB derivations, and frame-structure guidance for 5G-A and 6G.

  • Motivation and prior work: Existing pilot-based JSC schemes can rearrange subcarriers, interfere with data, or use parameters incompatible with 5G NR.The cited literature includes 802.11ad-based and high-frequency OFDM schemes, while direct use of modern 5G and 5G-A signals remains scarce.
  • PRS-based approach: PRS is applied as a sensing reference signal because its long sequence, good autocorrelation, rich resources, and flexible configuration support radar sensing.The approach uses 5G PRS without additional modification to communication signals.
  • Evaluation basis: The paper compares PRS with SS, DMRS, and CSI-RS and analyzes PRS sequence correlation and time-frequency resource mapping for radar sensing.The stated processing focuses on pilot subcarriers, avoiding changes to the 5G signal.
  • Analysis and contributions: The study derives CRLBs for PRS-based range, velocity, and positioning estimation and proposes multi-frame velocity measurement to improve accuracy.It also applies increased Fourier-transform points to improve range and velocity estimation while considering time-slot overhead.
  • Future frame structure: The proposed analysis informs 5G-A and 6G frame designs by addressing sensing-reference-signal configuration, overhead allocation, and the range–velocity estimation tradeoff.The frame-structure discussion targets different scenario requirements.

B. Time-Frequency Resource Mapping for PRS

PRS uses flexible, relatively rich time-frequency resources within the 5G NR framework, supporting configurable mappings and multiplexing across base stations. Compared with other downlink reference signals, its richer resources are expected to benefit radar sensing.

  • 5G NR resource configuration: 5G NR allocates PRS in granularities of 4 PRBs, ranging from 24 to 272 PRBs.A PRB contains 12 continuous frequency-domain subcarriers.
  • Reference-signal comparison: PRS has richer time-frequency resources than SS, DMRS, and CSI-RS, whose resources are limited or reduced by antenna-port multiplexing.The comparison treats an OFDM symbol on one subcarrier as the smallest resource element.
  • Configurable mapping: PRS supports Comb 2/4/6/12 in frequency and Symbol 2/4/6/12 configurations in time.These options support positioning-accuracy requirements while avoiding resource waste.
  • Configurable mapping: PRS time-domain mapping patterns are summarized in Table IV using the RE offset kPRS.The supplied passage identifies unsupported mapping modes with NA.
  • Two-dimensional mapping: The PRS mapping diagrams represent one 14-symbol slot across 12 consecutive subcarriers, or one PRB.The comb-like allocation multiplexes downlink PRS signals from multiple base stations on different subcarriers to control interference.

III. JSC SIGNAL MODEL

The paper models a 5G BS transmitting downlink PRS toward a target and using the echo for range and velocity estimation. The model represents PRS across time-frequency resources, derives its continuous-time OFDM signal, and separates range and Doppler information from received symbols.

  • Signal model: The PRS-based JSC model considers a 5G BS transmitting downlink PRS and estimating target range and velocity from the echo.An ambiguity function is also derived to demonstrate feasibility for radar sensing.
  • Signal model: For a Comb 4 example, PRS occupies M OFDM symbols and N subcarriers in a time slot, with PRS carried on a subset indexed by J.The corresponding mapped OFDM signal is illustrated in Fig. 4.
  • Continuous-time representation: The continuous-time PRS signal uses modulated symbols indexed by subcarrier k and OFDM symbol m, with total symbol duration Ts=T+TCP and subcarrier spacing ∆f=1/T.A rectangular function gates each OFDM symbol over its total duration.
  • Frequency mapping: Equally spaced PRS subcarriers are determined by the comb size, the first PRS subcarrier index k0, and the subcarrier spacing ∆f.For Comb 4 with four symbols, k0 takes 0, 2, 1, and 3 in sequence as related to the OFDM symbol index.
  • Received radar signal: The received echo model includes target range Rr, Doppler shift fd,r, attenuation factor ξ, and propagation speed c.The attenuation factor is regarded as constant during PRS transmission.
  • Range-Doppler extraction: Dividing received modulation symbols by transmitted symbols produces a matrix whose two vectors carry range and Doppler information.The model denotes the dyadic product by ⊗.

