3GPP TSG RAN WG1 Meeting #120bis												R1-2501750
Wuhan, China, April 7th – 11th, 2025

Agenda Item:		9.8.2	
Source:				InterDigital, Inc.
Title:						Remaining Issues on Extension of FR3 Channel Modelling 
Document for:		Discussion and Decision

Conclusions
In this contribution, we shared and discussed our views for Rel-19 FR3 NF channel modeling. Based on presented discussions, the following observations and proposals are made,

Observation 1: SnS and cluster consideration have a high impact on details related to other steps.
Proposal 1: Prioritize discussions on SnS and cluster consideration to facilitate progress on other topics.

Observation 2: For generating the distance  for the  non-direct paths, to ensure the support of multi bounce channel, two options can be considered. 
Proposal 2: In generating the distance  for the  non-direct paths,  is generated by , where  is a scaling factor and is generated by a Beta distribution with reusing the  and  defined for .

Observation 3: Based on the ray tracing evaluation for non-direct path for indoor office scenario, there is 1 cluster that has the property of .
Proposal 3: In addition to the agreed  cluster,  cluster should be defined with  and     for the indoor office scenario. 

Observation 4: The cluster visibility probability and visibility region are a function of the first bounce distance . The cluster visibility increases when the first bounce distance  increases for both indoor office and UMi scenarios. 
Proposal 4: Support the first bounce cluster location-based/distance-based, i.e.,  approach for determination of the cluster visibility region for non-direct paths.

Observation 5: The cluster visibility region is also a function of the first bounce distance . The -th cluster first bounce distance  can be reused for deriving non-direct path VP/VR. 
Proposal 5: The cluster visibility can be determined by its first bounce distance  when it is generated at the Step 7 and the scaling factor  agreed in RAN1 #120 meeting can be utilized for the determination of visibility probability and visibility region at BS side. 

Proposal 6: For the modelling of spatial non-stationarity, if the unified visible probability and visibility region-based approach is adopted, the following method is used to determine the visibility region: 
Visibility region is generated based on the first bounce distance , and the visibility (rectangular) area is determined as by the 4-th order polynomial function listed in Table 2.   

Proposal 7: For the modelling of spatial non-stationarity in Step 11, if the unified visible probability and visibility region-based approach is adopted, the power attenuation factor is generated as  dB when the visibility region of n-th cluster is non-visible. 

3GPP TSG RAN WG1 #120bis                                                       R1-2501755
Wuhan, China, April 7th – 11th, 2025

 
Agenda Item:	9.8.2
Source: 	LG Electronics
Title: 	Discussion on channel modelling adaptation/extension for 7-24GHz
Document for:	Discussion and decision

Conclusion
Based on the above discussion, the following proposals and observations are provided.

Observation 1: The currently agreed blocker-based approach cannot model the phenomenon that each element of an extremely massive antenna array may experience different impact by a cluster.

Proposal 1: The near- or far-field condition for the non-direct paths between BS and UE is determined by near-field probability.

Proposal 2: For the modelling of spatial non-stationarity, impact of incomplete scatterer should be considered.

Proposal 3: For the modelling of spatial non-stationarity at BS side, support VP/VR based approach.

3GPP TSG RAN WG1 #120bis                                                            R1-2501820
WuHan, China, April 7th – 11st, 2025

Source:	vivo, BUTP
Title:	Views on channel model adaptation/extension of TR38.901 for 7-24GHz
Agenda Item:	9.8.2
Document for:	Discussion and Decision

Conclusions
In this contribution, we have elaborated the realization of near-filed channel modeling in consideration of the antenna element-wise channel parameters associated with indirect paths. Then, the issue of spatial non-stationarity is addressed, with a detailed investigation of how to model such a property by visibility region. The observations and proposals are summarized as follows.
Observation 1: The SNS probability of cluster follows the truncated normal distribution.
Observation 2: For cluster without SNS, the power attenuation factor should be modeled as 1.
Observation 3: For cluster with SNS, the power attenuation factor could be modeled as 1 for the visible elements.
Observation 4: For cluster with SNS, the formula related to the minimum distance between Tx antenna element and the VR region could be used to model the power attenuation factor with the range [0 1] for invisible elements.

Proposal 1:  The distance  is generated independently for different sub-clusters in the same cluster.
Proposal 2: The generation of absolute delay of the cluster can wait for the conclusion of the discussion on agenda item 9.8.1.
Proposal 3: For the  clusters, the distance  is equal to speed of light times the absolute delay of the cluster.
Proposal 4:  For non-direct paths, it needs to define a constraint that the sum of  and  should be less than or equal to the absolute distance corresponding to the absolute delay of cluster.
Proposal 5: For near-field channel, the following formula is adopted to model the ray-wise angular domain parameters of non-direct path between TRP and UE as antenna element-wise channel parameter:




where  refers to the vector pointing from transmit antenna element s to the first-bounce scatterer for ray m of cluster n, and .
where  refers to the vector pointing from receive antenna element u to the last-bounce scatterer for ray m of cluster n, and .
Proposal 6: For step 1 in the modelling of spatial non-stationarity by stochastic-based approach at BS side
If the randomly generated parameter  (following the uniform distribution as  is smaller than , the cluster is impacted by the SNS, where the  is calculated by a truncated normal distribution within [0,1]
     where 
Proposal 7: For step 2 in the modelling of spatial non-stationarity by stochastic-based approach at BS side with the antenna dimension of 
The visible antenna elements are determined by a random rectangular VR based on visibility probability , where  is calculated by

One side of rectangular VR, , is generated as

The other side of rectangular VR, , is determined as

The four corner of the antenna array is randomly selected as the reference corner of the rectangular VR.
Proposal 8: For the step 3 in the modelling of spatial non-stationarity by stochastic-based approach at BS side
If the cluster without SNS, the power attenuation factor   is set to 1.
If the cluster with SNS, the power attenuation factor  is generated by

Proposal 9: For the modelling of spatial non-stationarity at UE side, three steps are used
Step 1: Determinate the grip case according to the typical proportion.
Step 2: Obtain the fixed linear per antenna element attenuation values  based on the grip case.
Step 3: Apply the fixed linear attenuation value  for the per element pair channel.
Proposal 10: The power attenuation value  and  are introduced into the channel impulse response to reflect the impact of spatial non-stationary at BS side and UE side.

