3GPP TSG-RAN WG1 Meeting #121	   R1-2503248
St Julian’s, Malta, 19-23 May, 2025

Agenda Item:	9.7.2
Source:	Huawei, HiSilicon
Title:	Channel modelling for ISAC 
Document for:	Discussion and Decision

Conclusions
In this contribution, we provided the mono-static RCS for AGV modelled with multiple scattering points and discussed remaining issues for pathloss, LOS probability for EO type-2 and the pathloss for the specular refection ray, spatial consistency and link level simulation. 
The observations and proposals are summarized as follows:
Observation 1:
The evaluation performance would be overestimated if it is LOS condition based on the probability equation while it should be NLOS condition instead because the building blocks the LOS ray between the TRP and ST/UT.
For ST/UE not affected by the EO type-2, the LOS probability model is the same and should be the same as the legacy even when the 3D EO type-2 is modelled, e.g., when located at the same street as TRP.

Observation 2:
The blockage model is an additional feature on top of the basic propagation path without changing the LOS/NLOS status.
The hard transition from LOS to NLOS or vice versa is the same for the case when the EO type-2 is not modelled and can also be avoided as the legacy procedure as well by modelling spatial consistency.

Proposal 1: AGV is also modelled with multiple scattering points, similar to vehicle modelled with five points.
The recommended five scattering points for AGV are located in front, left, back, right and roof.
The parameter values for the five scattering points for mono-static RCS are summarized in Table 1.


Proposal 2: When the path loss model of UMi, UMa and RMa scenario of TR38.901 is used for the target channel, one of the following options can be adopted: 
Option A：only the  is applied in the target channel irrespective of the breakpoint distance
Option B：use hUT 1.5 m for the breakpoint distance calculation.
Option C：set the effective environment height hE to 0.25m for the breakpoint distance calculation.

Proposal 3: When the EO type-2 is used to model the indirect paths of the channel, the LOS and NLOS condition is determined by the geometric location of the EO, Tx, Rx and ST.
If any of EO type-2 deployed is located on the line-of-sight of one link, the link is determined as NLOS condition, and otherwise the LOS/NLOS condition is determined based on the LOS probability equation.
Adopt the text proposal as follows accordingly in section 7.9.5.2:
-----------------------------------------------------Start of the text proposal ---------------------------------------------
[In Step 2 in Clause 7.9.4.1,]
The LOS/NLOS condition is determined by the geometry of the EO, Tx and Rx. 
If the EO blocks the direct path between the Tx and Rx, the link is determined as NLOS condition
Otherwise the LOS/NLOS condition is based on the LOS probability equation of the link. 
-----------------------------------------------------End of the text proposal --------------------------------------------- 
Proposal 4: 
The path loss calculation of the wall reflection is scaled by the LOS pathloss value without applying the stochastic K-factor, disregard whether the LOS ray exists or not, for both LOS/NLOS conditions.
.
Adopt the text proposal as follows accordingly in section 7.9.5.2:
-----------------------------------------------------Start of the text proposal ---------------------------------------------
In Step 10 in Clause 7.9.4.1, 
 for a NLOS ray specularly reflected by a type-2 EO, if present, in the SPST-SRX link and the STX-SPST link is determined as follows. 
-	If the STX-SPST link is in LOS condition, 
-	If the STX-SPST link is not in LOS condition, 
-	If the SPST-SRX link is in LOS condition, 
-	If the SPST-SRX link is not in LOS condition, 
----------------------------------------------- End of the text proposal -----------------------------------------------------

Proposal 5: 
When multiple scattering points of target is modelled, spatial consistency is always applied to the scattering points within a sensing target.
Adopt the text proposal in red as follows accordingly in section 7.9.4.1:
----------------------------------------------------- Start of Text Proposal--------------------------------------------------
Large scale parameters:
Step 2: Assign propagation condition (LOS/NLOS) for each pair of STX and SPST, and each pair of SPST and SRX according to Table 7.4.2-1 updated as necessary in Clause 7.9.3. LOS/NLOS condition of the SPs within the same ST are correlated.
Step 3: Calculate pathloss with formulas in Table 7.4.1-1 updated as necessary in Clause 7.9.3 for each STX-SPST link, and each SPST-SRX link.
Step 4: For each STX-SPST link and SPST-SRX link, generate large scale parameters, e.g. delay spread (DS), angular spreads (ASA, ASD, ZSA, ZSD), Ricean K factor (K) and shadow fading (SF) taking into account cross correlation according to Table 7.5-6 and using the procedure described in clause 3.3.1 of [14] with the square root matrix√(C_MxM (0)) being generated using the Cholesky decomposition and the following order of the large scale parameter vector: sM = [sSF, sK, sDS, sASD, sASA, sZSD, sZSA]T. 
The LSPs for links from co-sited sectors to a STX/SPST/SRX are the same. In addition, these LSPs for the links of STX/SPST/SRX on different floors are uncorrelated. The LSPs of the SPs within the same ST are correlated.
……
Step 12: Generate the cross polarization power ratios for paths in set R.
The cross polarization power ratios for each ray m of a cluster n in a STX-SPST link is generated using Step 9 of Clause 7.5, i.e., =.
The cross polarization power ratios for each ray m’ of a cluster n’ in a SPST-SRX link is generated using Step 9 of Clause 7.5 by replacing subscript n, m with n’, m’, i.e., .
For monostatic sensing mode,  is equal to  if  and . 
Generate the cross polarization power ratios (XPR) for each path  in set R at SPST p. XPR is log-Normal distributed. Draw XPR values as
	,	(7.9.4-3)
where  is Gaussian distributed with and  from Table 7.9.2.2. Note:  is independently drawn for each path in set R.

The outcome of Steps 1-12 shall be identical for all the links from co-sited sectors to a STX/ST/SRX. 
The outcome of Steps 1-12 shall be correlated for all the links for the SPs within the same sensing target. 
-----------------------------------------------------End of Text Proposal----------------------------------------------------

Proposal 6: 
Towards defining the channel for link level simulation, adding a new path for each ST on top of the current CDL and/or TDL table, where the new path is of parameters for normalized delay, power, angle (AOA, ZOA, AOD, ZOD) and velocity.
The Table 3,Table 4,Table 5,Table 6 and Table 7 under Annex 2 can be referred to for the CDL-A, CDL-B, CDL-C, CDL-D and CDL-E for the ISAC link level simulation, respectively.
The Table 8,Table 9,Table 10,Table 11 and Table 12 under Annex 2 can be referred to for the TDL-A, TDL-B, TDL-C, TDL-D and TDL-E for the ISAC link level simulation, respectively.
The parameter values of the sensing target path is up to the purpose of link level simulation and can be decided together with other simulation assumptions in the evaluation stage.

3GPP TSG RAN WG1 #121                                                            R1-2503373
St Julian, Malta, April 19th – 23rd, 2025

Source:	vivo, BUPT
Title:	Views on Rel-19 ISAC channel modelling
Agenda Item:	9.7.2
Document for:	Discussion and Decision

Conclusions
In this contribution, we have studied the target channel modeling, including the realization of concatenation in the target channel modeling, including the RCS modeling, and EO modeling, channel model formation, including two component channel modeling, as well as the enhancements of Doppler, spatial consistency as additional modelling components, and link-level based channel model for ISAC. The observations and proposals are summarized as follows.
Observation 1: In case that the specular phenomenon occurs, the amplitude gap between LOS path and specular paths is about 10dB, while the gap between the specular paths and non-specular paths is about 5dB.
Observation 2: The system performance is not greatly affected if the target channel is directly summated into the background channel without power normalization.
Observation 3: If assuming the azimuth angle  of the center point  and the elevation angle  of the scatterer P can be set to zero, i.e.,  and  the micro-Doppler frequency shift induced by the vibration of a scatterer point can be derived by Eq. 1.
Observation 4: The micro-Doppler frequency shift induced by the human arm/leg vibration can be calculated by either the instantaneous amplitude or the average amplitude.
Observation 5: The moving direction of human and frequency bands influence the micro-Doppler:
When the human body with the moving direction is facing the transmit and receive antennas, the sinusoidal function fluctuations due to the micro-Doppler are obvious, for all the six cases.
When the human body with the moving direction is perpendicular to the transmit and receive antennas, the sinusoidal function fluctuations due to the micro-Doppler are not that obvious, for all the six cases.
Micro-Doppler behavior for arm periodic swing is more obvious than that for leg periodic swing.
Different frequency bands influence micro-Doppler differently, i.e., the higher the frequency band, the more influence on micro-Doppler.
Observation 6: The agreement between the formula of micro-Doppler frequency  and the experiment results is very well.

Proposal 1: RAN1 defines a sensing target RCS as
The RCS of the SPST is defined as the hypothetical area that intercepts the incident electromagnetic wave and reinduces the scattered electromagnetic wave at the SPST. The RCS of the SPST is a scalar value in  (or dBsm), that is calculated by the power density isotropically observed at the receiver proportional to the impinged power density, multiplied by a spreading factor of , where  is the distance from the SPST to the receiver.
Proposal 2: In the multi-point RCS model, the UAV is divided into 5 scatter points, i.e., airframe body and four airfoils.
Proposal 3: In the multi-point RCS model, the human is divided into 6 scatter points, i.e., head, abdomen, two legs and two hands.
Proposal 4: Whether the location of scatter point of the sensing target changes is dependent on the specific use cases.
Proposal 5: The RCS value of the entire target is the linear summation of the RCSs from the segmented parts of the entire target.
Proposal 6: The region of two legs or two hands can be modelled by the same RCS.
Proposal 7: The region of four airfoils of UAV can be modelled by the same RCS.
Proposal 8: RAN1 utilizes the mechanism of spatial consistency to ensure the channel correlation among different scatter points in the same target.
Proposal 9: The reference TR in Table 2 for the indoor room scenario is supported.
Proposal 10: For type-2 EO, the indoor and the urban grid scenario are prioritized for the type-2 EO modelling.
Proposal 11: The effective polarization matrix of the type-2 EO reflection path should be modified to

Proposal 12: RAN1 modifies the steps to compute the effective polarization matrix of the type-2 EO reflection path based on Step 1-4.
Proposal 13: Power normalization is not required in the channel combination procedure.
Proposal 14: The combination of directly summation without power normalization can be applied to the all the scenario or sensing mode as a simple implementation.
Proposal 15: RAN1 should consider reducing the ISAC channel model complexity by reducing the number of clusters for TX-target and target-RX link in target channel.
Proposal 16: RAN1 agrees on the formula of micro-Doppler frequency shift in Eq. 1 or Eq. 2 at least for the human arm/leg target with the vibration of a scatterer point and the mono-static transmitter/receiver.
Proposal 17: RAN1 studies the formula of micro-Doppler at least for human target, and Eq. 1 or Eq. 2 can be a starting point.

3GPP TSG RAN WG1 Meeting #121	          	     	                                               R1-2503446
Malta, May 19th – 23rd, 2025
Agenda Item:	9.7.2
Source: 	EURECOM
Title: 	Discussion on ISAC channel modeling
Document for:	Discussion and decision

Conclusions
In this contribution, the following proposals are put forward:
Proposal 1: The number of scattering points depend on the size of the targets, type of channel as well the distance between the sensing target.
Proposal 2: RCS option 2 or RCS option 3 with B1 equal to 1 is used to model RCS of animals. In both options, A is mean RCS value. If Option 2 is used, B is modelled using log-normal distribution. If Option 3 is used, B2 is modelled using log-normal distribution.
Proposal 3: The small-scale parameters for the sensing channel such as RCS, echo angles, cross power ratio are generated after the general parameters are generated. Subsequently, channel coefficient for the sensing channel is generated then pathloss is calculated for each sensing cluster.
Proposal 4: Both NLOS and LOS rays are generated for the Tx-target and target-Rx links if LOS condition is determined.
Proposal 5: There are maximum two bounces between Tx and Rx.
Proposal 6: EO type 1 is not modelled in the propagation path Tx-target-Rx.
Proposal 7: A propagation path with more than one sensing target is not modelled.
Proposal 8: When an EO type 2 blocks a LOS path in a one link, that link is considered as NLOS path. Otherwise, LOS probability is calculated as in TR 38.901, 37.885.
Proposal 9: The target channel is modelled with Tx-target-Rx rays, Tx-target-EO type2/stochastic clutter-Rx rays, Tx- EO type 2/stochastic clutter- target -Rx rays.
Proposal 10: The number of clusters between the Tx and the Rx is the same as number of clusters in TR 38.901. The clusters are divided in two types LOS cluster and NLOS cluster where each NLOS cluster includes 20 rays as in TR 38.901. The number of rays in a cluster is the same as in the current TR.
Proposal 11: In power scaling factor equation, if an object is modelled by multiple scattering points, RCS component A of the object is the average value of RCS components A of all scattering points.
Proposal 12: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, hUT 1.5 m is used.
Proposal 13: EO-Type 2 can be modeled in background channel.
Proposal 14: Background channel with pathloss, LOS probability modelled with both EOs and stochastic clutters is generated from the channel model in TR 38.901.
Proposal 15: Power normalization is carried out on both target channel and background channel with equations (1) and (2) when target channel and background channel are combined to keep the same channel power as background channel without target.
Proposal 16: After concatenation, the power threshold for path dropping is -40dB for target channel.
Proposal 17: if the Tx and Rx are static, the mobility model in TR 38.901 is used.
Proposal 18: a micro-Doppler model is not necessary to model the frequency shift.
Proposal 19: For TRP-UE bistatic, UE-TRP bistatic, UE-UE bistatic, UE monostatic, if the UE is static when it is the Tx or Rx, new spatial consistent procedures need to be studied.
Proposal 20: when spatial consistency is enabled, the links between BS/UT and multiple scattering points of target are modelled as multiple targets with single scattering points.
Proposal 21: when spatial consistency is enabled, path dropping after concatenation is only performed at the beginning of the simulation.
Proposal 22: Extends the 2D spatial consistent procedures to support 3D spatial consistency in some scenarios such as UAV where the existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused to model 3D spatial consistency for ISAC channel.

3GPP TSG RAN WG1 #121		                                        R1-2503525
Saint Julian, Malta, May 19th – 23th, 2025

Agenda Item:     9.7.2
Source:	Spreadtrum, UNISOC
Title:	             Discussion on ISAC channel modelling
Document for:	Discussion and decision

Conclusion
In this contribution, we provided our views on the details of the ISAC channel modelling. The following proposals are made:
Proposal 1: When the EO type-2 is used to model the Tx-target link and target-Rx link in the target channel, option A is preferred.
Proposal 2: Spatial consistency should not be applied for the multiple scattering points of the same target.
Proposal 3: For spatially-consistent UT/BS/Targets mobility modelling, enhancements on Procedure A is preferred.
Proposal 4: Global grid (aka. Unified grid) + shifting UT/ST locations is preferred to model the correlation of links between two nodes of STs and UTs.

3GPP TSG RAN WG1 Meeting #121	R1-2503577
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda Item:	9.7.2
Title:		Discussion on ISAC channel modelling
Source:	Samsung
Document for:	Discussion and Decision

Conclusion
This contribution discusses the ISAC channel modelling including the framework, target and background environment modelling, spatial consistency and validation of channel modelling. The observations and proposals are summarized as follows: 
Proposal 1: Extend the working assumption on bistatic RCS for vehicle to large size UAV and AGV with necessary parameter adaptations to gain implementation benefit from the unified framework.
Proposal 2: RAN 1 consider the antenna pattern-like approach for RCS modeling of large size UAV 
Proposal 3: RAN 1 discuss the RCS values for large size UAV with table 1
Table 1. monostatic RCS of large size UAV with single scattering point

Proposal 4: Support the proposal formula considering its consistency, simplicity, and practicality for modelling RCS in both monostatic and bistatic configurations, particularly for the specified target types.
Proposal 5: Support option 1 for UAV of large size, human with RCS model 2, vehicle with single/multiple scattering points, and AGV.
Proposal 6: Support the value of the RCS component A, i.e., mean of monostatic RCS values in all 3D incident/scattered angles of a target to be applied to both monostatic and bistatic RCS of the target. 
Observation 1: Real-world B2/XPR discontinuities are influenced by too many variables to be resolved by angular correlation alone.
Observation 2: Introducing the angular correlation would overcomplicate the model without guaranteeing improved accuracy or practicality.
Proposal 7: Deprioritize the modelling of angular correlation for components B2 and XPR since it is not essential.
Proposal 8: Support applying the same value of B2/XPR to a path until the value is updated. 
Observation 3: Different target models have different values for components A/B1/B2, which leads to ambiguity in defining components A/B1/B2 for EO type-1.
Proposal 9: RAN1 to further study the types of target applicable to modelling the EO type-1 considering diverse scenarios and necessary modifications.
Proposal 10: Use same initial random phase for the same ray in Tx-target link and target-Rx link of a target for monostatic sensing.
Observation 4: -40dB threshold strikes a fair balance between complexity reduction (5% path drop) and channel modelling accuracy.
Observation 5: Retaining low-power paths via -40dB threshold is beneficial for testing robustness of distant target detection robustness due to near-far effect
Proposal 11: Consider -40dB as threshold for path dropping in target channel generation
Proposal 12: Consider use hUT 1.5 m for pathloss calculation (Option 5) when the height of a scattering point of target is less than 1.5m. 
Proposal 13: Consider generating the absolute delay model for sensing scenarios Urban grid, highway and HST based on the values of parameters for UMa and RMa.
Proposal 14: No need to perform power normalization between target channel and background channel.
Observation 6: The mobility of stochastic clutters can be reflected as slow-moving objects following a specific Doppler distribution with an average Doppler of zero, which has a significant impact on target channel modelling 
Proposal 15: RAN1 to consider reusing dual mobility model in TR 38.901 to reflect the effect of the stochastic clutters in target channel with necessary modifications, e.g., maximum speed and ratio of moving scatters.  
Proposal 16: the path dropping after concatenation of Tx-target and target-Rx link is not reperformed and the large/small scale parameters can be updated according to the procedure of spatial consistency. 
Proposal 17: No spatial consistency enabled for the case where the links originate from distinct communication scenarios or involve virtual reference points.
Observation 7: Enabling spatial consistency for channels of multiple scattering points may help reflect realistic scattering characteristics of scattering points and their associated parameters.
Proposal 18: When multiple scattering points of target is modelled, enabling spatial consistency is optional.
Proposal 19: When spatial consistency is not implemented, channels of the multiple scattering points of a target is independently generated. 
Observation 8: The validity of direct application of horizontal correlation distance to vertical or 3D correlation distance is unclear. 
Proposal 20: RAN1 to study on whether horizontal correlation can be directly reused to model 3D spatial consistency for ISAC channel. 
Proposal 21: EO type-2 should be modelled in background channel if modelled in target channel 
Observation 9: Option A may require new criteria to determine LOS blockage due to EO type-2.
Observation 10: the blockage model is particularly designed for human/vehicular blocking and its applicability to EO type-2 is unverified and may require more study. 
Proposal 22: RAN1 to further address the limitations of Option A and Option B and down-select from the two options.

3GPP TSG RAN WG1 #121	R1-2503646
St Julian’s, Malta, May 19th – 23th, 2025

Agenda Item:	9.7.2
Source:	Pengcheng Laboratory
Title:	Discussion on ISAC channel modelling
Document for:	Discussion and Decision

Conclusion
In this contribution, we have the following proposals regarding ISAC channel modelling.
Proposal 1: EO type-2 is modelled in background channel if modelled in target channel.
Proposal 2: The LOS condition of the Tx-target link and target-Rx link can be determined based on the following option.
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition. Otherwise, use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition.
Proposal 3: When multiple scattering points of target is modelled, spatial consistency is used.
Proposal 4: It is necessary to consider the Doppler frequency in background channel for monostatic sensing.
Proposal 5: The reference points to generate the UT monostatic background channel are stationary, and their Doppler frequencies are  by reusing Eq. (7.6-45) in Section 7.6.10 of TR38.901.
Proposal 6: In order to generate Tx-target link, target-Rx link and the background channel between a RSU-type UE and another node (TRP, pedestrian UE, vehicle UE, RSU-type UE), the following reference TRs should be adopted.

3GPP TSG RAN WG1 #121		 	      R1-2503698
St Julian’s, Malta, May 19th – 23rd, 2025

Source:           ZTE Corporation, Sanechips, CAICT
Title:             Discussion on channel modelling for ISAC
Agenda item:      9.7.2
Document for:     Discussion

Conclusion
In this contribution, we provide our measurements, RT simulations, observations, and analysis on ISAC channel modelling, and we have the following observations and proposals:
Proposal 1: For bi-static RCS of small size UAV and of human model 1, the following patterns can be adopted
The bistatic RCS of UAV with small size is modelled as 
The values/pattern of A*B1 is given by

Component A, i.e., : same as component A of mono-static RCS for UAV of small size
 dB, where  is the bi-static angle between incident ray and scattered ray,  is within 0 and 180 degree
The effect of forward scattering  is -Inf in Rel-19
Component B2: same as component B2 of mono-static RCS for UAV of small size
The bistatic RCS of Human with RCS model 1 is modelled as
The values/pattern of A*B1 is given by

Component A, i.e., : same as component A of mono-static RCS for Human with RCS model 1
 dB, where  is the bi-static angle between incident ray and scattered ray,  is within 0 and 180 degree
The effect of forward scattering  is -Inf in Rel-19
Component B2: same as component B2 of mono-static RCS for Human with RCS model 1

Proposal 2: The values of the parameters to generate UT mono-static background channel for UAV-type or RSU-type UT are provided 

Proposal 3: For TRP mono-static sensing, the spatial consistency of background channel is not considered. 
Proposal 4: For UT mono-static sensing, the following additional parameters for background channel spatial consistency should be introduced.
Proposal 5: The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused to model 3D spatial consistency for UAV scenario. 
Proposal 6: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, the option below is adopted.
- 	Option 5: use hUT 1.5 m for pathloss calculation
Proposal 7: Endorse the step 1-7 in section 6 for EO type 2 into TR for ISAC channel modeling, where the channel coefficient for the EO reflected path is given by

Proposal 8: The pathloss of NLOS ray due to EO type 2 in the STX-SPST link or SPST-SRX link is calculated by									
-	 is the total propagation distance due to EO type2.
Proposal 9: The following contents type-2 EO are considered into TR for ISAC channel modeling
Proposal 10: For ISAC LLS, one or more sensing clusters or taps are newly added in the existing LLS tables of TR 38.901 
Values of parameters of the added cluster(s) can be discussed in performance evaluation stage
Proposal 11: For micro-Doppler modeling, reuse the patterns defined in report from ETSI ISAC ISG in LS R1-2501370 as follows
The micro-Doppler motion displacement  induces phase variations on the scattering signals from the sensing target expressed as , where  is the carrier wavelength. The micro-Doppler motion displacement,  can be expressed through the micro-Doppler velocity function,  using the following expression: . 
Proposal 12: To support ISAC, update the procedure of the existing map-based hybrid channel modeling in TR 38.901. 
A step 14 for ISAC is newly added at the end of section 8 of TR 38.901

3GPP TSG-RAN WG1 Meeting #121	R1- 2503720
Malta, MT, May 19th – 23rd, 2025

Agenda Item:	9.7.2
Source:	Tejas Networks
Title:	Discussion on ISAC channel modelling
Document for:	Discussion and Decision
__________________________________________________________________________________

Discussion
The objectives of the study item “Channel modelling for integrated sensing and communication (ISAC) in 5G-NR” [1], listed in Rel-19, and endorsed in RAN1#103, are as follows: 

This contribution provides our view on some of these topics as captured in following subsections.









Physical object model
Bistatic RCS

In bistatic RCS modeling, accurately capturing both forward scattering and shadowing effects is essential for realistic system performance in sensing and communication scenarios. Modeling these effects ensures that the ISAC systems account for the diverse propagation behaviors in complex environments, improving detection and tracking capabilities. A combination of forward scattering and shadowing effects allows for a more comprehensive representation of the target channel, enhancing the system's ability to detect and localize objects in a variety of real-world scenarios.
Observation-1: Forward scattering and shadowing effects appear depending on the size of the object, its material properties, and the relative distance between the transmitter, receiver, and target. 
Forward scattering is dominant in the far field and enhances target detectability, while shadowing is pronounced especially when the receiver is within the near field of large objects. Ignoring either phenomenon risks oversimplifying the channel model and leads to unrealistic performance evaluations, particularly in high-frequency bands or complex urban environments. We agree that the reuse of blockage model B as an optional feature is reasonable, and it offers a practical way to represent shadowing effects. However, forward scattering and shadowing are two distinct and physically significant phenomena, and both need to be accurately modeled in order to reflect realistic channel behavior.
Proposal-1: The existing blockage model B can be used as an optional feature to model the blocking effect caused by a target, without special handling of RCS values in the forward scattering direction. 



Accurate modeling of the RCS for vehicles equipped with single or multiple SPST structures is critical for the performance evaluation of automotive radar systems. As vehicular technologies advance towards higher levels of autonomy and integrated services, the need for realistic and high-fidelity bistatic RCS modeling becomes increasingly important. Option A, B, and C have been introduced to extend RCS modeling towards bistatic conditions. Option A introduces a relatively simple adjustment by applying attenuation based on the bisector angle between incident and scattered directions, but it may lack fidelity in handling specular reflections and complex scattering behaviors. Option B introduces two peak modeling (back-scattering and specular reflection peaks) with attenuation factors tailored to different parts of the vehicle (front, back, sides), offering improved accuracy. However, it does not account for additional scattering effects outside these two peaks. Option C further enhances the modeling by additionally considering a scattering term and taking the maximum across all modeled contributions, providing a more comprehensive and flexible bistatic RCS representation. This makes Option C particularly suitable for ISAC applications where fine-grained accuracy across diverse vehicle orientations and dynamic environments is crucial. 
Observation-2: Option C enables accurate characterization of both back-scattering and specular reflection peaks, incorporates a flexible attenuation mechanism based on the incident-scatter angle, and includes provisions for another scatter effects, although set to -Inf by default for future extensibility.
Proposal-2: Considering the increasing complexity of Sensing and Communication integration in automotive systems, Option C for bistatic RCS modeling of vehicles with single/multiple SPST structures is preferred.