C. Ambiguity Function of PRS

PRS has a pushpin-type ambiguity function, indicating simultaneous high range and velocity resolution. The section also describes PRS-based range processing and accuracy–complexity tradeoffs.

  • Ambiguity function: The PRS ambiguity function is derived from the signal expression using the round-trip delay and normalized sinc-function formulation.The definition uses conjugation and delay-dependent time limits.
  • Ambiguity function: PRS exhibits a pushpin-type ambiguity function that provides high range and velocity resolution simultaneously.This supports the feasibility of PRS for radar sensing.
  • Range estimation: An NJ-point IFFT is applied to equally spaced PRS frequency-domain symbols to identify the range peak.The estimated range is obtained from the recorded peak index and averaged across columns.
  • Range estimation: Tuning the PRS comb size enables short-range, medium-range, and long-range ranging, while the guard interval can impose a stricter practical restriction.The maximum unambiguous range is constrained by the peak-index range and practical guard-interval requirements.
  • Range estimation: Increasing IFFT points through fractional factor ma improves ranging accuracy, but raises complexity without changing maximum unambiguous range.The stated complexity is o(ma × N · log2(maN)), and accuracy approaches CRLB only as ma increases indefinitely.

B. Velocity Estimation using PRS

PRS velocity estimation uses Doppler processing over PRS symbols, with a multiple-frame method that improves velocity resolution and reduces per-frame overhead. This benefit comes with a longer sensing refresh time.

  • Single-frame velocity estimation: Velocity is estimated by applying an MJ-point FFT to equally spaced PRS symbols in the time domain and averaging row-wise estimates.The FFT peak yields Doppler frequency, which is converted to velocity using the carrier-frequency relation.
  • Single-frame velocity estimation: Increasing FFT points improves velocity-measurement accuracy and can reduce the PRS symbols and time-slot overhead required.The maximum unambiguous velocity remains unchanged as FFT points increase.
  • Multiple-frame velocity estimation: Multiple-frame estimation extracts PRS symbols from continuously transmitted downlink frames and combines them for velocity measurement.The sensing refresh time is defined as the time required to achieve a velocity measurement, with Tf = 10 ms per frame.
  • Multiple-frame velocity estimation: Multiple-frame processing improves velocity resolution and estimation accuracy by increasing the total PRS symbols used for sensing.It reduces within-frame overhead but increases sensing refresh time.
  • Tradeoff: The longer refresh time can delay updates when a detected vehicle changes velocity during the measurement interval.The resulting tradeoff involves velocity resolution, per-frame overhead, and sensing refresh time.
  • Sensing–communication tradeoff: Reducing PRS symbol overhead increases communication resources within a frame and improves the communication rate.Sensing and communication are time-division multiplexed in the described downlink frame.

V. CRLB OF RADAR SENSING AND POSITIONING

This section derives the Cramer-Rao lower bounds for PRS-based radar sensing and positioning.

  • The CRLBs of PRS-based radar sensing and positioning are derived in this section.

A. CRLB of PRS based Radar Sensing

The PRS-based radar-sensing CRLB is derived for range and velocity estimation under large-sample assumptions. The analysis identifies a tradeoff between range and velocity accuracy as the signal duration changes.

  • CRLB derivation: Theorem 1 derives the CRLB for range and velocity estimated by PRS-based radar sensing when M ≫ 1 and N ≫ 1.
  • CRLB derivation: The derivation models known transmitted symbols, Gaussian noise, and unknown time-delay and Doppler parameters through the likelihood and Fisher information matrix.The CRLB matrix is obtained by inverting the Fisher information matrix.
  • CRLB derivation: Range and velocity CRLB expressions are converted from time delay and Doppler shift using the relations Rr = cτ/2 and the velocity–Doppler relation.
  • Performance tradeoff: Decreasing T lowers the range-estimation CRLB while increasing the velocity-estimation CRLB, establishing a range–velocity performance tradeoff.The parameters can therefore be configured to satisfy actual radar-sensing accuracy requirements.

