| R1-2503239 AIML UE features.docx |
3GPP TSG-RAN WG1 Meeting #121 Tdoc R1-2503239
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda Item: 9.15.1
Source: Ericsson
Title: UE Features for Rel-19 AI/ML for NR Air Interface
Document for: Discussion, Decision
|
Conclusion
Based on the discussion in the previous sections we propose the following:
Proposal 1 For the BM features, possible purposes of the features should not be part of any feature group naming (remove “for UE/NW-sided model inference” across all features)
Proposal 2 For the increased number of reported RSs for beam management feature, support candidate values M = 8, 16, 32, 64.
Proposal 3 For components of FG 58-1-2 and FG 58-1-3 in addition to the agreed components:
a. Support setB-subset-of-setA and setB-different-from-setA
b. Supported maximum number of predicted beams in each reporting instance, candidate value is 1,2 or 4
c. Support for UE-assisted performance monitoring
d. AI/ML PU (FFS on details)
Proposal 4 For components of FG 58-1-4 and FG 58-1-5 in addition to the agreed components:
a. Support SetB-equal-to-SetA, setB-subset-of-setA and setB-different-from-setA
b. Supported maximum number of predicted beams in each reporting instance, candidate value is 1, 2 or 4
c. Supported maximum number of predicted time instances, candidate value is 1,2,3 or 4
d. Support for UE-assisted performance monitoring
e. AI/ML PU (FFS on details)
Proposal 5 For components of FG 58-1-7 5 in addition to the agreed components:
a. Support SetB-equal-to-SetA, setB-subset-of-setA and setB-different-from-setA
Proposal 6 Move FG 58-2-1 to L2 Feature groups, the same as other positioning methods.
Proposal 7 Add the following as components of FG 58-2-1:
a. Support reporting the monitoring outcome with an indication that the target UE cannot perform the Case 1 positioning method.
b. Support LMF initiated activation / deactivation of UE-based positioning Case 1.
c. Support reporting the location estimate as generated by UE-based positioning Case 1.
Proposal 8 UE-based positioning Case 1 does not require separate FGs for model performance monitoring. Performance monitoring is a component of Case 1 main FG 58-2-1.
Proposal 9 Agree to the current placeholder values for the components in FG 58-2-3/3a/3b.
Proposal 10 RAN1 should discuss whether a new processing capability is needed for Case 1 before deciding whether to introduce FG for QCL processing of the DL PRS for Case 1.
Proposal 11 Do not Introduce additional FGs for label provisioning for Case 1.
Proposal 12 The components 8-11 of the basic feature group 58-3-1 for Rel-19 CSI prediction using UE-sided model is defined as follows:
8. Support X=1 CQI based on the first/earliest slot of the CSI reporting window and the first/earliest predicted PMI (TDCQI=’1-1’)
9. Value for APU occupation, when P/SP-CSI-RS is configured for CMR. FFS the candidate values.
10. Value for CPU occupation, when A-CSI-RS is configured for CMR. FFS the candidate values.
11. Scaling factor for active resource counting Kp. FFS the candidate values.
Proposal 13 Additional FGs related to performance monitoring, UE-sided data collection, etc., can be added pending further RAN1 agreements.
|
| R1-2503255.docx |
3GPP TSG-RAN WG1 Meeting #121 R1-2503255
St Julian’s, Malta, May 19 – 23, 2025
Agenda Item: 9.15.1
Source: Huawei, HiSilicon
Title: UE features for AI/ML for NR air interface
Document for: Discussion and Decision
|
Conclusions
In this contribution, we discussed the FGs for Rel-19 NR_AIML_air. The following proposals are provided.
Proposal 1: Update FG 58-1-1 as the cyan-highlighted parts in Table 1.
Proposal 2: Update FG 58-1-2 and FG 58-1-3 as the cyan-highlighted parts in Table 2.
Proposal 3: Update FG 58-1-4 and FG 58-1-5 as the cyan-highlighted parts in Table 3.
Proposal 4: Update FG 58-1-7 as the cyan-highlighted parts in Table 4.
Proposal 5: Update FG 58-2-1 to FG 58-2-3b as the cyan-highlighted parts in Table 5.
Proposal 6: Update FG 58-3-1 and FG 58-3-2 as the cyan-highlighted parts in Table 6.
Additional FGs could be added to emulate the additional FGs of Rel-18 CSI prediction FG 40-3-2-1b, 40-3-2-2, 40-3-2-3, 40-3-2-3a, 40-3-2-7, 40-3-2-8, 40-3-2-9, 40-3-2-11, and 40-3-2-12.
|
| R1-2503385_Discussion on UE features for AIML for NR Air Interface.docx |
3GPP TSG RAN WG1 #121 R1- 2503385
St Julian’s, Malta, May 19th – 23th, 2025
Source: vivo
Title: Discussion on UE features for AIML Air interface
Agenda Item: 9.15.1
Document for: Discussion and Decision
|
Conclusion
In this contribution we provide updates on UE features for AI/ML based beam management, positioning and CSI prediction.
Introduce the following Rel. 19 UE FGs for AI/ML based beam management in section 2.
Introduce the following Rel. 19 UE FGs for AI/ML based positioning in section 2.
Introduce the following Rel. 19 UE FGs for AI/ML based CSI prediction in section 2.
|
| R1-2503399-UE features AIML for NR Airinterface.docx |
3GPP TSG RAN WG1 #121 R1- 2503399
Malta, 19 – 23 May 2025
Agenda item: 9.15.1
Source: Nokia
Title: Views on UE features for AI/ML for NR Air Interface
Document for: Discussion and Decision
|
Conclusion
In this contribution, we presented our views, suggestions, and proposals on UE features for AI/ML for NR Air Interface, and we have made the following proposals
Proposal 1: Adopt the proposed changes to FGs and components in Table 1 to Rel-19 UE features for AI/ML support for beam management in NR Air Interface
Proposal 2: Adopt the proposed changes to FGs and components and a new FG in Table 2 to Rel-19 UE features for AI/ML Positioning for NR Air Interface
Proposal3: Adopt the proposed changes to FGs and components, and two new FGs in Table 3 to Rel-19 UE features for AI/ML techniques for CSI prediction in NR Air Interface
|
| R1-2503591 UE features for AIML for NR Air Interface_final.docx |
3GPP TSG RAN WG1 #121 R1-2503591
St Julian’s, Malta, May 19th –23rd, 2025
Agenda item: 9.15.1
Source: Samsung
Title: UE features for AI/ML for NR Air Interface
Document for: Discussion and Decision
|
Conclusion
The proposals made in this contribution are summarized below.
