3GPP TR 38.901 V16.1.

0 (2019-12)
Technical Report

3rd Generation Partnership Project;

Technical Specification Group Radio Access Network;
Study on channel model for frequencies from 0.5 to 100 GHz
(Release 16)

The present document has been developed within the 3rd Generation Partnership Project (3GPP TM) and may be further elaborated for the purposes of 3GPP.
The present document has not been subject to any approval process by the 3GPP Organizational Partners and shall not be implemented.
This Report is provided for future development work within 3GPP only. The Organizational Partners accept no liability for any use of this Specification.
Specifications and Reports for implementation of the 3GPP TM system should be obtained via the 3GPP Organizational Partners' Publications Offices.
Release 16 2 3GPP TR 38.901 V16.1.0 (2019-12)

Keywords
New Radio

3GPP

Postal address

3GPP support office address

650 Route des Lucioles - Sophia Antipolis
Valbonne - FRANCE
Tel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16

Internet
http://www.3gpp.org

Copyright Notification

No part may be reproduced except as authorized by written permission.

The copyright and the foregoing restriction extend to reproduction in all media.

© 2019, 3GPP Organizational Partners (ARIB, ATIS, CCSA, ETSI, TSDSI, TTA, TTC).
All rights reserved.

UMTS™ is a Trade Mark of ETSI registered for the benefit of its members
3GPP™ is a Trade Mark of ETSI registered for the benefit of its Members and of the 3GPP Organizational Partners
LTE™ is a Trade Mark of ETSI registered for the benefit of its Members and of the 3GPP Organizational Partners
GSM® and the GSM logo are registered and owned by the GSM Association

3GPP
Release 16 3 3GPP TR 38.901 V16.1.0 (2019-12)

Contents
Foreword .................................................................................................................................................... 5
1 Scope ................................................................................................................................................ 6
2 References ........................................................................................................................................ 6
3 Definitions, symbols and abbreviations ........................................................................................... 7
3.1 Definitions ................................................................................................................................................. 7
3.2 Symbols ..................................................................................................................................................... 7
3.3 Abbreviations............................................................................................................................................. 8
4 Introduction ...................................................................................................................................... 9
5 General ........................................................................................................................................... 10
6 Status/expectation of existing information on high frequencies .................................................... 10
6.1 Channel modelling works outside of 3GPP ............................................................................................. 10
6.2 Scenarios of interest................................................................................................................................. 12
6.3 Channel measurement capabilities ........................................................................................................... 13
6.4 Modelling objectives ............................................................................................................................... 13
7 Channel model(s) for 0.5-100 GHz ................................................................................................ 14
7.1 Coordinate system ................................................................................................................................... 14
7.1.1 Definition ........................................................................................................................................... 14
7.1.2 Local and global coordinate systems ................................................................................................. 15
7.1.3 Transformation from a LCS to a GCS................................................................................................ 15
7.1.4 Transformation from an LCS to a GCS for downtilt angle only ........................................................ 18
7.2 Scenarios.................................................................................................................................................. 19
7.3 Antenna modelling................................................................................................................................... 22
7.3.1 Antenna port mapping ........................................................................................................................ 23
7.3.2 Polarized antenna modelling .............................................................................................................. 23
7.4 Pathloss, LOS probability and penetration modelling ............................................................................. 24
7.4.1 Pathloss .............................................................................................................................................. 24
7.4.2 LOS probability.................................................................................................................................. 30
7.4.3 O2I penetration loss ........................................................................................................................... 31
7.4.3.1 O2I building penetration loss ....................................................................................................... 31
7.4.3.2 O2I car penetration loss ................................................................................................................ 32
7.4.4 Autocorrelation of shadow fading ...................................................................................................... 32
7.5 Fast fading model .................................................................................................................................... 33
7.6 Additional modelling components ........................................................................................................... 51
7.6.1 Oxygen absorption ............................................................................................................................. 51
7.6.2 Large bandwidth and large antenna array .......................................................................................... 52
7.6.2.1 Modelling of the propagation delay.............................................................................................. 52
7.6.2.2 Modelling of intra-cluster angular and delay spreads ................................................................... 52
7.6.3 Spatial consistency ............................................................................................................................. 54
7.6.3.1 Spatial consistency procedure ...................................................................................................... 54
7.6.3.2 Spatially-consistent UT/BS mobility modelling ........................................................................... 54
7.6.3.3 LOS/NLOS, indoor states and O2I parameters ............................................................................ 58
7.6.3.4 Applicability of spatial consistency .............................................................................................. 59
7.6.4 Blockage ............................................................................................................................................ 60
7.6.4.1 Blockage model A ........................................................................................................................ 61
7.6.4.2 Blockage model B ........................................................................................................................ 63
7.6.5 Correlation modelling for multi-frequency simulations ..................................................................... 65
7.6.5.1 Alternative channel generation method ........................................................................................................ 66
7.6.6 Time-varying Doppler shift................................................................................................................ 68
7.6.7 UT rotation ......................................................................................................................................... 68
7.6.8 Explicit ground reflection model........................................................................................................ 68
7.6.9 Absolute time of arrival ..................................................................................................................... 71
7.6.10 Dual mobility ..................................................................................................................................... 72
7.6.11 Sources of EM interference ................................................................................................................ 72
7.6.12 Embedded devices.............................................................................................................................. 72
3GPP
Release 16 4 3GPP TR 38.901 V16.1.0 (2019-12)