3GPP TSG RAN WG1 #120bis	R1-2501898
Wuhan, China, April 7th – 11th, 2025

Source:	 ZTE Corporation, Sanechips
Title:	 Discussion on channel model adaptation and extension
Agenda Item:	 9.8.2
Document for:  Discussion

Conclusion
In this contribution, we provide our analysis and proposals for the channel model adaptation/extension of TR38.901 at least for 7-24GHz.
For near-field channel model:
Observation 1: Considering the 3D locations of both TRP and UE will not be given in the link level evaluation, the antenna element-wise channel parameters of direct path for the near-field channel cannot be directly calculated following the same way of system level evaluation.
Proposal 1:  For near-field channel, for number of   non-direct path, the scaling factor  for UMi scenarios are confirmed, i.e.,   generated by a Beta distribution with .
Proposal 2: For near-field channel, for number of   non-direct path, the scaling factor .
Proposal 3: For near-field channel, for number of   non-direct path, the scaling factor 
Proposal 4: For near-field channel, the value of    for InH is 2
Proposal 5: The antenna element-wise channel parameters for the link level evaluation can be obtained following same procedure of non-direct path between TRP and UEs. And the detailed procedure to realize the near-field channel modeling in the link level evaluation is given as following.

Proposal 6: For the calibration parameters for the near-field channel, the singular values are the eigenvalues of the mean covariance matrix of either 1 PRB or 4 consecutive and non-overlapping PRBs.
For spatial non-stationarity:

Proposal 7:For the modelling of the attenuation for each antenna, introduce multiple sets of fixed values per case, each set corresponding to a range of frequency, and the following table can be considered..

Proposal 8: For the clusters of a UE, the ratio of clusters with SNS feature follows the normal distribution, e.g., ;

Proposal 9: The ratio of visible elements to the total number of antenna elements can be used as a parameter to measure the size of the visible region. For the ratio generation process, the following aspects should be considered:
For LOS rays, generate the ratio of visible elements based on the probability function:;
For NLOS clusters, generate the ratio of visible elements based on uniform distribution;

Proposal 10: The Detailed procedure to model the spatial non-stationarity in the link level simulations is given as following:

Proposal 11: For the calibration of spatial non-stationarity on the UT side, the BS antenna configuration with limited number of antenna ports/elements can be considered.

3GPP TSG-RAN WG1 Meeting #120-bis	R1- 2501936
Wuhan, China, April 7th – April 11th, 2025

Agenda Item:	9.8.2
Source:	NVIDIA
Title:	Channel model adaptation of TR 38.901 for 7-24 GHz
Document for:	Discussion
1	

Conclusion
For near-field channel, no changes are expected on following parameter for both direct and non-direct path between TRP and UE:
Delay
Agreement
For near-field channel, for the non-direct paths, the distance  between the TRP and the 1st point associated with cluster is generated by Option-2:
The distance  is generated by a scaling factor () multiplied by the absolute distance corresponding to the absolute delay of cluster, where the scaling factor () is generated by:
For UMa scenarios, a Beta distribution with []:
For UMi scenarios, a Beta distribution with []:
For Indoor Office scenarios, a Beta distribution with []
For the value of  and 
For UMa, the value of  is 2.
For UMi, the value of  is 2.
For InH, the value  is [75%, 10.9%] of the total number of clusters for InH
Note: the  clusters are the strongest of the total number of clusters in the channel
Note: the value of  is equal to total number of clusters minus.

Agreement
Confirm the following working assumption.
Working Assumption (made in RAN1#119)
For near-field channel, for the non-direct paths, the impact of spherical wavefront is optionally considered at UE side.
Agreement
For near-field channel, for number of   non-direct paths, the distance  between the UE and the 2nd point associated with cluster is generated by:
a scaling factor () multiplied by the absolute distance corresponding to the absolute delay of cluster.
Agreement
For the modelling of spatial non-stationarity at BS side, if physical blocker-based approach is adopted, the nearest K blockers from BS are selected.
For the modelling of spatial non-stationarity at BS side, if physical blocker-based approach is adopted, at least for blockage model B, the following parameters to define the building edge are introduced in the Table 7.6.4.2-5 in TR 38.901:
In the blockage model B, the following equation is used to calculate the attenuation caused by the building edge:
, where  represents one of , ,  and  defined in equation 7.6-29 in section 7.6.4.2 of TR 38.901.
Note: Corresponding Working Assumption made in RAN1#119 does not need to be confirmed.
Agreement
For near-field channel, the following formula is adopted to model the phase parameters of non-direct path between TRP and UE as antenna element-wise channel parameter:
For the TRP side, the element-wise phase is calculated as:

where  is the generated distance of the cluster n.  is the spherical unit vector with azimuth departure angle and elevation departure angle for ray m of cluster n.  is the vector pointing from reference point to transmit antenna element s, wherein the reference point is the physical center of the antenna array/center at Tx side.
For the UE side, the element-wise phase is calculated as:


where  is the generated distance of the cluster n.  is the spherical unit vector with azimuth arrival angle and elevation arrival angle for ray m of cluster n.  is the vector pointing from reference point to receive antenna element u, wherein the reference point is the physical center of the antenna array/center at Rx side.