Proposal 1: Support the following update to FG 58-1-1 with red highlighted.
Proposal 2: Support the following update to FG 58-1-2 with red highlighted.
Proposal 3: Support the following update to FG 58-1-3 with red highlighted.
Proposal 4: Support the following update to FG 58-1-4 with red highlighted.
Proposal 5: Support the following update to FG 58-1-5 with red highlighted.
Proposal 6: Support the following update to FG 58-1-7 with red highlighted.
Proposal 7: Support the following update to FG 58-2-1 with red highlighted.
Proposal 8: Support the following update to FG 58-2-2 with red highlighted.
.
Proposal 9: Support the following update to FG 58-3-1 with red highlighted.
Proposal 10: Support the following new FG 58-3-3
Proposal 11: Support the following new FG 58-3-4
|
| R1-2503652_Discussion on UE features for AIML for NR Air Interface.docx |
3GPP TSG RAN WG1 #121 R1-2503652
St Julian’s, Malta, May 19th – 23th, 2025
Title: Discussion on other aspects of AI/ML model and data
Source: ZTE Corporation, Sanechips
Agenda item: 9.15.1
Document for: Discussion/Decision
|
Conclusion
In this contribution, we provide our analysis and proposals for the UE feature design for Rel-19 AI/ML use cases for air interface.
General principle
Proposal 1: Consider the following two principles for Rel-19 UE feature design for AI/ML use cases for physical layer
Principle#1: Balance between UE capability report and applicability report
Principle#2: Balance between UE implementation flexibility and UE implementation fragmentation
AI beam prediction
Proposal 2: Update the FG 58-1-1 as following.
Proposal 3: Regarding FG58-1-2, update the RS type for set A and set B as following.
6. RS type of Set A and set B
Support of SSB as RS type for Set B
6a. Support of CSI-RS as RS type for Set B
Support of SSB as RS type for Set A
Support of CSI-RS as RS type for Set A
FFS: RS type for Set A
Proposal 4: Regarding FG58-1-2, discuss the following two methods to report the supported combinations of number of beams in set A and number of beams in set B considering the flexibility of different methods.
Method.1: Report the supported combinations of the number of resources for Set B and the number of resources for Set A via UE capability.
Method.2: Report the maximum number of resources for Set B/Set A via UE capability, and report the supported combinations of the number of resources for Set B and the number of resources for Set A via applicability report.
Proposal 5: Regarding FG58-1-2, add the following component 10 for performance monitoring.
Component 10: Supported options for performance monitoring for beam case 1 with UE side model
Proposal 6: Regarding FG58-1-4, add the following two components.
Supported maximum number of predicted time instances
Supported value(s) of time gap between predicted time instances
Proposal 7: Regarding FG58-1-7, we propose the following
Remove the “[for BM case 1]” in the name and component 1
Remove component 3
AI POS enhancement
Proposal 8: New UE feature for AI POS enhancement case 1 is needed, while no need to define any new UE feature for AI POS enhancement case 3a and case 3b.
Proposal 9: Support the following UE feature updates for AI POS enhancement in Rel-19:
Proposal 10: Consider the following UE features for AI POS enhancement in Rel-19.
Proposal 11: RAN1 further discusses whether new UE features are required for the following aspect specifically for the AI POS enhancement case 1.
Common DL PRS Processing Capability, e.g., FG 13-1
Support of PRS measurement for case 1 in RRC_IDLE
AI CSI prediction
Proposal 12: At least the following UE features from Rel-18 Doppler MIMO codebook can be reused for Rel-19 AI CSI prediction.
|
| R1-2503740_AI_UE_feature.docx |
3GPP TSG RAN WG1 #121 R1-2503740
St Julian, Malta, May 19th – 23rd, 2025
Agenda item: 9.15.1
Source: Ofinno
Title: Views on UE features for AI/ML for NR Air Interface
Document for: Discussion/Decision
Views on UE features for AI/ML beam management for NR Air Interface
In this contribution, we present our views on UE features for AI/ML for NR air interface. We propose:
[AI/ML Beam management]
Proposal#1:
Regarding component 14 in 58-1-4, remove bracket and add the wording for as follows
[14. Supported value(s) of time gap between predicted time instances and between reference time to the first future time instance].
Reason
In RAN1#120, it was agreed that the for BM-Case2 one inference report comprises inference results of N future times instances, where N = [1, 2, 4, 8]. Furthermore, in RAN1#120bis, it was agreed that the time gap between two consecutive future time instances is configured. Apparently, when N=1, there is no time gap between two future time instances. So, it would be desired to change “time gap of two future time instances” to “valid time duration of each predicted time instance” at least for N=1 case.
Related agreements:
Proposal#2:
Add the following component in 58-1-4 as:
14a: Supported combinations of value(s) of valid time duration for each predicted time instance and the number of future time instance(s)
Reason
Regarding inference, a UE may use a different model for inferencing based on different time gap and/or different number of future time instance(s). Based on the related agreement as shown in the above, the configured value of the time gap and the number of future time instances can be commonly used for all the inference models. Then, such UE may run the same number of inferences for the same number of future time instances regardless of the property of each inference model. To handle this issue, combinations between the time gap and the number of future time instance(s) can be considered. For example, the UE can support the following combinations where whole inference time (i.e. time gap value N) covers 80ms, i.e. {(10ms, 8), (20ms, 4), (40ms, 2), (80ms, 1)}
|
TDoc file conclusion not found |
| R1-2503781.docx |
3GPP TSG RAN WG1 #121 R1-2503781
St Julian’s, Malta, May 19th – 23th, 2025
Source: CATT, CICTCI
Title: Discussion on UE features for AI/ML for NR Air Interface
Agenda Item: 9.15.1
Document for: Discussion and Decision
|
Conclusion
In this contribution, we provided our views regarding RAN1 UE features for AI/ML for NR air interface based on the outcome previous meeting. Our suggestions are presented in Section 2, covering (1) Beam management, (2) Positioning accuracy enhancement, and (3) CSI prediction.
|
| R1-2503850.docx |
3GPP TSG RAN WG1 #121 R1-2503850
St Julian’s, Malta, May 19th – 23rd, 2025
Source: CMCC
Title: Discussion on UE features for AI/ML for NR air Interface
Agenda item: 9.15.1
Document for: Discussion & Decision
|
Conclusions
In this contribution, we discuss Rel-19 UE features for AI/ML for NR air Interface, and the following proposals are made.
