R1-2410018
discussion
AI/ML specification support for beam management
From Lenovo
Summary
This Lenovo contribution presents 26 proposals for AI/ML specification support in NR beam management, covering data collection, model inference for both UE-side and NW-side implementations, performance monitoring, and UE capability reporting across spatial-domain (BM-Case1) and temporal (BM-Case2) beam prediction use cases.
Position
Lenovo advocates for a comprehensive AI/ML beam management framework that supports both UE-side and NW-side inference with strong emphasis on: 1) UE autonomy in data collection and model training initiation, 2) unified configuration approaches supporting both BM-Case1 and BM-Case2 through common frameworks, 3) robust performance monitoring with multiple metric alternatives (Alt 1, 2, 3), and 4) practical overhead reduction through variable-size beam reports and differential RSRP quantization. They push for flexible associated ID mechanisms combined with performance monitoring rather than rigid pre-defined conditions, and emphasize hardware resource management through AI process units.
Key proposals
- Proposal 1 (Data Collection): Support UE initiated beam management procedure for data collection for UE-side model training
- Proposal 2 (Data Collection): Support beam report with variable size at least for NW-side model training
- Proposal 3 (UE-side Configuration): For UE-sided model, at least for BM Case-1, support to only configure resource set for Set B for inference results report
- Proposal 4 (UE-side Consistency): Support the consistency between training and inference for UE side model by combining an associated ID and performance monitoring
- Proposal 8 (UE-side BM-Case2): Considering the prediction window for BM-Case2 with Mode 1 and Mode 2 configurations based on CSI reference resource and RRC-configured values
- Proposal 11 (UE-side Resources): Introduce AI process units for beam report with AI/ML inference at UE-side
- Proposal 13 (UE-side Indication): For a beam report associated with AI inference, the UE indicates that the reported beams are predicted beams or measured beams in the beam report
- Proposal 15 (NW-side Inference): To Support NW-side AI/ML inference, the gNB can configure one or more CSI reports for the UE to report the L1-RSRPs of all the beams configured in the CMR associated with the CSI report
- Proposal 18 (NW-side Monitoring): For NW-side AI/ML model performance monitoring, support Tx beam repetition for the UE to report the best L1-RSRP of a Tx beam among all its Rx beams
- Proposal 19 (UE-side Monitoring): For UE-side AI/ML inference, support aperiodic beam measurement for performance monitoring and dynamic beam updating within the beam set associated with the aperiodic trigger state for beam measurement
- Proposal 23 (Performance Metrics): Additional support Alt 2 and Alt 3 as the performance metric(s) of AI/ML model monitoring using L1-RSRP difference information and predicted vs measured RSRP differences
- Proposal 25 (Event Triggering): Support event triggered beam report for hybrid performance monitoring for UE-side AI/ML model
- Proposal 26 (Capability Reporting): On applicability for inference for UE-side model, use associated IDs provided by NW with inference related parameters selected from specified candidate values