PaperlayerView exact revision on arXiv

Source-linked AI summary

Hybrid Precoding and Combining for mmWave Full-Duplex Joint Radar and Communication Systems under Self-Interference

Murat Bayraktar, Nuria González-Prelcic, Hao Chen

arXiv:2311.14942v2eess.SPcs.ITmath.IT

TL;DR

Full-duplex mmWave joint radar and communication must balance downlink performance and sensing while suppressing self-interference under hybrid-hardware constraints. This paper designs generalized-eigenvalue precoders and an analog combiner, reporting high radar SINR and accurate OFDM range and velocity estimates.

Abstract

from arXiv · show

In the context of integrated sensing and communication (ISAC), a full-duplex (FD) transceiver can operate as a monostatic radar while maintaining communication capabilities. This paper investigates the design of precoders and combiners for a joint radar and communication (JRC) system at mmWave frequencies. The primary goal of the design is to guarantee certain performance in terms of some sensing and communication metrics while minimizing the self-interference (SI) caused by FD operation and taking into account the hardware limitations coming from a hybrid MIMO architecture. Specifically, we introduce a generalized eigenvalue-based precoder design that considers the downlink user rate, the radar gain, and the SI suppression. Since the hybrid analog/digital architecture degrades the SI mitigation capability of the precoder, we further enhance SI suppression with the analog combiner. Our numerical results demonstrate that the proposed architecture achieves the required radar gain and SI mitigation while incurring a small loss in downlink spectral efficiency. Additionally, the numerical experiments also show that the use of orthogonal frequency division multiplexing (OFDM) radar with the proposed beamforming architecture results in highly accurate range and velocity estimates for the detected targets.

I. INTRODUCTION

Full-duplex mmWave joint radar and communication must balance sensing, downlink communication, and self-interference suppression. The paper designs hybrid precoders and combiners to pursue that trade-off using phase-shifter hardware.

I. INTRODUCTION

  • SI must be suppressed before the receive LNAs and ADCs to prevent saturation and clipping, making it a central constraint in hybrid FD-JRC design.Large-array mmWave hardware-only suppression can be costly and challenging.Source: §I · p. 1
  • FD-JRC precoders must balance downlink data rate and sensing performance while contributing to SI mitigation, and analog combiners must support radar processing and SI suppression.The combiner must also preserve target reflections for observation.Source: §I · p. 1
  • Earlier designs face implementation or modeling limits: one requires fine-grain attenuators, while another assumes downlink paths come from targets and uses limiting DFT codebooks.The first design also relies on null-space projection at the combiner.Source: §I · p. 1
  • The proposed design addresses multipath downlink and non-coincident target angles with phase-shifter hybrid precoders, generalized-eigenvalue beam design, and an SI-suppressing analog combiner.The paper reports a reasonable sensing–communication trade-off with this practical architecture.Source: §I · p. 2

II. SYSTEM MODEL

The system is a hybrid-architecture FD base station that serves a downlink mobile station while receiving target echoes. It assumes estimated channel and target angles, with range and velocity left to be estimated.

II. SYSTEM MODEL

  • The FD base station transmits data streams to the mobile station while its collocated receive side observes target reflections; each target is estimated in a separate short mmWave frame.The channel and target angles are assumed pre-estimated, so radar operation is framed as tracking range and velocity.Source: §II · p. 2
  • The base station uses collocated ULAs and OFDM, with hybrid transmit and receive chains; the mobile station also uses a hybrid architecture.The base station has NBS antennas and NBS,RF RF chains on each side, while the mobile station has NMS antennas and NMS,RF RF chains.Source: §II · p. 2
  • Overall precoders and combiners each combine frequency-flat analog stages with subcarrier-specific digital stages.The system model defines analog and digital components at both the base station and mobile station.Source: §II · p. 2
  • The target model uses point targets with shared AoA and AoD because the transmit and receive arrays are collocated, while the SI channel is modeled as near-field LoS.The downlink channel is represented by multiple geometric paths with path gain, delay, and arrival and departure angles.Source: §II · pp. 2–3

III. PRECODER/COMBINER DESIGN FOR FD-JRC

The design seeks high downlink spectral efficiency under transmit and receive radar-gain requirements while mitigating SI. It uses alternating optimization to handle the non-convex precoder and combiner problem.