3GPP TSG- RAN WG1 #120-bis Meeting 	                                                                      R1- 2501955
Wuhan, CN, April 07th – 11st, 2025

Agenda item:	9.8.2
Title:	Discussions on FR3 Channel Modelling
Source:	Lekha Wireless Solutions
Document for:	Discussion/ Decision


 

Conclusion
In this proposal, we discussed and provide our views on delays for direct and non-direct paths for FR3 near-field channels with following proposals:

Proposal 1: From the options shortlisted for delay of direct path, it is more credible to go with Option 2, as the delays must be relative to the distance of direct path between antenna elements at TRP and UE side.
Proposal 2: Considering the points mentioned above, we believe that the current presumptions available in TR 38.901 for non-direct paths are not sufficient for the delay modelling in near-field setup, hence, it is more suitable to go with Option 1 mentioned in observation 1.
Proposal 3: The entire channel realization process to capture the near-field propagation is shown as following,

4. 

3GPP TSG RAN WG1 #120bis		                               R1-2502005
Wuhan, China, April 7th – 11th, 2025

Source:	CATT
Title:	On channel model adaptation/extension for 7-24GHz
Agenda Item:	9.8.2
Document for:	Discussion and Decision

Conclusions
In this contribution, we provided our views on scenarios and modelling methodology for 7-24GHz. We also provided potential adaptation and extension of the channel model in TR 38.901 and the discussion for calibration assumptions.
Observation 1: Rayleigh distance, which is defined by evaluating the phase error between planar wave and spherical wave, may not be applicable for large scale MIMO systems using beamforming techniques.
Observation 2: To be consistent with assumptions for calibration, the CDL models could be selected from CDL-D and CDL-E.
Observation 3: CDL-D model is more suitable for urban macro cell scenarios and CDL-E model is more suitable for indoor scenarios.
Proposal 1:  For the modelling of spatial non-stationarity at BS side, unified VP and VR method should be adopted as the modelling methodology.
Proposal 2: For the modelling of spatial non-stationarity, if unified VP and VR based approach is adopted, Step 1 and Step 2 can be merged: Whether a cluster is impacted by SNS can be decided when generating the visibility region.
Proposal 3: For unified VR and VP based approach, the region of visible antennas in the antenna array can be generated from the intersection portion of a cluster specific 3-D region and the antenna array plane.
Proposal 4: For unified VR and VP based approach, a cluster is not impacted by SNS when all antennas are visible.
Proposal 5: For unified VR and VP based approach, when the region of visible antennas is the whole antenna array, the cluster specific power attenuation factor of an antenna element in the region is 1.
Proposal 6: For unified VR and VP based approach, to ensure consistency, when the region of visible antennas is part of the antenna array, the cluster specific power attenuation factor of an antenna element in the region is determined by the distance between the antenna element and the cluster and the ratio of the number of visible elements within the cluster to the total number of elements. The cluster specific power attenuation factor of an antenna element out the region is 0.
Proposal 7: Different implementations, e.g., procedures/equations, are used for near-field and far-field channel realization.
Proposal 8: Based on phase variations criteria and beamforming assumption, the effective Rayleigh distance, a scaled Rayleigh distance with a scaling factor can be used to define the near-field region. The proposed scaling factor is 0.4.
Proposal 9: The near-field or far-field condition for the non-direct paths between BS and UE follows the near-field or far-field condition for the direct path.
Proposal 10: To model near field based on the structure of existing stochastic model in TR 38.901, the followings are adopted:
If the boundary of near field and far field under LOS/NLOS is established, the boundary of near field and far field under LOS/NLOS is generated in the step of “Assign propagation condition”. 
The antenna element-wise phase can be reflected in the step of “Generate channel coefficient”
Proposal 11: To model of spatial non-stationarity based on the structure of existing stochastic model in TR 38.901:
The visibility region generation of the cluster, the determination of whether a cluster is impacted by SNS and the power attenuation factor generation can be applied in the step of “Generate cluster powers”.
The effect of power attenuation factor can be reflected in the step of “Generate channel coefficient”.
Proposal 12: CDL models are more suitable for link level simulation.
Proposal 13: In link level simulation, the CDL-D and CDL-E model can be selected based on specific scenarios.
Proposal 14: To model near field in link level simulation, the followings are adopted for link level simulation:
The antenna element-wise phase can be reflected in the step of “Generate channel coefficient”. 
Proposal 15: To model spatial non-stationarity in link level simulation, the followings are adopted for link level simulation:
The effect of power attenuation factor can be reflected in the step of “Generate channel coefficient”. 
Proposal 16: For calibration assumption for near-field channel modelling, adopt the followings:
For UT Antenna Configuration, it is suggested to remove Config2. 
For UT antenna pattern, it is suggested to use isotropic antenna pattern for Config 1. 
For UT Polarized antenna modelling, Model-2 in Clause 7.3.2 of TR38.901 can be assumed for Config 1.
The corresponding change of parameters for the calibration assumption table is as follows:
Proposal 17: For calibration assumption for BS side spatial non-stationarity, adopt the followings:
For UT Antenna Configuration, it is suggested to remove Config2. 
For UT antenna pattern, it is suggested to use isotropic antenna pattern for Config 1. 
For UT Polarized antenna modelling, Model-2 in Clause 7.3.2 of TR38.901 can be assumed for Config 1.
The corresponding change of parameters for the calibration assumption table is as follows:
Proposal 18: For calibration assumption for UT side spatial non-stationarity, adopt the followings:
For BS Antenna Configuration and port mapping, it is suggested to keep Option 4 only. 
For UT Antenna Configuration, it is suggested to remove Option 3. 
For UT antenna pattern, it is suggested to use directional antenna pattern. 
For UT Polarized antenna modelling, Model-2 in Clause 7.3.2 of TR38.901 can be assumed.
The corresponding change of parameters for the calibration assumption table is as follows:

3GPP TSG RAN WG1 #120-bis                               R1-2502187
Wuhan, CN, April 7th – 11th, 2025

Source:	BUPT, CMCC, X-NET
Title:	Discussion on near-field propagation and spatial non-stationary
Agenda Item:	9.8.2
Document for:	Discussion and Decision