7.7 Channel models for link-level evaluations .............................................................................................. 73

7.7.1 Clustered Delay Line (CDL) models ................................................................................................. 73
7.7.2 Tapped Delay Line (TDL) models ..................................................................................................... 76
7.7.3 Scaling of delays ................................................................................................................................ 79
7.7.4 Spatial filter for generating TDL channel model ............................................................................... 80
7.7.4.1 Exemplary filters/antenna patterns ............................................................................................... 80
7.7.4.2 Generation procedure ................................................................................................................... 81
7.7.5 Extension for MIMO simulations ...................................................................................................... 81
7.7.5.1 CDL extension: Scaling of angles ................................................................................................ 81
7.7.5.2 TDL extension: Applying a correlation matrix............................................................................. 82
7.7.6 K-factor for LOS channel models ...................................................................................................... 83
7.8 Channel model calibration ....................................................................................................................... 83
7.8.1 Large scale calibration ....................................................................................................................... 83
7.8.2 Full calibration ................................................................................................................................... 84
7.8.3 Calibration of additional features ....................................................................................................... 85
7.8.4 Calibration of the indoor factory scenario .......................................................................................... 87
8 Map-based hybrid channel model (Alternative channel model methodology) .............................. 88
8.1 Coordinate system ................................................................................................................................... 89
8.2 Scenarios.................................................................................................................................................. 89
8.3 Antenna modelling................................................................................................................................... 89
8.4 Channel generation .................................................................................................................................. 89

Annex A: Further parameter definitions ........................................................................................... 100

A.1 Calculation of angular spread ................................................................................................................ 100
A.2 Calculation of mean angle ..................................................................................................................... 100

Annex B: Change history ..................................................................................................................... 101

3GPP
Release 16 5 3GPP TR 38.901 V16.1.0 (2019-12)

Foreword
This Technical Report has been produced by the 3rd Generation Partnership Project (3GPP).

The contents of the present document are subject to continuing work within the TSG and may change following formal
TSG approval. Should the TSG modify the contents of the present document, it will be re-released by the TSG with an
identifying change of release date and an increase in version number as follows:

Version x.y.z

where:

x the first digit:

1 presented to TSG for information;

2 presented to TSG for approval;

3 or greater indicates TSG approved document under change control.

y the second digit is incremented for all changes of substance, i.e. technical enhancements, corrections,
updates, etc.

z the third digit is incremented when editorial only changes have been incorporated in the document.

3GPP
Release 16 6 3GPP TR 38.901 V16.1.0 (2019-12)

1 Scope
The present document captures the findings of the study item, "Study on channel model for frequency spectrum above 6
GHz" [2] and from further findings of the study item, "Study on New Radio Access Technology [22]" and the study
item "Study on Channel Modeling for Indoor Industrial Scenarios [23]". The channel models in the present document
address the frequency range 0.5-100 GHz. The purpose of this TR is to help TSG RAN WG1 to properly model and
evaluate the performance of physical layer techniques using the appropriate channel model(s). Therefore, the TR will be
kept up-to-date via CRs in the future.

This document relates to the 3GPP evaluation methodology and covers the modelling of the physical layer of both
Mobile Equipment and Access Network of 3GPP systems.