ISAC channel model
Reference TRs

In the context of ISAC (Integrated Sensing and Communication) channel modeling, adding low-energy clusters to the background channel is important to improve realism. Real-world environments are characterized not only by strong reflections but also by numerous weak reflections from small objects, irregular surfaces, and distant scatterers. These weak signals are particularly valuable for sensing applications, where detecting subtle environmental features can be crucial. Therefore, enriching the background channel with low-energy clusters helps in creating a more comprehensive multipath environment, leading to better evaluation of sensing-related performance. Current reference TRs, such as TR 38.901 and TR 37.885, are primarily communication-focused and emphasize stronger paths. However, for ISAC scenarios, neglecting low-energy multipaths could result in an incomplete background characterization. Without the inclusion of low-energy clusters, the background channel may lack the necessary richness and diversity, leading to unrealistic assumptions in ISAC system evaluation, particularly for target detection and tracking in complex environments.
Observation-3: Without the inclusion of low-energy clusters, the background channel may lack the necessary richness and diversity, leading to unrealistic assumptions in ISAC system evaluation, particularly for target detection and tracking in complex environments.
Proposal-3: To better reflect the weak but significant multipath components present in practical wireless channels, the generation of the background channel should include the low-energy clusters. 
Proposal-4: The addition of low-energy clusters should be scenario-dependent, with more clusters included in dense urban and indoor scenarios and fewer or no additional clusters in simpler environments like rural or highway settings.

Parameter values to generated background channel for monostatic

The agreement outlines the modeling of a mono-static sensing background channel, utilizing three reference points (RPs). The height and distance between the RPs and the Tx follow a Gamma distribution. Clarification is needed regarding whether the distances between the Tx and all RPs, as well as the heights of the RPs, are the same or vary.
Observation-4:  In the context of mono-static sensing, few options can be considered for modeling Doppler frequency in the background channel, depending on the specific use cases and scenarios. The choice of modeling approach should be based on the simulation requirements and intended application. 
Proposal-5: Two approaches are proposed for modeling Doppler frequency in the background channel for monostatic sensing systems.
Each of the three scattering points in the background channel moves in a random direction, with mobility modelled according to 3GPP TR 38.901.
A purely statistical model is employed, wherein the Doppler frequency is characterized using Jakes' method or other suitable spectral models.

Proposal-6: Companies are encouraged to provide results using standardized Doppler frequency models. Based on the calibration results, we recommend to define a set of standardized Doppler frequency models, each suited to different use cases and levels of simulation fidelity, and select the appropriate model based on the specific mono-static sensing application requirements. 
Proposal-7: Clarification should be provided on whether the distances between the Tx and all RPs, as well as the heights of the RPs, are identical or independently generated.


In mono-static sensing, it is agreed that the large-scale and small-scale parameters used to generate the Tx–target link are also applied to the target–Rx link, ensuring consistency between the two directions. This includes assuming reciprocity between the departure angle on one link and the arrival angle on the other. However, further study is required to determine whether this reciprocity should extend to the initial phase of the channel. 
Observations-5: In mono-static sensing, it is agreed that the large-scale and small-scale parameters used to generate the Tx–target link are also applied to the target–Rx link, ensuring consistency between the two directions. This includes assuming reciprocity between the departure angle on one link and the arrival angle on the other. The initial phase for the Tx-target and target-Rx link is reciprocal if the first and last matrix, in total polarization of a direct/indirect path (CPMtx,sp,rx= CPMsp,rx . CPMsp . CPMtx,sp), are symmetric.
Proposal-8: The initial phase used to generate the Tx-target link are respectively the same as that of the target-Rx link for monostatic sensing. 
Impact of height of target on LOS probability or pathloss

In the current agreement, the height of a scattering point on the target is used for computing LOS probability and pathloss in various sensing scenarios (UMi, UMa, RMa, and their AV variants), irrespective of the lower height bounds specified in the referenced TRs (e.g., TR 38.901). However, when the scattering point height drops below 1.5 m - particularly under 1 m - two key issues arise. First, in models such as UMi, reducing the user terminal height (hUT) forcing the use of the higher pathloss component PL₂, which may not realistically reflect near-ground propagation. Second, the LOS probability decreases significantly when hUT is reduced, especially in InF scenarios, possibly underestimating the likelihood of LOS conditions in short-range sensing tasks. Several companies reasonably suggest using PL₁ regardless of dBP, or setting a minimum dBP value to avoid unphysical results. This maintains consistency while allowing for short-range sensing where low targets are involved. While the LOS probability reduction at lower target heights is a natural effect due to increased likelihood of obstructions, it may not fully capture real-world situations. Some companies consider this behavior justified per TR assumptions, while others suggest using a fixed target height to reflect practical sensing needs better.
Observation-6:  The LOS probability reduction at lower target heights is a natural effect due to the increased likelihood of obstructions; however, it may not fully capture real-world situations. If the height of a scattering point for the target is below 1.5m, as observed in certain sensing scenarios (UMi, UMa, RMa), this reduction may underestimate the LOS probability, particularly in environments with complex urban structures or in the presence of lower-level obstructions (e.g., vehicles or small buildings). Further investigation is needed to account for these variations.
Proposal-9: Further study is needed to accurately model these variations and to better account for the impact of lower target heights in real-world scenarios.
Power normalization between target channel and background channel


In the context of generating the combined ISAC channel, two main modeling strategies are outlined. Option 1 involves a straightforward summation of the target and background channels without any power normalization. In contrast, Option 2 introduces normalization to ensure that the overall channel power remains consistent with the background. Among its sub-options, Alt 1 - as proposed - is to apply a scaling factor α uniformly to the sum of the target and background channels. There are differing views on whether power normalization should be applied when combining the target and background channels in ISAC modeling. Some support applying normalization to ensure consistent total received power, especially when targets are introduced. Others argue that normalization may not be necessary in all scenarios, such as when the background channel is already obstructed or when the target channel is deterministically modeled, since it could introduce inaccuracies. It was also noted that even with multiple targets, the overall channel characteristics might remain largely unaffected. To address these variations, some propose a flexible or adaptive normalization approach, such as using a linear programming based method to allocate power realistically.
Observation-7: Option 2 introduces normalization to ensure overall channel power remains consistent with the background.  Each option has its associated risks:
Alt 1: Applying power normalization to both the target and background channels ensures a balanced channel power, but it may lead to an underestimation or overestimation of the target's impact, particularly in cases where the target power varies significantly from the background. This could result in unrealistic channel conditions, impacting system performance and accuracy.
Alt 2: If the target's power exceeds that of the background, it may disproportionately affect the system's overall performance, potentially leading to inaccurate power balance or system overload.
Alt 3: If there are multiple targets, replacing background clusters with target channels could lead to an increased number of clusters, complicating the channel model and potentially resulting in higher computational complexity or instability in the channel representation.
Proposal-10: We propose Alt1, provided the characteristics (  ) of the channel should not change. 


EO type-2 
EO type-2 in background channel

The proposal suggests that EO type-2 be modeled as an optional feature in the background channel. It also proposes that the configuration of whether to model EO type-2 in both the target and background channels should be separate and configurable. This would provide flexibility depending on the specific needs of the scenario, allowing companies to decide whether to include EO type-2 in either or both channels based on the modeling requirements.
Observation-8: Some companies support the proposal, suggesting that EO type-2 should be considered in both the target and background channels as an optional feature, emphasizing its potential impact on target tracking and environment modeling. However, others argue that including EO type-2 in the background channel may be unnecessary, as they believe the current background channel already accounts for environmental effects.
Proposal-11: Considering that only one meeting remains and there is insufficient time to thoroughly model this aspect, we propose to drop it due to time constraints.
LOS condition considering EO type-2

The accurate modeling of LoS/NLoS conditions is essential for developing effective communication systems, as it influences the design and performance evaluation of various wireless technologies. Option A suggests that if a Type-2 EO obstructs the direct path in a link, the link should be classified as Non-Line-of-Sight (NLOS). Conversely, Option B recommends using existing LOS probability equations to determine the LOS/NLOS status, treating the impact of Type-2 EOs as additional attenuation factors without altering the fundamental LOS classification.
Observation-9: Option B maintains consistency with established methodologies in 3GPP TR 38.901, where blockage modeling is treated as an additive feature that does not alter the inherent LOS/NLOS state of a link. This approach allows for a more nuanced representation of the propagation environment by accounting for obstructions as factors that introduce additional attenuation rather than redefining the link's LOS status. This approach ensures consistency with the blockage model B in 3GPP TR 38.901, which provides a geometric method for capturing blocking effects by modeling blockers as rectangular screens and calculating attenuation using a knife-edge diffraction model. ​Furthermore, we recommend conducting comprehensive studies to assess the necessity and accuracy of modeling additional blockage effects from Type-2 EOs.
Proposal-12: Considering that only one meeting remains and there is insufficient time to thoroughly model this aspect, we propose to drop it due to time constraints.
EO type-2 for a link in NLOS condition

Observation-10: In NLOS conditions, signal propagation is often dominated by reflections from environmental objects. EO type-2 objects, characterized by their ability to produce strong specular reflections, can play a significant role in these scenarios. Modeling the specular reflected rays from EO type-2 objects enhances accuracy of the channel representation, particularly in sensing use cases where such reflections contribute valuable information about the environment. 
Proposal-13: Considering that only one meeting remains and there is insufficient time to thoroughly model this aspect, we propose to drop it due to time constraints.

Spatial consistency 
General question

Proposal-14: Without spatial consistency, two implementation approaches are possible in a single simulation drop.
The positions of the transmitter (Tx), target, and receiver (Rx) are fixed.
If positions of Tx/target/Rx are changed during the drop, then all channel parameters are regenerated independently to reflect the new positions.
Large-scale and small-scale parameters should remain correlated in the channel across time.

Proposal-15: Spatial consistency should be enabled for computing both large-scale and small-scale parameters. If spatial consistency is not implemented, it is upto the companies to determine how to generate these parameters. Regardless, the statistical characteristics of the channel should remain consistent over time.


When spatial consistency is not enabled in a simulation, the approach, here, is to maintain fixed positions for the transmitter (Tx), receiver (Rx), and target within a single simulation drop. This allows for the modeling of Doppler effects through velocity parameters without necessitating spatial correlation mechanisms. 
Observation-11: The proposal suggests that when spatial consistency is not enabled, it allows an approach to implement in single drop. The clarity is needed if the position has changed how to generate the parameter. 
Proposal-16: We assume that the Tx/Rx locations remain fixed. We believe that spatial consistency should be modeled for the ISAC link and in cases where this assumption is crucial, or if no agreement is reached, we propose the following approach:
For NLoS (Non-Line-of-Sight) conditions, the existing model should be used.
For LoS (Line-of-Sight) conditions, the parameters should be updated based on the target location.
RLPS (Remote Location Parameter Sets) and LSPs (Large-Scale Parameters) can remain unchanged across snapshots.
A soft LoS model from the existing 38.901 specification can be used for the Tx-Target and Target-Rx links.
SSPs (Small-Scale Parameters) should only be updated for the Tx-Target and Target-Rx links in LoS conditions, based on the location of the target.
This approach holds for the duration of a single simulation.


In spatially consistent channel modeling, it is essential to maintain the continuity and smooth evolution of channel characteristics as the transmitter (Tx), receiver (Rx), or target moves over space and time. A key aspect of this continuity involves the handling of multipath components—specifically, whether the set of propagation paths remains consistent during motion. In scenarios where path dropping is applied after concatenating the Tx-target and target-Rx links, it is important to distinguish between the selection of the path set and the generation of path-specific parameters.
Observation-12: While the set of paths remains fixed, the parameter values associated with each path (such as absolute delay, angle of arrival/departure, Doppler shift, and path power) should be updated as a function of the changing positions of Tx, target, and Rx. This approach allows the channel to evolve smoothly and realistically with movement.
Proposal-17: We agree with the proposal that when spatial consistency is enabled, the set of remaining paths, the subset of multipath components retained after path dropping and concatenation of the Tx-target and target-Rx links, should remain unchanged during the movement of the Tx, target and/or Rx.

Cases that spatial consistency are supported or not supported

Spatial consistency refers to the smooth evolution and correlation of channel parameters as user terminals (UTs), targets, and transceivers move within the same scenario. However, when channel links are generated by referencing different communication scenarios, the fundamental assumptions and statistical characteristics of the environments diverge. As such, spatial correlation across these scenario boundaries may no longer be valid or meaningful.
Observation-13: Industry feedback indicates strong consensus that spatial consistency should not be modeled for links spanning different channel scenarios, as the underlying large-scale parameters and propagation environments are inherently uncorrelated.
Proposal-18: We agree with the proposal to not to model spatial consistency for links that are generated across different communication scenarios.


Even when the height difference between the TRP and the UT/target is minimal, such as in UMi, InH and InF scenarios, spatial consistency is not be assumed between TRP-target/UT links and UT-target (UE-UE) links.  According to 3GPP TR 38.858, UE-UE links adopt identical angle spread statistics for both ends, while TRP-UT/target links follow TRP-side statistics for departure angles, which differ significantly. As in TR 38.901, the statistical parameters, in InF and InH scenarios, between TRP and UE differ even within the same scenario, and enforcing spatial consistency across these link types would result in unrealistic correlations.
Observation-14: Differences in angular spread parameters between TRP-target/UT and UT-UT links, modeling spatial consistency between these link types would introduce unrealistic correlations. Therefore, it's appropriate not to model spatial consistency between TRP-target/UT and UT-UT links in such scenarios.
Proposal-19: We agree with the proposal to not to model spatial consistency as long as the parameter sets between two links are different. 

Proposal-20: In the context of UE monostatic sensing, the spatial consistency of the background channels for different UEs or the same UE with mobility is a crucial aspect to consider for accurate channel modeling. The three primary parameters that need to be updated and correlated are: 
Distance between reference point and Tx
Height of reference point
LOS AOD between Tx and reference point
The spatial consistency for the background channels can be modeled in different ways, depending on the specific conditions and mobility of the objects in the scenario. The table below outlines the potential configurations for spatial consistency:

Spatial consistency between multiple scattering points

In sensing scenarios involving extended targets (e.g., humans, vehicles, UAVs), it is common to model the target using multiple scattering points to capture its reflective characteristics more accurately. The handling of spatial consistency across these scattering points is important to ensure realistic channel behavior. The current proposal suggests modeling each scattering point as if it were an independent single-point target when spatial consistency is enabled.
Observation-15: Our understanding of the second bullet is that multiple scattering points on the same target are treated as if they are individual targets, each with a single scattering point, for the purpose of spatial consistency modeling. Therefore, if scattering points on a single target fall within the defined correlation distance, spatial consistency should be applied between them. 
Proposal-21: When multiple scattering points on the same target are within the defined correlation distance, spatial consistency should be applied across those scattering points. 

Solutions for new spatial consistency
The selection of a spatial consistency model for the correlation of links between nodes of STs and UTs has been discussed. Various companies have expressed their preferences regarding different approaches, each balancing complexity, accuracy, and feasibility. Option 1 is widely regarded as a simpler and more efficient approach, as it requires only a single global reference grid, from which UT/ST-specific grids can be derived through shifts. In contrast, Option 2, which employs Cholesky decomposition, is deemed feasible only when dealing with a small number of UT/STs. For large-scale simulations, this method presents computational challenges, particularly in determining how many UT/STs should be included in the decomposition process. Given these technical considerations, the discussion has centered on striking a balance between model accuracy and computational efficiency/memory while ensuring flexibility in implementation.
Observation-16: Companies prioritizing simplicity and scalability can adopt Option 1, while those requiring more detailed and localized modeling can opt for Option 2 or its variations. By maintaining multiple options, the industry can support diverse implementation strategies while ensuring interoperability and innovation in spatial consistency modeling.
Proposal-22: It is recommended to retain all available options and establish a standardized procedure for selecting the most appropriate one based on specific needs and deployment scenarios.
Conclusions
We list all the observations and proposals here for quick reference
Observation-1: Forward scattering and shadowing effects appear depending on the size of the object, its material properties, and the relative distance between the transmitter, receiver, and target. 
Observation-2: Option C enables accurate characterization of both back-scattering and specular reflection peaks, incorporates a flexible attenuation mechanism based on the incident-scatter angle, and includes provisions for another scatter effects, although set to -Inf by default for future extensibility.
Observation-3: Without the inclusion of low-energy clusters, the background channel may lack the necessary richness and diversity, leading to unrealistic assumptions in ISAC system evaluation, particularly for target detection and tracking in complex environments.
Observation-4:  In the context of mono-static sensing, few options can be considered for modeling Doppler frequency in the background channel, depending on the specific use cases and scenarios. The choice of modeling approach should be based on the simulation requirements and intended application. 
Observations-5: In mono-static sensing, it is agreed that the large-scale and small-scale parameters used to generate the Tx–target link are also applied to the target–Rx link, ensuring consistency between the two directions. This includes assuming reciprocity between the departure angle on one link and the arrival angle on the other. The initial phase for the Tx-target and target-Rx link is reciprocal if the first and last matrix, in total polarization of a direct/indirect path (CPMtx,sp,rx= CPMsp,rx . CPMsp . CPMtx,sp), are symmetric.
Observation-6:  The LOS probability reduction at lower target heights is a natural effect due to the increased likelihood of obstructions; however, it may not fully capture real-world situations. If the height of a scattering point for the target is below 1.5m, as observed in certain sensing scenarios (UMi, UMa, RMa), this reduction may underestimate the LOS probability, particularly in environments with complex urban structures or in the presence of lower-level obstructions (e.g., vehicles or small buildings). Further investigation is needed to account for these variations.
Observation-7: Option 2 introduces normalization to ensure overall channel power remains consistent with the background.  Each option has its associated risks:
Alt 1: Applying power normalization to both the target and background channels ensures a balanced channel power, but it may lead to an underestimation or overestimation of the target's impact, particularly in cases where the target power varies significantly from the background. This could result in unrealistic channel conditions, impacting system performance and accuracy.
Alt 2: If the target's power exceeds that of the background, it may disproportionately affect the system's overall performance, potentially leading to inaccurate power balance or system overload.
Alt 3: If there are multiple targets, replacing background clusters with target channels could lead to an increased number of clusters, complicating the channel model and potentially resulting in higher computational complexity or instability in the channel representation.
Observation-8: Some companies support the proposal, suggesting that EO type-2 should be considered in both the target and background channels as an optional feature, emphasizing its potential impact on target tracking and environment modeling. However, others argue that including EO type-2 in the background channel may be unnecessary, as they believe the current background channel already accounts for environmental effects.
Observation-9: Option B maintains consistency with established methodologies in 3GPP TR 38.901, where blockage modeling is treated as an additive feature that does not alter the inherent LOS/NLOS state of a link. This approach allows for a more nuanced representation of the propagation environment by accounting for obstructions as factors that introduce additional attenuation rather than redefining the link's LOS status. This approach ensures consistency with the blockage model B in 3GPP TR 38.901, which provides a geometric method for capturing blocking effects by modeling blockers as rectangular screens and calculating attenuation using a knife-edge diffraction model. ​Furthermore, we recommend conducting comprehensive studies to assess the necessity and accuracy of modeling additional blockage effects from Type-2 EOs.
Observation-10: In NLOS conditions, signal propagation is often dominated by reflections from environmental objects. EO type-2 objects, characterized by their ability to produce strong specular reflections, can play a significant role in these scenarios. Modeling the specular reflected rays from EO type-2 objects enhances accuracy of the channel representation, particularly in sensing use cases where such reflections contribute valuable information about the environment. 
Observation-11: The proposal suggests that when spatial consistency is not enabled, it allows an approach to implement in single drop. The clarity is needed if the position has changed how to generate the parameter. 
Observation-12: While the set of paths remains fixed, the parameter values associated with each path (such as absolute delay, angle of arrival/departure, Doppler shift, and path power) should be updated as a function of the changing positions of Tx, target, and Rx. This approach allows the channel to evolve smoothly and realistically with movement.
Observation-13: Industry feedback indicates strong consensus that spatial consistency should not be modeled for links spanning different channel scenarios, as the underlying large-scale parameters and propagation environments are inherently uncorrelated.
Observation-14: Differences in angular spread parameters between TRP-target/UT and UT-UT links, modeling spatial consistency between these link types would introduce unrealistic correlations. Therefore, it's appropriate not to model spatial consistency between TRP-target/UT and UT-UT links in such scenarios.
Observation-15: Our understanding of the second bullet is that multiple scattering points on the same target are treated as if they are individual targets, each with a single scattering point, for the purpose of spatial consistency modeling. Therefore, if scattering points on a single target fall within the defined correlation distance, spatial consistency should be applied between them. 
Observation-16: Companies prioritizing simplicity and scalability can adopt Option 1, while those requiring more detailed and localized modeling can opt for Option 2 or its variations. By maintaining multiple options, the industry can support diverse implementation strategies while ensuring interoperability and innovation in spatial consistency modeling.


Proposal-1: The existing blockage model B can be used as an optional feature to model the blocking effect caused by a target, without special handling of RCS values in the forward scattering direction. 
Proposal-2: Considering the increasing complexity of Sensing and Communication integration in automotive systems, Option C for bistatic RCS modeling of vehicles with single/multiple SPST structures is preferred.
Proposal-3: To better reflect the weak but significant multipath components present in practical wireless channels, the generation of the background channel should include the low-energy clusters. 
Proposal-4: The addition of low-energy clusters should be scenario-dependent, with more clusters included in dense urban and indoor scenarios and fewer or no additional clusters in simpler environments like rural or highway settings.
Proposal-5: Two approaches are proposed for modeling Doppler frequency in the background channel for monostatic sensing systems.
Each of the three scattering points in the background channel moves in a random direction, with mobility modelled according to 3GPP TR 38.901.
A purely statistical model is employed, wherein the Doppler frequency is characterized using Jakes' method or other suitable spectral models.

Proposal-6: Companies are encouraged to provide results using standardized Doppler frequency models. Based on the calibration results, we recommend to define a set of standardized Doppler frequency models, each suited to different use cases and levels of simulation fidelity, and select the appropriate model based on the specific mono-static sensing application requirements. 
Proposal-7: Clarification should be provided on whether the distances between the Tx and all RPs, as well as the heights of the RPs, are identical or independently generated.
Proposal-8: The initial phase used to generate the Tx-target link are respectively the same as that of the target-Rx link for monostatic sensing.  
Proposal-9: Further study is needed to accurately model these variations and to better account for the impact of lower target heights in real-world scenarios.
Proposal-10: We propose Alt1, provided the characteristics (  ) of the channel should not change. 
Proposal-11: Considering that only one meeting remains and there is insufficient time to thoroughly model this aspect, we propose to drop it due to time constraints.
Proposal-12: Considering that only one meeting remains and there is insufficient time to thoroughly model this aspect, we propose to drop it due to time constraints.
Proposal-13: Considering that only one meeting remains and there is insufficient time to thoroughly model this aspect, we propose to drop it due to time constraints.
Proposal-14: Without spatial consistency, two implementation approaches are possible in a single simulation drop.
The positions of the transmitter (Tx), target, and receiver (Rx) are fixed.
If positions of Tx/target/Rx are changed during the drop, then all channel parameters are regenerated independently to reflect the new positions.
Large-scale and small-scale parameters should remain correlated in the channel across time.

Proposal-15: Spatial consistency should be enabled for computing both large-scale and small-scale parameters. If spatial consistency is not implemented, it is upto the companies to determine how to generate these parameters. Regardless, the statistical characteristics of the channel should remain consistent over time.
Proposal-16: We assume that the Tx/Rx locations remain fixed. We believe that spatial consistency should be modeled for the ISAC link and in cases where this assumption is crucial, or if no agreement is reached, we propose the following approach:
For NLoS (Non-Line-of-Sight) conditions, the existing model should be used.
For LoS (Line-of-Sight) conditions, the parameters should be updated based on the target location.
RLPS (Remote Location Parameter Sets) and LSPs (Large-Scale Parameters) can remain unchanged across snapshots.
A soft LoS model from the existing 38.901 specification can be used for the Tx-Target and Target-Rx links.
SSPs (Small-Scale Parameters) should only be updated for the Tx-Target and Target-Rx links in LoS conditions, based on the location of the target.
This approach holds for the duration of a single simulation.
Proposal-17: We agree with the proposal that when spatial consistency is enabled, the set of remaining paths, the subset of multipath components retained after path dropping and concatenation of the Tx-target and target-Rx links, should remain unchanged during the movement of the Tx, target and/or Rx.
Proposal-18: We agree with the proposal to not to model spatial consistency for links that are generated across different communication scenarios.
Proposal-19: We agree with the proposal to not to model spatial consistency as long as the parameter sets between two links are different. 
Proposal-20: In the context of UE monostatic sensing, the spatial consistency of the background channels for different UEs or the same UE with mobility is a crucial aspect to consider for accurate channel modeling. The three primary parameters that need to be updated and correlated are: 
Distance between reference point and Tx
Height of reference point
LOS AOD between Tx and reference point
The spatial consistency for the background channels can be modeled in different ways, depending on the specific conditions and mobility of the objects in the scenario. The table below outlines the potential configurations for spatial consistency:

Proposal-21: When multiple scattering points on the same target are within the defined correlation distance, spatial consistency should be applied across those scattering points. 
Proposal-22: It is recommended to retain all available options and establish a standardized procedure for selecting the most appropriate one based on specific needs and deployment scenarios.


References
[1] 3GPP TSG RAN WG1 #120, “RAN1 Chair’s Notes” version EOM1.

[2] 3GPP TSG RAN WG1 #120bis, “RAN1 Chair’s Notes” version EOM0.

[3] R1-2502553, Discussion/Decision document on “Summary #1 on ISAC channel modelling”.	

[4] 3GPP TR 38.901, “Study on channel model for frequencies from 0.5 to 100 GHz (Release 18)” V18.0.0 (2024-03)

TDoc file conclusion not found

3GPP TSG RAN WG1 #121	R1-2503726
St Julian’s, Malta, May 19th – 23th, 2025

Agenda Item:	9.7.2
Source:	Toyota ITC
Title:	Discussion on ISAC channel modelling
Document for:	Discussion and Decision

Conclusion
In this contribution we discussed our views on channel modelling for ISAC. Our proposals are summarized as follows: 
Proposal 1: How to select single or multiple scattering points for the target is depending on at least the size of the target.
Proposal 2: RAN1 to model a vehicle as multiple scattering points.
Proposal 3: TR 37.885 is adopted as the reference TR for channel models associated with RSU-type.
Proposal 4: In order to generate Tx-target link, target-Rx link and the background channel between a RSU-type UE and another node (TRP, normal/pedestrian UE, vehicle UE, RSU-type UE), the following rows are added to the existing table for reference TRs:

3GPP TSG RAN WG1 #121 		                                    		   R1-2503753
St Julian’s, Malta, May 19th – 23rd, 2025 

Agenda Item:	9.7.2
Source:	InterDigital, Inc.
Title:	Discussion on ISAC channel modeling
Document for:	Discussion and Decision

Conclusion
RCS details
Proposal 1: The parameterization of component A*B1 of human RCS model 2 can be represented by 4 rows, with parameters defined as follows:
Observation 1: Average RCS for human RCS model 2 (component A) depends on various parameters and conditions of the body such as:
Body size (kid, adult)
Body type (thin, fat)
Body position (standing, sitting)
Gender (male, female)
Attire (light, thick)
Angle, frequency, etc.

