AI/ML for NR Air Interface

agreement

For report content of inference results for UE-sided model, where the largest RSRP value is quantized to a 7-bit value in the range [-140, -44] dBm with 1dB step size, and the differential RSRP is quantized to a 4-bit value with 2 dB step size. Note: the model output is UE implementation and it doesn’t have to be RSRP subject to dBm value.

agreement

For report content of inference results for UE-sided model for BM-Case 1, the RSRP of predicted beam(s)in the report of inference results, is the predicted RSRP, where the predicted RSRP is based on AI/ML output. Note: how to capture it in the spec is a separate discussion.

agreement

For UE-side AI/ML model inference and BM-Case2, for the quantization of a RSRP value of inference results in a report over multiple future time instances, the largest RSRP value based on prediction of all time instances is the reference RSRP, and differential RSRPs in the report are computed relative to the reference RSRP. The time instance information of the beam with the largest RSRP are additionally indicated in the report.

agreement

For inference, for BM-Case 2 of UE-side model, The time gap between two consecutive future time instances is configured by RRC, and the number of future time instance(s) N is configured by RRC. time gap is [10ms, 20ms, 40ms, 80ms, 160ms] N = [1, 2, 4, 8] Reference time of the earliest time instance for the predicted results is based on the most recent occasion of the CSI-RS/SSB resource in Set B for measurement Where the most recent occasion of the CSI-RS/SSB resource of set B is the latest CSI-RS/SSB occasion no later than the corresponding CSI reference resource of the corresponding inference report. Presented in Thursday session

agreement

For UE-sided model, for configuring the resource for data collection purpose, support CSI-ReportConfig can used for configuring the resources for data collection purpose without CSI report. One CSI-ResourceConfigId is configured for Set A. One CSI-ResourceConfigId is configured for Set B. Note: UE performs measurement on all resources One or two associated IDs can be configured in CSI-ReportConfig When Set B is equal or a subset of set A (i.e., NZP-CSI-RS-ResourceId/SSB-Index in the resource set for Set B is within the NZP-CSI-RS-ResourceId/SSB-Index in the resource set for Set A), one associated ID is configured, Otherwise, one associated ID is configured for Set A and another one associated ID is configured for Set B FFS: whether/how to support 'aperiodic' CSI RS Note: This is not related to whether/how to support delivery/transmission of the collected data for training for UE-sided model.

agreement

For UE-sided model, in CSI-ReportConfig for inference One or two associated IDs can be configured in CSI-ReportConfig When Set B is equal or a subset of set A (i.e., NZP-CSI-RS-ResourceId/SSB-Index in the resource set for Set B is within the NZP-CSI-RS-ResourceId/SSB-Index in the resource set for Set A), one associated ID is configured, Otherwise, one associated ID is configured for Set A and another one associated ID is configured for Set B FFS: At least BM-Case 1, the applicability for 'aperiodic' CSI RS Final summary in R1-2501595.

agreement

For CSI prediction using UE-side model, at least for inference, Rel-18 CSI framework is reused. For CSI-RS resource type for CMR, periodic, semi-persistent, and aperiodic CSI-RS are supported. For inference report, for N4>=1, support Rel-18 codebook ('typeII-Doppler-r18')

agreement

For CSI prediction using UE-side model, if performance monitoring type 1 or 3 is supported, for calculation of monitoring metric, support Based on intermediate KPI Down select between SGCS and NMSE by RAN1#120bis FFS on the definition of SGCS/NMSE and how to calculate monitoring metric

agreement

For CSI prediction using UE-side model, for CSI processing criteria and timeline, at least for inference further study on Whether the CPU should be shared or separately counted between legacy CSI reporting and AI/ML-based CSI reporting Whether the Processing Unit should be shared or separately counted among AI/ML related features/functionalities. Whether new timeline is needed/updated for inference, and whether a different timeline is needed when functionality switches/activates. Whether legacy framework for active CSI-RS resource and port counting can be reused Note: Strive to study CSI processing criteria considering both BM and CSI case, and take the existing solutions as starting point. Presented in Friday session

agreement

For CSI prediction using UE-side model, for data collection for training, For resource configuration, For CSI-RS resource type for CMR, periodic and semi-persistent are supported. FFS on support of aperiodic CSI-RS resource FFS on whether to separately configure CSI-RS resource(s) used for measurements for model input and CSI-RS resource(s) used for measurements for ground-truth CSI FFS whether/how to enhance CSI-RS resource configuration to reduce CSI-RS signalling overhead. FFS on CSI report configuration Final summary in R1-2501626.