Proposal 1: Take the following modification (in red) for the feature of Rel-19 AI based beam management:
Proposal 2: Take the following modification (in red) for the feature of Rel-19 AI based beam management:
Proposal 3: Take the following modification (in red) for the feature of Rel-19 AI based beam management:
Proposal 4: Adopt the following FG for AI based beam management.
Proposal 5: For AI based positioning, model inference and model monitoring should be together supported.
Proposal 6: For AI based positioning, a new UE capability for the number of PRS resources should be defined.
Proposal 7: Adopt the following FGs for AI based positioning case1.
Proposal 8: Take the following modification (in red) for the feature of Rel-19 AI based CSI prediction:
Proposal 9: Introduce the following Rel-19 UE FGs for AI/ML based CSI prediction:
|
| R1-2503900.docx |
3GPP TSG RAN WG1 #121 R1-2503900 St Julian’s, Malta, May 19th – 23rd, 2025
Agenda Item: 9.15.1
Source: Xiaomi
Title: Discussion on UE features for AI/ML for NR Air Interface
Document for: Discussion
|
Conclusion
Based on the discussion, we propose the FGs need to be supported.
AI/ML for beam management
Proposal 2-1: Support the following FGs/rows for AI/ML based beam management.
AI/ML for positioning
Proposal 3-1: For UE 58-2-1, support the following update based on progress achieved in RAN1 120bis meeting(the update is highlighted as red text)
Proposal 3-2: Support UE feature for UE-based performance monitoring
Proposal 3-3: Support UE feature for PRS as follows
Proposal 3-4: For training data collection:
For Case of part A and part B generated by UE, there is no need to define related UE feature
For Case of part A generated by UE and part B generated by LMF, whether new UE feature is necessary is up to other WG
AI/ML for CSI prediction
Proposal 4-1: FG 58-3-1 and 58-3-2 is updated in the following Table.
Proposal 4-2: Support to introduce a new feature group on aperiodic CSI report timing relaxation at least for AI/ML model switches/activates.
Proposal 4-3: Support to introduce new feature on maximum number of aperiodic CSI-RS resources that can be configured in the same CSI report setting for AI/ML model based CSI prediction.
Proposal 4-4: Support to introduce feature group on performance monitoring for CSI prediction model.
Proposal 4-5: Support to introduce feature group on data collection for UE-side CSI prediction.
Proposal 4-6: The following feature groups on AI/ML-based CSI prediction in the Table could be supported.
|
| R1-2503984.DOCX |
3GPP TSG-RAN WG1 Meeting #121 R1-2503984
St Julian’s, Malta, May 19th – 23st, 2025
Agenda Item: 9.15.1
Source: LG Electronics
Title: Discussions on UE features for AI/ML for NR Air Interface
Document for: Discussion and Decision
|
Conclusion
In this contribution, we provided our views on Rel-19 UE features for AI/ML for NR Air Interface.
|
| R1-2504227.docx |
3GPP TSG RAN WG1 #121 R1-2504227
St Julian’s, Malta, May 19th–23rd, 2025
Source: OPPO
Title: UE features for AIML for air interface
Agenda Item: 9.15.1
Document for: Discussion and Decision
|
Conclusion
In this contribution, we discuss the potential UE features for AIML for air interface with the following proposals:
Proposal 1: The UE features of beam management can be updated according to the above UE feature table.
Proposal 2: Support information related to DL PRS and UL SRS as in the capability signaling of legacy methods, detailed capability value can be further discussed if needed.
Proposal 3: Introduce the Rel. 19 UE FGs for AI/ML based CSI prediction. |
| R1-2504352 Views on UE features for AI.docx |
3GPP TSG RAN WG1 #121 R1-2504352
St Julian’s, Malta, May 19th – 23th, 2025
Agenda item: 9.15.1
Source: Apple Inc.
Title: Views on UE features for AI/ML for NR Air interface
Document for: Discussion/Decision
AI based beam management
In RAN1 120bis, the following agreements are captured with FFS.
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agree
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
The UE feature features need to be clearly described.
Proposal 1-1: Remove the square brackets and keep the enclosed texts for 58-1-1, 58-1-2, 58-1-3, 58-1-4, 58-1-5, 58-1-7.
Proposal 1-2: Split component 1 and component 2 of 58-1-1 into two FGs.
Proposal 1-3: Introduce a UE FG for data collection for UE-side beam prediction for BM Case-2.
AI based positioning
The AI based positioning UE features was discussed in RAN1 120bis, and the following agreement was captured in [1].
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
In RAN2 #129, the following agreement was made:
Agreement
Introduce AI/ML positioning Case 1 as a new positioning method.
As such, the AI/ML positioning use case FGs and their components should be at least identical to that of the existing use cases. The FGs have been captured in FG 58-2-3, FG 58-2-3-a and FG 58-2-3-b in RAN1 120-bis. However, the components have not been agreed. Based on this, we make the following proposals.
Proposal 2-1:
FG 58-2-3
DL PRS Resources for UE-based positioning Case 1
Max number of DL PRS Resource Sets per TRP per frequency layer supported by UE.
Values = {1, 2}
Max number of TRPs across all positioning frequency layers per UE.
Values = {4, 6, 12, 16, 24, 32, 64, 128, 256}
Max number of positioning frequency layers UE supports
Values = {1, 2, 3, 4}
Proposal 2-2:
FG 58-2-3-a
DL PRS Resources for UE-based positioning Case 1 on a band
Max number of DL PRS Resources per DL PRS Resource Set
Values = {1, 2, 4, 8, 16, 32, 64}
Note: 16, 32, 64 are only applicable to FR2 bands
Max number of DL PRS Resources per positioning frequency layer.
Values = {6, 24, 32, 64, 96, 128, 256, 512, 1024}
Note: 6 is only applicable to FR1 bands
Proposal 2-3:
FG 58-2-3-b
DL PRS Resources for UE-based positioning Case 1 on a band combination
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1-only.
Values = {6, 24, 64, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR1 only BC.
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2-only.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR2 only BC
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1 in FR1/FR2 mixed operation.
