R1-2410031
discussion
Discussion on other aspects of AI/ML model and data on AI/ML for NR air-interface
From FUTUREWEI
FUTUREWEI's prior position on
9.1.4.2
at
RAN1#118bis
· AI-synthesized, paraphrased
verify sources →
Advocates for simplified model identification approaches with network-controlled model ID assignment and cell-scoped associated IDs, while deprioritizing complex offline cross-vendor collaboration mechanisms.
Summary
This Futurewei document discusses AI/ML model identification and transfer/delivery for NR air interface, presenting 14 proposals and 4 observations covering four model identification options and various model transfer cases. The document advocates for network-assigned model IDs, deprioritization of certain transfer cases, and exploration of two-sided models without explicit model identification.
Position
Futurewei advocates for network-controlled model identification with associated IDs limited to cell scope, argues against using associated IDs as model IDs, and pushes to deprioritize complex model transfer cases (z1, z2) that don't provide clear benefits over simpler OTT approaches. They uniquely propose exploring two-sided models without explicit model identification to reduce complexity, while supporting Case z4 for non-3GPP-transparent scenarios.
Key proposals
- Proposal 1 (Sec 2.1): For MI-Option 1, conclude that an associated ID is valid only within a cell, and the network assigns/manages associated IDs
- Proposal 2 (Sec 2.1): Conclude that associated ID is not model ID
- Proposal 3 (Sec 2.1): For MI-Option 1, conclude that RAN1 only discuss the case that model IDs are assigned in procedure AI-Example1. The procedure without Step D is not considered as model identification
- Proposal 5 (Sec 2.1): For MI-Option 1, revise Step C of AI-Example1 procedure to include cell ID/information along with associated ID(s)
- Proposal 7 (Sec 2.2): Clarify the boundary between MI-Option 1 and MI-Option 2, and the relationship between model ID and corresponding dataset used for model training
- Proposal 9 (Sec 2.2): For MI-Option 2, conclude that model IDs are assigned only by the NW
- Proposal 10 (Sec 2.3): Support MI-Option 3 with further study of its procedures and specification impact, based on model transfer Case z4
- Proposal 11 (Sec 2.4): RAN1 to discuss and decide whether MI-Option4 is a valid option for model identification
- Proposal 12 (Sec 2.4): Clarify the relationship between the standardized reference model and possible multiple derived models
- Proposal 13 (Sec 2.5): Study the feasibility of supporting two-sided models without model identification
- Proposal 14 (Sec 3): Deprioritize Case z1 as it does not provide clear benefit over Case y