R1-2409479
discussion
Discussion on AI/ML-based beam management
From ZTE
ZTE's prior position on
9.1.1
at
RAN1#118bis
· AI-synthesized, paraphrased
verify sources →
Strongly advocates for functionality-based lifecycle management over model-ID-based approaches and pushes for bitmap-based beam reporting methods to significantly reduce signaling overhead while leveraging existing UE capability frameworks.
Summary
ZTE proposes functionality-based LCM without model ID signaling for AI/ML beam management, emphasizing overhead reduction through bitmap-based beam reporting and threshold-based data omission. The document outlines specific enhancements for NW-side data collection, UE-side inference reporting, and performance monitoring mechanisms, totaling approximately 35 distinct proposals and observations across data collection, model inference, and performance monitoring sections.
Position
ZTE proposes utilizing functionality-based LCM without model ID based signaling for AI/ML beam management, arguing that model transfer challenges and existing associated ID support diminish the need for model-ID-based approaches. They support bitmap-based methods for beam information reporting to reduce overhead by up to 63.5% in typical settings, and propose threshold-based beam reporting with configurable minimum/maximum beam counts. For NW-side data collection, ZTE supports L1 signaling irrespective of purpose and recommends differential L1-RSRP reporting with larger quantization steps. Regarding UE-sided models, they prefer configuring associated IDs per CSI report configuration and support extending Rel-17 TCI state signaling for multiple future time instances in BM-Case2. For performance monitoring, ZTE supports beam prediction accuracy and RSRP prediction accuracy as primary metrics and requires failure detection to be based on consecutive monitoring results within a predefined window.
Key proposals
- Proposal (General Views): Functionality-based LCM without specifying any model ID based signaling should be utilized for AI/ML based beam management.
- Proposal (Data Collection - Issue 1): Support all three options for collected data content (L1-RSRPs with beam indication, L1-RSRPs only, beam indices only) to serve different NW-side LCM operations.
- Proposal (Data Collection - Issue 1): Support enhanced method (e.g., bitmap) for the indication of beam ID in UE reporting when partial beams are reported.
- Proposal (Data Collection - Issue 2): Configure one or two RS resource sets for beam measurement depending on whether Set B is a subset of Set A or distinct from it.
- Proposal (Data Collection - Issue 3): Support threshold-based beam reporting method with configurable minimum and maximum number of reported beam related information in a single report.
- Proposal (Data Collection - Issue 3): Support specification enhancements for data omission among samples (e.g., according to data quality) to reduce overhead.
- Proposal (Data Collection - Issue 3): Support differential L1-RSRP reporting with larger quantization step size applicable to the differential L1-RSRP.
- Proposal (Data Collection - Issue 4): At least support L1 signaling for NW-side data collection irrespective of the purpose (training, inference, monitoring).
- Proposal (Model Inference - Issue 1): Support enhancements to report information about measurements of multiple past time instances in one reporting instance for BM-Case2, including timestamp and reference beam indication.
- Proposal (Model Inference - Issue 2): For UE-sided model BM-Case1, consider Alt 1 (one CSI-ResourceConfigId for Set B) and Alt 3 (two CSI-ResourceConfigIds for Set A and Set B separately) for CSI-ReportConfig.
- Proposal (Model Inference - Issue 3): If Set B is a subset of Set A, indicate resources for Set B as a subset of Set A based on assistance information such as a bitmap mapping.
- Proposal (Model Inference - Issue 4): For UE-sided model inference results, support reporting of RSRP (Option 2) and probability information (Option 3), prioritizing measured RSRP over predicted RSRP if both are available.
- Proposal (Model Inference - Issue 6): Extend Rel-17 TCI state activation/indication signalling methods to activate/indicate N TCI states corresponding to N future time instances for BM-Case2.
- Proposal (Performance Monitoring - Issue 1): Support beam prediction accuracy related KPIs (Alt 1 and Alt 2) and RSRP prediction accuracy (Alt 3) as primary performance metrics.
- Proposal (Performance Monitoring - Issue 3): Model/functionality failure detection should be based on monitoring results of several consecutive times within a predefined monitoring window.