Values = {6, 24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2 in FR1/FR2 mixed operation.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands]
Two additional FGs need to be added to ensure support for all the FGs defined for each positioning method. These are on the PRS TEG associated information and for support in the RRC_INACTIVE state. As at now, there is no agreement on whether AI/ML positioning is or is not supported in RRC_INACTIVE. Unless there is an agreement not to support it, it is necessary to have an FG to allow for indication of support like the other positioning methods. Based on these arguments, we make the following proposals:
Proposal 2-4: Add a new FG 58-2-4
Support of PRS TEG association information for AI/ML based positioning
Support of reception of association between PRS and TRP Tx TEG for for AI/ML based positioning Case 1
Proposal 2-5: Add a new FG 58-2-5
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Finally, for Case 1, 3a and 3b, the following existing FGs should be applicable to AI/ML positioning
Proposal 2-6: For Case 1,
From the Rel-16 UE feature list for NR Positioning FG 13-1, 13-1a, 13-7, 13-7a and 13-8 are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning FG 27-3-2, 27-3-3, 27-6, 27-9, 27-10, 27-10a and 27-11 are applicable AI/ML-based positioning
Proposal 2-7: For Cases 3a and 3b
From the Rel-16 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
Proposal 2-8:
From the Rel-18 UE feature list for NR Positioning, the applicable FGs for AI/ML-based positioning are FFS.
AI based CSI prediction
The UE features was discussed in RAN1 120bis, and captured in [1].
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
On inference related UE capability, it was agreed in the RAN1 120bis that AI/ML PU is introduced. It is under further discussion on whether UE capability is introduced to indicate whether CPU or AI/ML PU is used, and how many AI/ML PU is used for CSI prediction. Whether additional UE capability is introduced can be further discussed in RAN1.
For feature group 58-3-2, component 6 and component 8 are duplicate of component 3, which can be removed.
Proposal 3-1: For feature group 58-3-2, remove component item 6 and 8.
In RAN1 120bis, separate UE capability for performance monitoring was agreed for AI based CSI prediction. On performance metric and reporting procedure itself, additional discussion and agreements are required.
Proposal 3-2: Add additional feature group 58-3-3, for CSI prediction for UE-sided performance monitoring when N4=1.
Proposal 3-3: Add additional feature group 58-3-4, for CSI prediction for UE-sided performance monitoring when N4>1.
Conclusion
In this contribution, we discussed UE feature for AI/ML. The proposals are:
For AI based beam management:
Proposal 1-1: Remove the square brackets and keep the enclosed texts for 58-1-1, 58-1-2, 58-1-3, 58-1-4, 58-1-5, 58-1-7.
Proposal 1-2: Split component 1 and component 2 of 58-1-1 into two FGs.
Proposal 1-3: Introduce a UE FG for data collection for UE-side beam prediction for BM Case-2.
For AI based positioning:
Proposal 2-1: For FG 58-2-3
DL PRS Resources for UE-based positioning Case 1
Max number of DL PRS Resource Sets per TRP per frequency layer supported by UE.
Values = {1, 2}
Max number of TRPs across all positioning frequency layers per UE.
Values = {4, 6, 12, 16, 24, 32, 64, 128, 256}
Max number of positioning frequency layers UE supports
Values = {1, 2, 3, 4}
Proposal 2-2: For FG 58-2-3-a
DL PRS Resources for UE-based positioning Case 1 on a band
Max number of DL PRS Resources per DL PRS Resource Set
Values = {1, 2, 4, 8, 16, 32, 64}
Note: 16, 32, 64 are only applicable to FR2 bands
Max number of DL PRS Resources per positioning frequency layer.
Values = {6, 24, 32, 64, 96, 128, 256, 512, 1024}
Note: 6 is only applicable to FR1 bands
Proposal 2-3: For FG 58-2-3-b
DL PRS Resources for UE-based positioning Case 1 on a band combination
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1-only.
Values = {6, 24, 64, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR1 only BC.
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2-only.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR2 only BC
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1 in FR1/FR2 mixed operation.
Values = {6, 24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2 in FR1/FR2 mixed operation.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands]
Proposal 2-4: Add a new FG 58-2-4
Support of PRS TEG association information for AI/ML based positioning
Support of reception of association between PRS and TRP Tx TEG for for AI/ML based positioning Case 1
Proposal 2-5: Add a new FG 58-2-5
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Proposal 2-6: For Case 1,
From the Rel-16 UE feature list for NR Positioning FG 13-1, 13-1a, 13-7, 13-7a and 13-8 are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning FG 27-3-2, 27-3-3, 27-6, 27-9, 27-10, 27-10a and 27-11 are applicable AI/ML-based positioning
Proposal 2-7: For Cases 3a and 3b
From the Rel-16 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
Proposal 2-8:
From the Rel-18 UE feature list for NR Positioning, the applicable FGs for AI/ML-based positioning are FFS.
For AI based CSI prediction:
Proposal 3-1: For feature group 58-3-2, remove component item 6 and 8.
Proposal 3-2: Add additional feature group 58-3-3, for CSI prediction for UE-sided performance monitoring when N4=1.
Proposal 3-3: Add additional feature group 58-3-4, for CSI prediction for UE-sided performance monitoring when N4>1.
Reference
[1] R1-2502959 Session Notes of AI 9.15.1 Ad-Hoc Chair (NTT DOCOMO, INC.)
|
TDoc file conclusion not found |
| R1-2504369 Summary AI R19 UE Features.docx |
3GPP TSG RAN WG1 #121 R1-2504369
St Julian’s, Malta, May 19th – 23th, 2025
Agenda Item: 9.15.1
Source: Moderator (AT&T)
Title: Summary of UE features for AI/ML for NR Air Interface
Document for: Discussion/Decision
|
Conclusion
Agreements reached during RAN1 #121 as part of this agenda item are summarized in [17].
|
| R1-2504419_Discussion_on_UE_features_for_AIML_use_cases.docx |
3GPP TSG RAN WG1 #121 R1-2504419 Malta, MT, May 19th – 23rd, 2025
Agenda item: 9.15.1
Source: Qualcomm Incorporated
Title: Discussion on UE features for AIML use cases
Document for: Discussion and Decision
Feature groups for AIML-based beam management
Proposal 1: Propose to confirm the parts in the following tables that are highlighted in green, for the AI/ML for BM use cases.
Proposal 2: Propose to add the text in red to the FGs for the AI/ML for BM use cases.
Feature groups for AIML-based positioning
RAN2 agreed to introduce AI/ML positioning Case 1 as a new positioning method, which means it needs to have dedicated listing of UE features and parameters that relate to capabilities of PRS, reporting, and receiving of assistance data (AD) (which is equivalent to other positioning methods). It was also agreed that UE in Case 1 can obtain most of ADs of UE-based DL-TdoA method.
Observation 1: The AI/ML positioning Case 1 is to be introduced as a new positioning method, requiring specific features related to capabilities of PRS, reporting, and receiving AD.
Observation 2: The AI/ML positioning Case 1 is agreed to share most of ADs and PRS configurations of UE-based DL-TDoA.