agreement

Send Reply LS R1-2501522 to SA2 about Case 2b with the content of Section 2 of R1-2501521 by revising “RAN1 would continue work and may provide additional input in the future to SA2 for Case 2b if any new agreement reached in RAN1.” To “RAN1 may provide additional input in the future to SA2 for Case 2b if any new agreement reached in RAN1.” Final LS is approved in R1-2501523.

agreement

Send Reply LS R1-2501524 to SA2 about Case 3b with the content of Section 3 of R1-2501521 by adding all the agreement made in this meeting for Case 3b. Final LS is approved in R1-2501525.

conclusion

For AI/ML based positioning Case 3a, regarding the time stamp in a measurement report from gNB to LMF, Existing IE “Time Stamp” in TS 38.455 can be reused from RAN1 perspective.

agreement

For Rel-19 AI/ML based positioning, for Case 3b, “FFS: k” in RAN1#119 agreement is resolved by supporting: k = {0...5}

agreement

For Rel-19 AI/ML based positioning, for Case 3b, “FFS: Nt' values” in RAN1#119 agreement is resolved by supporting: Nt' = {8, 16, 24}

agreement

For AI/ML based positioning Case 1, from RAN1 perspective, when the label data of location is generated by LMF and transferred from LMF to UE, label and quality indicator of label can be provided by reusing existing IEs. From RAN1 perspective, the existing IE can use one of the geographic shapes defined in TS 23.032. The location estimate uncertainty and confidence (if included with the geographic shapes) can serve as quality indicator of the label.

agreement

For AI/ML based positioning, When channel measurement is to be reported/sent, time stamp of channel measurement refers to the time instance when the channel measurement is performed. When label is to be reported/sent, time stamp of label is the time instance for which the label is valid. Presented in Wednesday session

agreement

For AI/ML based positioning Case 3b, with the existing definition of two channel measurement types (A) or type (B): (A) path-based measurement, i.e., measurement in the existing specifications (up to Rel-18), (B) Rel-19 enhanced measurement (see definition in RAN1#119 agreement for Case 3b). Descriptive info of the channel measurement is included together with the channel measurement values, either explicitly or implicitly. The descriptive info of the channel measurement includes at least: measurement type (A) or type (B) and measurement parameters. Regarding measurement parameters, For (A) path-based measurement, the measurement parameters include: number of additional paths, k. For (B) Rel-19 enhanced measurement, the measurement parameters include at least: Nt', k. FFS: Nt Note: Type (A) and Type (B) are used for discuss purpose only.

conclusion

For Rel-19 AI/ML based positioning Case 3b, If the gNB cannot report measurements according to measurement type (A) or measurement type (B) requested by LMF, the gNB responds to LMF that: the gNB cannot provide measurement with the requested measurement type. The gNB does not respond to LMF with a measurement report where the measurement type is different from requested. It is up to RAN3 to decide the signaling details. Note: It is RAN1 understanding that this is the same behavior as in legacy when the NG-RAN node cannot support the requested measurement.

agreement

RAN1 send an LS to RAN3 and RAN2 to inform them above agreements and conclusions.