Proposal 2: For human RCS model 2, support multiple values for A to reflect different human body characteristics. 
Proposal 3: RCS measurements provided by the companies should include details related to sensing type and scenario (ground based, airborne based, etc.).
Spatial consistency
Proposal 4: Initial random phase and XPR of  and  are link-correlated for spatial consistency modeling and spatially consistent mobility modeling of the same target.
Proposal 5: Initial random phase and XPR of  is link-correlated for spatially consistent mobility modeling of the same target only.
Proposal 6: The RCS parameter,  is link-correlated for spatially consistent mobility modeling of the same target only.
Observation 2: If XPR and initial random phase of  are modeled as link correlated for spatially consistency mobility modeling of a scattering point, it is not necessary to model angular correlation of these parameters as part of the RCS model for the target.
LOS probability for UAV targets; Case 7 & Case 9
Observation 3: The accurate LoS probability modelling for Case 7 and Case 9 requires measurement campaigns, which is not feasible due to time limitations in Rel. 19.
Proposal 7: Use validated 3GPP channel models with modifications to model the LoS probability for Case 7 and 9.
Observation 4: The LoS probability in TR 36.777 is modelled based on the height segments (low-UAV region, mid-UAV region and high-UAV region) of the aerial UE.
Proposal 8: The LoS probability for Case 7 and Case 9 is defined based on the aerial UE height segments, similarly to TR 36.777.
Proposal 9: Use Table 2 for defining the LoS probability for Case 9 between two aerial UE nodes for each scenario (UMi-AV, UMa-AV, RMa-AV).
Table 2: Table defining the LoS probability between two nodes (node 1 and node 2) as a function of height regions for each node
Observation 5: Case 7 can be considered as a special case of Case 9 with height of one node in the low-UAV region.
Proposal 10: Define the LoS probability for Case 7 between a normal UE and an aerial UE for each scenario (UMi-AV, UMa-AV, RMa-AV) using the following table
Shadow fading
Proposal 11: Use Table 3 for defining pathloss and shadow fading for Case 9 between two aerial UE nodes for each scenario (UMi-AV, UMa-AV, RMa-AV).
Table 3: Pathloss and shadow fading between two nodes as a function of height regions for each node
Proposal 12: Use the following table to define pathloss and shadow fading for Case 7 between a normal UE and an aerial UE for each scenario (UMi-AV, UMa-AV, RMa-AV).
Background channel
Proposal 13: Regarding combination of target and background channel, support Option 2 Alt1 “Power normalization on both target channel and background channel”

Proposal 14: Adopt a scaling parameter which defines a relative power difference between the background channel and target channel
Environment object
Proposal 15: When the EO type-2 is modeled in the target channel, adopt Option B, “Use the LOS probability equation to determine the LOS/NLOS condition of one link, and then the impacts of type-2 EO is modeled by a blockage model”
Proposal 16: An unintended target is a blockage factor in the target channel and further study whether Blockage model A or Blockage model B should be implemented
Power threshold for dropping concatenated links
Observation 6: For option 0 to generate indirect paths of NLOS ray+ NLOS ray in the target channel, we need to normalize ray power by number of rays per cluster(M). 
Proposal 17: Normalize rays in Option 0 by a scaling factor (e.g.,   where M is the number of rays in target-Rx link) to avoid power inflation through concatenation 
Observation 7: For Indoor Open Office scenario, nearly 40% and 3% of paths are dropped with the path dropping threshold of -25dB and -50dB, respectively.
Observation 8: For Indoor Factory scenario, nearly 45% and 4% of paths are dropped with the path dropping threshold of -25dB and -50dB, respectively.
Proposal 18: Do not adopt path dropping after concatenation
Modeling micro-Doppler
Proposal 19: Adopt micro-Doppler functions for humans and UAVs provided in Table 5.
Table 5: Micro-Doppler functions for micro-motions of humans and UAVs.
Sensing area
Proposal 20: Define a sensing coverage area for the ISAC channel model where the area includes only possible points for which the UE or network can localize an object within a specific accuracy requirement. 

TDoc file unavailable

3GPP TSG RAN WG1 #121			R1-2503761
St Julian’s, Malta, May 19th – 23th, 2025

Agenda item:	9.7.2
Title:	Discussion on ISAC channel modelling
Source:	SK Telecom
Document for:	Discussion and Decision
1. 

Conclusion
In order to simplify the large-scale channel model calibration, we propose RSU-type UE as the only UE option for Automotive sensing scenario.
We propose additional UT tx power for different UE type, and the value of tx power needs to be discussed between 23 – 41 dBm considering UE and BS tx power.
We propose the collaborative development and open sharing of a unified channel model source code—hosted on platforms such as GitHub—instead of relying solely on individual calibration templates.

3GPP TSG RAN WG1 #121                                                   R1-2503804
St Julian’s, Malta, May 19th – 23th, 2025

Source:	CATT, CICTCI
Title:	Discussion on ISAC channel modelling
Agenda Item:	9.7.2
Document for:	Discussion and Decision

Conclusion
In this contribution, we discuss the details related to channel modelling methodology. We have the following observation and proposals:
Only LOS condition is considered when modelling ground reflection in TR 38.901. Whether/how to model the impact of deterministic reflection in NLOS condition is absent in legacy TRs.
Support FL proposal 5.1-2 in R1-2503146 regarding references TRs for RSU-type UE. 
In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, use hUT 1.5 m for pathloss calculation.
In highway and urban grid sensing scenario, setting the environment height hE to 0.25m for modelling pathloss of V2B, P2B, B2R link.
Support simplified Option B in RAN1#119, i.e., when the EO type-2 is modelled in the target channel, using the existing LOS probability equations in TRs without any blockage model.
The RCS component B2 of two sensing targets is generated independently.
The RCS component B2 of different incident/scattered angles for a target is generated independently.
By default, the component B2 of RCS remains unchanged even if target position changes during simulation. If spatial consistency is considered, B2 varies according to updating periodicity related to correlated distance.
The XPR and initial phase of two sensing targets are generated independently.
The XPR and initial phase of different incident/scattered angles for a target are generated independently.
By default, XPR and initial random phase of target CPM remain unchanged even if target position changes during simulation. If spatial consistency is considered, XPR and initial random phase vary according to updating periodicity related to correlated distance.
The scaling factors in step 6 also apply to the scaling factor  and  in step 7 in section 7.5, TR 38.901, where the scaling factors need not be changed after cluster elimination in step 6 in section 7.5.
No power normalization of NLOS+NLOS indirect paths before path dropping for Option 3.
When the sensing Tx and sensing Rx are of the same type in Bi-static mode,
The ASA and ZSA of target-Rx link should obey the same statistical distribution characteristics of ASD and ZSD of Tx-target link, respectively.
The ASD and ZSD of target-Rx link should obey the same statistical distribution characteristics of ASA and ZSA of Tx-target link, respectively.
The initial phase of the Tx-target link and the initial phase of the target-Rx link are independent for monostatic sensing.
If EO type-2 is modelled when NLOS condition is determined for Tx-Target link or Target-Rx link:
The pathloss of Tx-Target link or Target-Rx link in NLOS condition with EO type-2 is calculated based on the LOS condition pathloss model.
On background channel for mono-static sensing, support Option 0 for the adjustment of absolute delay d3D.
Option 0: no scaling factor is applied to d3D
For TRP monostatic sensing, the reference points for generating the TRP monostatic background channel should be static and Doppler frequency should be 0.
For UT monostatic sensing, the reference points for generating the UT monostatic background channel should have the same velocity as UT.
The reference TR to generate the channel between real Tx and reference point in mono-static background channel should be defined clearly.
Further study the modelling of background channel for mono-static sensing in highway scenario, since there is no existing NLOS channel model in highway scenario according to TR 37.885. For example, 
For TRP based mono-static sensing, 
Option 1: reuse the channel with LOS condition.
Option 2: reuse RMa for highway FR1, and reuse UMa for highway FR2.
For UE based mono-static sensing, reuse the channel with LOS condition.
If Option 2 is adopted, alternative 1 is preferred: Power normalization on both target channel and background channel.
The existing spatial consistency model in TR 38.901 (i.e., site-specific correlation) is reused to model correlation of links between one RSU and different STs/UEs, instead of using the newly defined link correlation.
The parameter correlation and spatial consistency among multiple scattering points can be addressed independently.
Support the values of parameters for absolute delay , for both target channel and background channel:
For urban grid, the values of parameters for  of scenarios UMa are reused.
For highway, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively.
For HST, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively.
Note: If significant problem is found by validation, we can change the distribution of .

3GPP TSG RAN WG1 #121			        R1-2503842
St Julian’s, Malta, May 19th – 23st, 2025

Agenda item:	9.7.2
Source: 	CMCC, BUPT, SEU, PML
Title: 	Discussion on ISAC channel model
Document for:	Discussion and Decision

Conclusions
In this contribution, we provide our views on ZOA/ZOD generation for aerial-UE-to-aerial-UE fading channel, and the following proposal is made:
Proposal: Option-1, Option-2 or Option-3 can be considered for ZOA/ZOD generation for aerial-UE-to-aerial-UE fading channel.
Option 1
Reuse the procedure of TRP-aerial UE link generation defined in TR 36.777 by modifying the procedure of ZOA/ZOD generation defined in Step 7 of section 7.5 of TR 38.901 (as copied in Table 2) by the following procedure. 
For ZOA generation, equation (7.5-16) in TR 38.901, which is , is replaced by the following equation
And equation (7.5-17) in TR 38.901 is not adopted in the LOS case. 
ZOD generation follows the same procedure as ZOA. But reuse the distribution of ZSA for ZSD, considering symmetry. 
 and  is defined as in Figure 2 and Figure 3 for aerial-UE-to-aerial-UE link of RMa and UMa/UMi respectively. And  is the average height of rooftop. Once the RX aerial UEs is below the rooftop,  . Once the TX aerial UEs is below the rooftop, .
Option 2
Reuse the procedure of TRP-aerial UE link generation defined in TR 36.777 by setting the lower aerial UE as a TRP of same height, with an additional procedure that all rays which depart or arrive from above are removed if LOS.
Option 3
Reuse the procedure of TRP-aerial UE link generation defined in TR 36.777 by setting the lower aerial UE as a TRP of the same height, with an additional procedure that all rays which depart or arrive from above are removed with a given probability. 
The probability is defined as equal to the LOS probability (either LOS probability defined for TRP-aerial UE or possible new formula given for aerial-UE-to-aerial-UE LOS probability) between these two UEs. 

3GPP TSG RAN WG1 #121	                                                                                         R1-2503859
Malta, MT   May 19th – May 23th, 2025
Source:	                    BUPT, CMCC, VIVO, X-Net
Title:	                         ISAC Channel Modeling and Measurement Validation 
Agenda Item:	          9.7.2
Document for:	     Discussion and Decision

Conclusion 
In this contribution, we conduct field measurement validations and simulation analyses of ISAC channel, leading to a number of conclusions and modeling insights. The proposals are as follows:
Proposal 1: For AGV, the element  of cross polarization matrix of the scattering point  follows a normal distribution. The distributions are as follows: .
Proposal 2: We suggest using the formula to model the bistatic RCS of Vehicle, Large UAV, AGV and human model 2.

where


and  is applied to the  within 0~180 degrees. 
Proposal 3:  For Vehicle, Large UAV, AGV and human model 2, we suggest that .
Proposal 4: For vehicle, we suggest that  and .
Proposal 5: For Large UAV, we suggest that  and .
Proposal 6: For AGV, we suggest that  and .
Proposal 7: For human model 2, we suggest that  and .
Proposal 8: The large and small-scale parameters defined in TR 38.901/36.777/37.885 can be reused for background channel in TRP-UE and UE-TRP bi-static sensing. 
Proposal 9：For the method of generating the mono-static background channel using N reference points, N=3 can be considered for a 360° horizontal range.
Table. RP parameters for mono-static background channel modeling from BUPT
*The parameter values in UMa-TRP, UMi-TRP, UMa-UT, UMi-UT, and Indoor-UT have been inputted in the previous proposals and email discussions.
Proposal 10: In UMa-TRP scenarios, the following Gamma distribution parameters can be utilized for reference point generation:
     
     
Proposal 11: For monostatic background channel modeling, the O2I rate of monostatic Tx/Rx and each RP should be set as 0.
Proposal 12: In UMi-TRP scenarios, the following Gamma distribution parameters can be utilized for reference point generation:
     
     
Proposal 13: In UMa-UT scenarios, the following Gamma distribution parameters can be utilized for reference point generation:
     
     
Proposal 14: In UMi-UT scenarios, the following Gamma distribution parameters can be utilized for reference point generation:
     
     
Proposal 15: For monostatic background channel modeling of UMa-UT and UMi-UT scenarios, the ZOD offset of monostatic Tx/Rx and each RP should be set as 0.
Proposal 16: In Indoor-UT scenarios, the following Gamma distribution parameters can be utilized for reference point generation:
     
     
Proposal 17: In UMi-UAV scenarios, the following Gamma distribution parameters, which depend on UAV height, can be utilized for reference point generation:
  

    

Proposal 18: When considering the LLS channel model for ISAC in RAN1, it should be based on the premise that the sensing target cluster needs to have a deterministic component as the baseline for the evaluation of the sensing algorithm.
Proposal 19: For the CDL model of ISAC, two approaches are feasible: 
The target clusters only consider LOS + LOS rays. 
The power, delay, AoA, AoD, ZoA, and ZoD of the target clusters are all determined by the geometric position, and the extension values will be finalized during the evaluation.
Proposal 20: For targets such as small UAVs, large UAVs (in RMa and UMa scenarios), and humans, where the large-scale parameters between multiple points are highly correlated, it is recommended to reuse the spatial consistency results of a single point.
Proposal 21: For targets such as vehicle Type 1, vehicle Type 3, and large UAVs (in UMi scenarios), it is recommended to model the spatial consistency for multiple points separately.
Proposal 22: We suggest precompute the fixed spatial decay coefficients between target multiple points. Once spatial consistency for any single point is calculated, apply these coefficients to efficiently and accurately model ISAC multi-point spatial consistency.

3GPP TSG RAN WG1 #121	R1-2503893
St Julian’s, Malta, May 19th – 23rd, 2025

Agenda Item:	9.7.2
Source:	Xiaomi, BJTU, BUPT
Title:	Discussion on ISAC channel model
Document for:	Discussion and Decision

Conclusion
In this contribution, we provide the following proposals.
Proposal 1: For monostatic RCS with omni-directional RCS pattern, the RCS component A is calculated as the statistical expectation of the RCS raw data.
Proposal 2: For monostatic RCS with directional RCS pattern, the RCS component A is calculated by spherical integration as .
Proposal 3: The value of RCS component A is the average of monostatic RCS and commonly applied to bistatic and monostatic RCS.
Proposal 4: Do not support to model the correlation of B2/XPR/initial random phase for different targets.
Proposal 5: Do not support to model the angular correction of B2/XPR/initial random phase in different direct/indirect paths of same target.
Proposal 6: Support to model the correlation of B2/XPR/initial random phase for a direct/indirect path for same target based on Option 2, i.e., same value of B2/XPR/initial random phase applies to a path before the value of B2/XPR/initial random phase is updated, and the periodicity to update B2/XPR/initial random phase can be discussed in evaluation phase or up to company choice.
Proposal 7: For bistatic RCS of vehicle, support to confirm the working assumption, and the modelling method in working assumption can be extended to large UAV and AGV.
Proposal 8: For bistatic RCS of target exhibiting omni-directional RCS pattern, support the FL Proposal 4.2-1.
Proposal 9: For human monostatic and bistatic RCS, the RCS component A is defined as 3.65dB.
Proposal 10: For large UAV monostatic and bistatic RCS, the RCS component A is defined as -2.06dB.
Proposal 11: For vehicle monostatic and bistatic RCS, the RCS component A is defined as 12.53dB.
Proposal 12: For mono-static sensing, same initial random phase can be used for Tx-target link and target -Rx link.
Proposal 13: For UMi, UMa, RMa, InF, Indoor Office, Indoor Room, Highway, Urban grid and HST, current TRs that are referred to generate ISAC channel can be used directly when a scattering point of target is less than 1.5m.
Proposal 14: When the EO type-2 is modelled in the target channel, Option A is preferred to determine the LOS condition of the Tx-target link and target-Rx link.
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, and otherwise use the LOS probability equation to determine the LOS/NLOS condition
-	Reuse the LOS probability defined in existing TRs
-	Blockage by EO type-2 is not modeled
Proposal 15: When the height of a scattering point of target is less than 1.5m,
for InF, Indoor Office, Indoor Room, RMa, Highway, and Urban grid, existing TRs that are referred to generate ISAC channel can be reused directly;
for UMi and UMa, option 4 is preferred, i.e., use  in Table 7.4.1-1: Pathloss models in TR 38.901 when hUT is below 1.5m;
for HST, the pathloss model of RMa and UMa can be reused in FR1 and FR2, respectively.
Proposal 16: When path dropping after concatenation is considered for concatenation option 0 and option 3, the power threshold -40dB and -45dB can be used, and -45dB is preferable.
Proposal 17: The set of remaining indirect paths can be updated during movement of Tx, target, or Rx after a certain period, and it can be up to implementation in the SLS platform.
Proposal 18: Doppler effect  due to movement of stochastic clusters can be modelled by .
For RMa-AV, UMa-AV, UMi-AV, RMa, UMa, UMi, Indoor Office, Indoor Room, InF, and HST, dual mobility model in 7.6.10, TR 38.901 is reused to model Doppler effect .
For Highway and Urban grid, dual mobility model in 6.2.3, TR 37.885 is reused to model Doppler effect .
The maximum speed of moving scatterers  and ratio of moving scatterers among all scatterers  can be determined during the study of evaluation methodology for each scenario.
Proposal 19: To generate the absolute delay model for sensing scenarios UMi-AV, UMa-AV, RMa-AV, Urban grid, highway and HST
For UMi-AV, UMa-AV, RMa-AV, the values of parameters for  of scenarios UMi, UMa, and RMa are reused, respectively.
For Urban grid, the values of parameters for Δτ of scenarios UMa are reused.
For Highway, the values of parameters for Δτ of scenarios RMa and UMa are reused for FR1 and FR2, respectively.
For HST, the values of parameters for Δτ of scenarios RMa and UMa are reused for FR1 and FR2, respectively.
Proposal 20: The spatial correlations among random variables generating channels of links with different types of nodes are not modelled, i.e., BS-BS, BS-UT/ST, and UT/ST-UT
Note: ST and UT are considered as the same type of node.
Proposal 21: The spatial consistency between multiple scattering points of the same target is treated in the same way as spatial consistency between different targets with single scattering points.
Proposal 22: To enable spatial consistency procedure for 3D locations, the spatial consistency procedure is considered as a 3D random process, and the correlation distance is the 3D distance.

3GPP TSG-RAN WG1 Meeting #121	R1-2503955
St Julian’s, Malta, May 19th – 23th, 2025

Agenda item:		9.7.2
Source:	Nokia, Nokia Shanghai Bell
Title:	Discussion on ISAC channel modeling
WI code:	FS_Sensing_NR
Release:	Rel-19
Document for:		Discussion and Decision

Conclusions
In this contribution, we are making these proposals for ISAC channel modelling:
Proposal 1: The value of the RCS component A is the mean of monostatic RCS values in all 3D incident/scattered angles of a scattering point of a target, and it is applied to both monostatic and bistatic RCS of a target scattering point.
Proposal 2: The RCS component B2 of two different targets is generated independently.
Proposal 3: The RCS component B2 of different direct/indirect paths of a target in the target channel are generated independently.
Proposal 4: For the RCS component B2 of a direct/indirect path of a target in the target channel, the same value of B2 applies to a path before the value of B2 is updated. 
Proposal 5: For Rel-19 ISAC channel modeling, the XPR of two different scattering points related to two different targets shall be generated independently.
Proposal 6: For Rel-19 ISAC channel modeling, the initial random phase of two different scattering points related to two different targets shall be generated independently.
Proposal 7: For Rel-19 ISAC channel modeling, the XPR of a target for the different direct/indirect paths in the target channel are generated independently.
Proposal 8: Down prioritize study of RCS of EO type 1 objects in Rel-19 study.
Proposal 9: The initial random phase (generated in Step 10, section 7.5, TR38.901) is the same for the same ray in Tx-target link and target-Rx link of a target for monostatic sensing.
Proposal 10: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, use hUT 1.5 m for pathloss calculation.
Proposal 11: Support Option A: if type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition; otherwise use the LOS probability equation to determine the LOS/NLOS condition.
Proposal 12: In indoor scenarios, a limited number of walls, ceilings and floors may be modeled as EO-type 2. 
Proposal 13: In outdoor scenarios, EO type 2 objects are limited to ground bounce model specified in Section 7.6.8 of TR38.901.
Proposal 14: There is no need to model EO type-2 objects in the background channel if the related EO type-2 objects are modeled in the target channel.
Proposal 15: For targets with multiple scattering points, the micro-Doppler modeling shall be based on the dual mobility modeling of Section 7.6.10 of TR38.901 as baseline.
Proposal 16: LLS channel modeling is needed for ISAC channel modeling.
Proposal 17: LLS ISAC channel model is composed of a component of target channel and a component of background channel: .
Proposal 18: The CDL channel model can be used as the starting point for LLS target channel modeling.
Proposal 19: Add one “target cluster”, together with RCS model of a scattering points, to the CDL channel model for LLS target channel modeling.

3GPP TSG RAN WG1 #121		R1- 2503969
St Julian’s, Malta, May 19th – 23rd, 2025

Source:	Tiami Networks
Title:	Discussion on ISAC channel modeling
Agenda Item:	9.7.2
Document for:	Discussion and Decision

Conclusion

Proposal 1: RAN1 should consider ray-level full convolution (Option 0) for NLOS+NLOS paths in target channels, with 1-by-1 ray coupling (Option 3) and power threshold level-based path dropping to reduce complexity. Power normalization should be applied after the concatenation. 

Proposal 2: An RCS modeling framework should be developed that ensures consistency between monostatic and bistatic sensing modes across diverse target types, including dynamic multi-scattering-point targets like vehicles or AGVs. 

Proposa1 3: For targets with multiple scattering points such as car and human, determine a distance threshold to determine the number of scattering points i.e., single or multiple scattering points. 
Proposal 4: Consider spatial consistency or intra-target correlation models for multiple scattering points of a single target. 

Proposal 5: Consider generating very low energy clusters for the background channel to incorporate channels with energy levels comparable to the target channel power. To keep the complexity low, adding single ray clusters is recommended. 

Proposal 6: Limit spatial consistency modeling to LOS state and shadow fading for the target channel, treating NLOS parameters as uncorrelated to reduce complexity without sacrificing sensing accuracy.