Conclusions
Observation 1: Approximately 40% of the clusters are spatially stationary.
Proposal 1: The sum of the scaling factors generated for d1 and d2 for a given cluster should be less than or equal to 1 for k2 non-direct path.
Proposal 2: The determination of spatial non-stationarity in clusters (Step 1) is essential.
Proposal 3: For the modelling of spatial non-stationarity at BS side, if stochastic-based approach is adopted, for step 1,
If the randomly generated parameter  (following the uniform distribution as  is smaller than , the cluster is impacted by the SNS, where the  is calculated by a truncated normal distribution within [0,1]
     where 
Proposal 4: For the modelling of spatial non-stationarity at BS side with the antenna dimension of , if stochastic-based approach is adopted, for step 2,
The visible antenna elements are determined by a random rectangular VR based on visibility probability , where  is calculated by

The length of rectangular VR, , is uniformly generated from  with minimum value of 1,
The corresponding width, , is determined as  with minimum value of 1.
Proposal 5: For the modelling of spatial non-stationarity at BS side, if stochastic-based approach is adopted, for step 3,
If the cluster without SNS, the power attenuation factor   is set to 1.
If the cluster with SNS, the power attenuation factor  is generated by

3GPP TSG-RAN WG1 Meeting #120-bis                                                                                 	    R1-2502218
Wuhan, China, April 7-11, 2025

Agenda Item:	9.8.2
Source:	Huawei, HiSilicon 
Title:	Considerations on the 7-24GHz channel model extension 
Document for:	Discussion and Decision

Conclusions
In this contribution, we provide our views on the near-field and non-stationary channel modelling. Following observations and proposals are given:
Observation 1: Incomplete reflection, diffraction and scattering can result in spatial non-stationary effect. 
Proposal 1: For the near-field propagation model of non-direct path, support to generate the scaling factors  and  by the Beta distributions given in Table 1. 
Proposal 2: For the near-field propagation model of non-direct path, support to generate  and  according to formula (1) and (2), respectively.
Proposal 3: Support to set the ratio of non-direct path with spherical-wavefront phenomenon to one.
Proposal 4: Support to generate the SNS probability  by a truncated Gaussian distribution. 
Proposal 5: For the modelling of spatial non-stationarity, support to generate the visible probability by a function of cluster power, e.g., formula (4). 
Proposal 6: For the modelling of spatial non-stationarity, support to generate the visibility region as a rectangle sharing a corner with the antenna array with dimension, where, ,  is the antenna array height in vertical dimension and  is the antenna array width in horizontal dimension.
Proposal 7: For the modelling of spatial non-stationarity, support to generate the power attenuation factor  according to formula (5). 

3GPP TSG RAN WG1 #120bis		R1-2502287
Wuhan, China, April 7th – 11th, 2025	

Source:	OPPO
Title:	Channel model adaptation and extension for 7-24GHz
Agenda Item:	9.8.2
Document for:	Discussion and Decision

Conclusion
In this contribution, we discuss the channel model adaptation and extension to TR38.901 for near field (7-24GHz) with the following proposals:
Proposal 1: The near-field region can be defined based on Rayleigh distance and FFS on scaling factor, which only serves as a reference criterion for considering spherical wavefront or not.
Proposal 2: To align with the practical deployment, the distance between the TRP and the 1st point associated with the nth cluster is truncated with the lower bound  as

where  is the minimum distance between TRP and the 1st point associated with cluster in Table 4. 
Table 4 Minimum distance between TRP and the 1st point associated with cluster
Proposal 3: If the impact of spherical wavefront is considered at UE side, the following procedure can be used for generating  for the  clusters:
Step1: The distance  between TRP and 1st point associated with cluster n is generated.
Step2: If the impact of spherical wavefront is considered at UE side, the distance  between UE and the 2nd point associated with cluster n is generated independently of . 
Step3: If the impact of spherical wavefront is considered at UE side and the propagation delay on the two stages has exceeded the absolute delay defined in Section 7.6.9 in TR 38.901 multiplied by the speed of light, i.e. , the distances  and  are normalized into


where  and  are the normalized distances associated with the same cluster. Otherwise, equations ,  hold.
Step4: If the impact of spherical wavefront is NOT considered at UE side, the element-wise phase parameters of non-direct path at the TRP side is modeled based on . Otherwise, the element-wise phase parameters of non-direct path at the TRP side and UE side are modeled based on  and .
Proposal 4: For near-field channel, for the non-direct paths, the channel modelling of  clusters follow the same modelling method of direct path.
Proposal 5: For near-field channel, for the  non-direct paths, the locations of the mirror image of UE elements can be determined based on , , , , and distance . The modelling method of direct path can be applied with the locations of TRP elements and UE elements.
Proposal 6: For near-field channel, for link-level simulation, the absolute delay of the first cluster/LOS path is additionally modelled. The absolute delay for each cluster is obtained by adding the absolute delay of the first 1st cluster/LOS path and the normalized delay for each cluster.
Proposal 7: For near-field channel, for link-level simulation, the 3D distance between TRP and UE of direct path is generated with the absolute delay multiplied by the speed of light. The distance between TRP and 1st point associated with cluster and between UE and 2nd point associated with cluster are generated with assigned scaling factors following the same method as in the system-level simulation.
Proposal 8: For near-field channel, the following element-wise channel parameters of the link-level simulation for both direct and non-direct paths are modelled following the same method as in system-level simulation:
Phase
Angular domain parameters
Proposal 9: For near-field channel, the element-wise channel parameters of the specular path of the link-level simulation are modelled following the same method as the direct path in system-level simulation.
Proposal 10: For the modelling of spatial non-stationarity, adopt the existing knife edge attenuation model in TR 38.901 in blockage Model-B to model the power attenuation per element (i.e. physical blocker-based approach).
Proposal 11: For the modelling of spatial non-stationarity, if stochastic based approach is adopted, VR size is derived based on distance between antenna array of BS and cluster/UE.