agreement

For training data collection of AI/ML based positioning 3a/3b, if Part A and Part B are generated by different entities, for pairing between a Part A entry and a Part B entry, the following is needed: The time stamp of Part A (if Part A is transmitted) and the time stamp of Part B (if Part B is transmitted). FFS: other information is not precluded (e.g., if Part B is valid for a duration) Note: Purpose such as “training data collection” will not necessarily be specified in RAN1 specifications. RAN1 send an LS to RAN3 and RAN2 to inform the above. Friday decision: The draft LS doesn't seem to correctly capture the agreements. [Post-120-AI/ML-01] – Yufei (Ericsson) Email approval of Reply LS on AI/ML Positioning Case 3b by February 26 (Wednesday)

conclusion

Under 10% UCI loss, Case 2 shows small to large performance drop (by Comparing Scenario B and Scenario A). Signaling support to mitigate the UCI loss can improve the performance and helps preserving the gain of Case 2 over Case 0 (by comparing Scenario C and Scenario A and benchmark schemes).

agreement

Investigate the following approaches for signaling support for mitigating the impact of UCI loss NW-signaling to reset of historical CSI information at UE NW-triggered CSI retransmission

conclusion

For direction A, performance impact, if any, due to NW / UE data distribution mismatch with respect to UE side additional condition can be addressed.

agreement

In Options 3a-1 and 4-1, the exchanged dataset or the model parameters can be associated with an ID for pairing related discussion, then The same ID can be used for UE to collect UE-side target CSI for UE-side training The same ID can be used for applicability inquiry and reporting The same ID can be used for inference configuration FFS: whether ID/even same ID is needed for monitoring configuration FFS: where the ID is assigned Note: whether the purpose for pair will be specified will be discussed separately.

agreement

For inter-vendor-collaboration Options 3a-1 and 4-1 in Direction A, performance target is confirmed as additional information along with the exchanged dataset or the model parameters. FFS: type of performance metric FFS: input data for evaluating the performance

conclusion

Direction C with standardized model ensures a minimum performance requirement. The so called fully specified reference model discussed in RAN1 for Direction C is assumed to be equivalent to the to-be-specified model, if specified, for performance requirements in RAN4. Direction A with standardized model parameter / dataset exchange from NW-side to UE-side may achieves better performance.

conclusion

For NW-first training, in inter-vendor training collaboration Direction A and C, identification of root cause for performance issues can be achieved at least by the following Data distribution mismatch that is identified by either NW side or UE side Via target CSI and CSI feedback pair report, e.g., with the understanding that NW side runs inference using the NW side reference CSI generation part and compare with the inference using UE side CSI generation part. Note that this conclusion does not preclude other methods. Note that triggering of root cause detection may be after detecting performance degradation

agreement

For studying the standardized model structure, for temporal domain Case 0, in case of spatial-frequency domain input, adopt the following model structure as one example structure for study purpose, Encoder description: Decoder description: The decoder has a mirroring design as the encoder. Details are to be discussed.

agreement

For model structure scalability study for temporal domain Case 0, For the choice of token dimension and feature dimension, Alt 1: Use subband as the token dimension and Tx port as a feature dimension The number of tokens varies with the number of subbands. Alt 2: Use Tx port as the token dimension and subband as a feature dimension The number of tokens varies with the number of Tx ports. Alt 3: Use a fixed-size sub-block of Tx ports and subbands matrix (e.g., n_Tx_ports*m_Subbands) as a token and represent the input as a sequence of tokens. The number of tokens varies with the number of Tx ports and the number of subbands. For scalability over the feature dimension, Alt1: specific embedding layer for each feature size Alt2: a common embedding layer with padding (e.g., zero-padding or other techniques for padding values) For scalability over the token dimension, Alt1: positional embedding specific to each token index tokens out of token positions are used as input. Alt2: Padding at the input For scalability over payload configurations, Alt1: specific output linear layer for each payload configuration Alt2: truncation/masking of the output linear layer output Alt3: by varying quantization parameters Notes Other Alternatives are not precluded. Different Alternatives may be used in combination. Same/similar approach is applied at the decoder side. Evaluations to consider: Case 1 (scalable structure): Scalable model structure described above Using model structure as indicated in above diagram with fixed hyperparameters, companies may train a single parameter set or different parameter sets across different {number of Tx ports, CSI feedback payload size, bandwidth} assuming a common model structure. To report whether a single parameter set or different parameter sets were used across different {number of Tx ports, CSI feedback payload size, bandwidth}. (e.g., single parameter set across different payload sizes and bandwidths, different parameter set across different number of Tx ports) Case 2 (dedicated structure): Using model structure as indicated in above diagram with different hyperparameters, where the input and the output related hyperparameters are chosen optimally corresponding to each specific {number of Tx ports, CSI feedback payload size, bandwidth} without scalability considerations. Different parameter sets are trained across different number of Tx ports, CSI feedback payload sizes, and bandwidths. For each scalable model structure choice, to evaluate the SGCS performance of the non-AI/ML benchmark (e.g., Rel-16 eType2), Case 1, and Case 2, for each of {number of Tx ports, CSI feedback payload size, bandwidth}, and report the average gain (%) in SGCS of Case 1 and Case 2 over the non-AI/ML benchmark, as well as the loss (%) in the average gain of Case 1 w.r.t. Case 2. The average is performed by first calculating the SGCS gain (%) for each {number of Tx ports, CSI feedback payload size, bandwidth} and then averaging the SGCS gain (%) values over {number of Tx ports, CSI feedback payload size, bandwidth}.