3GPP TSG-RAN WG1 Meeting #121	R1- 2503991
St Julian’s, Malta, May 19th – 23rd, 2025

Agenda Item:	9.7.2
Source:	NVIDIA
Title:	Channel modelling for integrated sensing and communication with NR
Document for:	Discussion
1	

Conclusion
In the previous sections, we discuss general aspects of channel modelling for ISAC and make the following observations:
Observation 1: Wireless channel modelling needs to provide consistency and, above all, a correct representation of the frequency, spatial, and temporal correlations across base stations, devices, and objects in the environment.
Observation 2: Deterministic, physics-based modelling for wireless propagation, especially ray tracing, are essential for studying, evaluating, and developing key technologies in 5G-Advanced toward 6G, including ISAC, RIS, larger antenna arrays in new spectrum such as 7-24 GHz and sub-THz bands, AI/ML, etc.
Observation 3: WLAN sensing Task Group IEEE 802.11bf has embraced ray tracing based channel model for Wi-Fi sensing.
Observation 4: Though ray tracing has been considered in the map-based hybrid channel model in TR 38.901, the map-based hybrid model defined is not calibrated and has not been used in 3GPP simulation campaigns.
Observation 5: A common reference scenario defined for ray tracing not only can be used for ISAC evaluation but also for other key technologies toward 6G, including RIS, larger antenna arrays in new spectrum such as 7-24 GHz and sub-THz bands, AI/ML, to name a few.
Based on the discussion in the previous sections we propose the following:
Proposal 1: Even in geometry-based stochastic channel model, the sensing targets need to be modelled in a deterministic manner.
Proposal 2: In geometry‐based stochastic channel model, rays can be traced deterministically from the sensing transmitter to the sensing target and then to the sensing receiver, by modelling different types of interactions between the rays and surrounding objects such as reflections, diffractions and diffuse scatterings.
Proposal 3: In geometry‐based stochastic channel model, the LOS and NLOS conditions between Tx/Rx and sensing target should be deterministically determined based on geometrical locations of environment objects.
Proposal 4: In geometry‐based stochastic channel model, for modelling the target channel of a target with multiple scattering points, a LOS ray from Tx to each scattering point of the target and a LOS ray from each scattering point of the target to Rx are deterministically generated as follows.
For the i-th scattering point of the target, (AoA)_i/(ZoA)_i of the direct path at Rx, and (AoD)_i/(ZoD)_i of the direct path at the target are deterministically generated based on the locations of the i-th scattering point of the target and Rx. 
For the i-th scattering point of the target, (AoD)_i/(ZoD)_i of the direct path at Tx, and (AoA)_i/(ZoA)_i of the direct path at the target are deterministically generated based on the locations of Tx and the i-th scattering point of the target. 
The delay of the direct path associated with the i-th scattering point is deterministically generated as (d3D_tx_target_i + d3D_target_rx_i)/c.
The doppler of the direct path associated with the i-th scattering point of the target is deterministically generated by spherical unit vectors associated with (AoD)_i/(ZoD)_i at Tx, spherical unit vectors associated with (AoA)_i/(ZoA)_i at Rx, and velocities of Tx, i-th scattering point of the target and Rx. 
The power of the direct path associated with the i-th scattering point of the target is generated as the product of the power of the LOS ray from Tx to the i-th scattering point of the target, the power of the LOS ray from the i-th scattering point of the target to Rx, and the effect of RCS. 
Proposal 5: In geometry‐based stochastic channel model, when blockage or forward scattering between sensing targets is considered, a propagation path from Tx to Rx interacting with more than one sensing target should be modelled.
Proposal 6: In geometry‐based stochastic channel model, blockage and forward scattering between sensing targets should be modelled in the target channel.
Proposal 7: In geometry‐based stochastic channel model, environment objects are deterministically modelled in the ISAC channel.
Proposal 8: In geometry‐based stochastic channel model, environment objects (EO type-1) that share similar physical characteristics as a sensing target are deterministically modelled in the ISAC channel.
Proposal 9: In geometry‐based stochastic channel model, environment objects (EO type-2) that have large sizes, including building walls and ground, are deterministically modelled in the ISAC channel.
Proposal 10: In geometry‐based stochastic channel model, diffraction and scattering can be considered in addition to specular reflection for environment objects (EO type-2) in direct path.
Proposal 11: In geometry‐based stochastic channel model, spatial consistency should be considered for modelling environment objects based on the deterministic locations and movement (if any) of the environment objects.
Proposal 12: In geometry‐based stochastic channel model, define material properties for the buildings and the ground in the urban grid V2X scenario, by using Table 7.6.8-1 in TR 38.901 as a starting point.
Proposal 13: In geometry‐based stochastic channel model, deterministically model ground reflection in the ISAC channel modelling by using the explicit ground reflection model in Section 7.6.8 in TR 38.901 as a starting point.
Proposal 14: In geometry‐based stochastic channel model, deterministically model building wall reflection in the ISAC channel modelling by modifying the explicit ground reflection model in Section 7.6.8 in TR 38.901.
Proposal 15: Ray tracing based channel modelling should be investigated for ISAC. 
Proposal 16: Define a common reference scenario for ray tracing to be used in ISAC evaluation. 
Proposal 17: Select one the following options to define a common reference scenario for ray tracing to be used in ISAC evaluation:
Option 1: Real-scenario map that is a virtual representation of a real area on earth. 
Option 2: Synthetic-scenario map that is artificially constructed to mimic a certain environment such as urban macro, rural macro, indoor office, or indoor factory.
Proposal 18: Describe the scene geometry and the characteristics of the materials involved in the common reference scenario for ray tracing to be used in ISAC evaluation.
Proposal 19: Consider using the existing scenarios defined by METIS as a starting point for discussion.
Proposal 20: Consider the METIS indoor office scenario (including its scene description and material properties) as a reference scenario for ray tracing and map-based hybrid model defined in TR 38.901.
Proposal 21: Describe a reference indoor office scenario for ray tracing in TR 38.901 as follows.
Lengths and widths of the room, cubicles, and tables are given by the figure below:

Heights of the room, cubicles, and tables are 2.9 m, 1.5 m, 0.7 m, respectively.
Materials of the room, cubicles, and tables are concrete, wood, and wood, respectively.
Proposal 22: Update Table 7.6.8-1 on material properties based on the updated version of the ITU recommendation, i.e., ITU-R P.2040-3.
Proposal 23: For the urban grid defined for V2X in TR 37.885, the scene geometry can be described by including assumption on building height, and assumptions on the materials for the buildings and roads can be included.
Building height: 24m
Building material: Concrete
Ground material: Medium dry ground

3GPP TSG-RAN WG1 #121			R1-2504013
Malta, Malta, May 19th – 23rd, 2025

Agenda item:	9.7.2 
Title:	Discussion on ISAC Channel Modeling
Source:	          National Institute of Standards and Technology (NIST)
Document for:	Discussion and Decision

Conclusion
In this contribution, we discussed physical object modelling regarding bi-static RCS and angular and distance correlations of RCS component B2. We studied blockage models by considering target as a blocker, and validated the models against our measurements. Finally, we proposed a high-fidelity micro-Doppler regarding human reparation and heartbeat. Our observations and proposals are summarized as follows:

Proposal 1: When extending the monostatic UAV RCS model to develop its bistatic RCS, based on the work assumptions proposed in RAN1#120b, we recommend modifying the K1 and K2 parameters. Specifically, we suggest setting K1 to 4.09 and K2 to 0.8 to minimize fitting errors based on our measurements.

Proposal 2: The fitting performance of the bi-static RCS model degrades when the bistatic angle exceeds 90 degrees. This indicates that when the bistatic angle becomes large, particularly when it approaches 180 degrees, the bistatic RCS needs to be modeled differently. 

Proposal 3: RCS B2 angle correlation and distance correlation have been observed using measurement data, which suggests that B2 correlation should be considered when taget is moving relative to the TX or RX.

Observation 1: The original blockage model B does not account for the 3D shape of the blocker, which can reduce modeling accuracy for vehicles, particularly when the propagation occurs in the front-back direction.

Proposal 4: To model a 3D physical object using the Blockage B model, RAN1 could consider adjusting the screen size based on the physical object's size and orientation, following the procedures described in Section 3.1.1.

Observation 2: The Blockage model may overestimate the shadowing effect caused by a vehicle; for example, signal penetration through vehicle windows can reduce the attenuation from blockage.

Observation 3: The diffraction pattern can be modeled using the Fresnel integral rather than relying on the approximation provided by the arctan function. This approach allows us to better infer the shape of the blocker but may need more accurate screen model.  The rectangular screen model used in blockages A and B does not accurately represent most objects, as these objects often exhibit dielectric properties or have non-rectangular silhouettes.

Proposal 5: Equation 4-1 to 4-3 provide a high-fidelity model for body movement regarding human respiration and heartbeat. The corresponding micro-Doppler shift can be computed as follows:
,
where  and  are the unit radial vector from TX to the human subject and from RX to the human subject, respectively.

References

3GPP TSG RAN WG1 #121			R1-2504054
St, Julian’s Malta, May 19th – 23th, 2025

Agenda item:		9.7.2
Source:	China Telecom
Title:	Discussion on ISAC channel modelling
Document for:		Discussion

Conclusions
In this contribution, we discuss ISAC channel modelling related issues and have following proposals:
Proposal 1: For vehicle with single/multiple scattering points, support the following working assumption to generate bi-static RCS 
Proposal 2: For human with single/multiple scattering points, the bi-static RCS model of which can be applied by the mathematical model of its mono-static mode.
Proposal 3: For small UAV with single/multiple scattering points, the bi-static RCS model of which can be applied by the mathematical model of its mono-static mode.
Proposal 4: 
The bistatic RCS of UAV with small size is modelled as
Component A: same as component A of mono-static RCS for UAV of small size
Component B1:  where  is the 3D bi-static angle between incident angle and scattered angle
Component B2: same as component B2 of mono-static RCS for UAV of small size
The bistatic RCS of human with RCS model 1 is modelled as
Component A: same as component A of mono-static RCS for human with RCS model 1
Component B1:  where  is the 3D bi-static angle between incident angle and scattered angle
Component B2: same as component B2 of mono-static RCS for human with RCS model 1
Proposal 5: When the EO type-2 is modeled in the target channel, to determine the LOS condition of the Tx-target link and target-Rx link 
If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, otherwise use the LOS probability equation to determine the LOS/NLOS condition.
Proposal 6: Support Option 2-Alt 1 to generate the combined ISAC channel.

3GPP TSG-RAN WG1 Meeting #121		R1-2504069
St Julian's, Malta, 19 - 23 May, 2025

Agenda item:	9.7.2
Source:	Sony
Title:	Remaining Issues on Channel Modelling for ISAC  
Document for:	Discussion and Decision

Conclusion
In this contribution, we share our views on the development of ISAC channel model. We made the following observations:
Observation 1: EO-type 2 can represent various type of objects such as the ground, a building, wall, ceiling, etc. Each of these objects may have its own modelling characteristics/parameters.

In addition, we have also made the following proposals:
Proposal 1: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition. Otherwise, use the LOS probability equation defined in existing TR 38.901 to determine the LOS/NLOS condition.
Proposal 2: Define the EO-type 2 modelling parameters for different type of EO-type 2 (e.g.,  the ground, a building, wall, ceiling)
Proposal 3: The environment object (EO) that are common to both target channel and background channel, such as EO1, can be considered as an optional feature in the background channel modelling.
Proposal 4: Support the same A value for both monostatic and bistatic case for the same target type, as the RCS directivity effect is already covered by B1.
Proposal 5: Conclude that the modeling of angular correlation of component B2 is not required in Release-19.

3GPP TSG-RAN WG1 Meeting #121                                                              R1-2504110
Malta, May 19 – May 23, 2025

Source:	Panasonic
Title: 	Discussion on ISAC channel modelling	
Agenda Item:	9.7.2
Document for:	Discussion

Conclusions
Based on the discussion above, we propose the following:

Support Alt3: the target channel of a target will replace one cluster in the background channel without power normalization
If the type-2 EO blocks the LOS, the link is determined as NLOS condition. Otherwise, the LOS probability methodology defined in TR 38.901 is used to determine the LOS/NLOS condition. 
The LOS probability of the target channel and background channel are determined independently.
Define criteria how to determine whether a target example vehicle is modeled with a single or multiple scattering points

The RCS value of the entire target is the linear summation of the RCSs from the scattering points of the entire target.
Support specifying micro-Doppler functions for UAV and human. 
Study whether and how to consider the changes of relative locations (i.e., micro motion) of the multiple scattering points in the modelling of delay and angle of a direct/indirect path. 
Discuss whether the micro-Doppler patterns are per target or per scattering point. 
Additional aspects, for example crosstalk, of the direct Tx-to-Rx channel should be considered in the case of monostatic sensing mode. 

3GPP TSG RAN WG1 #121                                                  R1-2504119
St Julian’s, Malta, May 19th – 23th, 2025

Source:           CALTTA
Title:             Discussion on channel modelling for ISAC
Agenda item:      9.7.2
Document for:     Discussion

Conclusion
In this contribution, we provide our analysis on ISAC channel modelling, and we have the following proposals:
Proposal 1: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, use  in Table 7.4.1-1: Pathloss models in TR 38.901.
Proposal 2: The horizontal correlation distance from Table 7.6.3.1-2 in TR 38.901 should be repurposed to model 3D spatial consistency in UAV scenarios.

3GPP TSG RAN WG1 #121	R1-2504159
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda Item:	9.7.2
Source:	Pengcheng Laboratory
Title:	Discussion on ISAC channel modelling
Document for:	Discussion and Decision

Conclusion
In this contribution, we have the following proposals regarding ISAC channel modelling.
Proposal 1: EO type-2 is modelled in background channel if modelled in target channel.
Proposal 2: The LOS condition of the Tx-target link and target-Rx link can be determined based on the following option.
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition. Otherwise, use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition.
Proposal 3: When multiple scattering points of target is modelled, spatial consistency is used.
Proposal 4: It is necessary to consider the Doppler frequency in background channel for monostatic sensing.
Proposal 5: The reference points to generate the UT monostatic background channel are stationary, and their Doppler frequencies are  by reusing Eq. (7.6-45) in Section 7.6.10 of TR38.901.
Proposal 6: In order to generate Tx-target link, target-Rx link and the background channel between a RSU-type UE and another node (TRP, pedestrian UE, vehicle UE, RSU-type UE), the following reference TRs should be adopted.

3GPP TSG RAN WG1 #121			 R1-2504161
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda item:	9.7.2
Source:	Moderator (Xiaomi)
Title:	Summary #1 on ISAC channel modelling
Document for:	Discussion/Decision

Summary #5 on ISAC channel modelling	Moderator (Xiaomi)

3GPP TSG RAN WG1 #121			 R1-2504162
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda item:	9.7.2
Source:	Moderator (Xiaomi)
Title:	Summary #1 on ISAC channel modelling
Document for:	Discussion/Decision

Summary #5 on ISAC channel modelling	Moderator (Xiaomi)

3GPP TSG RAN WG1 #121			 R1-2504163
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda item:	9.7.2
Source:	Moderator (Xiaomi)
Title:	Summary #2 on ISAC channel modelling
Document for:	Discussion/Decision

Conclusion
The component XPR/initial random phase of two different targets are generated independently.


[Moderator’s note] Since the main concern is the impacts to Doppler of each direct/indirect path due to mobility of Tx/target/Rx, we may follow the majority view that no angular correlation of XPR among different direct/indirect paths of a target. 

[FL1][M] Proposal 4.7-2
The XPR of a target for the different direct/indirect paths in the target channel are generated independently



[Moderator’s note] Due to mobility of Tx/target/Rx, the incident/scattered angle of a direct/indirect path changes, which impacts the XPR values. 
Option 1 is simpler, but it also has clear drawback. 
Option 2 maximizes reusing existing principles to handle large/small scale parameters in 38.901
If Option 3 or 4 is selected, we open more discussions and cannot be completed in one meeting
Option 5 is quite different from existing solutions from 38.901. 

The moderator would like to check whether Option 2 can be agreeable, which seems the only solution which we could complete the issue in the limited remaining time. 
Note: Option 2 seems also the default behaviour since it keeps the similar principle as large/small scale parameters generation in existing 38.901. In a simulation with existing 38.901, once generated, such parameters are used for a while (multiple frames/slots) before update. Option 2 keeps the similar principle. 

[M] Proposal 4.7-3
On XPR of a target for a direct/indirect path in the target channel, the same value of XPR applies to a path before the value of XPR is updated. 
The periodicity to update XPR can be discussed in evaluation phase or up to company choice




[FL1][M] Proposal 4.7-3-rev1
On XPR of a target for a direct/indirect path in the target channel, the same value of XPR applies to a path before the value of XPR is updated. 
The periodicity to update XPR can be discussed in evaluation phase or up to company choice
Whether/how/when to update XPR can be discussed in evaluation phase or up to companies’ choices


Agreement
XPR of different direct/indirect paths of a target in the target channel are generated independently.
On the XPR of a direct/indirect path of a target in the target channel, the same value of XPR applies to a path before the value of XPR is updated.
Note: whether/how/when to update XPR can be discussed in evaluation phase or up to companies’ choices


[Moderator’s note] Since the main concern is the impacts to Doppler of each direct/indirect path due to mobility of Tx/target/Rx, we may follow the majority view that no angular correlation of initial random phasamong different direct/indirect paths of a target. 

[FL1][M] Proposal 4.7-4
The initial random phase of a target for the different direct/indirect paths in the target channel are generated independently



[Moderator’s note] Due to mobility of Tx/target/Rx, the incident/scattered angle of a direct/indirect path changes, which impacts the initial random phase at target. 
Option 1 is simpler, but it also has clear drawback. 
Option 2 maximizes reusing existing principles to handle large/small scale parameters in 38.901
If Option 3 or 4 is selected, we open more discussions and cannot be completed in one meeting
Option 5 is quite different from existing solutions from 38.901. 

The moderator would like to check whether Option 2 can be agreeable, which seems the only solution which we could complete the issue in the limited remaining time. 
Note: Option 2 seems also the default behaviour since it keeps the similar principle as large/small scale parameters generation in existing 38.901. In a simulation with existing 38.901, once generated, such parameters are used for a while (multiple frames/slots) before update. Option 2 keeps the similar principle. 

[M] Proposal 4.7-5
On the initial random phase of a target for a direct/indirect path in the target channel, the same value of initial random phase applies to a path before the value of the initial random phase is updated. 
The periodicity to update initial random phase can be discussed in evaluation phase or up to company choice



[FL1][M] Proposal 4.7-5-rev1
On the initial random phase of a target for a direct/indirect path in the target channel, the same value of initial random phase applies to a path before the value of the initial random phase is updated. 
The periodicity to update initial random phase can be discussed in evaluation phase or up to company choice
Whether/how/when to update initial random phase can be discussed in evaluation phase or up to companies’ choices


Agreement
Initial random phase of different direct/indirect paths of a target in the target channel are generated independently.
On the initial random phase of a direct/indirect path of a target in the target channel, the same value of initial random phase applies to a path before the value of initial random phase is updated.
Note: whether/how/when to update initial random phase can be discussed in evaluation phase or up to companies’ choices

RCS for other targets/EO type-1
[Moderator’s note] We agreed several meetings ago that EO type-1 is modelled as if target in the channel generation. If possible, it is good to define RCS model for other targets or EO type-1 in Rel-19. Otherwise, the additional targets/EO type-1 can be discussed in evaluation phase. 

The following two proposals are made based on inputs from Ericsson tdoc. 

[M] Proposal 4.8-1 
The bistatic RCS of a bird is modelled by,
Component A: uniformly distributed within range [-40, -20] dBsm
Component B1: 0 dB 
Component B2: standard deviation of B2 is 0 dB


[FL2][M] Proposal 4.8-1-rev1 
The bistatic RCS of a bird is modelled by,
Component A: uniformly distributed within range [-40, -20] dBsm
Component B1: 0 dB 
Component B2: standard deviation of B2 is 3 dB


[FL2][M] Proposal 4.8-2 
The bistatic RCS of a tree is modelled by,
The values/pattern of A*B1 of the bistatic RCS is given by:

 is the bistatic angle between the incident ray and scattering ray within the plane of incident direction () and scattering direction ().
 is 20 degree
 is 23 dBsm
 is 13 dBsm
Component B2: standard deviation of B2 is 3 dB



Collection on the setup for RCS measurement/simulation from companies 

[Moderator’s note] Based the email discussion [Post-120-ISAC-01] and confirmation by Chair, we would create a document to collect details on the setup for RCS measurement/simulation (i.e. vehicle size, frequency, distance, etc). The file includes separate sheets for different target types. Please consider adding your information for measurement/simulation. 
/Inbox/drafts/9.7(FS_Sensing_NR)/9.7.2 Channel Modelling/ Setup for RCS validation/

So far, ZTE and OPPO already input the related information.

If any additional comments, please provide it in following table


[FL1][L] Proposal 4.9-1 
The excel file which collects the setup for RCS measurement/simulation from all companies will be submitted with a tdoc number, and will be used as a reference for the CR to TR 38.901 for ISAC channel model. 


ISAC channel model
Reference TRs

Summary on company views

Ericsson
The LOS probability between an aerial UE and a normal UE is the following:

With 





,   
with parameter values 

Ericsson
For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE height of 22.5m-35m, which is similar to the corresponding BS height, use terrestrial UE-BS link of UMa scenario in TR 38.901 for urban areas and that of RMa scenario for rural areas with the BS representing the aerial UE.
For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE heights that are lower than 22.5m, use terrestrial UE-BS link of UMi scenario in TR 38.901 for urban areas with the BS representing the aerial UE.
For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE heights that are higher than 35m, use terrestrial UE-BS link of UMa in TR 38.901 for urban areas and that of RMa scenario for rural areas with the BS representing the aerial UE.


[Moderator’s note] In RAN1 #120bis, we discussed the above Case 7 but cannot conclude detailed model on LOS probability, pathloss and shadowing. Ericsson suggests the following proposal to model UE-aerial UE link. Let’s check if it is agreeable. 

[H] Proposal 5.1-1 
The LOS probability between an aerial UE and a normal UE is the following:

With 





,   
with parameter values 

For aerial UE heights similar to the corresponding macro BS height (25 m in Uma, 35 m in Rma), use the Uma or Rma model in TR 38.901 with the BS representing the aerial UE. 
For aerial UE heights that are higher than the corresponding macro BS height (25 m in Uma, 35 m in Rma), use the Uma or Rma model in TR 38.901 with the BS representing the aerial UE. 




[FL1][H] Proposal 5.1-1-rev1 
The LOS probability between an aerial UE and a normal UE is generated by:

With 



,   
with parameter values 

The pathloss and shadow fading between an aerial UE and a normal UE are generated using TRP-aerial UE link of UMi-AV in Annex A and B of TR 36.777 by setting hBS =1.5m for FR1



[FL2][H] Proposal 5.1-1-rev2 
To generate the channel between an aerial UE and a normal UE,
The LOS probability is generated by:

The pathloss and shadow fading are generated using TRP-aerial UE link of UMi-AV in Annex A and B of TR 36.777 by setting hBS =1.5m for FR1
Note: 
The height ranges of low-UAV, Mid-UAV and High-UAV are defined as in Table B-1: LOS probability in TR 36.777
The second height range for UMi-AV is further divided into 2 regions, i.e., [22.5, 100] and [100, 300]



[Moderator’s note] revised in Tuesday offline session
[FL2][H] Proposal 5.1-1-rev3 
To generate the channel between an aerial UE and a normal UE,
The LOS probability is generated by:

The pathloss and shadow fading are generated using TRP-aerial UE link of UMi-AV in Annex A and B of TR 36.777 by setting hBS =1.5m for FR1
Note: 
The height ranges of low-UAV, Mid-UAV and High-UAV are defined following the applicability range in terms of aerial UE height in Table B-1: LOS probability in TR 36.777
The second height range for UMi-AV is further divided into 2 regions, i.e., [22.5, 100] and [100, 300]




[Moderator’s note] Another open issue on the large scale parameters for link between two aerial UEs (Case 9). To be honest, there is no validation at all. The only possible approach for us is to reuse certain existing proposals/solutions from other cases. The following proposal is made the by moderator. 
The LOS probability between two aerials Ues are generated by updating Table B-1: LOS probability in TR 36.777, with following updates (marked as red in the Table)
Condition is revised to be based max(h1, h2), where h1, h2 are the height of the two aerial Ues
The formula for the first range of height is changed to that agreed for Proposal 5.1-1
In the third height range of each scenario, LOS probability is 100%
In the second height range, the proposal is to reuse 36.777, which is the only reference we have
In the first height range, we may use the agreement for Proposal 5.1-1 since it targets low Tx/Rx

Please check if the proposal can be agreeable or any suggested update are welcome? However, please being constructive since we are short of any validations. 

Note: using UAV to sense another UAV is not a typical case, at least in the quite early stage that we even never study using TRP to sense UAV yet. If a solution cannot be agreeable, the moderator would suggest we remove this case from Rel-19. 

[FL1][H] Proposal 5.1-2 
To generate the channel between a first aerial UE with height h1 and a second aerial UE with height h2, h1<=h2,
The LOS probability between the two aerial Ues is generated by:
The pathloss and shadow fading between two aerial Ues are generated using TRP-aerial UE link of UMi-AV in Annex A and B of TR 36.777 by setting height of TRP equal to the height of the first aerial UE. 



[FL2][H] Proposal 5.1-2-rev1 
To generate the channel between a first aerial UE with height h1 and a second aerial UE with height h2, abs(h1-hBS) <= abs(h2-hBS),
The LOS probability between the two aerial Ues is generated by:

The pathloss and shadow fading between two aerial Ues are generated using TRP-aerial UE link of UMi-AV in Annex A and B of TR 36.777 by setting height of TRP equal to the height of the first aerial UE. 
Note: 
The height ranges of low-UAV, Mid-UAV and High-UAV are defined as in Table B-1: LOS probability in TR 36.777
The second height range for UMi-AV is further divided into 2 regions, i.e., [22.5, 100] and [100, 300]


[Moderator’s note] revised in Tuesday offline session
[FL2][H] Proposal 5.1-2-rev2 
To generate the channel between a first aerial UE with height h1 and a second aerial UE with height h2, abs(h1-hBS) <= abs(h2-hBS),
The LOS probability between the two aerial Ues is generated by:

The pathloss and shadow fading between two aerial Ues are generated using TRP-aerial UE link of UMi-AV in Annex A and B of TR 36.777 by setting height of TRP equal to the height of the first aerial UE. 
Note: 
The height ranges of low-UAV, Mid-UAV and High-UAV are defined following the applicability range in terms of aerial UE height in Table B-1: LOS probability in TR 36.777
The second height range for UMi-AV is further divided into 2 regions, i.e., [22.5, 100] and [100, 300]




[Moderator’s note] RSU type UE is agreed in 9.7.1. Please check the following proposal on reference TRs for RSU type UE.

[FL1][H] Proposal 5.1-3 
In order to generate Tx-target link, target-Rx link and the background channel between a RSU-type UE and another node (TRP, pedestrian UE, vehicle UE, RSU-type UE), the following reference TRs are adopted




Agreement
In order to generate Tx-target link, target-Rx link and the background channel between a RSU-type UE and another node (TRP, pedestrian UE, vehicle UE, RSU-type UE), the following reference TRs are adopted


Background channel for monostatic

[Moderator’s note] In RAN1 #120bis, the above agreement is made with basic intention to have reciprocity for the Tx-target link and target-Rx link. Initial random phase for the two links should have same property otherwise it breaks the reciprocity. Please check the following proposal. 

[FL1][H] Proposal 5.2-1 
The initial random phase (generated in Step 10, section 7.5, TR38.901) is the same for the same ray in Tx-target link and target-Rx link of a target for monostatic sensing.


Agreement
The initial random phase (generated in Step 10, section 7.5, TR38.901) is the same for the same ray in Tx-target link and target-Rx link of a target for monostatic sensing.


[Moderator’s note] In [Post-120bis-ISAC-02], the parameters for UT monostatic sensing for Uma-AV is agreed. We still need values for parameters for UT monostatic sensing for Umi-AV and Rma-AV. 

[H] Question 5.2-2 
Please inputs on the remaining values of parameters for UT monostatic sensing for Umi-AV and Rma-AV with aerial UE as sensing transmitter or receiver.




