R1-2410892
discussion
FL summary #5 for AI/ML in beam management
From Samsung
Summary
This is Samsung's FL summary #5 for AI/ML in beam management (Tdoc R1-2410892), containing over 100 proposals addressing UE-sided and NW-sided models, performance monitoring, configuration frameworks, and beam indication mechanisms across multiple technical issues.
Position
Samsung, as the moderator, advocates for a comprehensive AI/ML beam management framework supporting both UE-sided and NW-sided models with practical implementation considerations. They push for flexible configuration options (supporting multiple alternatives rather than premature down-selection), reuse of existing CSI frameworks where possible, and clear separation between AI/ML and legacy processing. Samsung opposes overly complex signaling mechanisms and premature standardization of implementation-specific details, favoring solutions that provide network flexibility while maintaining UE implementation freedom.
Key proposals
- Proposal 2.1-1a (Sec 2.1): Introduce beam accuracy indicator (BAI) for beam prediction accuracy in CSI reports for UE-assisted performance monitoring
- Proposal 2.2-1a (Sec 2.2): Support dedicated resource sets for monitoring configured in separate CSI report configuration linked to inference configuration
- Proposal 3.3-1b (Sec 3.3): Configure associated ID within CSI framework at CSI-ReportConfig level with at least one ID for Set A and Set B consistency
- Proposal 4.1b (Sec 4.1): Support ranking information of predicted Top K beams conveyed by order of beam information based on model output
- Proposal 5.4 (Sec 5.4): Support larger quantization steps (3-4 dB) than legacy for differential L1-RSRP reporting to reduce overhead
- Proposal 6.1 (Sec 6.1): Support reporting measurement results of multiple time instances in one report for NW-sided BM-Case 2
- Proposal 7.1 (Sec 7.1): Study extension of TCI state activation methods to indicate N joint TCI states for N future time instances
- Proposal 8.1 (Sec 8.1): Use existing CPU mechanism as starting point for AI/ML-based CSI processing with separate counting between legacy and AI/ML
- Proposal C (Sec 9): Send LS to RAN2 defining three content types for NW-sided model training data collection
- Proposal 1.1B Direction A (Sec 1.1): Support both CSI-ReportConfig and inference parameter sets for applicability reporting
- Proposal 1.3B (Sec 1.2): Reply that RAN1 hasn't discussed 'functionality' term in Release 19 for associated ID questions
- Proposal 3.1a (Sec 3.1): Support reference time Option 3 based on latest Set B transmission occasion for BM-Case 2
- Proposal 5.5 (Sec 5.5): Support L1-RSRP omission reporting up to M beams within X dB gap to largest measured value
- Proposal 7.2-1a (Sec 7.2): Study Top-K beam measurements in existing CSI framework with dynamic resource configuration
- Proposal E (Sec 9): Define Top K beam prediction accuracy linking inference results to performance monitoring occasions