From RAN1 perspective, UE features for AI/ML positioning Case 1 need to include equivalent features of UE-based DL-TdoA method, including features related to PRS resource capabilities, PRS processing capabilities, PRS QCL processing capabilities, on-demand PRS capabilities, capabilities related to reporting LocationEstimate, capabilities for receiving assistance data (AD), capabilities related to measurement gap (MG)/PRS processing window (PPW) operations, capabilities related to operation in INACTIVE/IDLE states, and capabilities related to PRS bandwidth aggregation (BWA), as follow:
RAN1 related features: RS configuration related features and RAN1 ADs
Supported PRS resource configurations
Supported PRS QCL processing
Supported PRS processing
Supported PRS in INACTIVE/IDLE modes
Supported PRS outside MG and in PPW
Supported PRS BWA
AD related features:
Support for receiving AD information (e.g., agreed ADs from AD info#1 to #16 of previous agreements)
RAN2 related features: On-demand PRS and reporting related features:
Supported on-demand PRS
Supported on-demand PRS with BWA
Support for Periodical reporting
Support for Ten milliseconds response time
Support for Periodic reporting interval in milliseconds
Support for Scheduled location request
Support for Multiple location estimates in same measurement report
Supported reporting shapes
Above parameters and features are common to existing positioning methods (e.g., UE-based DL-TdoA and UE-based DL-AoD). There can also be other information related to applicability due processing requirements as we discuss in coming section. These can be reconsidered for Case 1. RAN1 can conclude that parameters and parameters of legacy methods can be included for Case 1 (as applicable).
Our understanding is that the AD and RS configurations (except on-demand PRS) related features can be discussed by RAN1, while reporting features can be decided by RAN2. We propose including related FGs and highlighting recommendations to RAN2 regarding the reporting features.
RAN1 UE features
Last meeting RAN1 agreed on initial listing for many of the above features. The following features from legacy UE-based DL-TdoA positioning and common UE features are also relevant to Case 1 and need to be introduced:
Supported PRS processing for Case 1
Supported QCL processing for Case 1
Support Case 1 outside MG and in PPW
Support Case 1 with low latency MG activation
Support Case 1 in IDLE and INACTIVE states
Support Case 1with PRS bandwidth aggregation
RAN1 AD features:
Support of PRS TEG association information
LOS/NLOS indicator for UE-based positioning assistance data
Supported PRS processing for Case 1
For PRS processing, the AIML model can learn features that are bandwidth (BW) dependent and specific. Therefore, it may not support bandwidth similar to legacy methods. In addition, the features of buffering assumptions, measurement period parameters (e.g., T and N), and number of PRSs to be processed in one slot can be impacted by AIML model complexity, making them different from those assumed by legacy methods. AIML processing can have processing and load balancing different from legacy methods. The UE may need extra processing (e.g., preprocessing for measurements as model input) which is not necessarily needed for the legacy methods. Therefore, the PRS processing features of Case 1 need to be differentiated from legacy methods. We propose a separate feature group for PRS processing to be introduced. The content can borrow existing features but need to have its own FG.
Proposal 1: Introduce the following Rel. 19 UE FG for UE-based positioning Case 1:
Supported QCL processing for Case 1
For QCL processing, the AI/ML model learns spatial features with respect to TRPs. Whether UE can support a specific source for deriving QCL relations when obtaining relevant PRS measurements depends on model development and dataset used for training. The AIML model can be sensitive to the QCL source assumption. To ensure consistency between training and inference, the UE features for QCL processing need to be separate and not necessarily common to legacy methods. We propose separate FGs to indicate QCL processing for Case 1. Once again, the same content can be reused but separate FGs are needed:
Support of SSB from neighbour cell as QCL source of a DL PRS for UE-based positioning Case 1
Support of DL PRS from serving/neighbour cell as QCL source of a DL PRS for UE-based positioning Case 1
Proposal 2: Introduce the following Rel. 19 UE FGs for UE-based positioning Case 1:
Support Case 1 outside MG and in PPW
For AI/ML Case 1, UE can also indicate whether it supports positioning outside MG and in PPW, similar to other positioning methods. These capabilities need to be differentiated for AI/ML as it may not be common to legacy methods. We propose the following new FGs:
DL PRS Processing Capability outside MG - buffering capability for UE-based positioning Case 1
DL PRS measurement outside MG and in a PRS processing window for UE-based positioning Case 1
Support of more than one activated PRS processing windows across all active DL BWPs for UE-based positioning Case 1
Proposal 3: Introduce the following Rel. 19 UE FGs for UE-based positioning Case 1:
Support Case 1 with low latency MG activation
For AI/ML Case 1, UE can also indicate whether it supports low latency MG activation, similar to other positioning methods. We propose the following change to existing FG:
Proposal 4: Introduce the following change for UE-based positioning Case 1:
Support Case 1 in IDLE and INACTIVE states
For AI/ML Case 1, UE can also indicate whether it supports AI/ML positioning in IDLE or INACTIVE modes along with corresponding PRS resource and processing features. We propose the following changes to existing FGs and new FGs:
Support of PRS measurement in RRC_INACTIVE state for UE-based positioning Case 1
DL PRS processing capabilities in RRC inactive state for UE-based positioning Case 1
Support of PRS measurement in RRC_IDLE for UE-based positioning Case 1
Proposal 5: Introduce the following Rel. 19 UE FGs for UE-based positioning Case 1:
Support Case 1 with PRS bandwidth aggregation
RAN1 agreed to support assistance data for PRS/PRS bandwidth aggregation similar to UE-based DL-TdoA. Therefore, it is expected that their corresponding UE features that are provided for UE-based DL-TdoA need to be stated for Case 1. We propose to introduce the following FGs corresponding to PRS bandwidth aggregation features for Case 1:
DL PRS processing capabilities for aggregated PRS processing of 2 PFLs in intra-band contiguous within a MG for RRC_CONNECTED - UE-based positioning Case 1
DL PRS processing capabilities for aggregated PRS processing of 3 PFLs in intra-band contiguous within a MG for RRC_CONNECTED - UE-based positioning Case 1
DL PRS processing capabilities for aggregated PRS processing of 2 PFLs in intra-band contiguous for RRC_IDLE and RRC_INACTIVE - UE-based positioning Case 1
DL PRS processing capabilities for aggregated PRS processing of 3 PFLs in intra-band contiguous for RRC_IDLE and RRC_INACTIVE - UE-based positioning Case 1
PRS bandwidth aggregation with two PFL combinations - UE-based positioning Case 1
PRS bandwidth aggregation in RRC_CONNECTED UE-based positioning Case 1
PRS bandwidth aggregation in RRC_ INACTIVE UE-based positioning Case 1
PRS bandwidth aggregation in RRC_IDLE - UE-based positioning Case 1
Proposal 6: Introduce the following Rel. 19 UE FGs for UE-based positioning Case 1:
Other supported AD features
RAN1 agreed to support assistance data for PRS TEG and LOS/NLOS indicator AD as in UE-based DL-TdoA. Therefore, it is expected that their corresponding RAN1 UE features to be provided for Case 1. We propose to introduce FGs corresponding to these RAN1 AD features for Case 1, as follows:
Proposal 7: Introduce the following Rel. 19 UE FGs for UE-based positioning Case 1:
Proposal 8: Introduce the following change for UE-based positioning Case 1:
RAN2 UE features: Recommendations from RAN1
RAN1 agreed on ADs of UE-based DL-TdoA to be provided to UE for Case 1. Some of these require UE features that were initially introduced by RAN2. RAN1 needs to alert RAN2 about these relevant features:
Proposal 8: In AI/ML positioning Case 1, for UE features, RAN1 sends an LS to RAN2 with the following text clarifications regarding other remaining features for Case 1 that RAN1 agreed to support:
For UE-based positioning Case 1, RAN1 agreed on providing UE with all assistance data (AD) of UE-based DL-TdoA (e.g., on-demand PRS, integrity, etc.), other than TRP location. It is RAN1 understanding that some of these ADs were originally introduced by RAN2 and require corresponding RAN2 UE features to be introduced.
