[Paper Review] Near-Field Beam Management for Extremely Large-Scale Array Communications

Changsheng You, Yunpu Zhang+6 more|arXiv (Cornell University)|Jun 28, 2023

Antenna Design and Optimization15 citations

TL;DR

The paper surveys near-field beam management for XL-arrays, covering beam training, tracking, and scheduling, and proposes hierarchical/polar-domain codebooks and methods to reduce overhead and improve multi-user performance.

ABSTRACT

Extremely large-scale arrays (XL-arrays) have emerged as a promising technology to achieve super-high spectral efficiency and spatial resolution in future wireless systems. The large aperture of XL-arrays means that spherical rather than planar wavefronts must be considered, and a paradigm shift from far-field to near-field communications is necessary. Unlike existing works that have mainly considered far-field beam management, we study the new near-field beam management for XL-arrays. We first provide an overview of near-field communications and introduce various applications of XL-arrays in both outdoor and indoor scenarios. Then, three typical near-field beam management methods for XL-arrays are discussed: near-field beam training, beam tracking, and beam scheduling. We point out their main design issues and propose promising solutions to address them. Moreover, other important directions in near-field communications are also highlighted to motivate future research.

Motivation & Objective

• Motivate the shift from far-field to near-field modeling for XL-arrays in 6G-era high-frequency systems.
• Introduce practical applications and challenges of near-field XL-array communications.
• Survey three core near-field beam management tasks: training, tracking, and scheduling, and propose promising solutions.
• Highlight open problems and future directions in channel modeling, ML-based design, and hardware-aware implementations.

Proposed method

• Discuss near-field energy-spread and the need for polar-domain (angle-distance) codebooks.
• Propose 2D near-field beam training with 2D codebooks, and develop hierarchical codebooks to reduce training overhead.
• Present fast near-field beam training methods, including two-phase and two-stage hierarchical approaches, to achieve low overhead.
• Describe wideband near-field beam training using true-time-delay (TTD) to mitigate beam-split effects.
• Outline near-field beam tracking approaches, including single- and multi-array collaboration, with EKF/UKF-based and side-information-aided prediction.
• Explain near-/mixed-field beam scheduling, introducing LDMA and the impact of range-angle dimensions on multi-user access and mixed-field interference.

Experimental results

Research questions

• RQ1 How to design efficient near-field codebooks and beam training procedures that account for spherical wavefronts in XL-arrays?
• RQ2 How can near-field beam tracking maintain reliable links under high mobility and with large polar-domain codebooks?
• RQ3 How should scheduling and power allocation exploit the range domain (LDMA) and mixed near-/far-field scenarios in XL-array systems?
• RQ4 What are practical considerations for hardware, ML integration, and ISAC/IRS integration in near-field XL-array beam management?

Key findings

• Two-phase near-field beam training can achieve high success rates close to exhaustive search with substantially lower overhead (518 vs 3072 symbols).
• Two-stage hierarchical near-field training significantly reduces overhead (24 symbols) with only modest loss in performance compared to exhaustive search.
• Far-field hierarchical training underperforms due to model mismatch in near-field channels.
• Near-field beam focusing enables LDMA, allowing simultaneous service of multiple users at similar angles by exploiting range information, and mixed-field scheduling shows potential for balanced interference and performance in hybrid scenarios.
• Collaborative near-field tracking with multiple XL-arrays can avoid explicit range tracking by combining angle estimates from multiple arrays to locate the user.