3GPP TSG RAN WG1 #120bis	        	R1-2502328
Wuhan, China, April 7th – 11th, 2025

Agenda Item:					    9.8.2
Source:	Sony
Title:	Further discussion of channel model adaptation/extension of TR38.901 for 7–24GHz 
Document for:	Discussion and decision

Conclusions
We made the following observations and proposals:
 Harmonizing the attenuation values in the agreement table might be difficult. Therefore, if further convergence is pursued, aligning the evaluation condition shall be guaranteed first.
 For the modeling of spatial non-stationarities at the UE side, better accuracy may be obtained if different sets of attenuation values are provided for different frequency ranges. Examples of subranges can be: i) < 1 GHz, ii) 2–6 GHz, iii) 6–10 GHz, and iv) 10–15 GHz. 
 For the modeling of spatial non-stationarities at the UE side, provide different sets of attenuation values where each set has a specific frequency range of applicability. For example, one set can be given for simulations up to 10 GHz and another for simulations in the 10–15 GHz frequency range. 
 For the modeling of spatial non-stationarities at the UE side, when measuring or simulating attenuation values due to self-blockage by the user’s hand and/or head, the choice of smartphone materials and the placement of the antennas have a strong influence on the observed attenuation values. In particular, the larger the gap s between the device rim and the antennas, the smaller the experienced attenuation due to self-blockage. Moreover, electrically insulating materials, such as plastic, placed between the device rim and the antennas tend to reduce the experienced attenuation values. Conversely, smartphone designs endowed with a metal rim in contact with the antennas tend to suffer from larger attenuation losses, as the antennas are likely to have direct contact with the user’s fingers.
 The experimental method used impacts the accuracy of the obtained attenuation data. Some examples of methods, sorted from most to least accurate, are measurements with a commercial device, measurements with a prototype, full-wave electromagnetic simulations, and raytracing simulations. Raytracing does not consider effects due to the gap between the antennas and the device rim, or whether the antennas are integrated with the device rim. 
 To compare attenuation values provided by different companies in a meaningful way, those attenuation values need to be computed in the same way. For example, when extracting attenuation values from 3D antenna radiation patterns, the extraction process can be based on the total antenna efficiency obtained from the measured radiation pattern.
 For the modeling of spatial non-stationarities at the UE side, to meaningfully compare different sets of attenuation values, companies should ensure that i) the sets of attenuation values stem from similar settings, including the antenna placements, user grip, the gap between the antennas and the smartphone rim, and the smartphone case materials, ii) the experimental methods are adequate, i.e., measurements are performed with sufficient accuracy to accept comparison, and iii) the magnitudes being compared have been obtained from the measurement data in similar ways, e.g., total antenna efficiencies obtained by integrating the measured antenna radiation pattern over similar solid angles.
 Attenuation value sets for the same frequency range can be merged if they are comparable according to the principles sketched in the proposal above, i.e., if they stem from similar settings, accurate experimental methods, and like computations. If these conditions are not fulfilled, multiple data sets can be provided for the same frequency range. During the working item phase, companies can then select the most suitable set of attenuation values for simulation.
 For the modeling of spatial non-stationarities at the UE side, companies are strongly encouraged to provide further attenuation value sets for accurate harmonization.
 For a fixed gap  between the smartphone rim and the antennas, the experienced attenuation losses due to self-blockage decrease with frequency. This somewhat counterintuitive fact follows from the fact that the electrical length of the gap, , increases with the frequency.
 Despite having been obtained for different frequency ranges, the attenuation values of set 2 and 3, alt-1 come rather close, except for antenna index 5 and the dual-hand grip case.
 For the modeling of spatial non-stationarities at the UE side, FFS values in the table are not used for simulations of up to 4 antenna ports, as they correspond to antenna placements that are not part of the 4-element configuration agreed for calibration.
 For the modeling of spatial non-stationarities at the UE side, if simulations with more than 4 antenna ports are desired, FFS values in the table can be replaced by numerical values according to one of the following methods: i) Interpolation of the attenuation values of the two nearest antenna placements, and ii) setting the value to zero. Modified tables corresponding to each method are provided for convenience in our contribution.

3GPP TSG RAN WG1 #120bis	R1-2502340
Wuhan, China, Apr. 7th – 11th, 2025

Source:	Lenovo
Title:	Channel model extension for 7-24 GHz
Agenda Item:	9.8.2
Document for:	Discussion and Decision

Conclusion
This contribution studied aspects on channel model extension for 7-24 GHz. We have the following proposals:
For antenna element-wise modeling of delay for the near-field channel for the direct and non-direct path(s), further clarification is needed on the working assumption regarding the applicability of the additional modeling component in Clause 7.6.2 of TR 38.901 to the near-field channel with antenna element-wise delay modeling.
For non-angular parameters in near-field channels, the parameters are modeled to have antenna group-wise parameters, i.e., antenna elements of a sub-array of the BS antenna array comprising multiple antenna elements are modeled to have the same parameter values.

3GPP TSG RAN WG1 Meeting #120bis	R1-2502342
Wuhan, China, April 7th – 11st, 2025

Source:	Intel Corporation
Title:	Discussion on channel model adaptation/extension
Agenda item:	9.8.2
Document for:	Discussion

Conclusions
In this contribution, we discussed near-field communications and other channel model adaptations/extensions. The following is a summary of proposals and observations made.
Proposal 1: RAN1 to consider defining soft criteria for leveraging near-field models and state that near-field modelling is applicable for evaluation scenarios that contain links that meet such criteria.
FFS: criteria for near-field model consideration.
Proposal 2: For study of near-/far-field conditions for small-scale channel parameters, RAN1 considers different near-/far-field conditions for different parameters of the channel.
Proposal 3: Define the near-/far-field condition for direct path channel phase as follows: The distance  between the origin of the transmit-side GCS and the origin of the receive-side GCS is smaller/larger than the far-field distance , where the far-field distance  is defined as the smallest distance such that for any  and ,  for .
FFS: the value of .
Proposal 4: Define the near-/far-field condition for non-direct path channel phase based on same principle for direct path based on distance  and 
The distance  and  between the origin of the transmit/receive-side GCS and the k-th cluster is smaller/larger than the far-field distance , where the far-field distance  is defined as the smallest distance such that for any  and ,  for .
FFS: the value of .