B. CRLB of PRS based Positioning

The paper derives the CRLB for range estimation in PRS-based positioning from the PRS signal model and Gaussian-noise delay-estimation results.

  • CRLB derivation: Theorem 2 gives the CRLB for PRS positioning.The derivation proceeds from the PRS signal model, subcarrier power assumptions, and delay-estimation CRLB results.
  • Signal model: The base station generates PRS, maps it to physical resource units, and obtains the OFDM baseband signal through one-symbol IFFT.
  • Signal model: Assuming equal power across PRS-bearing subcarriers, the relative power weight is determined uniformly over those subcarriers.
  • CRLB derivation: The positioning delay CRLB is derived using standard Gaussian-white-noise estimation results, yielding the corresponding range CRLB.

VI. SIMULATIONS AND NUMERICAL RESULTS

The simulations evaluate PRS-based range sensing under representative 5G parameters and compare accuracy across processing choices, reference signals, comb patterns, SNR, and fractional factors.

  • Simulation setup: The simulation section evaluates PRS-based range and velocity processing, compares performance with CRLBs, and informs 5G-A and 6G frame-structure suggestions.
  • Range estimation: With 1000 Monte Carlo simulations, fractional processing reduced range-estimation RMSE from 1.17 m to 0.33 m.The paper reports that the resulting ranging accuracy reaches the centimeter level.
  • Range estimation: The simulations compare PRS-based sensing with other 5G downlink reference signals and examine the effects of comb structure and SNR on range accuracy.The compared signals are SS, DMRS, and PRS; CSI-RS is excluded because it occupies few resource elements in one PRB.
  • Range estimation: Comb 2 performs better at low and medium SNRs, while Comb 4 offers lower overhead and higher throughput in small-delay-spread, high-SNR scenarios.
  • Range estimation: Range-estimation RMSE decreases with SNR and approaches the root CRLB; fractional factor ma makes convergence faster.Considering accuracy and complexity, the paper recommends ma greater than 10.

B. Velocity Estimation

The velocity-estimation simulations show that fractional processing improves accuracy, while SNR and subcarrier spacing strongly affect performance.

  • Simulation setup: The velocity-processing simulations use continuous PRS transmission with M = 128 symbols and evaluate the correlation peak for a target moving at 15 m/s.
  • Velocity estimation: After 1000 Monte Carlo simulations, fractional processing reduced velocity-estimation RMSE from 2.24 m/s to 0.69 m/s.The paper reports that the improved accuracy reaches the cm/s level.
  • Velocity resolution: Decreasing subcarrier spacing improves velocity-estimation accuracy, although the improvement remains limited even when spacing changes from 240 kHz to 120 kHz.
  • Velocity estimation: Velocity-estimation RMSE decreases as SNR increases, and fractional processing makes the RMSE curve converge faster toward the root CRLB.

C. Velocity Measurement using Multiple Frames

The paper proposes measuring velocity across multiple frames to improve accuracy while managing sensing refresh time and communication overhead, then discusses corresponding 5G-A and 6G frame designs.

  • Multiple-frame measurement: Velocity resolution, per-frame PRS overhead, and sensing refresh time form a tradeoff: higher resolution can require more overhead or a longer refresh time.
  • Multiple-frame measurement: Using more frames improves velocity-estimation accuracy and anti-noise performance; with three frames, the sensing refresh time is 0.03 s and vehicle travel is 0.45 m at 54 km/h.The assumed PRS symbol overhead per frame is 22.9%.
  • Conclusion: The paper concludes that PRS-based sensing is feasible for radar sensing and positioning and proposes future 5G-A and 6G frame-structure directions.
  • 5G-A and 6G frame design: The proposed frame design separates sensing and communication subframes, while communication subframes may also support sensing when sensing demand is high.
  • 5G-A and 6G frame design: The sensing reference signal is intended to support waveform reconfiguration and flexible time-frequency allocation for different sensing requirements.
  • 5G-A and 6G frame design: Resource allocation creates a range–velocity estimation tradeoff, and chirp or other frequency modulation schemes could increase resolution and accuracy through a larger time-bandwidth product.
Loading 2211.11488v1…