This document is intended to capture the channel model(s) for frequencies from 0.5GHz up to 100GHz.

2 References
The following documents contain provisions which, through reference in this text, constitute provisions of the present
document.

- References are either specific (identified by date of publication, edition number, version number, etc.) or
non-specific.

- For a specific reference, subsequent revisions do not apply.

- For a non-specific reference, the latest version applies. In the case of a reference to a 3GPP document (including
a GSM document), a non-specific reference implicitly refers to the latest version of that document in the same
Release as the present document.

[1] 3GPP TR 21.905: "Vocabulary for 3GPP Specifications".

[2] 3GPP TD RP-151606: "Study on channel model for frequency spectrum above 6 GHz".

[3] 3GPP TR 36.873 (V12.2.0): "Study on 3D channel model for LTE".

[4] 3GPP RP-151847: "Report of RAN email discussion about >6GHz channel modelling", Samsung.

[5] 3GPP TD R1-163408: "Additional Considerations on Building Penetration Loss Modelling for 5G
System Performance Evaluation", Straight Path Communications.

[6] ICT-317669-METIS/D1.4: "METIS channel model, METIS 2020, Feb, 2015".

[7] Glassner, A S: "An introduction to ray tracing. Elsevier, 1989".

[8] McKown, J. W., Hamilton, R. L.: "Ray tracing as a design tool for radio networks, Network,
IEEE, 1991(6): 27-30".

[9] Kurner, T., Cichon, D. J., Wiesbeck, W.: "Concepts and results for 3D digital terrain-based wave
propagation models: An overview", IEEE J.Select. Areas Commun., vol. 11, pp. 1002–1012, 1993.

[10] Born, M., Wolf, E.: "Principles of optics: electromagnetic theory of propagation, interference and
diffraction of light", CUP Archive, 2000.

[11] Friis, H.: "A note on a simple transmission formula", proc. IRE, vol. 34, no. 5, pp. 254–256, 1946.

[12] Kouyoumjian, R.G., Pathak, P.H.: "A uniform geometrical theory of diffraction for an edge in a
perfectly conducting surface" Proc. IEEE, vol. 62, pp. 1448–1461, Nov. 1974.

[13] Pathak, P.H., Burnside, W., Marhefka, R.: "A Uniform GTD Analysis of the Diffraction of
Electromagnetic Waves by a Smooth Convex Surface", IEEE Transactions on Antennas and
Propagation, vol. 28, no. 5, pp. 631–642, 1980.

[14] IST-WINNER II Deliverable 1.1.2 v.1.2, "WINNER II Channel Models", IST-WINNER2, Tech.
Rep., 2007 (http://www.ist-winner.org/deliverables.html).
3GPP
Release 16 7 3GPP TR 38.901 V16.1.0 (2019-12)

[15] 3GPP TR36.101: "User Equipment (UE) radio transmission and reception".

[16] 3GPP TR36.104: "Base Station (BS) radio transmission and reception".

[17] Asplund, H., Medbo, J., Göransson, B., Karlsson, J., Sköld, J.: "A simplified approach to applying
the 3GPP spatial channel model", in Proc. of PIMRC 2006.

[18] ITU-R Rec. P.1816: "The prediction of the time and the spatial profile for broadband land mobile
services using UHF and SHF bands".

[19] ITU-R Rec. P.2040-1: "Effects of building materials and structures on radiowave propagation
above about 100 MHz", International Telecommunication Union Radiocommunication Sector
ITU-R, 07/2015.

[20] ITU-R Rec. P.527-3: "Electrical characteristics of the surface of the earth", International
Telecommunication Union Radiocommunication Sector ITU-R, 03/1992.

[21] Jordan, E.C., Balmain, K.G.: "Electromagnetic Waves and Radiating Systems", Prentice-Hall Inc.,
1968.

[22] 3GPP TD RP-162469: "Study on New Radio (NR) Access Technology".

[23] 3GPP TD RP-182138: "SID on Channel Modeling for Indoor Industrial Scenarios".