3GPP TSG-RAN WG1 Meeting #120bis		R1-2502381
Wuhan, China, April 7th – 11th, 2025

Agenda item:	9.8.2 
Title:	Discussion on channel model adaptation/extension of TR38.901 for 7-24GHz
Source:	Samsung 
Document for:	Discussion and Decision 

Conclusion
The impact between BS and clusters may differ depending on whether the UE’s location is in the near-field or far-field region
RAN1 consider the impact not only between BS and UE but also between BS and cluster taking into account UE’ location
For distance calculation between BS and cluster and between cluster and UE in non-direct path modelling, since it is next step after cluster generation, LOS angle is calculated based on reference point 
For angular domain parameters for direct paths, the LOS angles between BS and UE are calculated based on antenna element-wise method
RAN1 clarify whether LOS angle applies differently for each parameter generation
For same implementation method, the consistency makes that the model can be universally applied without the need for transition points. However, this comes at the complexity for computation. Non-planar wavefronts are significant in near-field conditions, while far-field condition is adequately modelled using plane wavefront. So, by using a more complex equation across all field condition, the model may introduce unnecessary computational overhead in far-field conditions
For different implementation method, it enhances computational efficiency by applying the appropriate model to each region. However, it introduces challenges in managing the transition between near- and far-field regions
RAN1 discuss same/different implementation method taking into account channel consistency and computational complexity
RAN1 consider the updates of channel parameter with weighting vector for antenna array  
RAN1 defined a sufficient number of antenna elements and corresponding antenna port numbers as basic assumptions. Also, UE distribution includes distances ranging from very close to far away. Even within these assumptions, differences in metrics (e.g., singular values) due to phase difference will be sufficiently shown
RAN1 consider the consolidated frequency band and BS antenna configuration: 7 GHz for frequency band and (M, N, P, Mg, Ng; MP, NP) = (24, 32, 2, 1, 1; 8, 32), dH = 0.5λ, dV = 0.7λ (UMi) for BS antenna configuration
RAN1 consider the Config 1 for UE antenna configuration as a basic assumption
RAN1 agreed to update the angular domain parameter in the previous meetings. If only phase term is considered in calibration, there may be ambiguity in implementing the channel model
Angular dependency cannot be neglected, as different elements within the array will experience varying angular relationships with the UE. It means that the radiation power contribution of each element to the total received signal will differ, depending on its position and angle with the respect to the UE
RAN1 consider both phase term and angular domain parameter for calibration
Spatial non-stationarity occurs when a large-scale antenna array in BS is introduced or the UE is very close to the BS. 
RAN1 discuss the condition of occurrence for spatial non-stationarity
Considering the presence of geographical blockers between the BS and UE, it is crucial to examine power attenuation per antenna element within the BS
RAN1 consider the blockage model B of blocker-based approach for the modelling of spatial non-stationarity as starting point
The characteristics of spatial non-stationarity at least includes the imbalance of power due to shadow effects per element within large antenna array at the BS
In the existing blockage mode B, the projected distance ( and ) between Tx/Rx and the edge of blocker and distance (r) between Tx and Rx is calculated based on reference point of Tx and Rx
RAN1 consider the location of each element of antenna array in the BS when the projected distance and reference distance is calculated
If the need for Model A arises, it would be necessary to first discuss how the blocking areas for both the building edges and individual antenna elements will be established under non-self blocking, similar to what was mentioned during the discussions about Model B

3GPP TSG RAN WG1 #120bis			R1-2502626
Wuhan, China, April 7th – 11st, 2025
 
Agenda Item:	9.8.2
Source:	Apple
Title:	Channel Model Adaptation and Extension of TR 38.901 for 7-24 GHz
Document for:	Discussion/Decision

Conclusion
In this contribution, we provided our views on channel model adaptation/extension of TR 38.901 for 7-24 GHz. Our proposals are as follows:

Proposal 1: For near-field channel model study, for  clusters in non-direct paths between TRP and UE, the distance  is generated by a scaling factor () multiplied by the absolute distance corresponding to the absolute delay of the cluster, where the scaling factor is generated from Beta distribution: 
For UMa scenarios, a Beta distribution with 
For UMi scenarios, a Beta distribution with 
For indoor office scenarios, a Beta distribution with . 

Proposal 2: For near-field channel model study, for non-direct paths between TRP and UE, is 10.9% of the total number of clusters for InH. 

Proposal 3: For near-field channel model study, for non-direct paths between TRP and UE, the distance is generated by a scaling factor () multiplied by the absolute distance corresponding to the absolute delay of the cluster, where the scaling factor is generated by 
For  strongest clusters,  (i.e., there is no second point associated with the cluster)
For the remaining  clusters, , where A is a Beta distribution.  

Proposal 4: For near-field channel model study, if the generated  for a non-direct path is within the Rayleigh distance, then this non-direct path has spherical wavefront. Otherwise, this non-direct path has planar wavefront.

Proposal 5: For the modelling of spatial non-stationarity with unified VP and VR, in Step 1, compare a uniformly generated parameter between 0 and 1 with a probability value. If the generated parameter is smaller than the probability value, then the cluster is impacted by spatial non-stationarity. Otherwise, it is not impacted by spatial non-stationarity. 
The probability value is generated by a truncated normal distribution within [0,1]. 

Proposal 6: For the modelling of spatial non-stationarity with unified VP and VR, in Step 2, the VR is determined based on the distance  between TRP and the first bounce cluster. 
The larger the distance, the larger the visible region.

Proposal 7: For the modelling of spatial non-stationarity with unified VP and VR, in Step 3, the power attenuation factor is equal to VR over the total number of TRP antennas. 