agreement

For model structure scalability study for temporal domain Case 0, for the choice of {number of Tx ports, CSI feedback payload size, bandwidth}, take the following as baseline values. Number of Tx ports [N1, N2, P] = [2, 8, 2] (32 CSI-RS ports) [N1, N2, P] = [4,4,2] (32 CSI-RS ports) [N1, N2, P] = [2, 4, 2] (16 CSI-RS ports) CSI feedback payload size Multiple values, encouraged to pick at least one value from each of small, medium, and large payload regions X, Y, and Z. Bandwidth & subband size All subbands A subset of all the subbands.

agreement

Capture the following observations into TR: For the evaluation of performance impact due to UE-side / NW-side data distribution mismatch with respect to UE-side additional condition (issue 4 and 6), When dataset A include dataset B, for case 2, option 3a-1, with NW-side target CSI sharing 1 source [Panasonic] observes minor performance loss (-0.22% ~ -1.09%) relative to case 1 with NW side target CSI sharing. When dataset A include dataset B, for case 2, option 3a-1, without NW-side target CSI sharing, 1 2 source [ZTE,vivo] observes minor performance loss (-0.67 -0.006% ~ -0.97%) related to case 1 without NW side target CSI sharing. 1 source [OPPO] observes moderate performance loss (-3.3% ~ -4%) relative to case 1 without NW side target CSI sharing When dataset A include dataset B, for case 3 (Direction B), 1 2 source [ZTE,vivo] observes minor performance loss (-0.6% ~ -0.9% -0.9% ~ +0.01%) relative to case 1. When dataset A does not include dataset B, for case 2, option 3a-1, with NW-side target CSI sharing 3 sources [QC, Apple, Samsung] observe similar performance (-0.4% ~ +0.24%) as case 1 for Alt1 UE training. 3 sources [vivo, Xiaomi, ETRI] observe minor performance loss up to -2.5% relative to case 1 for Alt1 training. 1 2 source [Ericsson, ZTE] observe moderate performance loss of (-3.8 ~ -8.3%) relative to case 1 for Alt1 training. 1 source [OPPO] observe significant performance loss (-51% ~ -62.5%) relative to case 1 for Alt1 UE training. 2 sources [CATT, Futurewei] observe minor performance loss (-1.62% -2.78% ~ -3.2%) relative to case 1 for Alt2 UE training 1 source [Apple] observe moderate loss of -6.7% relative to case 1 for Alt2 UE training. When dataset A does not include dataset B, for case 2, option 3a-1, without NW-side target CSI sharing 4 5 sources [CATT, ZTE, Xiaomi, ETRI, vivo] observe minor to moderate performance loss (-0.4% ~ -3.9%) relative case 1. 2 3 source [QC, Ericsson, Apple] observe moderate performance loss (-6.7% ~ -8.6%) relative to case 1. 1 source [OPPO] observes significant performance loss of -62.1% relative to case 1. When dataset A does not include dataset B, for case 2, option 4-1, 4 sources [QC, Apple, Xiaomi, ETRI] observe zero to minor performance loss (-2.4% ~ 0%) relative to case 1 for Alt1 UE training. 1 source [Ericsson] observe minor to moderate performance loss (-2.9% ~ -7.9%) relative to case 1 for Alt1 UE training depending on whether dataset A applies augmentation using various phase normalization methods. 4 sources [CATT, Xiaomi, Futurewei, ETRI] observe minor performance loss (-1.41% ~ -3.52%) relative to case 1 for Alt2 UE training. 1 source [Apple] observes moderate loss of -7.9% relative to case 1 for Alt2 UE training. 1 source [vivo] observes moderate loss of -7.7% relative to case 1 for 3a-1 for Alt1 UE training when backbone are different. When dataset A does not include dataset B, for case 3, Direction B, 4 sources [CATT, vivo, ZTE, ETRI] observe minor loss to positive gain (-3.7% ~ 1%) relative to case 1. 1 source [QC] observes significant loss (-20%) relative to case 1.