3 Definitions, symbols and abbreviations

3.1 Definitions
For the purposes of the present document, the terms and definitions given in TR 21.905 [1] apply.

3.2 Symbols
For the purposes of the present document, the following symbols apply:

A antenna radiation power pattern

Amax maximum attenuation
d2D 2D distance between Tx and Rx
d3D 3D distance between Tx and Rx
dH antenna element spacing in horizontal direction
dV antenna element spacing in vertical direction
f frequency
fc center frequency / carrier frequency
ˆ
Frx,u,θ Receive antenna element u field pattern in the direction of the spherical basis vector θ
ˆ
Frx,u,ϕ Receive antenna element u field pattern in the direction of the spherical basis vector φ
ˆ
Ftx,s,θ Transmit antenna element s field pattern in the direction of the spherical basis vector θ
ˆ
Frx,s,ϕ Transmit antenna element s field pattern in the direction of the spherical basis vector φ
hBS antenna height for BS
hUT antenna height for UT
rˆrx , n , m spherical unit vector of cluster n, ray m, for receiver
rˆtx , n , m spherical unit vector of cluster n, ray m, for transmitter
α bearing angle
β downtilt angle
γ slant angle
λ wavelength
3GPP
Release 16 8 3GPP TR 38.901 V16.1.0 (2019-12)

κ cross-polarization power ratio in linear scale

µlgASA mean value of 10-base logarithm of azimuth angle spread of arrival
µlgASD mean value of 10-base logarithm of azimuth angle spread of departure
µlgDS mean value of 10-base logarithm of delay spread
µlgZSA mean value of 10-base logarithm of zenith angle spread of arrival
µlgZSD mean value of 10-base logarithm of zenith angle spread of departure
PrLOS LOS probability
SLAV side-lobe attenuation in vertical direction
σlgASA standard deviation of 10-base logarithm of azimuth angle spread of arrival
σlgASD standard deviation of 10-base logarithm of azimuth angle spread of departure
σlgDS standard deviation value of 10-base logarithm of delay spread
σlgZSA standard deviation of 10-base logarithm of zenith angle spread of arrival
σlgZSD standard deviation of 10-base logarithm of zenith angle spread of departure
σ SF standard deviation of SF
φ azimuth angle
θ zenith angle
φˆ spherical basis vector (unit vector) for GCS
φˆ′ spherical basis vector (unit vector) for LCS
φ3dB horizontal 3 dB beamwidth of an antenna
θˆ spherical basis vector (unit vector), orthogonal to φˆ , for GCS
θˆ′ spherical basis vector (unit vector), orthogonal to φˆ′ , for LCS
θ etilt electrical steering angle in vertical direction
θ 3dB vertical 3 dB beamwidth of an antenna
ψ Angular displacement between two pairs of unit vectors

3.3 Abbreviations
For the purposes of the present document, the abbreviations given in TR 21.905 [1] and the following apply. An
abbreviation defined in the present document takes precedence over the definition of the same abbreviation, if any, in
TR 21.905 [1].

2D two-dimensional
3D three-dimensional
AOA Azimuth angle Of Arrival
AOD Azimuth angle Of Departure
AS Angular Spread
ASA Azimuth angle Spread of Arrival
ASD Azimuth angle Spread of Departure
BF Beamforming
BS Base Station
BP Breakpoint
BW Beamwidth
CDF Cumulative Distribution Function
CDL Clustered Delay Line
CRS Common Reference Signal
D2D Device-to-Device
DFT Discrete Fourier Transform
DS Delay Spread
GCS Global Coordinate System
IID Independent and identically distributed
InF Indoor Factory
InF-SL Indoor Factory with Sparse clutter and Low base station height (both Tx and Rx are below the
average height of the clutter)
3GPP
Release 16 9 3GPP TR 38.901 V16.1.0 (2019-12)

