R1-2409481
discussion
Discussion on specification support for AI CSI prediction
From ZTE
ZTE's prior position on
9.1.3
at
RAN1#118bis
· AI-synthesized, paraphrased
verify sources →
Proposes a systematic approach that prioritizes identifying additional conditions before developing detailed consistency solutions. Supports reusing proven AI beam prediction mechanisms but opposes introducing new associated IDs for scenarios/carrier frequency.
Summary
ZTE presents simulation results evaluating the generalization capability of AI-based CSI prediction models across different down tilt angles and TXRU mappings. The document contains two observations regarding model performance and one proposal concluding that neither down tilt angle nor TXRU mapping should be treated as additional conditions for AI CSI prediction due to sufficient model generalization.
Position
ZTE argues that down tilt angle and TXRU mapping should not be defined as network-side additional conditions requiring an associated ID for AI CSI prediction. They present technical evidence showing that AI models generalize well across different down tilt angles, with only a 0.28% SGCS impact when using datasets from different angles. Regarding TXRU mappings, ZTE acknowledges that models trained on one configuration may suffer performance loss in another, but proposes that this can be mitigated by using a mixed dataset rather than introducing specification-level consistency constraints. Consequently, they conclude that model generalization is sufficient to guarantee performance without adding complexity to the specification for these specific parameters.
Key proposals
- Proposal 1 (Sec Discussion on the consistency issue): Conclude that neither down tilt angle nor TXRU mapping is considered as additional condition for AI CSI prediction because the performance can be guaranteed by model generalization.