For UE-based positioning Case 1, from RAN1 perspective, the following other remaining RAN2 UE features corresponding to UE-based DL-TdoA need also to be introduced by RAN2.
Feature group for AIML-based CSI prediction
In the last meeting, supporting N4=1 and N4 > 1 are introduced as separate UE features for CSI prediction via UE side model.
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Regarding the remaining issues of these two rows, we propose the following
For the highlighted (yellow) text of row 58-3-1,
Confirm component 8 and component 11
Component 9 and 10 to replaced by reporting the O_CPU and / or A_CPU values.
Confirm the per band and per band-combination granularity
Adding component for timeline requirement for FG58-3-1
For the highlighted (yellow) text of row 58-3-1,
Add O_CPU value and / or A_CPU values to the components of row 58-3-2.
Adding component for timeline requirement for FG58-3-2
For other aspects, in our view, R19 AIML based CSI prediction is a new CSI report using legacy R18 codebook configurations, thus, other UE features related to R18 codebook should be added for R19 CSI prediction based on UE side model. Following is proposed
For R19 AIML-based CSI prediction UE feature, copy FG 40-3-2-1b (max number of A-CSI-RS resource), FG 40-3-2-2 family (R16 eType II based prediction R=2), FG 40-3-2-3 family (CQI), FG 40-3-2-7 (starting location), FG 40-3-2-8, FG 40-3-2-9 and FG 40-3-2-10 by changing the feature group name to CSI prediction based on UE side model.
Additionally, similar to legacy UE capability signalling for concurrent codebooks FG 16-8 and FG 23-9-5, concurrent processing of AIML based CSI prediction and legacy non-AI CSI codebooks should be signaled so as to avoid UE underreporting its capability.
Support UE capability signalling for CSI-RS triplets concurrent processing of AIML-based CSI predication and other legacy CSI reports. The signalling granularity should be per-band and also per band-combination.
In the last meeting, it was agreed that UE report total number of AIML processing units for processing AIML CSI-related use cases. This signaling includes the number of AIML PUs per CC and across all CCs. A dedicated feature group should be added according to this agreement.
Introduce dedicated FG for UE to report total number of AIML processing units per CC and across all CCs for AIML CSI-related use cases.
In summary, based on proposals above, the feature groups for AIML-based CSI prediction should contain the followings.
|
TDoc file conclusion not found |
| R1-2504523_Discussion on UE features for AIML for NR Air Interface.docx |
3GPP TSG RAN WG1 #121 R1-2504523
St Julian’s, Malta, May 19th – 23th, 2025
Source: NTT DOCOMO, INC.
Title: Discussion on UE features for AI/ML for NR Air Interface
Agenda Item: 9.15.1
Document for: Discussion and Decision
|
Conclusion
In this contribution, we discussed on UE features for AI/ML for NR Air Interface. Based on the discussion, we made following proposals.
FG common to AI/ML-based CSI reporting
Proposal 1: Introduce the following FG for AI/ML-based CSI reporting, and make it as prerequisite FG for all FG of AI/ML-based CSI reporting for inference.
FGs for AI/ML in beam management
Proposal 2: Update FG 58-1-1 as follows.
Proposal 3: Update FG 58-1-7 as follows.
Proposal 4: Update FG 58-1-2 and 58-1-3 as follows.
Proposal 5: Update FG 58-1-4 and 58-1-5 as follows.
Proposal 6: Introduce the following FG for UE assisted performance monitoring.
FGs for AI/ML for positioning accuracy enhancement
Proposal 7: FG 58-2-1 should be modified as follows.
Proposal 8: FG 58-2-2 should be modified as follows.
Proposal 9: FG 58-2-3, 58-2-3a, and 58-2-3b should follow the elements of existing FGs 13-2/3/4 series. The FGs can be modified as follows.
FGs for AI/ML for CSI prediction
Proposal 10
Merge the components of FG58-3-2 into FG58-3-1 to fully utilize AI/ML's prediction capability, i.e., support N4 ≥ 1 in FG58-3-1 for AI/ML-based CSI prediction.
Proposal 11
Make FG58-0-1 (the newly introduced FG based on FG2-35) one of the prerequisites of FG58-3-1.
Proposal 12
Revisit the components of FG58-3-1 corresponding to CPU and AI/ML PU based on the progress of down-selection.
Copy and reuse the components of FG40-3-2-11 (timeline relaxation) as components of FG58-3-1.
The details of the components can be modified based on the RAN1 progress on timeline discussions.
Proposal 13
Copy and merge the component of FG40-3-2-1b into FG58-3-1.
Copy and reuse the FG40-3-2 family including the following,
FG40-3-2-1a-1 and rename it as DD unit size when A-CSI-RS is configured for CMR N4>1 for CSI prediction for UE-sided inference.
FG40-3-2-2 and rename it as Support R=2 for CSI prediction for UE-sided inference.
FG40-3-2-3 and rename it as Support X=1 based on first and last slot of Wcsi for CSI prediction for UE-sided inference
FG40-3-2-3a and rename it as Support X=2 CQI based on 2 slots for CSI prediction for UE-sided inference.
