Ericsson · 8.1
Maintenance on Artificial Intelligence (AI)/Machine Learning (ML) for NR Air Interface ·
RAN1#124bis · Source verification
the AI's delta
new
vs RAN1#124
First tracked appearance at RAN1_124bis
AI-synthesized from contributions · all text is paraphrased
Every position summary on this site is generated by an AI from the actual Tdoc contributions. This page shows you the exact source documents the AI read to produce the summary above, so you can verify it yourself. Click any Tdoc ID to view its detail page, or click "3gpp.org ↗" to read the original on the official 3GPP server.
Contributions at RAN1#124bis · 2 docs
Remaining Issues of Rel-19 AIML
Position extracted by AI
Ericsson supports zero-padding the RS-PAI to 3 bits when nroftransmissionOccasion-r19 results in 1-bit or 2-bit payloads, ensuring compliance with PUCCH constraints that prohibit 1 or 2 information bits apart from HARQ-ACK/SR. They oppose restricting the network configuration by prohibiting nroftransmissionOccasion-r19 values of 1 or 3, arguing that such restrictions would hinder the network's ability to detect inaccuracies in specific monitoring occasions. They argue that limiting the network to supporting at least seven occasions reduces visibility into error instances, whereas zero-padding preserves the network's functionality for performance metric calculation. They present a technical case that equivalent information can be obtained via UE-reported measurements on Set A beams, but emphasize that if RS-PAI is used, the zero-padding solution is preferred to maintain granular monitoring capabilities.
Summary
This Ericsson contribution addresses remaining specification issues for Rel-19 AI/ML beam management, specifically focusing on the reporting of the RS-Prediction Accuracy Indicator (RS-PAI) on PUCCH. It identifies a conflict where certain configurations result in 1-bit or 2-bit RS-PAI payloads, which are unsupported by current NR specifications for PUCCH. The document presents one primary proposal to resolve this by applying zero-padding to ensure the payload meets the minimum bit requirement.
Session Notes of AI 8.1
Position extracted by AI
Ericsson, as the Ad-Hoc Chair, facilitates the endorsement of specification corrections for Rel-19 AI/ML features. They support zero-padding UCI bit sequences to 3 bits in TS 38.212 to ensure PUCCH compatibility when RS-PAI payload is small. They endorse updating TS 38.214 Clause 5.2.1.6 to include periodic CSI reports for UE-side data collection in CPU duration calculations. They confirm that for UE-sided models, data collection resources are restricted to always-on SSB and P/SP CSI-RS. They agree to reuse existing CPU occupation time definitions for CSI-ReportConfig with reportQuantity 'none' and no TRS-info. They note the absence of consensus on PDSCH rate matching around NZP CSI-RS in Set A for inference.
Summary
This document serves as the session notes for AI 8.1 at TSG RAN WG1 #124bis, documenting agreements on specification maintenance for AI/ML in the NR air interface. It records the endorsement of two Technical Proposals (TPs) regarding zero-padding for PUCCH CSI reporting and CPU duration definitions for UE-side data collection, while noting a lack of consensus on PDSCH rate matching around NZP CSI-RS. The document contains 0 formal proposals from Ericsson, as it primarily captures agreed-upon specification changes and open issues.
Prior contributions
Ericsson has no prior contributions to 8.1 in the meetings currently tracked. This is either a new contributor to this sub-topic or the earliest meeting in our history.
How this was derived
The AI extracted the "position extracted" field above directly from each Tdoc during summarization.
For the delta summary at the top, the AI compared Ericsson's consolidated stance at RAN1#124bis
against their stance at RAN1#124 and classified the change as
new.
Always verify critical claims against the original Tdocs linked above.