agreement

Capture the following observations into TR: For the evaluation of performance impact due to mismatch between the distribution of the dataset used for reference model(s) training, UE-side data distribution, and NW-side data distribution (Issue 9), For case 2 (model trained on dataset S but tested on dataset B), 2 sources [vivo, QC] observe significant performance loss (-7.2% ~ -17.4%) relative to case 1, where dataset B consists of actual field data. 1 source [ETRI] observes significant performance loss of -37.3% relative to case 1, where dataset S and B are different by TxRU mapping. 3 resources [ZTE, Panasonic, Ericsson] observe moderate performance loss (-2.12% ~ -4.75%) relative to case 1, where dataset S and B are generated from different scenarios, antenna layout or UE location. E.g., Uma and InH respectively, or dataset S and B are different by antenna layout and indoor/outdoor ratio. For case 2A (model further finetuned on dataset B), If finetune at NW and fix encoder at UE For using actual field data as dataset B When dataset A / B mismatch is considered but dataset B does not contain A, 1 source [vivo] observe moderate loss (-4.63% ~ -7.66%) relative to case 1. The improvement compared to case 2 is minor to moderate (2.58% ~ 9.17%). When dataset A / B mismatch is not considered, 2 1 sources [vivo, QC] observes minor to moderate loss of (-3.5% ~ -7.66%) loss relative to case 1. The improvement compared to case 2 is moderate to significant (2.6% ~ 13.9%) depending on the specific scenario where the field data is collected. For using synthetic data as dataset B When dataset A / B mismatch is considered but dataset B does not contain A, 1 source [ZTE] observes moderate loss to case 1 (-4.32% ~ -4.75%). The improvement compared to case 2 is negative (-0.68% ~ -1.02%). When dataset A includes or is the same as dataset B, 3 sources [ZTE, Panasonic, Ericsson] observe minor to moderate loss relative to case 1 (-0.97% ~ -4.67%). The improvement compared to case 2 is negative to minor (-0.61.01% ~ 1.6149%). Dataset S and B are Uma and InH respectively, or dataset S and B are different by antenna layout and indoor/outdoor ratio. If finetune at UE and fix decoder at NW For using actual field data as dataset B, 2 sources [vivo, QC] observe moderate to significant loss (-6.8% ~ -12%) relative to case 1. The improvement compared to case 2 is minor moderate (0.4% ~ 7.1%) depending on the specific scenario where the field data is collected. For using synthetic data as dataset B 2 source [Panasonic, Ericsson] observe minor to moderate loss (-1.56% ~ -3.81%) relative to case 1. The improvement compared to case 2 is marginal (0.03% ~ 0.75%). Dataset S and B are different by antenna layout and indoor/outdoor ratio. 1 source [ETRI] observes significant loss (-23.8%) relative to case 1. The improvement compared to case 2 is significant (13.5%). Dataset S and B are different by TxRU mapping If finetune at both UE and both NW sides For using actual field data as dataset B When dataset A / B mismatch is considered and dataset A does not contain B, 1 source [vivo] observes significant loss (-8.5% ~ -44.85%) relative to case 1 depending on specific scenarios for data collection. The improvement compared to case 2 is negative (-28% ~ -1.3%). The loss relative to finetune at one side is negative (-1.72% ~ -37.19%). When dataset A / B mismatch is not considered and A is equal to B, 1 2 sources [vivo, QC] observes moderate to significant loss of (-7.3% ~ -56.1%) relative to case 1. The improvement compared to case 2 is negative to significant (-39.3% ~ 10.1%). The loss relative to finetune at one side is minor to significant (-1.7 ~ 48.4%) negative to moderate (-3.82~ 4.67%). For using synthetic data as dataset B When dataset A / B mismatch is modelled by different scenarios (i.e., by NW side condition), If the input is in spatial-frequency domain, 1 source [Samsung] observes minor to significant loss (-29.0% ~ -30.5%) relative to case 1 depending on whether data Set S is used in the finetune. If the input is in angle-delay domain, 1 source [Samsung] observes similar performance to case 1 (-0.3% ~-1.3%) When dataset A includes B or is equal to B or has same distribution as B, If the input is in spatial-frequency domain, and dataset A does not contain B, 2 sources [Panasonic, Ericsson] observe minor to moderate loss (-1.1% ~ -3.69%) relative case 1 depending on the modelling of synthetic data. The improvement compared to case 2 is minor (0.57% ~ 1.3%). Compared to finetune at one side, the performance improvement is similar to minor (-0.35% ~ 1%). If the input is in spatial-frequency domain, 1 source [Samsung] observes minor to significant loss (-3.67% ~ -11.18%) relative to case 1 depending on whether data Set S is used in the finetune. If the input is in angle-delay domain, 1 source [Samsung] observes similar performance to case 1 (-0.89% ~1.51%). When dataset A / B mismatch is modelled by Rx antenna spacing and dataset A does not contain B and dataset A has same distribution as S (i.e., synthetic/field data distribution mismatch modeled only at the UE side (via UE-side additional condition) but not modeled at the NW-side), 1 source [Samsung] observes similar performance (-0.89% ~ 0.21%) as case 1.