Proposal 8: For the modelling of spatial non-stationarity at UE side, support the fixed value of attenuation for handheld device is given by set-1 @ 6GHz, i.e.,
one hand grip: [0, 0, 0, 6.15, 7.31, 0, 0, 0] dB for candidate antenna locations 1-8, respectively
dual hand grip: [7.31, 0, 7.31, 6.15, 0, 0, 0, 6.15] dB for candidate antenna locations 1-8, respectively
head and one hand grip: [10.57, 7.08, 0, 0, 6.55, 10.12, 10.57, 10.67] dB for candidate antenna locations 1-8, respectively. 

Agenda item:	9.8.2
Source: 	Qualcomm Incorporated
Title: 	Channel model adaptation or extension of TR38.901 for 7-24GHz
Document for:	Discussion/Decision

Conclusion 
We make the following observations and proposals in this document:
On spherical wavefront modeling:
Observation 1: The stochastic generation of points associated with a cluster can lead to scenarios where high-power clusters are positioned close to the gNB. This proximity amplifies the effects of the spherical wavefront, resulting in a significant divergence between SWM and PWM channels, even when UEs are far from the gNB.
Proposal 1: Additional mitigation measures may be necessary to ensure channels associated with UEs far from the gNB do not exhibit strong spherical wavefront properties. Suggest establishing a minimum distance threshold for  for UEs beyond a certain distance from the gNB.
On spatial non-stationarity
Proposal 2: Spatial non-stationarity may arise due to incomplete scattering (depending on the dimensions of a reflectors/scatterer) or due to blockage (a physical entity obstructing the view of a portion of the array). To model spatial non-stationarity at the gNB side support a physical blockage approach and a stochastic approach. 
If spatial non-stationarity arises due to blockage, then reuse the existing blockage framework in 38.901 (with potential modifications) to model SnS. 
If spatial non-stationarity arises due to incomplete scattering and/or blockage, a stochastic approach based on visibility regions can be used to model SnS.
Proposal 3: If spatial non-stationarity is to be modeled, the concept of visibility region or blockage region is applied on a per cluster basis. 
Proposal 4: If spatial non-stationarity is to be modeled using visibility regions at the gNB side, consider the probability of spatial non-stationarity impacting a UE at any given location.
Proposal 5: If spatial non-stationarity is to be modeled using a stochastic approach, partial blockage of the direct path between a UE and gNB is also modelled.  
Proposal 6: For the modelling of spatial non-stationarity, if the unified visible probability and visibility region is adopted, for Step 1 (whether a cluster is impacted by SnS) consider the following methodology:
For each user, generate a probability  reflecting the number of clusters that are expected to be impacted by SnS according to a certain distribution. For each cluster of this user, generate a random parameter  drawn from a uniform distribution over the unit interval. If  is smaller than  then this cluster is assumed to be impacted by SnS.
Proposal 7: For the modelling of spatial non-stationarity, if the unified visible probability and visibility region, is adopted, for the Step 2 (generate the visibility region) consider the following guidelines:
Visibility regions are assumed to be rectangular in shape with same orientation as the antenna array. The size of visibility regions is not dependent on exact cluster location but can be dependent on cluster power. The placement/location of the visibility region on the panel is determined in a stochastic manner.
Proposal 8: For the modelling of spatial non-stationarity, if the unified visible probability and visibility region is adopted, for Step 3 (calculate the power attenuation factor) consider the following methodology:
For portions of the array that fall outside the visibility region of a cluster, the cluster power decreases linearly in dB domain from the edge of the visibility region with a certain roll off factor.
Proposal 9: For the modelling of spatial non-stationarity at UE side, consider the following: 
Introduce frequency-range specific losses for the impacted antennas
Blockage loss associated with the impacted antennas for each case are between 3-5 dB for 6-15 GHz range. 
Blockage loss associated with the impacted antennas for each case are between 3-10 dB for sub-6 GHz range.

3GPP TSG-RAN WG1 Meeting #120bis	Tdoc R1-2502884 
Wuhan, China, April 7th – 11th, 2025

Agenda Item:	9.8.2
Source:	Ericsson
Title:	Discussion on adaptation and extension of channel model
Document for:	Discussion

Conclusion
In the previous sections we made the following observations: 
Observation 1	A wavefront is characterized by two radii of curvature in the two principal planes, and can be planar, spherical, cylindrical or other depending on the values of these two radii.
Observation 2	Most wavefronts in the UMa scenario are approximately either planar or spherical.
Observation 3	Companies have different preferences on how “a point associated with the cluster” should be interpreted.
Observation 4	The interpretation that this point coincides in direction and distance with the first (or last) interaction point in the propagation path is NOT consistent with the occurrence of common interactions such as specular reflections or diffractions.
Observation 5	There is a need to characterize the wavefront curvature in typical environments.
Observation 6	The radius of curvature is to a first approximation independent of the frequency in the UMa scenario, while it has a weak frequency dependence in the indoor-office scenario.
Observation 7	Some 10% of the rays in the UMa scenario have a radius of curvature that is equal to the path length, i.e. a relative radius of curvature that is 1, while the rest of the paths have a radius of curvature that is <1.
Observation 8	Some 75% of the in the indoor-office scenario have a radius of curvature that is equal to the path length, i.e. a relative radius of curvature that is 1, while the rest of the paths have a radius of curvature that is <1.
Observation 9	Some 31% of the in the indoor-factory scenario have a radius of curvature that is equal to the path length, i.e. a relative radius of curvature that is 1, while the rest of the paths have a radius of curvature that is <1.
Observation 10	The relative radius of curvature at the UE side  can be directly determined from the relative radius of curvature at the gNB side  as per the above equation.
Observation 11	For rays with a relative radius of curvature , a Beta distribution with parameters  = 3.81,  = 1.62 provides a good fit to the stochastic distribution of relative radius of curvature in the UMa scenario.
Observation 12	The shape of the distribution of the relative radius of curvature is consistent with the expectation that more scattering occurs near the UE than the gNB in an UMa scenario.
Observation 13	Companies’ different interpretations of “the distance  between the TRP and the 1st point associated with cluster” from the RAN1#120 agreement can explain diverging parameter value proposals for the beta distribution.
Observation 14	For rays with a relative radius of curvature , a Beta distribution with parameters = 1.42,  = 1.39 provides a good fit to the stochastic distribution of relative radius of curvature in the indoor-office scenario.
Observation 15	The shape of the distribution of the relative radius of curvature is consistent with the expectation that scattering occurs with equal probability both near the UE and the gNB in an indoor-office scenario.
Observation 16	For rays with a relative radius of curvature , a beta distribution with parameters = 1.38,  = 1.26 provides a good fit to the stochastic distribution of relative radius of curvature in the indoor-factory scenario.
Observation 17	The shape of the distribution of the relative radius of curvature is consistent with the expectation that scattering occurs with equal probability both near the UE and the gNB in an indoor-factory scenario.
Observation 18	Stronger paths are more likely to have a relative radius of curvature .