InF-DL Indoor Factory with Dense clutter and Low base station height (both Tx and Rx are below the
average height of the clutter)
InF-SH Indoor Factory with Sparse clutter and High base station height (Tx or Rx elevated above the
clutter)
InF-DH Indoor Factory with Dense clutter and High base station height (Tx or Rx elevated above the
clutter)
InF-HH Indoor Factory with High Tx and High Rx (both elevated above the clutter)
InH Indoor Hotspot
IRR Infrared Reflecting
ISD Intersite Distance
K Ricean K factor
LCS Local Coordinate System
LOS Line Of Sight
MIMO Multiple-Input-Multiple-Output
MPC Multipath Component
NLOS Non-LOS
O2I Outdoor-to-Indoor
O2O Outdoor-to-Outdoor
OFDM Orthogonal Frequency-Division Multiplexing
PAS Power angular spectrum
PL Path Loss
PRB Physical Resource Block
RCS Radar cross-section
RMa Rural Macro
RMS Root Mean Square
RSRP Reference Signal Received Power
Rx Receiver
SCM Spatial Channel Model
SINR Signal-to-Interference-plus-Noise Ratio
SIR Signal-to-Interference Ratio
SSCM Statistical Spatial Channel Model
SF Shadow Fading
SLA Sidelobe Attenuation
TDL Tapped Delay Line
TOA Time Of Arrival
TRP Transmission Reception Point
Tx Transmitter
UMa Urban Macro
UMi Urban Micro
UT User Terminal
UTD Uniform Theory of Diffraction
V2V Vehicle-to-Vehicle
XPR Cross-Polarization Ratio
ZOA Zenith angle Of Arrival
ZOD Zenith angle Of Departure
ZSA Zenith angle Spread of Arrival
ZSD Zenith angle Spread of Departure

4 Introduction
At 3GPP TSG RAN #69 meeting the Study Item Description on "Study on channel model for frequency spectrum
above 6 GHz" was approved [2]. This study item covers the identification of the status/expectation of existing
information on high frequencies (e.g. spectrum allocation, scenarios of interest, measurements, etc), and the channel
model(s) for frequencies up to 100 GHz. This technical report documents the channel model(s). The new channel model
has to a large degree been aligned with earlier channel models for <6 GHz such as the 3D SCM model (3GPP TR
36.873) or IMT-Advanced (ITU-R M.2135). The new model supports comparisons across frequency bands over the
range 0.5-100 GHz. The modelling methods defined in this technical report are generally applicable over the range 0.5-
100 GHz, unless explicitly mentioned otherwise in this technical report for specific modelling method, involved
parameters and/or scenario.
3GPP
Release 16 10 3GPP TR 38.901 V16.1.0 (2019-12)

Subsequently, at the 3GPP TSG RAN #81 meeting the Study Item Description "Study on Channel Modeling for Indoor
Industrial Scenarios" was approved [23]. The findings from this study item is also captured in the present technical
report. The Industrial channel model was developed by considering new measurements and information in the literature.
An overview list of all such contributions and sources is available in tdoc R1-1909706.

The channel model is applicable for link and system level simulations in the following conditions:

- For system level simulations, supported scenarios are urban microcell street canyon, urban macrocell, indoor
office, rural macrocell, and indoor factory.

- Bandwidth is supported up to 10% of the center frequency but no larger than 2GHz.

- Mobility of either one end of the link or both ends of the link is supported

- For the stochastic model, spatial consistency is supported by correlation of LSPs and SSPs as well as
LOS/NLOS state.

- Large array support is based on far field assumption and stationary channel over the size of the array.

5 General

6 Status/expectation of existing information on high

frequencies

6.1 Channel modelling works outside of 3GPP

This clause summarizes the channel modelling work outside of 3GPP based on the input from companies.

Groups and projects with channel models:

- METIS (Mobile and wireless communications Enablers for the Twenty-twenty Information Society)

- MiWEBA (Millimetre-Wave Evolution for Backhaul and Access)

- ITU-R M

- COST2100

- IEEE 802.11

- NYU WIRELESS: interdisciplinary academic research center

- Fraunhofer HHI has developed the QuaDRiGa channel model, Matlab implementation is available at
http://quadriga-channel-model.de

Groups and projects which intend to develop channel models:

- 5G mmWave Channel Model Alliance: NIST initiated, North America based

- mmMAGIC (Millimetre-Wave Based Mobile Radio Access Network for Fifth Generation Integrated
Communications): Europe based

- IMT-2020 5G promotion association: China based

METIS Channel Models:

3GPP
Release 16 11 3GPP TR 38.901 V16.1.0 (2019-12)

- Identified 5G requirements (e.g., wide frequency range, high bandwidth, massive MIMO, 3-D and accurate
polarization modelling)

- Performed channel measurements at various bands between 2GHz and 60 GHz

- Provided different channel model methodologies (map-based model, stochastic model or hybrid model). For
stochastic model, the proposed channel is focused on outdoor square, Indoor cafeteria and indoor shopping mall
scenarios.