agreement

Study performance-complexity trade-off by comparing different AI/ML models, e.g. by optimizing existing designs, and/or by comparing different precoder representation in (spatial-frequency and angle-delay) or (spatial-frequency-time and angle-delay-doppler), by considering the following aspects Performance comparison between different AI/ML designs and benchmark schemes Complexity numbers (FLOP, calculated/projected latency or power consumption if available, measured latency or power consumption if available) of different AI/ML designs and benchmark schemes

agreement

From RAN1 perspective, for the study of delivery/transfer Case z4, if the known structured model is specified in 3GPP, at least consider the following for the open format. Define an open format within 3GPP.

conclusion

From RAN1 perspective, the study progress of model identification and model transfer/delivery in Agenda item 9.1.4.2 are sufficient. New work/discussion in AI 9.1.4.2 can be triggered by LS from other working group(s) or other TSG(s), if any The work on pCR (if any) to capture the output of AI 9.1.4.2 to be done in future meeting(s) Other on-demand work (if any) From RAN1 perspective, there is no consensus in Rel-19 on the need of standardized solution for model delivery/transfer of one-sided model In RAN1, if needed, whether model identification is needed for CSI compression or not is to be decided in Agenda item 9.1.4.1 In RAN1, if needed, whether standardized solution(s) for model transfer/delivery are needed for CSI compression or not is to be decided in Agenda item 9.1.4.1 Final summary in R1-2501406.

Evolved positions

New contributors this meeting

Evolved positions

New contributors this meeting

Evolved positions

New contributors this meeting

Company positions

Open issues

Company positions

Open issues

Company positions

Open issues