[FL1][H] Proposal 5.2-2 
For Umi-AV and Rma-AV with aerial UE as sensing transmitter or receiver, the values of parameters to generate background channel for UT monostatic sensing are provided in the following table 

Note 1: Distributions of height and distance of reference point are not subject to geographical constraints on TRP for the corresponding deployment scenario.
Note 2: The reference points for generating the UT monostatic background channel have the same velocity as UT.
Note 3: In the UT monostatic sensing in Uma and Umi scenario, the ZOD offset should be set as 0


Agreement
For UMi-AV and RMa-AV with aerial UE as sensing transmitter or receiver, the values of parameters to generate background channel for UT monostatic sensing are provided in the following table 

Note 1: Distributions of height and distance of reference point are not subject to geographical constraints on TRP for the corresponding deployment scenario.
Note 2: The reference points for generating the UT monostatic background channel have the same velocity as UT.
Note 3: In the UT monostatic sensing in UMa and UMi scenario, the ZOD offset in the background channel should be set as 0


[Moderator’s note] The issue handling d3D is delayed as remaining issue from the email discussion [Post-120bis-ISAC-02]. The following options are available. 
Option 0: no scaling factor is applied to d3D
Option 1: An offset is applied to d3D, i.e., d3D-c1
Option 2: A scaling factor d_s is multiplied to d3D, i.e., d3D*d_s. d_s is a value within range [0, 1].
Option 3: An offset is applied to d3D, i.e., d3D-

From the email discussions, all interested companies except one are OK with Option 0 or 1. The moderator would like to check if Option 3 can be possible compromise solution. 

[H] Proposal 5.2-3 
To generate the background channel for monostatic sensing, ‘d3D-+’ is used to model the absolute delay between the Tx and each reference point.


[H] Proposal 5.2-3-rev1 
To generate the background channel for TRP monostatic sensing and UT monostatic sensing, ‘ +’ is used to model the absolute delay between the Tx and each reference point.



[FL1][H] Proposal 5.2-3-rev2 
To generate the background channel for TRP monostatic sensing and UT monostatic sensing, ‘ +’ is used to model the absolute delay between the Tx and each reference point.


Agreement
To generate the background channel for TRP monostatic sensing and UT monostatic sensing, ‘ +’ is used to model the absolute delay between the Tx and each reference point.


[Moderator’s note] In the email discussion [Post-120bis-ISAC-03], two sub-bullets under Step 4 for Clause 7.9.4.2 in the CR to Tr 38.901 are put in brackets. 
There should be no problem for the first sub-bullet
For the second sub-bullet, there was a comment that  should be per RP. However, based on the early agreement that a separate  is generated for the background channel, the same  should be applied to the background channel


[FL1][H] Proposal 5.2-4 
Remove the brackets for first sub-bullet under Step 4 for Clause 7.9.4.2 in the CR to TR 38.901.
On the absolute delay of the background channel for monostatic sensing, the same  is generated and applied to the 3 channels between the STX/SRX and the 3 RPs.


[FL2][H] Proposal 5.2-4-rev1 
Remove the brackets for first sub-bullet under Step 4 for Clause 7.9.4.2 in the CR to TR 38.901.
On the absolute delay of the background channel for monostatic sensing, the same three  are independently generated and respectively applied to the 3 channels between the STX/SRX and the 3 RPs.




Power threshold for path dropping in target channel

[Moderator’s note] There were discussions on the motivation to have -40dB as threshold for path dropping in target in RAN1 #120bis, please check tdoc from Xiaomi and Apple. According to Xiaomi’s results, 
For concatenation Option 0, with a probability of about 90% in LOS condition Case 2 and LOS condition Case 3
with X = -25dB, the power drop ratio of indirect paths is up to 30% 
with X = -40dB, the power drop ratio of indirect paths is up to 5%.


[FL1][M] Proposal 5.3-1 
Power threshold for path dropping after concatenation is -40dB for target channel



[FL1][M] Proposal 5.3-1-rev1 
Power threshold for path dropping after concatenation is up to -40dB for target channel. Up to company to choose a value in the implementation


[Moderator’s note] Agreed in Tuesday online session
Agreement
Power threshold for path dropping after concatenation is up to -40dB for target channel for option 3. Up to company to choose a value in the implementation.
Power threshold for path dropping after concatenation is up to -25dB for target channel for option 0. Up to company to choose a value in the implementation.
For calibrations for both option 0 and option 3, power threshold for path dropping after concatenation is -40dB for target channel.

Impact of height of target on LOS probability or pathloss

[Moderator’s note] In reality, the height of scattering point of a target may be less than 1.5m. However, some concern raised if we directly use the height of scattering point to calculate pathloss by extension of applicable range of existing pathloss formula. 
Some companies have concern on the large calculated pathloss since hUT<1m results in negative breakpoint distance,  in Table 7.4.1-1 Pathloss models in TR 38.901 will apply. Then, the proposal is to enforce the use of 
Some other companies have concern on extension of application range of pathloss formula without validation results. 
Some companies comment that there is no much difference for pathloss around 1.5m (the assumed breakpoint distance should be positive in the comment) 

Which option (4 vs. 5) is preferred? 

[H] Question 5.4-1 
Which option (4 vs. 5) is preferred? 
Option 4: use  in Table 7.4.1-1: Pathloss models in TR 38.901
Option 5: use hUT 1.5 m for pathloss calculation



[FL1][H] Proposal 5.4-1 
In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, 
use hUT 1.5 m for breakpoint distance (dBP) calculation
Note: hUT 1.5 m is only used for dBP calculation. The exact h_UT of the scattering point is still used to determine all other parameters of ISAC channel, e.g., delay, AOD/ZOD/AOA/ZOA, etc. 


Agreement
In sensing scenario UMi, UMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, 
use hUT 1.5 m for breakpoint distance (dBP) calculation
Note: hUT 1.5 m is only used for dBP calculation. The exact h_UT of the scattering point is still used to determine all other parameters of ISAC channel, e.g., delay, AOD/ZOD/AOA/ZOA, etc. 

Absolute delay

[Moderator’s note] In RAN1 #120bis, we confirm the WA on absolute delay as agreement. However, we don’t have the values of parameters for  for Urban grid, Highway, HST. Since the channel model for Urban grid, Highway, HST in the existing TRs are derived based on UMa and RMa, it is straightforward we could reuse the values of parameters for  of UMa and RMa in such scenarios. 

[FL1][H] Proposal 5.5-1
To generate the absolute delay model for sensing scenarios Urban grid, highway and HST, for both target channel and background channel
For Urban grid, the values of parameters for  of scenarios UMa are reused. 
For Highway, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
For HST, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
Note: If significant problem is found by validation, we can change the distribution of 



[FL1][H] Proposal 5.5-1-rev1
To generate the absolute delay model for sensing scenarios Urban grid, highway and HST, for both target channel and background channel
For Urban grid, the values of parameters for  of scenarios UMa are reused. 
For Highway, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
For HST, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
Note: no measurements on  of the 3 scenarios are submitted in Rel-19. 


[Moderator’s note] Agreed in Tuesday online session
Agreement
To generate the absolute delay model for sensing scenarios Urban grid, highway and HST, for both target channel and background channel
For Urban grid, the values of parameters for  of scenarios UMa are reused. 
For Highway, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
For HST, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
Note: no measurements on  of the 3 scenarios are submitted in Rel-19. 


[Moderator’s note] As commented during the offline discussion, it needs clarification what is the absolute delay model for scenario UMi-AV, UMa-AV and RMa-AV? 
[FL2][H] Proposal 5.5-2
To generate the absolute delay model for sensing scenarios UMi-AV, UMa-AV and RMa-AV, for both target channel and background channel, 
For the TRP-TRP link and TRP-UE link, the values of parameters for  of scenarios UMi, UMa and RMa are respectively reused. 
For the UE-UE link, the values of parameters for  of scenarios UMi are reused. 
For the TRP-UAV link, the values of parameters for  of scenarios UMi, UMa and RMa are respectively reused.
For the UE-UAV link, the values of parameters for  of scenarios UMi are reused.
For the UAV-UAV link, the values of parameters for  of scenarios UMi are reused.
Note: no measurements on  of the scenarios UMi-AV, UMa-AV and RMa-AV are submitted in Rel-19. 


Power normalization between target channel and background channel

[Moderator’s note] The original proposal is per majority view, however, no much supports are observed. Please note that we already have Option 1, i.e., no power normalization for the combined channel. If an agreement cannot be made for Option 2, we have to drop Option 2 from Rel-19. 

For reference, this is the list of supporting companies for each alternative
Alt 1: Power normalization on both target channel and background channel
Apple, Nokia, CATT, LGE, OPPO, QC, CT, Spreadtrum
Alt 2: Only the power of the background channel is scaled down to make total power normalized
EURECOM, ZTE, IDC, OPPO
Alt 3: the target channel of a target will replace one cluster in the background channel
BUPT (indoor/outdoor scenarios), Panasonic, vivo


[M] Proposal 5.6-1 
Power normalization applied on both target channel and background channel, i.e., , where  is such that  



[M] Proposal 5.6-1-rev1 
Please provide your preference 
Alt 1A
A power scaling factor  is applied on both target channel and background channel, subjected to

Where, 
 are the powers of target channel and background channel of one implementation/sample of the ISAC channel. 
Alt 1B
A power scaling factor  is applied on both target channel and background channel, 

Alt 1C
A power scaling factor  is applied on both target channel and background channel, 


[Moderator’s note] If Alt 1B/1C is agreeable, we need to run simulation to come up with value . For simplicity, we may only consider the large scale parameters, e.g., the power of the target channel can be estimated using power scaling factor 


On the other hand, if majority view of Alt 1 is not agreeable, we may end up with conclusion that power normal over target/background channel are not supported in Rel-19. 



[FL2][M] Proposal 5.6-1-rev2 
To do power normalization of target channel and background channel of ISAC channel
A power scaling factor  is applied on both target channel and background channel, subjected to

Where, 
 equals to the power scaling factor of target channel, i.e.,
.
 is the nlos power  calculated by the large scale parameters of background channel and the k factor, i.e., . 



[FL2][M] Proposal 5.6-1-rev3 
To do power normalization of target channel and background channel of ISAC channel
A power scaling factor  is applied on both target channel and background channel, such as the following holds:
If  , then

Otherwise,

Where, 
 equals to the power scaling factor of target channel, i.e.,
.
 is the power of the NLOS clusters calculated by the large scale parameters of background channel, i.e., . 


[Moderator’s note] Regarding ‘FFS condition to select option, e.g. depending on scenario, sensing mode, number of target/EO type-2’, there is a good consensus it is not necessary based on companies’ inputs in RAN1#120bis. 
Yes: 
No: ZTE, CATT, Apple, vivo, QC, SS, Ericsson, Xiaomi
[FL2][L] Proposal 5.6-2 
A condition is not specified to select between Option 1 ‘no power normalization’ and Option 2 ‘with power normalization’ in the combination of target channel and background channel. 


Doppler of moving scatters

[Moderator’s note] based on inputs from RAN1 #120bis, the maximum speed and ratio of moving scatters are scenario dependent and may be modelled by EO type-1. 

[M] Proposal 5.7-1 
Dual mobility model in 7.6.10, TR 38.901 is reused as start point to model Doppler effect  due to movement of stochastic clusters, i.e., 
Necessary changes on maximum speed and ratio of moving scatters can be discussed in evaluation phase. 


[Closed][FL1][M] Proposal 5.7-1-rev1 
Dual mobility model in 7.6.10, TR 38.901 is reused as start point to model Doppler effect  due to movement of stochastic clusters, i.e., 
Necessary changes on maximum speed and ratio of moving scatters can be discussed in evaluation phase. 


Blockage model A/B

[Moderator’s note] It is generally OK that an existing additional feature in existing R 38.901 can be reused for ISAC. One new point for this issue is to clarify the target can be considered as a blocker, which happens in the reality. Therefore, the moderator makes the following proposal. However, if it is not agreeable, the moderator suggest we could discuss such details of blockage model A/B in evaluation phase. 

[M] Proposal 5.8-1
The existing blockage model A/B is reused to model the blocking effect due to a target as an optional feature 
Applicable to the LOS/NLOS rays in the background channel of the target
Applicable to the LOS/NLOS rays in the Tx-target and target-Rx link of the other target


[FL2][M] Proposal 5.8-1-rev1
The existing blockage model A/B is reused to model the blocking effect due to a target as an optional feature 
Applicable to the LOS/NLOS rays in the background channel of the target
Applicable to the LOS/NLOS rays in the Tx-target and target-Rx link of the other target
Note: The location of the target as a blocker is known before applying the blockage model A/B.


Micro-Doppler

[Moderator’s note] A placeholder for micro-Doppler was agreed in RAN1 #119. If there is time, it is nice to discuss certain detailed functions to handle micro-Doppler. However, it is the moderator’s understanding that the discussion is mostly informative since we don’t know the exact interested use case/target in future evaluation. Decision on micro-Doppler in evaluation phase sounds better.  
[FL2][M] Question 5.9-1
Companies are encouraged to input on the proper functions/models for micro-Doppler


EO type-2 
EO type-2 in background channel 

[Moderator’s note] It is a long discussion whether EO type-2 can be modeled in background channel, but no conclusion yet. The following arguments are observed

From proponent
In the real propagation with target and wall (EO type-2) present, it is possible that the signal reflected by EO type-2 can be observed in both target channel and background channel. 
For bistatic sensing, if there is no LOS ray, specular reflected ray of EO type-2 in background channel may serve as reference for measurement of time difference
From opponent
GBSM channel in existing 38.901 is validated assuming possible presence of wall (i.e., EO type-2). If further model EO type-2 in background channel, it just increases the power and change the distribution of background channel comparing with existing channel for communication.

Based on discussion from RAN1 #120bis, 
Option 1: EO type-2 is modelled in background channel if modelled in target channel
Supported by (7): HW, LG, Lenovo, SS, ZTE, Sony, vivo(InH)
Option 2: EO type-2 is not modelled in background channel
Supported by (10): CATT, QC, vivo(others), SPRD, Ericsson, IDCC, DOCOMO, OPPO, Xiaomi, Nokia

I soft the wording of Option 1, let’s check if following proposal is agreeable. 
[FL2][H] Proposal 6.1-1 
EO type-2 can be modelled in background channel if modelled in target channel


LOS condition considering EO type-2

[Moderator’s note] This is a key assumption to model EO type-2. If we cannot agree on a solution, we may end up with removing the feature of EO type-2. 
Option A reflect the real channel propagation. If a LOS ray is blocked, it is NLOS condition. However, it changes the behavior in current TR
Option B binds EO type-2 with blockage model B, which is not desired by majority of companies. 
Option C is likely kind of compromise, which reuses existing solution for LOS condition determination by LOS probability and doesn’t tight EO type-2 with blockage mode. 

If the majority view of Option A cannot be agreeable, Option C may be the outcome. Without an agreed solution, EO type-2 may need to be removed from the Rel-19 the study item.
 Comments are welcome, especially whether or not a proponent of Option A can be fine with Option C. 

[FL2][H] Proposal 6.2-1 
Down-select in RAN1#121 one option from the following options to determine the LOS condition of the Tx-target link and target-Rx link?
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
Supported by (12): HW, ZTE, Sony, QC, LG, Xiaomi, MTK, Nokia, CMCC, Spreadtrum, OPPO, vivo
Option B: Use the LOS probability equation to determine the LOS/NLOS condition of one link, and then the impacts of type-2 EO is modelled by a blockage model
Supported by (6): Ericsson, CATT (without blockage model), Lenovo, IDCC, DOCOMO, BUPT
Option C: Use the LOS probability equation to determine the LOS/NLOS condition of one link,
Supported by (7): CATT, Ericsson, Nokia, Lenovo, DOCOMO, IDCC, 
Note: If EO type-2 is agreed to be modelled in background channel, the agreed option is extended to LOS condition determination for background channel. 


EO type-2 for a link in NLOS condition

[Moderator’s note] The existing ground reflection model in section 7.6.8, TR 38.901 is defined assuming LOS condition between Tx and Rx. This is the default behavior since we agreed to use section 7.6.8 of TR 38.901 as reference. Therefore, EO type-2 is at least supported in LOS condition. 

In last meeting, Huawei and ZTE discussed the issue and propose to solve it under NLOS condition. The method is to derive power of EO type-2 based on power of LOS ray temporally generated assuming LOS condition. If this is agreeable, we allow such enhancement. 

[FL1][M] Proposal 6.3-1 
The pathloss of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link is calculated by , where 
 is a pathloss assuming a LOS condition for the STX-SPST link or SPST-SRX link
 is the total propagation distance due to EO type-2



[FL1][M] Proposal 6.3-1-rev1 
The pathloss of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link is calculated by , where 
 is a pathloss assuming a LOS condition for the STX-SPST link or SPST-SRX link
 is the total propagation distance due to EO type-2


[Moderator’s note] Proposal revised in the online session and a related agreement is made

[FL1][M] Proposal 6.3-1-rev1 
The of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link, is calculated by , where 
 is the pathloss assuming a LOS condition for the STX-SPST link or SPST-SRX link
 is the total propagation distance due to EO type-2
Note: impact of polarization matrix of EO type-2 is not included in the above

Agreement
EO type-2 can be modelled in NLOS condition.


[Moderator’s note] Based on the discussion online, the moderator prepares a TP to capture power metric for EO type-2
[FL2][M] Proposal 6.3-2
The follow TP is used generate the power (except for the impact of polarization matrix of EO type-2) of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link. 




Whether/how to normalize the power of target channel, background channel or the combined channel if EO type-2 is modelled?

[Moderator’s note] In existing ground reflection model in section 7.6.8, TR 38.901, the power of the ground reflected ray is added to the generated fast fading channel without further power normalization. This is the default behavior since we agreed to use section 7.6.8 of TR 38.901 as reference. 

Based on discussion from RAN1 #120bis, all interested companies agree to reuse existing rule from ground reflection model in 7.6.8, TR 38.901. 
Supported: ZTE, LGE, HW, CATT, vivo, QC, BUPT, Ericsson, Xiaomi
Not supported: 

Therefore, the moderator suggest we don’t spend more time on this issue.  

Spatial consistency 
General question

[Moderator’s note] We agreed that path dropping after concatenation based on a power threshold. During movement of Tx/target/Rx, the power of each direct/indirect path will change. The set of paths with power higher than threshold change accordingly. However, it is time consuming if path dropping is redo every time when Tx/target/Rx moves, especially for Option 0 for concatenation.  

Just echo one comments from OPPO, the intention of proposal is to do path dropping only in the beginning of simulation, while the parameter values, e.g. absolute delay, for each path can be updated based on location of Tx, target and/or Rx

The following are summary on views from companies. Not sure 
Supported by: ZTE, CATT, LGE, vivo, QC, Xiaomi, OPPO
Not Supported by: Ericsson
Neutral: BUPT (neglectable impact)

Note: Cluster dropping with power -25 dB lower than maximum cluster power is supported in Step 6 in TR 38.901. however, whether to update the cluster dropping is not explicitly captured TR. The similar handling may be applicable for path dropping, if we don’t have time to clearly make an agreement. 
[Closed][FL1][L] Proposal 7.1-1 for conclusion
When spatial consistency is enabled, the path dropping after concatenation of Tx-target and target-Rx link is not reperformed even if Tx, target, Rx positions change during simulation.
Note: the large scale/small scale parameter values can still be updated according to the procedure of spatial consistency. 


Cases that spatial consistency are supported or not supported

[Moderator’s note] Based on the discussion in last meeting, it should be quite agreeable if the two links are generated using different parameter value sets of channel model. For background channel for TRP monostatic sensing, it is straightforward there is no spatial consistency for the virtual receivers/reference points of different TRPs. 
Based on the inputs from RAN1 #120bis, all received comments are supporting the proposal. 

[FL1][H] Proposal 7.2-1
Spatial consistency is not modelled for
the links that are generated referring to channel models with parameter values of different communication scenarios
E.g., between TRP-target/UT link in one scenario and target/UT-UT link in another scenario
the background channels for TRP monostatic sensing of different TRPs
Supported by: ZTE, CATT, HW, LGE, vivo, QC, BUPT, Xiaomi, OPPO, Ericsson (bullet 1)


Agreement
Spatial consistency is not modelled for
the links that are generated referring to channel models with parameter values of different communication scenarios
E.g., between TRP-target/UT link in one scenario and target/UT-UT link in another scenario
the background channels for TRP monostatic sensing of different TRPs


[Moderator’s note] For sensing scenario UMi, InH and InF, it is likely we use the channel model referring the same communication scenario to generate TRP-UT/ST link and UT-ST/UT link. However, as discussed by BUPT, the ASD/ZSD may still be different for the two links. Therefore, it is proposed to not model spatial consistency for the links. 
Based on the inputs from RAN1 #120bis, all received comments are supporting the proposal.

[FL1][H] Proposal 7.2-2
Spatial consistency is not modelled between TRP-target/UT link and target/UT-UT link for sensing scenario UMi, InH and InF
Supported by: CATT, LGE, vivo, QC, BUPT, Xiaomi, DOCOMO, OPPO, Huawei


Agreement
Spatial consistency is not modelled between TRP-target/UT link and target/UT-UT link for sensing scenario UMi, InH and InF.


[Moderator’s note] Assuming Proposal 7.2-1/2 are agreeable, one remaining issue is the spatial consistency between TRP-TRP link and other links with UT/ST in at least one end. With similar reason as Proposal 7.2-2, ASD/ZSD/ASA/ZSA for TRP-TRP link are set according to TRP statistics, while for other links with UT/ST, ASD/ZSD/ASA/ZSA of at least one end (i.e., the ST/UT) doesn’t follow TRP statistics. Therefore, the moderator makes the following proposal to not model spatial consistency for the links.

[FL1][H] Proposal 7.2-3
Spatial consistency is not modelled between TRP-TRP link and any other links for ISAC channel.


Agreement
Spatial consistency is not modelled between TRP-TRP link and any other links for ISAC channel.

Additional parameters for spatial consistency

[Moderator’s note] On background channel for UE monostatic sensing, we need a way to maintain spatial consistency of generated background channel during the movement of UE. Please check following proposal from ZTE. 

[H] Proposal 7.3-1
On background channel for UE monostatic sensing, spatial consistency is supported for the following parameters. 




[FL1][H] Proposal 7.3-1-rev1
On background channel for UE monostatic sensing, spatial consistency, if enabled, is supported for the following additional parameters. 




[FL1][H] Proposal 7.3-1-rev2 for conclusion
On background channel modelling,
Spatial consistency is not supported for TRP monostatic sensing across different TRPs 
Spatial consistency is not supported for UE monostatic sensing across different Ues
Additionally spatial consistency is not supported for different RPs for same TRP for TRP monostatic sensing
Additionally spatial consistency is not supported for different RPs for same UE for UE monostatic sensing


[Moderator’s note] Agreed in Tuesday online session
Agreement
On background channel modelling,
Spatial consistency is not supported for TRP monostatic sensing across different TRPs 
Spatial consistency is not supported for UE monostatic sensing across different UEs
Spatial consistency is not supported across different Reference Points for same TRP for TRP monostatic sensing
Spatial consistency is not supported across different Reference Points for same UE for UE monostatic sensing

Spatial consistency between multiple scattering points

Huawei
The XPR correlation between SPs can be modelled as the spatial consistency.

[Moderator’s note] different views are observed on the relation between multiple scattering point for a target and spatial consistency. 
Option 1: When spatial consistency is NOT implemented, channels of the multiple scattering points of a target is independently generated
Supported by: OPPO, QC, CATT, Spreadtrum, SS, CMCC, Nokia
Option 2: When multiple scattering points of target is modelled, spatial consistency is used
Supported by: Huawei, ZTE, vivo, BUPT, DOCOMO, Ericsson, LG, IDCC, 

It is quite controversial. However, either option doesn’t require new feature for ISAC channel model. I mean, the Rel-19 SI anyway supports both components, multiple scattering points and spatial consistency. The two options are essentially second level work to combine the two components. 

Since RAN1 already makes the two components and it seems hard to get a consensus on either option, the moderator suggests RAN1 to clarify the selected option in evaluation phase. 

[H] Proposal 7.4-1
Delete the following sentence from the Step 2 in section 7.9.4.1 in the draft CR on ISAC channel model to TR 38.901. 
“The propagation conditions for different STX-SPST links and SPST-SRX links are uncorrelated.”
Whether to enable spatial consistency when multiple scattering points for a target is used can be discussed in evaluation phase



[Closed][FL1][M] Proposal 7.4-1-rev1 for conclusion
Delete the following sentence from the Step 2 in section 7.9.4.1 in the draft CR on ISAC channel model to TR 38.901. 
“The propagation conditions for different STX-SPST links and SPST-SRX links are uncorrelated.”
Whether to enable spatial consistency when multiple scattering points for a target is used can be discussed in evaluation phase



[Moderator’s note] The agreement from RAN1 #120 is to separately generated LOS condition/pathloss and LOS ray/NLOS clusters for the links of multiple scattering points. From inputs in RAN1 #120bits, all companies are OK with model spatial consistency of the links between STX/SRX and the multiple scattering points. 

Our existing agreement on spatial consistency is applicable to links between STX/SRX and targets. Therefore, the following proposal is made. 

[FL1][H] Proposal 7.4-2
Spatial consistency, if enabled, for the links between BS/UT and multiple scattering points of a target are modelled as if multiple targets. 


Agreement
Spatial consistency can be enabled for multiple scattering points of a target. 
Spatial consistency, if enabled, for the links between BS/UT and multiple scattering points of a target are modelled as if multiple scattering points are multiple targets.

3D spatial consistency

[Moderator’s note] This is an essential feature to simulate 3D correlation when UAV can have a vertical speed. On the other hand, if there is not conclusion, we end up with limitation on UAV trajectory evaluation in evaluation phase. 

Based on discussion in last meeting, there is a good support to reuse existing correlation distance. I avoid using either vertical or 3D correlation but only capture the basic intention.
 
[H] Proposal 7.5-1
Option 1: The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused as vertical correlation distance for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’
Option 2: 3GPP assumes UAV height cannot change in a simulation drop, if spatial consistency is enabled.
Option 3: The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused to model 3D spatial consistency for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’



[FL1] [H] Proposal 7.5-1-rev1
The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is used as the correlation distance for 3D spatial consistency for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’


Agreement
The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is used as the correlation distance for 3D spatial consistency for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’.

Solutions for new spatial consistency
[Moderator’s note] We don’t have time to align the understanding on exact solution to model spatial consistency for ST/UT links. However, it is always possible two or many may align their views offline. 

Procedure A vs. B for spatial consistency

[Moderator’s note] Based on inputs in RAN1 #120bis, all companies agree that procedure A/B are both inherited, with one company commented it is the default behaviour and no need for an agreement. With such understanding, we don’t need to spend time on this issue again. 

