The article discusses the challenges of beam training in multiuser XL-MIMO systems and proposes a novel three-phase graph neural network (GNN)-based beam training scheme. The scheme aims to reduce pilot overhead, exploit user correlation, and mitigate beam conflicts in near-field communication. It involves UL pilot transmission, GNN-based estimation, near-field beam allocation, and hybrid TPC design. Simulation results show improved performance compared to traditional methods.
לשפה אחרת
מתוכן המקור
arxiv.org
תובנות מפתח מזוקקות מ:
by Wang Liu,Cun... ב- arxiv.org 03-27-2024
https://arxiv.org/pdf/2402.13597.pdfשאלות מעמיקות