III. PRECODER/COMBINER DESIGN FOR FD-JRC

  • Communication is measured by per-subcarrier spectral efficiency, while sensing metrics include transmit and receive radar gain and receiver-chain radar SINR.The section defines the communication metric and the radar metrics used to express the design trade-off.Source: §III · p. 3
  • The design maximizes downlink spectral efficiency subject to high transmit and receive radar gains, SI mitigation, and unit-magnitude constraints on analog stages.The radar SINR is closely related to target-estimation quality.Source: §III · p. 3
  • The method alternates among design variables because the joint optimization is non-convex, with the BS combiner's unit-magnitude constraint a major source of difficulty.Fully digital solutions are later decomposed into hybrid approximations.Source: §III · p. 3

A. Precoder Optimization for the BS

The BS precoder is designed to preserve communication performance while suppressing SI and meeting target-direction gain requirements. The fully digital design is then converted to a hybrid implementation, with the BS combiner addressing residual SI.

A. Precoder Optimization for the BS

  • The fully digital precoder uses a generalized eigenvalue problem to maximize spectral efficiency while mitigating the SI covariance.The SI term is fixed while spectral efficiency is maximized, and the resulting eigenvectors provide the precoder directions.Source: A. Precoder Optimization for the BS · p. 4
  • To meet the target-gain constraint, the design coherently combines SI-suppressing precoders with beams having high gain toward the target.A second generalized eigenvalue problem supplies target-direction beams that also suppress SI.Source: A. Precoder Optimization for the BS · p. 4
  • The fully digital precoders are normalized to satisfy the transmit-power constraint and decomposed across subcarriers into a frequency-flat analog precoder and digital precoders.The hybrid approximation uses low-complexity phase extraction with alternating minimization.Source: A. Precoder Optimization for the BS · p. 4
  • Hybrid decomposition degrades the fully digital solution's SI suppression, so the BS combiner is used to suppress the remaining SI.The fully digital precoder provides perfect SI suppression before hybrid decomposition.Source: A. Precoder Optimization for the BS · p. 4

B. Combiner Optimization for the MS

The MS combiners are selected from the downlink channel’s right singular vectors, then represented using the same hybrid decomposition algorithm used for the precoders.

B. Combiner Optimization for the MS

C. Combiner Optimization for the BS

The BS combiner is designed to reduce SI across subcarriers while retaining target-direction gain, using an iterative relaxation of its unit-modulus constraint.

C. Combiner Optimization for the BS

  • The BS combiner minimizes total SI across subcarriers while keeping target-direction gain above a threshold.Source: C. Combiner Optimization for the BS · p. 4
  • To handle nonconvexity, the method limits radar-gain loss, relaxes unit modulus, and applies block coordinate descent to randomly selected combiner entries.It also limits changes from the preceding iteration; optimized entries are normalized to unity afterward to restore the unit-modulus constraint.Source: C. Combiner Optimization for the BS · p. 4
  • The iterative procedure terminates when its cost converges, and its convex subproblem is solved using CVX.The alternating-optimization solution is stated to converge to local minima.Source: C. Combiner Optimization for the BS · pp. 4–5

IV. NUMERICAL RESULTS

The numerical evaluation uses a 28 GHz hybrid FD-JRC setup with 32 BS antennas, four streams, and OFDM parameters following 5G NR.