Link level channel model
Summary on company views

Option 1: Based on existing CDL channel model: ZTE, CATT, vivo, DOCOMO
Alt 1: add one or more clusters representing target(s): HW, Apple, SK Telecom, BUPT, Xiaomi, QC

Alt 2: Update the parameters of one or more cluster to represent target(s)

Option 2: Reducing the system level channel model for ISAC to single Tx and single Rx: CMCC, MTK
Option 3: The Tx-target link and target-Rx link are modelled by separate CDL channels, concatenation is then performed to get the target channel; the background channel is modelled by another CDL channel. 
Ericsson
Option 4: Not supported: QC

Parameter values for the target cluster to be decided in evaluation stage: Huawei
certain modification, e.g., the delay, angle, velocity or doppler parameter to the target cluster: vivo

CDL only: Ericsson
CDL & TDL: 

Ericsson: For link-level ISAC channel models,
H_ISAC=H_target+H_background still holds.
legacy CDL models are reused to model Tx-target and target-Rx links of target channel and Tx-Rx channel of background channel.
target channel is generated by concatenating Tx-target and target-Rx link

[Moderator’s note] For better understanding, please proponent companies clarify the 3 options. 
[FL1][M] Question 8-1
For Option 1, please clarify whether/how to apply RCS of a target. Comments are other details are welcome too. 


[FL1][M] Question 8-2
For Option 2 if preferred by any company, please clarify 
How to generate the location/distance of the Tx and Rx in link level simulation?
Any limitation on the relative location target and Tx/Rx?
Comments are other details are welcome too.


[FL1][M] Question 8-3
For Option 3, please clarify 
how to control the power of target channel by concatenating two CDL channels?
Do you prefer to allow both LOS condition and NLOS condition for each CDL channel?
Comments are other details are welcome too.


[Moderator’s note] Hope the early questions are helpful to align companies’ understandings. 
Due to limited time, suggest to agree on the principle in Rel-19. Given the diverse purpose/scenario for simulation, suggest to keep exact parameter values open and discuss them in evaluation phase. 
With them, can we agree on the following proposal from RAN1 #120bis?

[M] Proposal 3-4
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing CDL channel model
Values of parameters of the added cluster(s) can be discussed in performance evaluation stage
TDL channel model from exiting 38.901 is not enhanced for ISAC channel model


[M] Proposal 3-4-rev1
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing CDL LLS channel model
A new parameter of velocity of the cluster(s) of target(s) is added
Values of parameters of the added cluster(s) of target(s) can be discussed in performance evaluation stage
TDL channel model from exiting 38.901 is not enhanced for ISAC channel model



[M] Proposal 3-4-rev2
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing CDL LLS channel model
A new parameter of velocity is added for the cluster(s) of target(s), and if necessary, for at least some of the existing clusters in existing LLS channel model. 
Values of parameters of the added cluster(s) of target(s) can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)
TDL channel model from exiting 38.901 is not enhanced for ISAC channel model



[FL1][M] Proposal 8-4-rev3
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model
A new parameter of velocity is added for the cluster(s) of target(s), and if necessary, for at least some of the existing cluster(s) in existing LLS channel model. 
Values of parameters of the added cluster(s) of target(s), and if necessary, the velocity of at least some of the existing cluster(s), can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)


[FL2][M] Proposal 8-4-rev4
Down select one from the following 3 options in RAN1 #121
Option 1: 
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model
A parameter of velocity is added for the cluster(s) of target(s), and if necessary, for at least some of the existing cluster(s) in existing LLS channel model. 
Values of parameters of the added cluster(s) of target(s), and if necessary, the velocity of at least some of the existing cluster(s), can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)

Option 2: 
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model
A parameter of velocity is added for the cluster(s) of target(s), and if necessary, use dual mobility model in section 7.6.10 in TR 38.901 to model Doppler of the moving scatters. 
Values of parameters of the added cluster(s) of target(s), and if necessary, the parameters for the moving scatters, can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)

Option 3: LLS channel model is not supported in Rel-19.


Hybrid channel model
[FL1][L] Question 9.1-1  
Please provide your view whether/how ISAC channel based on map-based hybrid channel model should be supported I Rel-19. If so, how to achieve it? 

3GPP TSG RAN WG1 #121			 R1-2504164
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda item:	9.7.2
Source:	Moderator (Xiaomi)
Title:	Summary #3 on ISAC channel modelling
Document for:	Discussion/Decision

Conclusion
No further study on power normalization of target channel and background channel of ISAC channel in Rel-19
Note: sub-section “7.9.5.3 Power normalization across target channel and background channel” in the TR remains as a placeholder with the following text.
To combine the target channel and the background channel, an alternative scheme of power normalization can be applied to keep the same/similar channel power as the background channel without sensing target.



[Moderator’s note] Regarding ‘FFS condition to select option, e.g. depending on scenario, sensing mode, number of target/EO type-2’, there is a good consensus it is not necessary based on companies’ inputs in RAN1#120bis. 
Yes: 
No: ZTE, CATT, Apple, vivo, QC, SS, Ericsson, Xiaomi
[Closed][FL2][L] Proposal 5.6-2 
A condition is not specified to select between Option 1 ‘no power normalization’ and Option 2 ‘with power normalization’ in the combination of target channel and background channel. 


Doppler of moving scatters

[Moderator’s note] based on inputs from RAN1 #120bis, the maximum speed and ratio of moving scatters are scenario dependent and may be modelled by EO type-1. 

[M] Proposal 5.7-1 
Dual mobility model in 7.6.10, TR 38.901 is reused as start point to model Doppler effect  due to movement of stochastic clusters, i.e., 
Necessary changes on maximum speed and ratio of moving scatters can be discussed in evaluation phase. 


[Closed][FL1][M] Proposal 5.7-1-rev1 
Dual mobility model in 7.6.10, TR 38.901 is reused as start point to model Doppler effect  due to movement of stochastic clusters, i.e., 
Necessary changes on maximum speed and ratio of moving scatters can be discussed in evaluation phase. 


Blockage model A/B

[Moderator’s note] It is generally OK that an existing additional feature in existing R 38.901 can be reused for ISAC. One new point for this issue is to clarify the target can be considered as a blocker, which happens in the reality. Therefore, the moderator makes the following proposal. However, if it is not agreeable, the moderator suggest we could discuss such details of blockage model A/B in evaluation phase. 

[M] Proposal 5.8-1
The existing blockage model A/B is reused to model the blocking effect due to a target as an optional feature 
Applicable to the LOS/NLOS rays in the background channel of the target
Applicable to the LOS/NLOS rays in the Tx-target and target-Rx link of the other target



[FL2][M] Proposal 5.8-1-rev1
The existing blockage model A/B is reused to model the blocking effect due to a target as an optional feature 
Applicable to the LOS/NLOS rays in the background channel of the target
Applicable to the LOS/NLOS rays in the Tx-target and target-Rx link of the other target
Note: The location of the target as a blocker is known before applying the blockage model A/B.


[FL4][M] Proposal 5.8-1-rev2
The existing blockage model A/B is reused to model the blocking effect due to a target as an optional feature 
Applicable to the LOS/NLOS rays in the background channel of the target
Applicable to the LOS/NLOS rays in the Tx-target and target-Rx link of the other target
Note: The location, orientation and size of the target as a blocker is known before applying the blockage model A/B.


Micro-Doppler

[Moderator’s note] A placeholder for micro-Doppler was agreed in RAN1 #119. If there is time, it is nice to discuss certain detailed functions to handle micro-Doppler. However, it is the moderator’s understanding that the discussion is mostly informative since we don’t know the exact interested use case/target in future evaluation. Decision on micro-Doppler in evaluation phase sounds better.  

[FL3][M] Question 5.9-1
Companies are encouraged to input on the proper functions/models for micro-Doppler



[FL4][M] Proposal 5.9-1-rev1 
Exact formula for micro-Doppler is not further studied in Rel-19. 


EO type-2 
EO type-2 in background channel 

[Moderator’s note] It is a long discussion whether EO type-2 can be modeled in background channel, but no conclusion yet. The following arguments are observed

From proponent
In the real propagation with target and wall (EO type-2) present, it is possible that the signal reflected by EO type-2 can be observed in both target channel and background channel. 
For bistatic sensing, if there is no LOS ray, specular reflected ray of EO type-2 in background channel may serve as reference for measurement of time difference
From opponent
GBSM channel in existing 38.901 is validated assuming possible presence of wall (i.e., EO type-2). If further model EO type-2 in background channel, it just increases the power and change the distribution of background channel comparing with existing channel for communication.

Based on discussion from RAN1 #120bis, 
Option 1: EO type-2 is modelled in background channel if modelled in target channel
Supported by (7): HW, LG, Lenovo, SS, ZTE, Sony, vivo(InH)
Option 2: EO type-2 is not modelled in background channel
Supported by (10): CATT, QC, vivo(others), SPRD, Ericsson, IDCC, DOCOMO, OPPO, Xiaomi, Nokia

I soft the wording of Option 1, let’s check if following proposal is agreeable. 
[FL4][H] Proposal 6.1-1 
EO type-2 can be modelled in background channel if modelled in target channel


LOS condition considering EO type-2

[Moderator’s note] This is a key assumption to model EO type-2. If we cannot agree on a solution, we may end up with removing the feature of EO type-2. 
Option A reflect the real channel propagation. If a LOS ray is blocked, it is NLOS condition. However, it changes the behavior in current TR
Option B binds EO type-2 with blockage model B, which is not desired by majority of companies. 
Option C is likely kind of compromise, which reuses existing solution for LOS condition determination by LOS probability and doesn’t tight EO type-2 with blockage mode. 

If the majority view of Option A cannot be agreeable, Option C may be the outcome. Without an agreed solution, EO type-2 may need to be removed from the Rel-19 the study item.
 Comments are welcome, especially whether or not a proponent of Option A can be fine with Option C. 

[FL2][H] Proposal 6.2-1 
Down-select in RAN1#121 one option from the following options to determine the LOS condition of the Tx-target link and target-Rx link?
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
Supported by (12): HW, ZTE, Sony, QC, LG, Xiaomi, MTK, Nokia, CMCC, Spreadtrum, OPPO, vivo
Option B: Use the LOS probability equation to determine the LOS/NLOS condition of one link, and then the impacts of type-2 EO is modelled by a blockage model
Supported by (6): Ericsson, CATT (without blockage model), Lenovo, IDCC, DOCOMO, BUPT
Option C: Use the LOS probability equation to determine the LOS/NLOS condition of one link,
Supported by (7): CATT, Ericsson, Nokia, Lenovo, DOCOMO, IDCC, 
Note: If EO type-2 is agreed to be modelled in background channel, the agreed option is extended to LOS condition determination for background channel. 



[FL2][H] Proposal 6.2-1-rev1
Down selection among one from the following 3 options in RAN1 #121 

Option 1: 
To determine the LOS condition of the Tx-target link and target-Rx link, 
Option A: If type-2 EO is in the LOS ray of one link, the NLOS probability is 100%, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
 Supported by QC, HW

Option 2: 
To determine the LOS condition of the Tx-target link and target-Rx link, 
For urban grid scenario modelling EO type-2: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
For other scenarios:  The LOS probability equation defined in existing TRs to determine the LOS/NLOS condition. 
This can be revisited and updated when necessary in the evaluation phase when the details of EO type-2 deployment is settled. 
Supported by HW 

Option 3: 
To determine the LOS condition of the Tx-target link and target-Rx link, the following two options are agreed as candidate solutions:
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
Option C: Use the LOS probability equation to determine the LOS/NLOS condition of one link.
Supported by CATT

Note: Common to all 3 options, if EO type-2 is agreed to be modelled in background channel, the agreed option is extended to LOS condition determination for background channel when EO type-2 is present. 


[FL4][H] Proposal 6.2-1-rev2
Down selection among one from the following options in RAN1 #121 

Option 2: 
To determine the LOS condition of the Tx-target link and target-Rx link, when EO type-2 is modelled, 
For urban grid scenario: If type-2 EO is in the LOS ray of one link, the LOS probability is p, p=0, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
For other scenarios: The LOS probability equation defined in existing TRs to determine the LOS/NLOS condition. 
This can be revisited and updated when necessary in the evaluation phase when the details of EO type-2 deployment is settled. 
Supported by HW, QC, ZTE 

Option 3: 
To determine the LOS condition of the Tx-target link and target-Rx link, when EO type-2 is modelled, the following two options are agreed as candidate solutions:
Option A: If type-2 EO is in the LOS ray of one link, the link is determined as NLOS condition, and otherwise use the LOS probability equation defined in existing TRs to determine the LOS/NLOS condition
Option C: Use the LOS probability equation to determine the LOS/NLOS condition of one link.
Supported by CATT

Option 4: 
Support legacy stochastic LOS/NLOS condition of Type-2 EO in section 7.9.
Support deterministic NLOS condition of Type-2 EO, in the section of map-based hybrid channel of 38.901.
Supported by Ericsson


EO type-2 for a link in NLOS condition

[Moderator’s note] The existing ground reflection model in section 7.6.8, TR 38.901 is defined assuming LOS condition between Tx and Rx. This is the default behavior since we agreed to use section 7.6.8 of TR 38.901 as reference. Therefore, EO type-2 is at least supported in LOS condition. 

In last meeting, Huawei and ZTE discussed the issue and propose to solve it under NLOS condition. The method is to derive power of EO type-2 based on power of LOS ray temporally generated assuming LOS condition. If this is agreeable, we allow such enhancement. 

[FL1][M] Proposal 6.3-1 
The pathloss of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link is calculated by , where 
 is a pathloss assuming a LOS condition for the STX-SPST link or SPST-SRX link
 is the total propagation distance due to EO type-2



[FL1][M] Proposal 6.3-1-rev1 
The pathloss of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link is calculated by , where 
 is a pathloss assuming a LOS condition for the STX-SPST link or SPST-SRX link
 is the total propagation distance due to EO type-2


[Moderator’s note] Proposal revised in the online session and a related agreement is made

[FL1][M] Proposal 6.3-1-rev1 
The of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link, is calculated by , where 
 is the pathloss assuming a LOS condition for the STX-SPST link or SPST-SRX link
 is the total propagation distance due to EO type-2
Note: impact of polarization matrix of EO type-2 is not included in the above

Agreement
EO type-2 can be modelled in NLOS condition.


[Moderator’s note] Based on the discussion online, the moderator prepares a TP to capture power metric for EO type-2

[FL2][M] Proposal 6.3-2
The follow TP is used generate the power (except for the impact of polarization matrix of EO type-2) of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link. 


[Moderator’s note] Agreed in Wednesday online session
Agreement
The follow TP is used generate the power (except for the impact of polarization matrix of EO type-2) of the ray specular reflected by an EO type 2 in the STX-SPST link or SPST-SRX link. 

Whether/how to normalize the power of target channel, background channel or the combined channel if EO type-2 is modelled?

[Moderator’s note] In existing ground reflection model in section 7.6.8, TR 38.901, the power of the ground reflected ray is added to the generated fast fading channel without further power normalization. This is the default behavior since we agreed to use section 7.6.8 of TR 38.901 as reference. 

Based on discussion from RAN1 #120bis, all interested companies agree to reuse existing rule from ground reflection model in 7.6.8, TR 38.901. 
Supported: ZTE, LGE, HW, CATT, vivo, QC, BUPT, Ericsson, Xiaomi
Not supported: 

Therefore, the moderator suggest we don’t spend more time on this issue.  

Spatial consistency 
General question

[Moderator’s note] We agreed that path dropping after concatenation based on a power threshold. During movement of Tx/target/Rx, the power of each direct/indirect path will change. The set of paths with power higher than threshold change accordingly. However, it is time consuming if path dropping is redo every time when Tx/target/Rx moves, especially for Option 0 for concatenation.  

Just echo one comments from OPPO, the intention of proposal is to do path dropping only in the beginning of simulation, while the parameter values, e.g. absolute delay, for each path can be updated based on location of Tx, target and/or Rx

The following are summary on views from companies. Not sure 
Supported by: ZTE, CATT, LGE, vivo, QC, Xiaomi, OPPO
Not Supported by: Ericsson
Neutral: BUPT (neglectable impact)

Note: Cluster dropping with power -25 dB lower than maximum cluster power is supported in Step 6 in TR 38.901. however, whether to update the cluster dropping is not explicitly captured TR. The similar handling may be applicable for path dropping, if we don’t have time to clearly make an agreement. 
[Closed][FL1][L] Proposal 7.1-1 for conclusion
When spatial consistency is enabled, the path dropping after concatenation of Tx-target and target-Rx link is not reperformed even if Tx, target, Rx positions change during simulation.
Note: the large scale/small scale parameter values can still be updated according to the procedure of spatial consistency. 


Cases that spatial consistency are supported or not supported

[Moderator’s note] Based on the discussion in last meeting, it should be quite agreeable if the two links are generated using different parameter value sets of channel model. For background channel for TRP monostatic sensing, it is straightforward there is no spatial consistency for the virtual receivers/reference points of different TRPs. 
Based on the inputs from RAN1 #120bis, all received comments are supporting the proposal. 

[FL1][H] Proposal 7.2-1
Spatial consistency is not modelled for
the links that are generated referring to channel models with parameter values of different communication scenarios
E.g., between TRP-target/UT link in one scenario and target/UT-UT link in another scenario
the background channels for TRP monostatic sensing of different TRPs
Supported by: ZTE, CATT, HW, LGE, vivo, QC, BUPT, Xiaomi, OPPO, Ericsson (bullet 1)


Agreement
Spatial consistency is not modelled for
the links that are generated referring to channel models with parameter values of different communication scenarios
E.g., between TRP-target/UT link in one scenario and target/UT-UT link in another scenario
the background channels for TRP monostatic sensing of different TRPs


[Moderator’s note] For sensing scenario UMi, InH and InF, it is likely we use the channel model referring the same communication scenario to generate TRP-UT/ST link and UT-ST/UT link. However, as discussed by BUPT, the ASD/ZSD may still be different for the two links. Therefore, it is proposed to not model spatial consistency for the links. 
Based on the inputs from RAN1 #120bis, all received comments are supporting the proposal.

[FL1][H] Proposal 7.2-2
Spatial consistency is not modelled between TRP-target/UT link and target/UT-UT link for sensing scenario UMi, InH and InF
Supported by: CATT, LGE, vivo, QC, BUPT, Xiaomi, DOCOMO, OPPO, Huawei


Agreement
Spatial consistency is not modelled between TRP-target/UT link and target/UT-UT link for sensing scenario UMi, InH and InF.


[Moderator’s note] Assuming Proposal 7.2-1/2 are agreeable, one remaining issue is the spatial consistency between TRP-TRP link and other links with UT/ST in at least one end. With similar reason as Proposal 7.2-2, ASD/ZSD/ASA/ZSA for TRP-TRP link are set according to TRP statistics, while for other links with UT/ST, ASD/ZSD/ASA/ZSA of at least one end (i.e., the ST/UT) doesn’t follow TRP statistics. Therefore, the moderator makes the following proposal to not model spatial consistency for the links.

[FL1][H] Proposal 7.2-3
Spatial consistency is not modelled between TRP-TRP link and any other links for ISAC channel.


Agreement
Spatial consistency is not modelled between TRP-TRP link and any other links for ISAC channel.

Additional parameters for spatial consistency

[Moderator’s note] On background channel for UE monostatic sensing, we need a way to maintain spatial consistency of generated background channel during the movement of UE. Please check following proposal from ZTE. 

[H] Proposal 7.3-1
On background channel for UE monostatic sensing, spatial consistency is supported for the following parameters. 




[FL1][H] Proposal 7.3-1-rev1
On background channel for UE monostatic sensing, spatial consistency, if enabled, is supported for the following additional parameters. 




[FL1][H] Proposal 7.3-1-rev2 for conclusion
On background channel modelling,
Spatial consistency is not supported for TRP monostatic sensing across different TRPs 
Spatial consistency is not supported for UE monostatic sensing across different Ues
Additionally spatial consistency is not supported for different RPs for same TRP for TRP monostatic sensing
Additionally spatial consistency is not supported for different RPs for same UE for UE monostatic sensing


[Moderator’s note] Agreed in Tuesday online session
Agreement
On background channel modelling,
Spatial consistency is not supported for TRP monostatic sensing across different TRPs 
Spatial consistency is not supported for UE monostatic sensing across different UEs
Spatial consistency is not supported across different Reference Points for same TRP for TRP monostatic sensing
Spatial consistency is not supported across different Reference Points for same UE for UE monostatic sensing

Spatial consistency between multiple scattering points

Huawei
The XPR correlation between SPs can be modelled as the spatial consistency.

[Moderator’s note] different views are observed on the relation between multiple scattering point for a target and spatial consistency. 
Option 1: When spatial consistency is NOT implemented, channels of the multiple scattering points of a target is independently generated
Supported by: OPPO, QC, CATT, Spreadtrum, SS, CMCC, Nokia
Option 2: When multiple scattering points of target is modelled, spatial consistency is used
Supported by: Huawei, ZTE, vivo, BUPT, DOCOMO, Ericsson, LG, IDCC, 

It is quite controversial. However, either option doesn’t require new feature for ISAC channel model. I mean, the Rel-19 SI anyway supports both components, multiple scattering points and spatial consistency. The two options are essentially second level work to combine the two components. 

Since RAN1 already makes the two components and it seems hard to get a consensus on either option, the moderator suggests RAN1 to clarify the selected option in evaluation phase. 

[H] Proposal 7.4-1
Delete the following sentence from the Step 2 in section 7.9.4.1 in the draft CR on ISAC channel model to TR 38.901. 
“The propagation conditions for different STX-SPST links and SPST-SRX links are uncorrelated.”
Whether to enable spatial consistency when multiple scattering points for a target is used can be discussed in evaluation phase



[Closed][FL1][M] Proposal 7.4-1-rev1 for conclusion
Delete the following sentence from the Step 2 in section 7.9.4.1 in the draft CR on ISAC channel model to TR 38.901. 
“The propagation conditions for different STX-SPST links and SPST-SRX links are uncorrelated.”
Whether to enable spatial consistency when multiple scattering points for a target is used can be discussed in evaluation phase



[Moderator’s note] The agreement from RAN1 #120 is to separately generated LOS condition/pathloss and LOS ray/NLOS clusters for the links of multiple scattering points. From inputs in RAN1 #120bits, all companies are OK with model spatial consistency of the links between STX/SRX and the multiple scattering points. 

Our existing agreement on spatial consistency is applicable to links between STX/SRX and targets. Therefore, the following proposal is made. 

[FL1][H] Proposal 7.4-2
Spatial consistency, if enabled, for the links between BS/UT and multiple scattering points of a target are modelled as if multiple targets. 


Agreement
Spatial consistency can be enabled for multiple scattering points of a target. 
Spatial consistency, if enabled, for the links between BS/UT and multiple scattering points of a target are modelled as if multiple scattering points are multiple targets.

3D spatial consistency

[Moderator’s note] This is an essential feature to simulate 3D correlation when UAV can have a vertical speed. On the other hand, if there is not conclusion, we end up with limitation on UAV trajectory evaluation in evaluation phase. 

Based on discussion in last meeting, there is a good support to reuse existing correlation distance. I avoid using either vertical or 3D correlation but only capture the basic intention.
 
[H] Proposal 7.5-1
Option 1: The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused as vertical correlation distance for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’
Option 2: 3GPP assumes UAV height cannot change in a simulation drop, if spatial consistency is enabled.
Option 3: The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused to model 3D spatial consistency for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’



[FL1] [H] Proposal 7.5-1-rev1
The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is used as the correlation distance for 3D spatial consistency for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’


Agreement
The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is used as the correlation distance for 3D spatial consistency for ISAC channel at least for UAV scenario, within same ‘Applicability range in terms of aerial UE height (defined in 36.777)’.

Solutions for new spatial consistency
[Moderator’s note] We don’t have time to align the understanding on exact solution to model spatial consistency for ST/UT links. However, it is always possible two or many may align their views offline. 

Procedure A vs. B for spatial consistency

[Moderator’s note] Based on inputs in RAN1 #120bis, all companies agree that procedure A/B are both inherited, with one company commented it is the default behaviour and no need for an agreement. With such understanding, we don’t need to spend time on this issue again. 

Link level channel model
Summary on company views

Option 1: Based on existing CDL channel model: ZTE, CATT, vivo, DOCOMO
Alt 1: add one or more clusters representing target(s): HW, Apple, SK Telecom, BUPT, Xiaomi, QC

Alt 2: Update the parameters of one or more cluster to represent target(s)

Option 2: Reducing the system level channel model for ISAC to single Tx and single Rx: CMCC, MTK
Option 3: The Tx-target link and target-Rx link are modelled by separate CDL channels, concatenation is then performed to get the target channel; the background channel is modelled by another CDL channel. 
Ericsson
Option 4: Not supported: QC

Parameter values for the target cluster to be decided in evaluation stage: Huawei
certain modification, e.g., the delay, angle, velocity or doppler parameter to the target cluster: vivo

CDL only: Ericsson
CDL & TDL: 

Ericsson: For link-level ISAC channel models,
H_ISAC=H_target+H_background still holds.
legacy CDL models are reused to model Tx-target and target-Rx links of target channel and Tx-Rx channel of background channel.
target channel is generated by concatenating Tx-target and target-Rx link

[Moderator’s note] For better understanding, please proponent companies clarify the 3 options. 
[FL1][M] Question 8-1
For Option 1, please clarify whether/how to apply RCS of a target. Comments are other details are welcome too. 


[FL1][M] Question 8-2
For Option 2 if preferred by any company, please clarify 
How to generate the location/distance of the Tx and Rx in link level simulation?
Any limitation on the relative location target and Tx/Rx?
Comments are other details are welcome too.


[FL1][M] Question 8-3
For Option 3, please clarify 
how to control the power of target channel by concatenating two CDL channels?
Do you prefer to allow both LOS condition and NLOS condition for each CDL channel?
Comments are other details are welcome too.


[Moderator’s note] Hope the early questions are helpful to align companies’ understandings. 
Due to limited time, suggest to agree on the principle in Rel-19. Given the diverse purpose/scenario for simulation, suggest to keep exact parameter values open and discuss them in evaluation phase. 
With them, can we agree on the following proposal from RAN1 #120bis?

