Near-Field Velocity Sensing and Predictive Beamforming

TL;DR

This work introduces near-field velocity sensing to jointly estimate radial and transverse target velocities from spherical-wave echoes, enabling velocity-aware, pilot-free predictive beamforming. It formulates a maximum-likelihood approach that jointly recovers $\beta$ and $\boldsymbol{v}=[v_r,v_\theta]^T$ and derives a gradient-based solution, leveraging the dependence of near-field Doppler on both velocity components. Building on this sensing capability, the paper presents a predictive beamforming framework that uses predicted location and velocity to design Doppler-compensated beamformers without channel estimation or motion-model priors, achieving high data rates. Numerical results demonstrate accurate velocity estimation, robust trajectory prediction, and near-optimal performance of the predictive beamformer with Doppler compensation, highlighting practical benefits for ISAC in near-field regimes.

Abstract

The novel concept of near-field velocity sensing is proposed. In contrast to far-field velocity sensing, near-field velocity sensing enables the simultaneous estimation of both radial and transverse velocities of a moving target. A maximum-likelihood-based method is proposed for jointly estimating the radial and transverse velocities from the echo signals. Assisted by near-field velocity sensing, a predictive beamforming framework is proposed for a moving communication user, which requires no channel estimation but achieves seamless data transmission. Finally, numerical examples validate the proposed approaches.

Figures (6)

• Figure 1: Illustration of the near-field velocity sensing.
• Figure 2: Impact of the distance on the ML function $g(\mathbf{Y}, \boldsymbol{\eta}, \boldsymbol{v})$.
• Figure 3: Convergence behavior of the proposed gradient ascent method.
• Figure 4: Predicted trajectory of the moving user.
• Figure 5: Estimated velocities of the moving user at different times.

Theorems & Definitions (2)

• Remark 1
• Remark 2