Based on the discussion in the previous sections we propose the following:
Proposal 1	Confirm that “the distance  between the TRP and the 1st point associated with cluster” should be understood as the distance to a hypothetical source of a spherical wave, and NOT the distance to the first interaction point.
Proposal 2	Don’t merge proposed parameter values and distributions derived under differing interpretations.
Proposal 3	Model the two strongest clusters in the UMa scenario with a relative radius of curvature .
Proposal 4	Model the strongest clusters in the indoor-office scenario with a relative radius of curvature , where  in LOS and  in NLOS.
Proposal 5	Model the strongest clusters in the indoor-factory scenario with a relative radius of curvature .
Proposal 6	Model wavefront curvature in the UMa, InH, and InF scenarios using steps R1—R6. Note: the curvature at the gNB side  is to be understood as the “distance between the BS/UE and a point associated with cluster”, i.e. . Similarly, the curvature at the UE side  is to be understood as .
Proposal 7	Adapt the existing blockage model B in TR 38.901 so that it can optionally determine the blockage per antenna element.
Proposal 8	Continue to study whether the blockage model in TR 38.901 can be enhanced to model also phase.
Proposal 9	Blockage by building edges should only be considered for links involving base stations that are in close vicinity to the building edge.
 

3GPP TSG RAN WG1 #120bis	R1-2502900
Wuhan, China, 7 – 11 April 2025

Agenda item:		9.8.2
Source:	Nokia
Title:	Discussion on Channel Model Adaptation/Extension of TR38.901 for 7-24GHz
WI code:	FS_NR_7_24GHz_CHmod
Release:	Rel-19
Document for:		Discussion and Decision

Conclusion
In this paper we elaborated on the open issue related to modelling of UE spatial non-stationarity modelling and shared our view on the reliability of near-file model parameters fitting.
The following observations and proposals were made:
Observation 1: Antenna operating at frequencies below 1 GHz will behave differently to user interaction than antennas operating at frequencies higher than 1 GHz. For the antennas operating at frequencies above 6GHz may have different electrical distance from the blockers (hands) also impacting the attenuation.
Proposal 1: Define a three frequency ranges for the values of attenuation for candidate antenna locations: below 1 GHz with two antenna locations (4 and 8), from 1 GHz to 6 GHz with four antenna locations (1, 3, 5, 7) and from 6 to 20 GHz all 8 antenna locations (1-8).

Table 6: The fixed values of proposed attenuation for candidate antenna locations for handheld device.
Proposal 2: RAN1 to define the fixed values of attenuation for candidate antenna locations depending on the frequency range as described in the Table 6 above.

Proposal 3: For calibration of SNS at UE side, RAN1 to use the same parameters and statistics as for full calibration, i.e.,
BS Antenna Configuration: Mg = Ng = 1, M = 8, N = 16, P = 2, dH = dV = 0.5λ (Option 4)
UE UT Antenna Configuration: 4 antenna port with single/linear polarization for calibration based on handheld device antenna model using candidate antenna locations (1,7,3,5) as described in Clause 7.3
UT antenna pattern: Based on directional antenna for UT described in Clause 7.3
UT Polarized antenna modelling: Vertically polarized reference antenna pattern, based on directional antenna for UT described in Clause 7.3
Metric: CDF of the ratio between the 2nd, 3rd ,4th, ..., xth (smallest) PRB singular value and the 1st PRB (largest) singular value (serving cell) at t=0 plotted in 10*log10 scale. Note: The PRB singular values of a PRB are the eigenvalues of the mean covariance matrix in the PRB.

Observation 2: The parameters of the models may change significantly when only a few different maps/configurations are utilized in different ray-tracing simulations. However, there is no measurements available that could justify the behavior of the spherical wavefront, especially after a single non-specula reflection.

Proposal 4: RAN1 to add a note in the TR 38.901, Table 7.6-13-1: Parameters for Uma, Umi and Indoor-Office, that the parameters were derived based on the limited number of ray-tracing simulations.

3GPP TSG RAN WG1 #120bis                                                     R1-2503024
Wuhan, China, April 7th – 11th, 2025

Title 	:  Summary#1 on the channel model adaptation and extension
Source 	:  Moderator (ZTE)
Agenda item	:  9.8.2
Document for:  Discussion and Decision

Conclusion

3GPP TSG RAN WG1 #120bis                                                     R1-2503025
Wuhan, China, April 7th – 11th, 2025

Title 	:  Summary#2 on the channel model adaptation and extension
Source 	:  Moderator (ZTE)
Agenda item	:  9.8.2
Document for:  Discussion and Decision

Conclusion

3GPP TSG RAN WG1 #120bis                                                     R1-2503026
Wuhan, China, April 7th – 11th, 2025

Title 	:  Summary#2 on the channel model adaptation and extension
Source 	:  Moderator (ZTE)
Agenda item	:  9.8.2
Document for:  Discussion and Decision

Conclusion

3GPP TSG RAN WG1 #120bis                                                     R1-2503027
Wuhan, China, April 7th – 11th, 2025

Title 	:  Summary#2 on the channel model adaptation and extension
Source 	:  Moderator (ZTE)
Agenda item	:  9.8.2
Document for:  Discussion and Decision

Conclusion