MiWEBA Channel Models:

- Addressed various challenges: Shadowing, spatial consistency, environment dynamics, spherical wave
modelling, dual mobility Doppler model, ratio between diffuse and specular reflections, polarization

- Proposed Quasi-deterministic channel model

- Performed channel measurements at 60 GHz

- Focused on university campus, street canyon, hotel lobby, backhaul, and D2D scenarios.

ITU-R M Channel Models:

- Addressed the propagation loss and atmospheric loss on mmW

- Introduced enabling antenna array technology and semiconductor technology

- Proposed deployment scenarios, focused on dense urban environment for high data rate service: indoor shopping
mall, indoor enterprise, in home, urban hotspot in a square/street, mobility in city.

COST2100 and COST IC1004 Channel Models:

- Geometry-based stochastic channel model that reproduce the stochastic properties of MIMO channels over time,
frequency and space. It is a cluster-level model where the statistics of the large scale parameters are always
guaranteed in each series of channel instances.

NYU WIRELESS Channel Models:

- Conducted many urban propagation measurements on 28/38/60/73 GHz bands for both outdoor and indoor
channels, measurements are continuing.

- Proposed 3 areas for 5G mmWave channel modelling which are small modifications or extensions from 3GPP's
current below 6GHz channel models

- 1) LOS/NLOS/blockage modelling (a squared exponential term); 2). Wideband power delay profiles (time
clusters and spatial lobes for a simple extension to the existing 3GPP SSCM model); 3). Physics-based path loss
model (using the existing 3GPP path loss equations, but simply replacing the "floating" optimization parameter
with a deterministic 1 m "close-in" free space reference term in order to provide a standard and stable definition
of "path loss exponent" across all different parties, scenarios, and frequencies).

802.11 ad/ay Channel Models:

- Conducted ray-tracing methodology on 60 GHz band indoor channels, including conference room, cubicle,
living room scenarios

- Intra cluster parameters were proposed in terms of ray excess delay and ray power distribution

- Human blockage models were proposed in terms of blockage probability and blockage attenuation

5G mmWave Channel Model Alliance:

- Will provide a venue to promote fundamental research into measurement, analysis, identification of physical
parameters, and statistical representations of mmWave propagation channels.

- Divided into six collaborative working groups that include a Steering Committee; Modelling Methodology
Group; Measurement Methodology Group; and groups that focus on defining and parameterizing Indoor,
Outdoor, and Emerging Usage Scenarios.

- Sponsored by Communications Technology Research Laboratory within the NIST.

mmMAGIC:
- Brings together major infrastructure vendors, major European operators, leading research institutes and
universities, measurement equipment vendors and one SME.
3GPP
Release 16 12 3GPP TR 38.901 V16.1.0 (2019-12)

- Will undertake extensive radio channel measurements in the 6-100 GHz range.

- Will develop and validate advanced channel models that will be used for rigorous validation and feasibility
analysis of the proposed concepts and system, as well as for usage in regulatory and standards fora.

IMT-2020 5G promotion association

- Jointly established by three ministries of China based on the original IMT-Advanced promotion group

- Members including the main operators, vendors, universities and research institutes in China

- The major platform to promote 5G technology research in China and to facilitate international communication
and cooperation

QuaDRiGa (Fraunhofer HHI)

- QuaDRiGa (QUAsi Deterministic RadIo channel GenerAtor) was developed at the Fraunhofer Heinrich Hertz
Institute within the Wireless Communications and Networks Department to enable the modelling of MIMO
radio channels for specific network configurations, such as indoor, satellite or heterogeneous configurations.

- Besides being a fully-fledged 3D geometry-based stochastic channel model (well aligned with TR36.873),
QuaDRiGa contains a collection of features created in SCM(e) and WINNER channel models along with novel
modelling approaches which provide features to enable quasi-deterministic multi-link tracking of users
(receiver) movements in changing environments. QuaDRiGa supports Massive MIMO modelling enabled
through a new multi-bounce scattering approach and spherical wave propagation. It will be continuously
extended with features required by 5G and frequencies beyond 6 GHz. The QuaDRiGa model is supported by
data from extensive channel measurement campaigns at 10 / 28 / 43 / 60 / 82 GHz performed by the same group.

6.2 Scenarios of interest

Brief description of the key scenarios of interest identified (see note):

(1) UMi (Street canyon, open area) with O2O and O2I: This is similar to 3D-UMi scenario, where the BSs are
mounted below rooftop levels of surrounding buildings. UMi open area is intended to capture real-life scenarios
such as a city or station square. The width of the typical open area is in the order of 50 to 100 m.

