Near-Field Beam Training with DFT Codebook

Xun Wu,Changsheng You,Jiapeng Li,Yunpu Zhang,Li Chen,Kaifeng Han
DOI: https://doi.org/10.1109/wcnc57260.2024.10571054
2024-01-01
Abstract:Prior works on near-field beam training mostly assume dedicated polar-domain codebooks and on-grid range estimation, however, this may incur large training overhead and deteriorated estimation accuracy. In this paper, we propose a new and efficient beam training scheme with off-grid range esti-mation based on conventional discrete Fourier transform (DFT) codebook, which greatly reduces the beam training overhead. In particular, we first analyze the received beam pattern at the user when far-field beamforming vectors are used for beam scanning, and reveal an interesting result that this beam pattern contains useful user angle and range information. Then, an efficient scheme was proposed to jointly estimate the user angle and range using DFT codebook. This scheme estimates the user angle based on a defined angular support and resolves the user range by leveraging an approximated angular support width. Finally, numerical simulations show that our proposed scheme significantly reduces the near-field beam training overhead and improves the range estimation accuracy compared with various benchmark schemes.
What problem does this paper attempt to address?