IV. NUMERICAL RESULTS

  • The simulated FD base station has two collocated 32-antenna ULAs, four RF chains, and a transmit–receive array separation of 6λ.The mobile station has 16 antennas and four RF chains; the system uses four streams.Source: §IV · p. 5
  • The system operates at 28 GHz over 100 MHz bandwidth, with 792 active subcarriers and 14 OFDM symbols at 120 kHz subcarrier spacing.The symbol duration is 8.92 μs, and the stated room-temperature noise power is −93.8 dBm.Source: §IV · p. 5
  • The BS receive-gain threshold is τR = 0.7NBS, with combiner-design parameters ϵ1 = 0.3 and ϵ2 = 0.1.Source: §IV · p. 5

A. Spectral Efficiency and Radar SINR Evaluation

The proposed design maintains radar SINR at high SI-to-noise levels, where the coherent-eigenvector benchmark fails to meet the required SINR. Its higher radar-gain threshold brings some spectral-efficiency loss, while the phase-shifter combiner matches the NSP benchmark with proposed precoding.

A. Spectral Efficiency and Radar SINR Evaluation

  • The evaluation averages 100 realizations for a five-path downlink channel, comparing the proposed design with SI-unaware optimal-precoding and coherent-eigenvector benchmarks.The mobile station is randomly placed 50 m from the BS, and the point target is 40 m away at azimuth 45°.Source: A. Spectral Efficiency and Radar SINR… · p. 5
  • At τT = 0.3√NBS, spectral-efficiency loss is negligible; increasing the threshold to τT = 0.35√NBS causes greater degradation for the proposed method.The proposed method also suppresses SI while maintaining the required transmit gain; the coherent-eigenvector method allocates more power to the target direction.Source: Figure 2 · p. 5, A. Spectral Efficiency and Radar SINR… · p. 5
  • The proposed precoder provides high radar SINR even at high SI-to-noise levels, while the coherent-eigenvector method fails to provide the required SINR.With the proposed precoder, the phase-shifter combiner attains the same performance as the NSP combiner even at high SI-to-noise levels.Source: Figure 3 · p. 5, A. Spectral Efficiency and Radar SINR… · p. 5

B. Range and Velocity Estimation for the Targets

The section uses OFDM subcarrier and symbol phase progression to form a range-velocity image and recover target estimates. In the stated simulation, the maps show highly accurate range and velocity estimates.

B. Range and Velocity Estimation for the Targets

  • OFDM subcarriers encode round-trip delay in progressive phase, while OFDM symbols encode Doppler, enabling range-velocity processing.The method applies a DFT across OFDM symbols followed by an inverse DFT across subcarriers to obtain a sparse image.Source: B. Range and Velocity Estimation for… · pp. 5–6
  • The maximum entry in the sparse image identifies the range and velocity indices used to recover each target’s range and velocity.The image is formed from a matrix of subcarrier-based OFDM radar observations.Source: B. Range and Velocity Estimation for… · pp. 5–6
  • Highly accurate range and velocity estimates are observed in the angle-range and range-velocity maps for four point targets.The simulation uses an MS at 50 m, four targets with 10 m^2 radar cross-section, and previously estimated angle errors below 0.5°.Source: B. Range and Velocity Estimation for… · p. 6, Figure 4 · p. 6

V. CONCLUSION

The paper proposes generalized eigenvalue-based precoding for FD-JRC, paired with analog-combiner optimization for further SI suppression. Simulations report effective SI suppression and highly accurate OFDM-based range and velocity estimation.

V. CONCLUSION

  • Generalized eigenvalue-based precoding is designed to suppress SI, maximize downlink spectral efficiency, and guarantee radar SINR.The analog combiner is further optimized for SI suppression using block coordinate descent and convex relaxation techniques.Source: §V · p. 6
  • OFDM-based radar processing with the proposed architecture achieves highly accurate range and velocity estimates.The conclusion also reports that the design is effective against SI.Source: §V · p. 6

Related papers

Loading 2311.14942v2…