R1-2601790 discussion

Discussion on beam management for downlink and uplink for 6GR air interface

From FUTUREWEI
Status: not treated
WI: FS_6G_Radio
Agenda: 10.5.2.4
Release: Rel-20
Source: 3gpp.org ↗
FUTUREWEI's prior position on 10.5.2.4 at RAN1#124 · AI-synthesized, paraphrased
verify sources →
Proposes adopting the 5G NR MIMO/RS/CSI framework as the starting point for 6G, arguing no fundamental paradigm shift is needed while supporting substantial evolution for upper midband around 7 GHz and 15 GHz. Presents a mixed antenna architecture combining low-resolution 1-bit all-digital receiver arrays for one-shot beam acquisition with analog/hybrid arrays for data transmission, supported by feasibility results showing near-ideal beamforming gain. Requires QCL/TCI enhancements to enable UL signals (SRS) as QCL source RS for DL TCI states, extending the Rel-17/18 unified TCI framework. Proposes AI/ML as an integral part of the 6G MIMO framework from day-1, identifying cross-frequency beam prediction (FR1-to-FR2) as a candidate use case requiring study of data collection configuration, inference-related resource configuration, and LCM operations including fallback and model switching.

Summary

Futurewei presents 3 Observations and 7 Proposals on 6G beam management, covering one-shot beam acquisition, QCL/TCI enhancements using UL signals, UE-initiated beam management latency reduction, AI/ML cross-frequency beam prediction, and simulation assumptions.

Position

Futurewei proposes a mixed antenna architecture combining a low-resolution all-digital receiver array for one-shot beam acquisition with an analog/hybrid array for data transmission to eliminate beam sweeping procedures. They require enabling UL signals (e.g., SRS) as QCL source RS for DL TCI states, departing from 5G NR's restriction to only DL signals. For UEIBM, they propose a network confirmation mechanism that updates TCI states and activates new beams without RRC reconfiguration or MAC-CE signaling, targeting approximately 30 ms latency reduction. On AI/ML, they support cross-frequency beam prediction using angular-delay domain FR1 channel inputs to predict mmWave beams, demonstrating feasibility for co-located scenarios with Transformer-based architectures. They propose specific LLS/SLS simulation parameters prioritizing around 7 GHz and 30 GHz with beam acquisition latency and overhead as primary performance metrics.

Key proposals

Your notes

Private to your account