R1-2410029
discussion
Discussion on specification support for AI/ML-based beam management
From FUTUREWEI
FUTUREWEI's prior position on
9.1.1
at
RAN1#118bis
· AI-synthesized, paraphrased
verify sources →
Strongly advocates for reusing existing CSI frameworks to minimize specification complexity and opposes introducing new complex metrics like probability information and confidence levels in favor of simpler, more testable approaches.
Summary
FUTUREWEI presents 11 proposals for AI/ML-based beam management specification support in NR Release 19, covering performance monitoring, model inference, data collection, and assistance information. The document focuses on reusing existing CSI framework while avoiding overly complex new metrics and signaling mechanisms.
Position
FUTUREWEI advocates FOR maximizing reuse of existing CSI framework to minimize specification effort, supporting simpler approaches that avoid complex new metrics like probability and confidence information. They push AGAINST supporting multiple alternatives that would increase complexity, specifically opposing Opt 3/4 for inference reporting and other performance monitoring alternatives beyond Alt 1. They favor practical solutions like using RS ID as implicit beam ID rather than defining new beam identifiers.
Key proposals
- Proposal 1 (Sec 2.1): For UE-assisted performance monitoring, only support Alt 1 (Top 1 or Top K beam prediction accuracy) without other alternatives, using full Set A for monitoring with longer period than Set B to reduce overhead
- Proposal 2 (Sec 2.1): Support dedicated CSI report configuration for performance monitoring with separate resource sets and report configuration from inference
- Proposal 3 (Sec 2.2.1): For network-sided model beam reports, support L1-RSRPs of up to M beams within X dB gap to largest measured value, increase maximum M from 4 to 8, use CRI/SSBRI as beam information
- Proposal 4 (Sec 2.2.2): For UE-sided model configuration, use CSI-ReportConfig for inference results reporting, consider Alt 2 (one CSI-ResourceConfigId for both Set A and B) and Alt 3 (separate CSI-ResourceConfigIds)
- Proposal 5 (Sec 2.2.2): Do NOT support Opt 3 (probability information) and Opt 4 (confidence information) in inference results reporting to avoid complex new metrics
- Proposal 6 (Sec 2.2.2): Support Option B for RSRP reporting - use measured L1-RSRP when available, predicted RSRP otherwise
- Proposal 7 (Sec 2.3): For data collection, support network providing multiple DL RS transmission configurations to UE, with UE reporting preferred ones to reduce signaling overhead
- Proposal 8 (Sec 2.3): Use RS ID as implicit beam ID indication and reuse existing L1-RSRP reporting framework as much as possible
- Proposal 9 (Sec 2.4): Confirm associated ID can be configured within CSI framework for consistency across training and inference
- Proposal 10 (Sec 2.5): For UE-side model applicability, further consider Option 1 (lowest latency) and Option 3 (lowest signaling overhead)
- Proposal 11 (Sec 3): Consider provided draft answers when responding to RAN2 LS on applicable functionality reporting for beam management UE-sided model