[M] Proposal 3-4
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing CDL channel model
Values of parameters of the added cluster(s) can be discussed in performance evaluation stage
TDL channel model from exiting 38.901 is not enhanced for ISAC channel model


[M] Proposal 3-4-rev1
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing CDL LLS channel model
A new parameter of velocity of the cluster(s) of target(s) is added
Values of parameters of the added cluster(s) of target(s) can be discussed in performance evaluation stage
TDL channel model from exiting 38.901 is not enhanced for ISAC channel model



[M] Proposal 3-4-rev2
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing CDL LLS channel model
A new parameter of velocity is added for the cluster(s) of target(s), and if necessary, for at least some of the existing clusters in existing LLS channel model. 
Values of parameters of the added cluster(s) of target(s) can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)
TDL channel model from exiting 38.901 is not enhanced for ISAC channel model



[FL1][M] Proposal 8-4-rev3
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model
A new parameter of velocity is added for the cluster(s) of target(s), and if necessary, for at least some of the existing cluster(s) in existing LLS channel model. 
Values of parameters of the added cluster(s) of target(s), and if necessary, the velocity of at least some of the existing cluster(s), can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)



[FL3][M] Proposal 8-4-rev4
Down select one from the following 3 options in RAN1 #121
Option 1: 
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model
A parameter of velocity is added for the cluster(s) of target(s), and if necessary, for at least some of the existing cluster(s) in existing LLS channel model. 
Values of parameters of the added cluster(s) of target(s), and if necessary, the velocity of at least some of the existing cluster(s), can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)

Option 2: 
The link-level channel for ISAC is generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model
A parameter of velocity is added for the cluster(s) of target(s), and if necessary, use dual mobility model in section 7.6.10 in TR 38.901 to model Doppler of the moving scatters. 
Values of parameters of the added cluster(s) of target(s), and if necessary, the parameters for the moving scatters, can be discussed in performance evaluation stage
Note: the effect of RCS of target is reflected by the power of the cluster(s) of target(s)

Option 3: LLS channel model is not supported in Rel-19.

Option 4: add a general sentence as follows in subsection 7.9.6 on LLS channel model for ISAC
The link-level channel for ISAC can be generated by adding one or more cluster(s) representing target(s) to the existing LLS channel model.


[FL4][M] Proposal 8-4-rev5
Down select one from the following 2 options in RAN1 #121
Option 3: LLS channel model is not supported in Rel-19.

Option 4: 
add a general sentence as follows in subsection 7.9.6 on LLS channel model for ISAC
The target channel of link-level channel for ISAC can be generated by adding and/or reusing one or more cluster(s) representing target(s) in the existing LLS channel model.
No further study on LLS channel model for ISAC in Rel-19. 
Supported: ZTE, Huawei, SKT, Apple, Xiaomi, QC, CATT, CMCC, LGE, MTK, NIST, DOCOMO, Sony


Hybrid channel model
[FL4][L] Question 9.1-1  
Please provide your view whether/how ISAC channel based on map-based hybrid channel model should be supported I Rel-19. If so, how to achieve it? 

3GPP TSG RAN WG1 #121			 R1-2504165
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda item:	9.7.2
Source:	Moderator (Xiaomi)
Title:	Summary #5 on ISAC channel modelling
Document for:	Discussion/Decision

Conclusion (proposed online)
Delete subsection 7.9.6 from the draft CR. For ISAC, no enhancement to LLS channel model is introduced in Rel-19.

Based on further offline discussion, the following wording may be acceptable. 
[FL5][M] Proposal 8-4-rev6 for conclusion
Delete subsection 7.9.6 from the draft CR. For ISAC, no enhancement to existing LLS channel model is introduced in Rel-19.

Hybrid channel model
[Closed][FL5][L] Question 9.1-1  
Please provide your view whether/how ISAC channel based on map-based hybrid channel model should be supported I Rel-19. If so, how to achieve it? 

TDoc file unavailable

3GPP TSG RAN WG1 #121                                                                                  R1-2504221
St Julian’s, Malta, May 19th – 23rd, 2025

Source:	OPPO
Title:	Study on ISAC channel modelling
Agenda Item:	9.7.2
Document for:	Discussion and Decision

Conclusions
This contribution is concluded with the following observations and proposals:
Proposal 1: The mono-static RCS (model-2) of a standing human body is dependent on both zenith angle and azimuth angle, and is modeled as 

Where,


with parameters defined as:

Proposal 2: The bi-static RCS of a standing human body is modeled as following 

where
k1=2.75 and k2=1 
The angles () represent the incident angle and the angles () represent the scattering angle. 
 is the mono-static RCS model-2 formula, and the angles () are the projections of the bisector angle on the vertical plane and the horizontal plane, respectively.
 is the bistatic angle between the incident ray and scattering ray. 

Proposal 3: Maximum speed and ratio of moving scatters depends on scenarios. 
Speed of 95%-100% moving scatters follows uniform distribution between 0 and 180km/h for UAV case 
Speed of 95%-100% moving scatters is 0 for indoor case 
Speed of 50% moving scatters is 3-60km/h and speed of remaining moving scatters is 0 for UMa/UMi.

Proposal 4: The power normalization in combining the target channel(s) and background channel is formulated as, where
 for  are power of K target channels between STX and SRX
There are two options on 
Option 1: , i.e., the power over both LOS ray (if exist) and NLOS rays in background channel. In this option, the K-factor for the whole STX-SRX channel may change due to power normalization. 
Option 2: , i.e., the power over NLOS rays only in background channel.  In this option, the K-factor (KR,bg) in the whole STX-SRX channel remains unchanged due to power normalization.
The power normalization coefficient for the background channel is , where =0.1 is adopted.
The power normalization coefficient for target channels is  

3GPP TSG RAN WG1 #121                                                                          R1-2504240
St Julian's, Malta, 19 - 23 May, 2025
Agenda Item:	9.7.2
Source:	Lenovo
Title:	Discussion on Channel Modelling for ISAC
Document for:	Discussion and decision

Conclusion
The list of the main observations and proposals within this paper are summarized as following: 




Proposal 1. The Initial background channel is defined as the propagation channel between the sensing Tx node and the sensing Rx node in the absence of the sensing targets.
Observation  1. Considering the sensing target object or EOs with pre-determined characteristics as an integral part of the background environment, as proposed in Option 1, invalidates the statistical modelling of [38.901, Subsection 7.5]. For example, assuming an AGV with a known position within the InF scenario, leads to a conditional channel statistics, and deviates the scenario from the initial InF statistics. 
Observation  2. Considering the sensing target object or EOs as an external add-on to an initial background/environment channel, as in Option 2, facilitates utilizing [38.901, Subsection 7.5] for the initial background/environment channel generation. However, as the target/EO is not an integral part of the background environment, the mutual impact (e.g., blockage) of the sensing target and environment needs to be further taken into account. 
Proposal 2. Consider the sensing target (and if considered, EOs) as an external add-on object to an initial background/environment channel, where no additional power normalization is needed between the background and target channel components. 
Proposal 3. Any methodology including a power normalization should explicitly define the modelled scenario to avoid confusion
Observation 3.The paths of the initial background channel, i.e., the channel without targets presence, include both paths of the background/environment channel, when the said paths are not impacted/blocked by the target, or paths of the target channel, when the sensing target collides with a path of the initial background channel.
Proposal 4. A high-level procedure as described in the following figure can be utilized to generate the ISAC channel by first generating the paths of the initial background channel and subsequently introducing the impact (e.g., blockage, additional single or higher order target reflections) of the sensing targets.   

Observation  4. The “Blockage model B” of [38.901, Subsection 7.6.4] provides a geometrical modelling for blockage of a background/environment channel by an object via a knife-edge diffraction model, where a determined attenuation is applied on the blocked ray. Other than the energy attenuation, the ray parameters (e.g., angle, delay) are not impacted in the proposed “Blockage model B”. 
Observation  5. When a propagation path is blocked by a target, both the “Blockage model B” of [38.901, Subsection 7.6.4] as well as the bistatic RCS framework predict the received power at a receiver. 
Proposal 5. The bistatic RCS framework shall remain consistent with the existing blockage modelling in TR 38.901, in terms of the resulting modelled Rx power.  
Proposal 6. When a propagation path is blocked by a target, the “Blockage model B” is utilized to determine the received energy associated with the impacted path by the target. 
Proposal 7. When knowledge of the EOs are available, the LOS condition for the sensing Tx/Rx-target path is determined according to the available LOS probability formula, and the geometrical blockage modelling is applied subsequently as described in “Blockage model B” of [TR 38.901, Subsection 7.6.4] to capture the energy impact of the EO on the blocked path.
Proposal 8. If EO is modelled in the target channel, it should be also modelled in the background channel.
Proposal 9. The TR 36.885 is used as a reference TR for channel modelling assumptions of the UE-type RSU to another UE-type RSU or another UE type.  
Observation  6. The existing link-level channel modelling methods do not capture the clutter mobility/doppler and do not capture the impact of low-energy clusters (i.e., clutter density in delay/angle domain), hence leading to an artificial sensing environment.  
Proposal 10. Update the legacy link-level channel modelling methods with the following additions: 
	Addition of a cluster as sensing target cluster to the other/legacy clusters
	Addition of doppler frequency shift following the statistical pattern matching each scenario (in terms of the clutter velocity and percentage of mobile clutters)
	Addition of low-energy clusters which encompass the full delay-angle domain.
Proposal 11. Treat spatial consistency process of the sensing cluster, or sensing channel, as separate procedure than the background/environment channel, wherein each of the channel generation Steps 1-4 of Subsection 3.1 are evolved separately with necessary consistency requirements.
Proposal 12. For the background/environment channel, take the spatial consistency procedures of [38.901, Subsection 7.6.3] as a starting point and enhance to support 3-D movement of the target, as well as the 3-D movement of the sensing Tx, sensing Rx nodes. 
Proposal 13. Define, for the B2 term describing RCS of a sensing target, a temporal correlation distance, as well as one or more angular correlation distances between two angles of incidence or angles of scattering. 

3GPP TSG RAN WG1 #121		R1-2504269
St Julian’s, Malta, May 19th - 23rd, 2025
Source: 	MediaTek Inc.
Title:	Discussion on ISAC channel modelling
Agenda item:	9.7.2
Document for:	Discussion

Conclusion 
In this contribution, it discusses the ISAC channel modelling issues with following proposals:
Proposal 1: For sensing LLS channel generation, using the simplified SLS, i.e., single Tx and single Rx, to generate the channel coefficient, then fed into the LLS simulator.
Proposal 2: It is suggested not to consider the relative position change of multiple scattering points belonging to the same sensing target in the whole simulation.
Proposal 3: It is suggested not to model a propagation path from Tx to Rx interacting with more than one sensing target in the sensing channel modelling.
Proposal 4: The following mechanisms are suggested to determine the LOS condition between Tx/Rx and each multiple scattering points:
If EO is in the LOS ray of one link, the corresponding link is determined as NLOS link, otherwise, LOS probability equation is used to determine the LOS condition.
For each link, i.e., Tx-scattering point link or scattering point-Rx link, the LOS probability is used separately.
Proposal 5: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, using hUT 1.5 m for pathloss calculation to align wiht the previous applicability range.
Proposal 6: For the combined ISAC channel, the power normalization between sensing target channel model and background model is not performed.
Proposal 7: The initial random phase (generated in Step 10, section 7.5, TR38.901) is the same for the same ray in Tx-target link and target-Rx link of a target for monostatic sensing.
Proposal 8: For indirect path modelling in sensing target modelling, reducing the number of rays per cluster and/or reducing the number of clusters are supported.
Proposal 9: EO modelling can be enabled/disabled for ISAC channel modelling, which is up to the sensing evaluation scenario.
Proposal 10: From ISAC channel modelling perspective, a common channel modelling framework is defined for all six sensing modes, i.e.,  shall include both monostatic mode ISAC channel and bistatic mode ISAC channel.

3GPP TSG RAN WG1 #121                                                                                   R1-2504337
St Julian’s, Malta, May 19th – 23th, 2025

Agenda Item:	9.7.2

Source:	Apple Inc.
Title:	Discussion on ISAC channel modelling
Document for:	Discussion/Decision

Conclusion
In the contribution, the following observations and proposals are made: 

Proposal 1: The Monostatic RCS for human with RCS model 2 is dependent on both elevation angle and horizontal angle

The values/pattern A*B1, i.e.,  is deterministic based on incident/scattered angles

Where,
,
,

The values/pattern of component A*B1 are generated by the following parameters :




The standard deviation of component B2 is 3.94 dB

Observation 1:  With bistatic scattering RCS decreases as angular separation increases, until TX and RX approach the LOS path in which case we see an increase due to forward scattering. This is seen in both electromagnetic simulations and in actual measurements.

Observation 2: The Forward scattering/diffraction region starts at about 70 degrees. Note that in R19, it has been agreed not to model the forward scattering region.

Proposal 2: For the model for the RCS in the bi-static case:
For an incident direction, the bistatic RCS in the backscatter direction is equal to monostatic RCS
For human with RCS model 1, the value of component A of the bistatic RCS is the same as that of the monostatic RCS
A is the mean of monostatic RCS values in all 3D incident/scattered angles, and is applied to both monostatic and bistatic RCS

Proposal 3: The bistatic RCS of Human with RCS model 1 is modelled as
The values/pattern of A*B1 is given by

Component A, i.e., : same as component A of mono-static RCS for Human with RCS model 1
Component B2: same as component B2 of mono-static RCS for Human with RCS model 1
The AttenuateFactor can be modelled as a linear or approximately linear function


Proposal 4: The bistatic RCS of Human with RCS model 2 is modelled using the bi-static RCS for vehicle formula with the following updates:
k1=0.5714 and k2=0.1
The values of k1 and k2 map to an approximately linear function

Proposal 5: To decide between modeling a target with single or multiple scattering points depends on the target type/size, the distance between sensing Tx/Rx and the target, the use case and deployment scenario and evaluation methodology

Proposal 6: 
A single scattering point with appropriate parameters may be used to represent a target in the SLS as default. 
Multiple scattering points may be used in an LLS or for use cases and scenarios where there may be a need to differentiate between target types or to track the movement of a target
FFS: the abstraction method from the multi-point model to the single point model.

Proposal 7: If a target is modelled with multiple scattering points, 
RAN1 assumes the relative locations of the multiple scattering points of a target are not changed 
Translational motion of the target can be modelled while rotational motion of the target should not be considered.
RAN1 assumes no rays are scattered from one scattering point to another scattering point of the same target

Proposal 8: RAN1 should limit interaction between more than one sensing target to limit the complexity of system level simulations subject to validation by measurements.

Observation 3: A power threshold of 30 dB shows little change in delay spread and RSRP when compared with full concatenation with a substantial drop in complexity (68.997 taps vs. 123 taps on average) vs 20 dB.

Observation 4: A power threshold of 30 dB shows little RSRP difference with and without normalization. 

Proposal 9: Any indirect path with power metric less than 30 dB is dropped 

Proposal 10: For the component B2,
The component B2 of two different targets are generated independently.
If Tx, target, Rx positions do not change during a simulation, the RCS component B2 of different incident/scattered angles for a target are generated independently


Proposal 11: The initial random phase  and XPR of two different targets are generated independently.

Proposal 12: For evaluations where there is movement, model 2 (angular dependent RCS) is used to maintain the spatial consistency. 
B1 captures the changes based on any bi-static angle change in a pre-determined manner. 
Parameter B2 can either be kept constant or changed based on a coherent time/distance. 

Proposal 13: The LLS is needed in the case of studies on signal processing methods e.g. waveform
It should be based on the existing CDL channel model
To model an ISAC channel, we add one or more clusters representing the target(s). The RCS is added to the sensing target cluster. 
Note there may need to be a discussion on the CDL parameters for the monostatic background channel

Proposal 14. To address the issues that are FFS:
Dual mobility model in 7.6.10, TR 38.901 is reused to model Doppler effect  due to movement of stochastic clusters, i.e., 
FFS1: the maximum speed of moving scatterers should not exceed maximum speed of target, Tx or Rx
FFS2: the ratio of moving scatterers among all scatterers depends on evaluation goal.

3GPP TSG RAN WG1 #121			R1-2504405
Malta, Malta, May 19th – 23rd, 2025

Agenda item:	9.7.2
Source: 	Qualcomm Incorporated
Title: 	Discussion on ISAC Channel modeling
Document for:		Discussion

Conclusion
We have made the following proposals:
Proposal 1: Support the following power normalization between the target and the background channel:
, where  is such that 

Proposal 2: To generate the absolute delay model for sensing scenarios Urban grid, highway and HST, for both target channel and background channel:
For Urban grid, the values of parameters for  of scenarios UMa are reused for both FR1 and FR2. 
For Highway, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 
For HST, the values of parameters for  of scenarios RMa and UMa are reused for FR1 and FR2 respectively. 

Proposal 3: Do not introduce a power threshold for path dropping after concatenation, i.e. X=-Inf. 

Proposal 4: With regards to angular correlation of XPR
The B2/XPR/initial random phase of two different targets are generated independently.
The B2/XPR/initial random phase of a target for the different direct/indirect paths in the target channel are generated independently.

Proposal 5: On the B2 / XPR / initial phase values of a direct/indirect path of a target in the target channel, the same value of B2 / XPR / initial phase values apply for a simulation. 
Note: Up to company choice to decide to re-initialize the above parameters for the purpose of simulation coverage, similar to the handling of the initial phase in today’s simulations. 

Proposal 6: With regards to the EO Type-2 in the target channel and the determination of the LOS condition, support Option A.
Proposal 7: EO type-2 is not modelled in background channel.
Proposal 8: With regards to doppler modelling, support specifying a function for micro-doppler modelling for the UAV and human outdoor sensing (e.g. VRUs) scenario in this release. 

Proposal 9: In order to generate Tx-target link, target-Rx link and the background channel between a RSU-type UE and another node (TRP, pedestrian UE, vehicle UE, RSU-type UE), the following reference TRs are adopted:

Proposal 10: The WA on bistatic RCS for vehicle is reused for large size UAV and AGV (the parameters k1 and k2 can be different for different target type). 

Proposal 11: Confirm the WA on bistatic RCS for vehicle.

Proposal 12: The value of the RCS component A for a vehicle is 10 dBsm.

Proposal  13: When spatial consistency is enabled, the path dropping after concatenation of Tx-target and target-Rx link is not reperformed even if Tx, target, Rx positions change during simulation.
Note: the large scale/small scale parameter values can still be updated according to the procedure of spatial consistency. 

Proposal 14: Clarify the definition of the uniform distribution for determining effective environment height in UMa pathloss calculations. Based on the above options, below are provided two alternative updated definitions:
Include final value: The distribution is defined as , where  is defined as  and  is the largest integer such that .
Exclude final value: The distribution is defined as , where  is defined as  and  is the largest integer such that .

Proposal 15: Modify the UMa breakpoint distance calculation to account for UTs in the height range m. Below are provided two alternative solutions which provide different potential outcomes for effective environment heights in this range:
Update   to 
,
With this option, for any m, the effective environment height will be 1 meter with probability 1.
Add an additional case for UTs in the height range m: in this case, set  = 1 m with probability  and 12 m otherwise.

3GPP TSG-RAN WG1 Meeting #121	Tdoc R1-2504455
St Julian’s, Malta, May 19th – 23rd, 2025

Agenda Item:	9.7.2
Source:	Ericsson
Title:	Discussion on ISAC Channel Modelling
Document for:	Discussion, Decision

Conclusion
In the previous sections, we made the following observations: 
Observation 1	In some sensing scenarios, the shadows are significant, and their existence may be a strong indicator of target presence. Examples are intruder detection or vehicular sensing of other vehicles and obstacles.
Observation 2	Shadow behind a sufficiently large target means the combined target and background channel needs to have lower energy than the background channel in the absence of any target.
Observation 3	It was agreed that sensing target is in the far field of sensing Tx/Rx. However, sensing Rx is likely to locate in the near field of a large target, where shadow effect needs to be taken into account.
Observation 4	For a car of size  and a frequency of 6 GHz, the far field from the car occurs at a distance of 1 km. A sensing Rx with its distance to the vehicle smaller than 1km is in the car’s near field.
Observation 5	A method of modelling the shadows is to apply the existing blockage model to background channel, where a target casts a shadow for sensing Rx, so that  is reduced in the shadow region.
Observation 6	Interaction between targets is critical, when one target’s shadows have significant impacts on another target, i.e., when incident or scattered rays of the second target are severely attenuated.
Observation 7	The blockage of one target can be considered in the Tx-target and/or target-Rx links of another target channel depending on sensing mode.
Observation 8	There is no discussion on types of unintended targets and their modelling, including RCS, CPM, and velocity.
Observation 9	Trees moving in the wind will challenge sensing algorithms in distinguishing an intended target from the unintended targets in frequency domain due to their similar Doppler shifts.
Observation 10	Without modelling of unintended objects, the SID’s objective of distinguishing targets from unintended objects and clutter/scattering patterns cannot be fulfilled, and the study is incomplete.
Observation 11	There is no discussion on  and  of moving stochastic clusters, and therefore they cannot serve as unintended objects or generate comparable physical characteristics as sensing targets.
Observation 12	Measurements show that there are many multipath coming from the direction of the trees and trees are strong clusters.
Observation 13	Discontinuities in the RCS cause signals that are not bandlimited and non-physical artifacts in the Doppler spectrum.
Observation 14	With Option 1, there are no rapid random variations of RCS, which are observed in measured channels. It obviates the need of combining observations over time to ensure a good probability of detection, which is necessary in reality. 
Observation 15	With Option 2, discontinuous RCS leads to non-physical artifacts in the channel, including very high Doppler frequencies and spurious out-of-band artifacts, which may negatively affect the accuracy of detection.
Observation 16	On the other hand, the finding that Doppler artifacts only occur in the target channel but not in the background channel may be used by a sensing algorithm to artificially inflate the target detection probability and reduce the false alarm rate. Hence, it is very important to not introduce such artifacts in the channel model.
Observation 17	With bi-static RCS WA, mono-static RCS is a very small subset of bi-static RCS, with the ratio of 1 : 360*180, with 1-degree angular granularity.
Observation 18	Bi-static RCS values are more widely used than mono-static RCS in target channel.
Observation 19	Mono-static RCS component of A*B1 is in a large value range for angle-dependent RCS. Mean of mono-static RCS doesn’t reflect RCS of the direct path with a particular incident angle.
Observation 20	When objects move and rotate, their scattering properties stay the same in the local coordinate system of the object, similar to the field patterns of BS and UE. This is why scattering (RCS) and cross-polarisation (CPM) is modelled in the coordinate system of the object.
Observation 21	Transformation between LCS and GCS is necessary to make sure that Tx, target, and Rx can relate one spherical unit vector in GCS with the same definition of  and .
Observation 22	If  is defined in the Global Coordinate System (GCS), rotation of the target will not be supported by the ISAC channel model. Vehicle, large UAV, AGV, and human can only move in the direction of the positive x-axis.
Observation 23	If  is defined in a Local Coordinate System (LCS), the channel model can support rotation of the targets by reusing the legacy transformation procedure in 38.901, which is used to support arbitrary orientation of BS and UE antenna panel and Type-2 EO as well.
Observation 24	Existing communication channel in 38.901 supports arbitrary orientation of BS and UE antenna array in their LCS. If the orientation of the LCS is different from the GCS orientation, a transformation between LCS and GCS is needed.
Observation 25	If the target’s mobility is in the x-y plane only, both and  are identity matrices.
Observation 26	If the target’s movement involves rotation around the x axis or y axis,  will generate two different scaling factors on the main diagonal and also two different factors on the antidiagonal in the matrix, and the transformation procedure cannot be skipped.
Observation 27	Both RCS component B2 and XPR of target are generated for each path, namely they are angle specific.
Observation 28	UMa measurements show that the multipath dispersion is very large, covering the range from close to zero absolute delay up to many microseconds.
Observation 29	Absolute delay of the first path of mono-static background channel model for UMa scenario is 0.36us, corresponding to a path length of 108m. Targets with distance to BS smaller than 54m would be sensed without the interference of background channel.
Observation 30	The maximum delay of RMa scenario is only 2.26us with a propagation path length of 678m.
Observation 31	If correlation distance in Table 7.6.9-1 is conformed, the same  used for the three reference points means they have the same level of correlation, which instead puts a restriction on their positions.
Observation 32	Doppler due to moving scatterers (clusters) in the environment is modeled in TR 38.901 clause 7.6.10 using a maximum speed  of the scatterers and a proportion  of the scatterers (clusters) that are mobile.
Observation 33	The existing guidance on  and  in TR 38.901 does not cover sensing scenarios.
Observation 34	For sensing, Doppler due to moving scatterers (clusters) in the environment can affect both the target channel and the background channel.
Observation 35	Moving scatterers in the background channel can have a profound impact on sensing performance.
Observation 36	It is a necessity to model Doppler variations from other objects than the target in the ISAC channel, otherwise results on target detection reliability and false alarm rates will be wildly inaccurate.
Observation 37	In a measurement of a SMa/RMa BS-BS background channel, several paths (clusters) show non-zero Doppler shifts.
Observation 38	In a measurement of a UMa BS-BS background channel, several paths (clusters) show power variations over time. The power variations could tentatively be connected to moving objects in the environment that had LOS to both Tx and Rx.
Observation 39	Since stochastic clusters have been used to generate multipath propagation in UE-BS communication channel, the need to model Type-2 EO in background channel is not clear.
Observation 40	Type-2 EO if modelled in background channel would lead to the consequence that the received power of the sensing Rx’s in the ISAC channel model in the absence of sensing targets is different from that in the legacy UE-BS communication channel.
Observation 41	The impact of a blocker in existing channel models is restricted to adjusting the amplitudes of paths that have previously been generated assuming a certain LOS state.
Observation 42	The presence of the blocker does not trigger a re-evaluation of the LOS state nor regeneration of paths in existing channel models.
Observation 43	Option B is in line with existing procedure for blocking in TR 38.901, where the blocking does not change the LOS state.
Observation 44	Option A allows the blocker to determine NLOS condition. Not only does it go against the existing blockage procedure, but also the spatial consistency between BS-target link and BS-UE channel is broken, when the target and UE are close to each other.
Observation 45	Legacy communication channel avoids a hard transition between LOS and NLOS states due to UT mobility by using soft LOS state.
Observation 46	Option A may cause a hard transition in the channel response as the targets moves, similar to UT movement, which legacy communication channel tries to avoid.
Observation 47	With Option A, simulation stops with the change of LOS/NLOS state. It puts a constraint on targets’ movement and hinders and disables target tracking.
Observation 48	Option A is more suitable for the map-based hybrid channel model or ray-tracing than the stochastic channel stated in section 7.5 of 38.901.
Observation 49	Specular reflection is a distance-dependent phenomenon, and it occurs when the object is of the same size as or larger than the first Fresnel zone.
 Based on the discussion in the previous sections we propose the following:
Proposal 1	Support blockage model B in 39.901 to model shadows, including
	modelling blockage in the background channel of the target
	modelling interactions between the targets when needed by introducing blockage in the Tx-target and target-Rx link of another target
Proposal 2	Study modelling of birds and trees.
Proposal 3	For the bistatic RCS of a bird, A is uniformly distributed between -40 to -20 dBsm, B=0dB, standard deviation of B2=3dB.
Proposal 4	Use the following bistatic RCS model of a single scattering point for a tree: A*B1 with a constant level of 10 dBm2 and a single lobe in the forward direction with A*B1 increasing to 23 dBm2. The B2 standard deviation is approximately 3 dB.
Proposal 5	For tree RCS model with two scattering points, consider a (larger) fraction of the RCS to correspond to a fixed scattering point representing static parts of the tree such as the trunk, and another (smaller) fraction of the RCS to correspond to moving scattering points representing leaves and branches.
Proposal 6	Study RCS of truck/bus, child, and animal based on agreed sizes in AI 9.7.1 and RCS of AGV of both sizes
Proposal 7	Since micro-Doppler is essential for distinguishing wanted and unwanted targets, methods to model micro-Doppler are needed for the completion of the Study Item.
Proposal 8	To model the Doppler signature (micro-Doppler) of various targets and Type-1 EOs, the target is represented by one (or more) fixed scattering points and one (or more) scattering points that are moving in relation to the fixed scattering point
Proposal 9	If angular discontinuity issue in bi-static RCS model is not solved, add a note: angular discontinuity of bi-static RCS happens when bi-sector changes direction and the incurred band-unlimited Doppler artifacts may lead to unintended sensing performance loss or other unexpected simulation behavior.
Proposal 10	Support angular correlation of B2 and model the stochastic B2 with C random scattering centres as .
Proposal 11	The offset  can be taken from a uniform grid with spacing 0.7 wavelengths and the number of scattering centres C can be chosen as 5.  Other methods to select these parameters that result in equivalent properties exist.
Proposal 12	If RCS Component A is the mean value of mono-static RCS, it should be noted that such RCS component A may not reflect the actual path loss in the target channel.
Proposal 13	If the CPM of the target is defined in the LCS, a rotation of the target can be supported by replacing  with .
Proposal 14	If the target’s mobility is in the x-y plane only, the transformation procedure can be skipped. Otherwise, it cannot.
Proposal 15	Targets can be treated in the same way as Type-2 EO by reusing the transformation procedure in 38.901.
Proposal 16	B2 and XPR of a target, which are the random variables, should be considered for spatial consistency to address target movements in a simulation by reusing the legacy correlation distances.
Proposal 17	Angle-specific random variables, B2 and XPM of the target, should be spatially consistent between adjacent angles, i.e. the random variables should not change abruptly for small changes in the angles.
Proposal 18	If low-power clusters is used for mono-static sensing mode, delay spread of low-power clusters can be increased by for example 5~10 times for RMa scenario.
Proposal 19	In order to reuse the legacy procedure of absolute delay including correlation distance,  for each of the three reference points is independently generated considering their relative positions and correlation distance.
Proposal 20	Further study how to use the ‘correlation distance in the horizontal plane [m]’ for reference points of different heights.
Proposal 21	The measurements show that the multipath with Doppler can be weaker than the 25 dB cluster dropping threshold. This requires low-power clusters to be mandatory.
Proposal 22	Model movement in the environment using unintended targets modelled by Type-1 EOs.
Proposal 23	Type-2 EO is modelled in the target channel only.
Proposal 24	In addition to the walls modelled in urban grid, a Type-2 EO can be modelled with a random orientation and a physical size in other outdoor scenarios and modelled reusing the room size in indoor scenario.
Proposal 25	Type-2 EO has no impact on LOS/NLOS condition.
Proposal 26	The legacy soft LOS state is applicable to ISAC channel model.
Proposal 27	Specular reflection is modelled, if Type-2 EO is of the same size as or larger than the first Fresnel zone. Otherwise, it is not modelled.
Proposal 28	It needs evaluation whether BS-aerial UE in 36.777 can be reused for terrestrial UE-UAV link and terrestrial UE-aerial UE channel, especially for indoor UEs.
Proposal 29	The LOS probability between an aerial UE and a normal UE is the following:
Proposal 30	For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE height of 22.5m-35m, which is similar to the corresponding BS height, use terrestrial UE-BS link of UMa scenario in TR 38.901 for urban areas and that of RMa scenario for rural areas with the BS representing the aerial UE.
Proposal 31	For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE heights that are lower than 22.5m, use terrestrial UE-BS link of UMi scenario in TR 38.901 for urban areas with the BS representing the aerial UE.
Proposal 32	For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE heights that are higher than 35m, use terrestrial UE-BS link of UMa in TR 38.901 for urban areas and that of RMa scenario for rural areas with the BS representing the aerial UE.
Proposal 33	For link-level ISAC channel models,
	 still holds.
	legacy CDL models are reused to model Tx-target and target-Rx links of target channel and Tx-Rx channel of background channel.
	target channel is generated by concatenating Tx-target and target-Rx link