FG40-3-2-7 and rename it as Support I = (n – nCSI,ref) for CSI reference slot for CSI prediction for UE-sided inference.
FG40-3-2-8 and rename it as Support of L=6 for CSI prediction for UE-sided inference.
FG40-3-2-9 and rename it as Supported maximum periodicity of CMR when configured as periodic CSI-RS for CSI prediction for UE-sided inference.
Proposal 14
Introduce the following FGs for the inference report of AI/ML CSI prediction.
Proposal 15
Introduce the following FG for the training data collection of AI/ML CSI prediction.
Proposal 16
Introduce the following FG for the performance monitoring of AI/ML CSI prediction.
|
| R1-2504690 Views on UE features for AI.docx |
3GPP TSG RAN WG1 #121 R1-2504690
St Julian’s, Malta, May 19th – 23th, 2025
Agenda item: 9.15.1
Source: Apple Inc.
Title: Views on UE features for AI/ML for NR Air interface
Document for: Discussion/Decision
AI based beam management
In RAN1 120bis, the following agreements are captured with FFS.
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agree
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
On many FFS points, our view is that the UE feature features need to be clearly described.
Proposal 1-1:
Remove the square brackets and keep the enclosed texts (“for NW-side model inference”, “for inference”, “for BM Case 1”) for 58-1-1, 58-1-2, 58-1-3, 58-1-4, 58-1-5, 58-1-7.
Proposal 1-2:
Split component 1 and component 2 of 58-1-1 into two FGs.
Proposal 1-3:
For 58-1-2, remove the bracket for Components 2, 3, 4, 5, 7.
Remove proposed components 7a/7b as they are redundant given Component 7.
For Component 8, a UE may report support for any or all of the CSI-RS resources types for set B measurement.
For Component 9, a UE may report support for any or all the report types.
For Component 10, list the performance monitoring options for which a UE may report support any of all.
For Component 11, consider combining Component 7 and Component 11 so a triplet is reported {set B size, set A size, { setB-subset-of-setA, setB-different-from-setA, both}}.
For Component 12, change the description to “Supported maximum number of reported predicted beams in each reporting instance”.
Use separate FGs for Inference reporting, data collection and performance monitoring.
Proposal 1-4:
For 58-1-4, consider similar changes as proposed for 58-1-2.
Consider combining Components 7, 12, 14, 15.
Proposal 1-5:
Introduce a UE FG for data collection for UE-side beam prediction for BM Case-2.
AI based positioning
The AI based positioning UE features was discussed in RAN1 120bis, and the following agreement was captured in [1].
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
In RAN2 #129, the following agreement was made:
Agreement
Introduce AI/ML positioning Case 1 as a new positioning method.
As such, the AI/ML positioning use case FGs and their components should be at least identical to that of the existing use cases. The FGs have been captured in FG 58-2-3, FG 58-2-3-a and FG 58-2-3-b in RAN1 120-bis. However, the components have not been agreed. Based on this, we make the following proposals.
Proposal 2-1:
FG 58-2-3
DL PRS Resources for UE-based positioning Case 1
Max number of DL PRS Resource Sets per TRP per frequency layer supported by UE.
Values = {1, 2}
Max number of TRPs across all positioning frequency layers per UE.
Values = {4, 6, 12, 16, 24, 32, 64, 128, 256}
Max number of positioning frequency layers UE supports
Values = {1, 2, 3, 4}
Proposal 2-2:
FG 58-2-3-a
DL PRS Resources for UE-based positioning Case 1 on a band
Max number of DL PRS Resources per DL PRS Resource Set
Values = {1, 2, 4, 8, 16, 32, 64}
Note: 16, 32, 64 are only applicable to FR2 bands
Max number of DL PRS Resources per positioning frequency layer.
Values = {6, 24, 32, 64, 96, 128, 256, 512, 1024}
Note: 6 is only applicable to FR1 bands
Proposal 2-3:
FG 58-2-3-b
DL PRS Resources for UE-based positioning Case 1 on a band combination
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1-only.
Values = {6, 24, 64, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR1 only BC.
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2-only.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR2 only BC
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1 in FR1/FR2 mixed operation.
Values = {6, 24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2 in FR1/FR2 mixed operation.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands]
Two additional FGs need to be added to ensure support for all the FGs defined for each positioning method. These are on the PRS TEG associated information and for support in the RRC_INACTIVE state. As at now, there is no agreement on whether AI/ML positioning is or is not supported in RRC_INACTIVE. Unless there is an agreement not to support it, it is necessary to have an FG to allow for indication of support like the other positioning methods. Based on these arguments, we make the following proposals:
Proposal 2-4: Add a new FG 58-2-4
Support of PRS TEG association information for AI/ML based positioning
Support of reception of association between PRS and TRP Tx TEG for for AI/ML based positioning Case 1
Proposal 2-5: Add a new FG 58-2-5
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Finally, for Case 1, 3a and 3b, the following existing FGs should be applicable to AI/ML positioning
Proposal 2-6: For Case 1,
From the Rel-16 UE feature list for NR Positioning FG 13-1, 13-1a, 13-7, 13-7a and 13-8 are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning FG 27-3-2, 27-3-3, 27-6, 27-9, 27-10, 27-10a and 27-11 are applicable AI/ML-based positioning
Proposal 2-7: For Cases 3a and 3b
From the Rel-16 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
Proposal 2-8:
From the Rel-18 UE feature list for NR Positioning, the applicable FGs for AI/ML-based positioning are FFS.
AI based CSI prediction
The UE features was discussed in RAN1 120bis, and captured in [1].
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
On inference related UE capability, it was agreed in the RAN1 120bis that AI/ML PU is introduced. It is under further discussion on whether UE capability is introduced to indicate whether CPU or AI/ML PU is used, and how many AI/ML PU is used for CSI prediction. Whether additional UE capability is introduced can be further discussed in RAN1.
For feature group 58-3-2, component 6 and component 8 are duplicate of component 3, which can be removed.
Proposal 3-1: For feature group 58-3-2, remove component item 6 and 8.
In RAN1 120bis, separate UE capability for performance monitoring was agreed for AI based CSI prediction. On performance metric and reporting procedure itself, additional discussion and agreements are required.
Proposal 3-2: Add additional feature group 58-3-3, for CSI prediction for UE-sided performance monitoring when N4=1.
Proposal 3-3: Add additional feature group 58-3-4, for CSI prediction for UE-sided performance monitoring when N4>1.