3GPP TSG RAN WG1 #121	R1-2504511
St Julian’s, Malta, May 19th – 23th, 2025

Title 	: Discussion on ISAC Channel Modelling
Source 	:  NTT DOCOMO, INC.
Agenda item	:  9.7.2
Document for:  Discussion

Conclusion
In this contribution, we provided our views and discussions on the ISAC channel modelling. The following proposals are made:
Proposal 1: The working assumption for bistatic RCS of vehicle can be extended for use in the cases of large size UAVs and AGVs. 

Proposal 2: For bistatic, the following values of component A, B1, B2 can be used for RCS model 1 of human:
Component A: -9.4 dBsm
Component B1: 0 dB 
Component B2: using log-normal distribution
Mean: 0 dB
Standard deviation: 4.2 dB

Proposal 3: The following values of component A, B1, B2 can be used for bistatic RCS model 2 of human:
Component A*B1: using a similar formular as the case for bistatic RCS of vehicles with small values k1 and k2 of Attenuatorfactor
FFS: k1 and k2
Component B2: using log-normal distribution
Mean: 0 dB
Standard deviation: 4.2 dB

Proposal 4: The values/patterns of monostatic/bistatic RCSs of hazardous objects can be borrowed from those of the other sensing targets (UAV, vehicle, AGV, human) depending on the shapes and sizes of hazardous objects

Proposal 5: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m: 
Use the Option 5 for pathloss calculation:
Option 5: use hUT 1.5 m for pathloss calculation
Use the height of a scattering point directly to existing LOS probability formulas in TRs

Proposal 6: When the EO type-2 is modelled in the target channel, use the LOS probability equation to determine the LOS/NLOS condition of one link, and then the impacts of type-2 EO is modelled by blockage model B.

Proposal 7: EO type-2 should not be modelled in background channel.

Proposal 8: For model link-level channel for ISAC, the following option 1 should be used
Option 1: Based on existing CDL channel model, 
Alt 1: add one or more clusters representing target(s)
Alt 2: Update the parameters of one or more cluster to represent target(s)

Proposal 9: For model link-level channel for ISAC based on existing CDL channel model,
Add absolute delay  to the top of the table of CDL channel model
Apply the RCS values and velocities to sensing cluster(s)
Two types of sensing clusters can be used:
One with short delay and strong power
One with long delay and weak power

3GPP TSG RAN WG1 Meeting #121  	          	     		R1-2504567
St Julian’s, Malta, May 19-23, 2025
_____________________________________________________________________Agenda item: 9.7.2
Source: LG Electronics
Title: 	Discussion on ISAC channel modeling
Document for: Discussion and decision

Conclusions
In this contribution, the issues related to the ISAC channel modeling were discussed. The following observations and proposals were made as conclusions.
Observation 1: The angular distance between two incident directions at which RCSs decorrelates significantly is quite small.
Observation 2: The working assumption is applicable for vehicle and AGV but requires further validation for human and UAV.
Observation 3: UAV scenarios for ISAC are not homogeneous, i.e. have different channels characteristics for significantly different object heights. Developing the universal approach that will cover both aerial and ground object and ensure spatial consistency between any points in space may be difficult, so we may divide the scenarios in two different cases: same-level TX-ST- RX and different levels TX-RX and ST
Observation 4: Additional studies needed for defining the vertical correlation distance for Urban/Vehicular scenarios, as well as both horizontal and vertical correlation distances for the high-altitude UAV cases. Initial proposals can be found in Table 3.
Observation 5: Sum-of-Sinusoids spatial correlation modeling approach considered here is suitable for full 3D modeling of mutually correlated TX-RX, TX-ST and ST-RX links, with arbitrary movement of ST and RX. Although we cannot require mandatory methodology of spatial consistence modeling, the described approach can be recommended as the baseline for calibrations. 
Proposal 1: RCS component A for each type of the target in Table 1 is supported, where, for a given A*B1, the value of component A is determined as an average of multivariable function. In particular, for the monostatic case adopt .
Proposal 2: Regardless of whether the RCS is introduced as the angular independent or angular independent random variable, simulate the RCS as a random variable whose realizations are uncorrelated in the angular domain. To provide this, generate the components B2 for any pair of direct/indirect paths as the independent lognormal variables.
Proposal 3: Model the bistatic RCS of human as a product of the linear monostatic RCS  (model 1 or 2) and the deterministic function of the bistatic angle so that  in linear domain.

Proposal 4: Model the monostatic human RCS model 2 as the product of the angular dependent component A*B1 parameterized according to Table 2 and the component B2 having the standard deviation σ = 1.47 dB.
Proposal 5: Model the monostatic RCS of animal as the product of the angular independent component A = 1.5 dBsm, the component B1=1 and the component B2 having the standard deviation σ = 3.94 dB.
Proposal 6: Consider adoption of the fully 3D spatial correlation model, supporting full 3D mobility for both the ST and UE for vehicular/pedestrian/low altitude UAV (same-level case).
Simulations of arbitrary mutual movements of the ST and RX units should be done with proper correlations between TX-ST, TX-RX and RX-ST links if they all located in the similar propagation conditions.
For high-altitude UAV links, consider introduction of TX-UAV link model with different propagation and spatial correlation properties. 

3GPP TSG RAN WG1 #121		 	       R1-2504706
St Julian’s, Malta, May 19th – 23rd, 2025

Source:           ZTE Corporation, Sanechips, CAICT
Title:             Discussion on channel modelling for ISAC
Agenda item:      9.7.2
Document for:     Discussion

Conclusion
In this contribution, we provide our measurements, RT simulations, observations, and analysis on ISAC channel modelling, and we have the following observations and proposals:
Proposal 1: For bi-static RCS of small size UAV and of human model 1, the following patterns can be adopted
The bistatic RCS of UAV with small size is modelled as 
The values/pattern of A*B1 is given by

Component A, i.e., : same as component A of mono-static RCS for UAV of small size
 dB, where  is the bi-static angle between incident ray and scattered ray,  is within 0 and 180 degree
The effect of forward scattering  is -Inf in Rel-19
Component B2: same as component B2 of mono-static RCS for UAV of small size
The bistatic RCS of Human with RCS model 1 is modelled as
The values/pattern of A*B1 is given by

Component A, i.e., : same as component A of mono-static RCS for Human with RCS model 1
 dB, where  is the bi-static angle between incident ray and scattered ray,  is within 0 and 180 degree
The effect of forward scattering  is -Inf in Rel-19
Component B2: same as component B2 of mono-static RCS for Human with RCS model 1

Proposal 2: The values of the parameters to generate UT mono-static background channel for UAV-type or RSU-type UT are provided 

Proposal 3: For TRP mono-static sensing, the spatial consistency of background channel is not considered. 
Proposal 4: For UT mono-static sensing, the following additional parameters for background channel spatial consistency should be introduced.
Proposal 5: The existing horizontal correlation distance in Table 7.6.3.1-2 in TR38.901 is reused to model 3D spatial consistency for UAV scenario. 
Proposal 6: In sensing scenario UMi, UMa, RMa, if the height of a scattering point of target is less than 1.5m, for pathloss calculation, the option below is adopted.
- 	Option 5: use hUT 1.5 m for pathloss calculation
Proposal 7: Endorse the step 1-7 in section 6 for EO type 2 into TR for ISAC channel modeling, where the channel coefficient for the EO reflected path is given by

Proposal 8: The pathloss of NLOS ray due to EO type 2 in the STX-SPST link or SPST-SRX link is calculated by									
-	 is the total propagation distance due to EO type2.
Proposal 9: The following contents type-2 EO are considered into TR for ISAC channel modeling
Proposal 10: For ISAC LLS, one or more sensing clusters or taps are newly added in the existing LLS tables of TR 38.901 
Values of parameters of the added cluster(s) can be discussed in performance evaluation stage
Proposal 11: For micro-Doppler modeling, reuse the patterns defined in report from ETSI ISAC ISG in LS R1-2501370 as follows
The micro-Doppler motion displacement  induces phase variations on the scattering signals from the sensing target expressed as , where  is the carrier wavelength. The micro-Doppler motion displacement,  can be expressed through the micro-Doppler velocity function,  using the following expression: . 
Proposal 12: To support ISAC, update the procedure of the existing map-based hybrid channel modeling in TR 38.901. 
A step 14 for ISAC is newly added at the end of section 8 of TR 38.901

3GPP TSG-RAN WG1 Meeting #121	Tdoc R1-2504707 
St Julian’s, Malta, May 19th – 23rd, 2025

Agenda Item:	9.7.2
Source:	Ericsson
Title:	Discussion on ISAC Channel Modelling
Document for:	Discussion, Decision

Conclusion
In the previous sections, we made the following observations: 
Observation 1	In some sensing scenarios, the shadows are significant, and their existence may be a strong indicator of target presence. Examples are intruder detection or vehicular sensing of other vehicles and obstacles.
Observation 2	Shadow behind a sufficiently large target means the combined target and background channel needs to have lower energy than the background channel in the absence of any target.
Observation 3	It was agreed that sensing target is in the far field of sensing Tx/Rx. However, sensing Rx is likely to locate in the near field of a large target, where shadow effect needs to be taken into account.
Observation 4	For a car of size  and a frequency of 6 GHz, the far field from the car occurs at a distance of 1 km. A sensing Rx with its distance to the vehicle smaller than 1km is in the car’s near field.
Observation 5	A method of modelling the shadows is to apply the existing blockage model to background channel, where a target casts a shadow for sensing Rx, so that  is reduced in the shadow region.
Observation 6	Interaction between targets is critical, when one target’s shadows have significant impacts on another target, i.e., when incident or scattered rays of the second target are severely attenuated.
Observation 7	The blockage of one target can be considered in the Tx-target and/or target-Rx links of another target channel depending on sensing mode.
Observation 8	There is no discussion on types of unintended targets and their modelling, including RCS, CPM, and velocity.
Observation 9	Trees moving in the wind will challenge sensing algorithms in distinguishing an intended target from the unintended targets in frequency domain due to their similar Doppler shifts.
Observation 10	Without modelling of unintended objects, the SID’s objective of distinguishing targets from unintended objects and clutter/scattering patterns cannot be fulfilled, and the study is incomplete.
Observation 11	There is no discussion on  and  of moving stochastic clusters, and therefore they cannot serve as unintended objects or generate comparable physical characteristics as sensing targets.
Observation 12	Measurements show that there are many multipath coming from the direction of the trees and trees are strong clusters.
Observation 13	Discontinuities in the RCS cause signals that are not bandlimited and non-physical artifacts in the Doppler spectrum.
Observation 14	With Option 1, there are no rapid random variations of RCS, which are observed in measured channels. It obviates the need of combining observations over time to ensure a good probability of detection, which is necessary in reality.
Observation 15	With Option 2, discontinuous RCS leads to non-physical artifacts in the channel, including very high Doppler frequencies and spurious out-of-band artifacts, which may negatively affect the accuracy of detection.
Observation 16	On the other hand, the finding that Doppler artifacts only occur in the target channel but not in the background channel may be used by a sensing algorithm to artificially inflate the target detection probability and reduce the false alarm rate. Hence, it is very important to not introduce such artifacts in the channel model.
Observation 17	With bi-static RCS WA, mono-static RCS is a very small subset of bi-static RCS, with the ratio of 1 : 360*180, with 1-degree angular granularity.
Observation 18	Bi-static RCS values are more widely used than mono-static RCS in target channel.
Observation 19	Mono-static RCS component of A*B1 is in a large value range for angle-dependent RCS. Mean of mono-static RCS doesn’t reflect RCS of the direct path with a particular incident angle.
Observation 20	Existing communication channel in 38.901 supports arbitrary orientation of BS and UE antenna array in their LCS. If the orientation of the LCS is different from the GCS orientation, the transformation procedure between LCS and GCS is provided.
Observation 21	If z axis of the LCS of the target is always the same as z axis of the GCS during the movement, both and  are identity matrices and  can be used without additional procedure.
Observation 22	If z axis of the LCS of the target is not always the same as z axis of the GCS during target movement,  will generate unequal power of HH and VV for the direct path, and the transformation procedure cannot be skipped.
Observation 23	Both RCS component B2 and XPR of target are generated for each path, namely they are angle specific.
Observation 24	UMa measurements show that the multipath dispersion is very large, covering the range from close to zero absolute delay up to many microseconds.
Observation 25	Absolute delay of the first path of mono-static background channel model for UMa scenario is 0.36us, corresponding to a path length of 108m. Targets with distance to BS smaller than 54m would be sensed without the interference of background channel.
Observation 26	The maximum delay of RMa scenario is only 2.26us with a propagation path length of 678m.
Observation 27	If correlation distance in Table 7.6.9-1 is conformed, the same  used for the three reference points means they have the same level of correlation, which instead puts a restriction on their positions.
Observation 28	Doppler due to moving scatterers (clusters) in the environment is modeled in TR 38.901 clause 7.6.10 using a maximum speed  of the scatterers and a proportion  of the scatterers (clusters) that are mobile.
Observation 29	The existing guidance on  and  in TR 38.901 does not cover sensing scenarios.
Observation 30	For sensing, Doppler due to moving scatterers (clusters) in the environment can affect both the target channel and the background channel.
Observation 31	Moving scatterers in the background channel can have a profound impact on sensing performance.
Observation 32	It is a necessity to model Doppler variations from other objects than the target in the ISAC channel, otherwise results on target detection reliability and false alarm rates will be wildly inaccurate.
Observation 33	In a measurement of a SMa/RMa BS-BS background channel, several paths (clusters) show non-zero Doppler shifts.
Observation 34	In a measurement of a UMa BS-BS background channel, several paths (clusters) show power variations over time. The power variations could tentatively be connected to moving objects in the environment that had LOS to both Tx and Rx.
Observation 35	Since stochastic clusters have been used to generate multipath propagation in UE-BS communication channel, the need to model Type-2 EO in background channel is not clear.
Observation 36	Type-2 EO if modelled in background channel would lead to the consequence that the received power of the sensing Rx’s in the ISAC channel model in the absence of sensing targets is different from that in the legacy UE-BS communication channel.
Observation 37	The impact of a blocker in existing channel models is restricted to adjusting the amplitudes of paths that have previously been generated assuming a certain LOS state.
Observation 38	The presence of the blocker does not trigger a re-evaluation of the LOS state nor regeneration of paths in existing channel models.
Observation 39	Option B is in line with existing procedure for blocking in TR 38.901, where the blocking does not change the LOS state.
Observation 40	Option A allows the blocker to determine NLOS condition. Not only does it go against the existing blockage procedure, but also the spatial consistency between BS-target link and BS-UE channel is broken, when the target and UE are close to each other.
Observation 41	Legacy communication channel avoids a hard transition between LOS and NLOS states due to UT mobility by using soft LOS state.
Observation 42	Option A may cause a hard transition in the channel response as the targets moves, similar to UT movement, which legacy communication channel tries to avoid.
Observation 43	With Option A, simulation stops with the change of LOS/NLOS state. It puts a constraint on targets’ movement and hinders and disables target tracking.
Observation 44	Option A is more suitable for the map-based hybrid channel model or ray-tracing than the stochastic channel stated in section 7.5 of 38.901.
Observation 45	Specular reflection is a distance-dependent phenomenon, and it occurs when the object is of the same size as or larger than the first Fresnel zone.
 Based on the discussion in the previous sections we propose the following:
Proposal 1	Support blockage model B in 39.901 to model shadows, including
	modelling blockage in the background channel of the target
	modelling interactions between the targets when needed by introducing blockage in the Tx-target and target-Rx link of another target
Proposal 2	Study modelling of birds and trees.
Proposal 3	For the bistatic RCS of a bird, A is uniformly distributed between -40 to -20 dBsm, B=0dB, standard deviation of B2=3dB.
Proposal 4	Use the following bistatic RCS model of a single scattering point for a tree: A*B1 with a constant level of 10 dBm2 and a single lobe in the forward direction with A*B1 increasing to 23 dBm2. The B2 standard deviation is approximately 3 dB.
Proposal 5	For tree RCS model with two scattering points, consider a (larger) fraction of the RCS to correspond to a fixed scattering point representing static parts of the tree such as the trunk, and another (smaller) fraction of the RCS to correspond to moving scattering points representing leaves and branches.
Proposal 6	Study RCS of truck/bus, child, and animal based on agreed sizes in AI 9.7.1 and RCS of AGV of both sizes
Proposal 7	Since micro-Doppler is essential for distinguishing wanted and unwanted targets, methods to model micro-Doppler are needed for the completion of the Study Item.
Proposal 8	To model the Doppler signature (micro-Doppler) of various targets and Type-1 EOs, the target is represented by one (or more) fixed scattering points and one (or more) scattering points that are moving in relation to the fixed scattering point
Proposal 9	If angular discontinuity issue in bi-static RCS model is not solved, add a note: angular discontinuity of bi-static RCS happens when bi-sector changes direction and the incurred band-unlimited Doppler artifacts may lead to unintended sensing performance loss or other unexpected simulation behavior.
Proposal 10	Support angular correlation of B2 and model the stochastic B2 with C random scattering centres as .
Proposal 11	The offset  can be taken from a uniform grid with spacing 0.7 wavelengths and the number of scattering centres C can be chosen as 5.  Other methods to select these parameters that result in equivalent properties exist.
Proposal 12	If RCS Component A is the mean value of mono-static RCS, it should be noted that such RCS component A may not reflect the actual path loss in the target channel.
Proposal 13	If the CPM of the target is defined in the LCS, a rotation of the target can be supported by replacing  with .
Proposal 14	If CPM of target is modelled in GCS, it should be noted that it doesn’t work for the case where z axis of the LCS of the target is different from that of GCS during target movement.
Proposal 15	If CPM of target is modelled in LCS,  can be used without additional procedure for the case where the LCS of the target is always the same as z axis of the GCS during the movement, and for other cases  is used instead of , where 1 is derived using equation 7.1-15 in 38.901 with α, β and γ, and spherical position () in equation 7.9.4-12 in draft CR and 2 is derived in the same way with spherical position () in equation 7.9.4-11.
Proposal 16	Targets can be treated in the same way as Type-2 EO by reusing the transformation procedure in 38.901.
Proposal 17	B2 and XPR of a target, which are the random variables, should be considered for spatial consistency to address target movements in a simulation by reusing the legacy correlation distances.
Proposal 18	Angle-specific random variables, B2 and XPM of the target, should be spatially consistent between adjacent angles, i.e. the random variables should not change abruptly for small changes in the angles.
Proposal 19	If low-power clusters is used for mono-static sensing mode, delay spread of low-power clusters can be increased by for example 5~10 times for RMa scenario.
Proposal 20	In order to reuse the legacy procedure of absolute delay including correlation distance,  for each of the three reference points is independently generated considering their relative positions and correlation distance.
Proposal 21	Further study how to use the ‘correlation distance in the horizontal plane [m]’ for reference points of different heights.
Proposal 22	The measurements show that the multipath with Doppler can be weaker than the 25 dB cluster dropping threshold. This requires low-power clusters to be mandatory.
Proposal 23	Model movement in the environment using unintended targets modelled by Type-1 EOs.
Proposal 24	Type-2 EO is modelled in the target channel only.
Proposal 25	In addition to the walls modelled in urban grid, a Type-2 EO can be modelled with a random orientation and a physical size in other outdoor scenarios and modelled reusing the room size in indoor scenario.
Proposal 26	Type-2 EO has no impact on LOS/NLOS condition.
Proposal 27	The legacy soft LOS state is applicable to ISAC channel model.
Proposal 28	Specular reflection is modelled, if Type-2 EO is of the same size as or larger than the first Fresnel zone. Otherwise, it is not modelled.
Proposal 29	It needs evaluation whether BS-aerial UE in 36.777 can be reused for terrestrial UE-UAV link and terrestrial UE-aerial UE channel, especially for indoor UEs.
Proposal 30	The LOS probability between an aerial UE and a normal UE is the following:
Proposal 31	For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE height of 22.5m-35m, which is similar to the corresponding BS height, use terrestrial UE-BS link of UMa scenario in TR 38.901 for urban areas and that of RMa scenario for rural areas with the BS representing the aerial UE.
Proposal 32	For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE heights that are lower than 22.5m, use terrestrial UE-BS link of UMi scenario in TR 38.901 for urban areas with the BS representing the aerial UE.
Proposal 33	For path loss and shadow fading of terrestrial UE-aerial UE link, for aerial UE heights that are higher than 35m, use terrestrial UE-BS link of UMa in TR 38.901 for urban areas and that of RMa scenario for rural areas with the BS representing the aerial UE.
Proposal 34	For link-level ISAC channel models,
	 still holds.
	legacy CDL models are reused to model Tx-target and target-Rx links of target channel and Tx-Rx channel of background channel.
	target channel is generated by concatenating Tx-target and target-Rx link

3GPP TSG RAN WG1 #121			 R1-2504945
St Julian’s, Malta, May 19th – 23rd, 2025
Agenda item:	9.7.2
Source:	Moderator (Xiaomi)
Title:	Summary #6 on ISAC channel modelling
Document for:	Discussion/Decision

Conclusion
There is no consensus to introduce an exact formula for micro-Doppler in Rel-19. The placeholder in the channel impulse response is kept in the draft CR.

TDoc file unavailable