Conclusion
In this contribution, we discussed UE feature for AI/ML. The proposals are:
For AI based beam management:
Proposal 1-1:
Remove the square brackets and keep the enclosed texts (“for NW-side model inference”, “for inference”, “for BM Case 1”) for 58-1-1, 58-1-2, 58-1-3, 58-1-4, 58-1-5, 58-1-7.
Proposal 1-2:
Split component 1 and component 2 of 58-1-1 into two FGs.
Proposal 1-3:
For 58-1-2, remove the bracket for Components 2, 3, 4, 5, 7.
Remove proposed components 7a/7b as they are redundant given Component 7.
For Component 8, a UE may report support for any or all of the CSI-RS resources types for set B measurement.
For Component 9, a UE may report support for any or all the report types.
For Component 10, list the performance monitoring options for which a UE may report support any of all.
For Component 11, consider combining Component 7 and Component 11 so a triplet is reported {set B size, set A size, { setB-subset-of-setA, setB-different-from-setA, both}}.
For Component 12, change the description to “Supported maximum number of reported predicted beams in each reporting instance”.
Use separate FGs for Inference reporting, data collection and performance monitoring.
Proposal 1-4:
For 58-1-4, consider similar changes as proposed for 58-1-2.
Consider combining Components 7, 12, 14, 15.
Proposal 1-5:
Introduce a UE FG for data collection for UE-side beam prediction for BM Case-2.
For AI based positioning:
Proposal 2-1: For FG 58-2-3
DL PRS Resources for UE-based positioning Case 1
Max number of DL PRS Resource Sets per TRP per frequency layer supported by UE.
Values = {1, 2}
Max number of TRPs across all positioning frequency layers per UE.
Values = {4, 6, 12, 16, 24, 32, 64, 128, 256}
Max number of positioning frequency layers UE supports
Values = {1, 2, 3, 4}
Proposal 2-2: For FG 58-2-3-a
DL PRS Resources for UE-based positioning Case 1 on a band
Max number of DL PRS Resources per DL PRS Resource Set
Values = {1, 2, 4, 8, 16, 32, 64}
Note: 16, 32, 64 are only applicable to FR2 bands
Max number of DL PRS Resources per positioning frequency layer.
Values = {6, 24, 32, 64, 96, 128, 256, 512, 1024}
Note: 6 is only applicable to FR1 bands
Proposal 2-3: For FG 58-2-3-b
DL PRS Resources for UE-based positioning Case 1 on a band combination
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1-only.
Values = {6, 24, 64, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR1 only BC.
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2-only.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for FR2 only BC
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR1 in FR1/FR2 mixed operation.
Values = {6, 24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands
Max number of DL PRS Resources supported by UE across all frequency layers, TRPs and DL PRS Resource Sets for FR2 in FR1/FR2 mixed operation.
Values = {24, 64, 96, 128, 192, 256, 512, 1024, 2048}
Note this is reported for BC containing FR1 and FR2 bands]
Proposal 2-4: Add a new FG 58-2-4
Support of PRS TEG association information for AI/ML based positioning
Support of reception of association between PRS and TRP Tx TEG for for AI/ML based positioning Case 1
Proposal 2-5: Add a new FG 58-2-5
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Support of PRS measurement in RRC_INACTIVE state for AI/ML based positioning Case 1
Proposal 2-6: For Case 1,
From the Rel-16 UE feature list for NR Positioning FG 13-1, 13-1a, 13-7, 13-7a and 13-8 are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning FG 27-3-2, 27-3-3, 27-6, 27-9, 27-10, 27-10a and 27-11 are applicable AI/ML-based positioning
Proposal 2-7: For Cases 3a and 3b
From the Rel-16 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
From the Rel-17 UE feature list for NR Positioning all FGs on SRS transmission are applicable AI/ML-based positioning
Proposal 2-8:
From the Rel-18 UE feature list for NR Positioning, the applicable FGs for AI/ML-based positioning are FFS.
For AI based CSI prediction:
Proposal 3-1: For feature group 58-3-2, remove component item 6 and 8.
Proposal 3-2: Add additional feature group 58-3-3, for CSI prediction for UE-sided performance monitoring when N4=1.
Proposal 3-3: Add additional feature group 58-3-4, for CSI prediction for UE-sided performance monitoring when N4>1.
Reference
[1] R1-2502959 Session Notes of AI 9.15.1 Ad-Hoc Chair (NTT DOCOMO, INC.)
|
TDoc file conclusion not found |
| R1-2504925 Session_Notes_AI_9-15-1.docx |
3GPP TSG RAN WG1 #121 R1-2504925
St Julian’s, Malta, May 19th – 23th, 2025
Source: Ad-Hoc Chair (AT&T)
Title: Session Notes of AI 9.15.1
Agenda Item: 9.15.1
Document for: Endorsement
9.15.1 UE features for AI/ML for NR Air Interface
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Note: It is RAN1 understanding that RAN2 will include listing of corresponding ADs as per RAN1 agreement for providing all assistance data of UE-based DL-TdoA)
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Adopt the following changes highlighted in chromatic fonts, while keeping the yellow highlighting, if any, as shown
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
Agreement: Introduce the following Rel. 19 UE FGs (yellow highlighting, if any, shows text that’s not yet agreed)
R1-2503239 UE Features for Rel-19 AI/ML for NR Air Interface Ericsson
R1-2503255 UE features for AI/ML for NR air interface Huawei, HiSilicon
R1-2503385 Discussion on UE features for AI/ML for NR Air Interface vivo
R1-2503399 UE features for AI/ML for NR Air Interface Nokia
R1-2503591 UE features for AI/ML for NR Air Interface Samsung
R1-2503652 Discussion on UE features for AI/ML for NR Air Interface ZTE Corporation, Sanechips
R1-2503740 Views on UE features for AI/ML for NR Air Interface Ofinno
R1-2503781 Discussion on UE features for AI/ML for NR Air Interface CATT, CICTCI
R1-2503850 Discussion on UE features for AI/ML for NR air Interface CMCC
R1-2503900 Discussion on UE features for AI/ML for NR Air Interface Xiaomi
R1-2503984 Discussions on UE features for AI/ML for NR Air Interface LG Electronics
R1-2504227 UE features for AIML for NR air interface OPPO
R1-2504352 On Rel-19 UE features for AI/ML for NR air interface Apple
R1-2504369 Summary of UE features for AI/ML for NR Air Interface Moderator (AT&T)
R1-2504419 UE features for AI/ML for NR air interface Qualcomm Incorporated
R1-2504523 Discussion on UE features for AI/ML for NR Air Interface NTT DOCOMO, INC.
|
TDoc file conclusion not found |