Example: [Tx height:10m, Rx height: 1.5-2.5 m, ISD: 200m]

(2) UMa with O2O and O2I: This is similar to 3D-UMa scenario, where the BSs are mounted above rooftop levels
of surrounding buildings.

Example: [Tx height:25m, Rx height: 1.5-2.5 m, ISD: 500m]

(3) Indoor: This scenario is intended to capture various typical indoor deployment scenarios, including office
environments, and shopping malls. The typical office environment is comprised of open cubicle areas, walled
offices, open areas, corridors etc. The BSs are mounted at a height of 2-3 m either on the ceilings or walls. The
shopping malls are often 1-5 stories high and may include an open area (or "atrium") shared by several floors.
The BSs are mounted at a height of approximately 3 m on the walls or ceilings of the corridors and shops.

Example: [Tx height: 2-3m, Rx height: 1.5m, area: 500 square meters]

(4) Backhaul, including outdoor above roof top backhaul in urban area and street canyon scenario where small cell
BSs are placed at lamp posts.

(5) D2D/V2V. Device-to-device access in open area, street canyon, and indoor scenarios. V2V is a special case
where the devices are mobile.

(6) Other scenarios such as Stadium (open-roof) and Gym (close-roof).

(7) Indoor industrial scenarios

Note: The scenarios of interest are based on the plenary email discussion and different from the supported
scenarios in clause 7. The indoor industrial scenarios were identified at a later stage in the 3GPP TSG
RAN #81 meeting.

3GPP
Release 16 13 3GPP TR 38.901 V16.1.0 (2019-12)

6.3 Channel measurement capabilities

The measurement capability as reported by each company is summarized in the following table.

Table 6.3-1: Channel measurement capabilities

6 - 20 GHz 20 - 30 GHz 30 - 60 GHz >60 GHz

Urban macro CMCC Nokia/Aalborg NYU
Nokia/Aalborg
Urban micro Aalto University AT&T AT&T AT&T
CMCC Aalto University Huawei Aalto University
Ericsson CMCC Intel/Fraunhofer HHI Huawei
Intel/Fraunhofer HHI Huawei NTT DOCOMO Intel/Fraunhofer HHI
Nokia/Aalborg Intel/Fraunhofer HHI Qualcomm NYU
NTT DOCOMO Nokia/Aalborg CATT
Orange NTT DOCOMO ETRI
NYU ITRI/CCU
Qualcomm ZTE
Samsung
CATT
KT
ETRI
ITRI/CCU
ZTE
Indoor Aalto University AT&T AT&T AT&T
CMCC Alcatel-Lucent Ericsson Aalto University
Ericsson Aalto University Huawei Huawei
Huawei BUPT Intel/Fraunhofer HHI Intel/Fraunhofer HHI
Intel/Fraunhofer HHI CMCC NTT DOCOMO NYU
Nokia/Aalborg Huawei NYU
NTT DOCOMO Intel/Fraunhofer HHI Qualcomm
Orange Nokia/Aalborg CATT
NTT DOCOMO ETRI
NYU ITRI/CCU
Qualcomm ZTE
Samsung
CATT
KT
ETRI
ITRI/CCU
ZTE
O2I Ericsson AT&T AT&T AT&T
Huawei Alcatel-Lucent Ericsson Huawei
Intel/Fraunhofer HHI Ericsson Huawei Intel/Fraunhofer HHI
Nokia/Aalborg Huawei Intel/Fraunhofer HHI
NTT DOCOMO Intel/Fraunhofer HHI NTT DOCOMO
Orange NTT DOCOMO
NYU
Samsung
KT

6.4 Modelling objectives

The requirements for channel modelling are as follows.

- Channel model SI should take into account the outcome of RAN-level discussion in the '5G' requirement study
item

- Complexity in terms of Description, Generating channel coefficients, development complexity and Simulation
time should be considered.

- Support frequency range up to 100 GHz.

3GPP
Release 16 14 3GPP TR 38.901 V16.1.0 (2019-12)

- The critical path of the SI is 6 – 100 GHz

- Take care of mmW propagation aspects such as blocking and atmosphere attenuation.

- The model should be consistent in space, time and frequency

- Support large channel bandwidths (up to 10% of carrier frequency)

- Aim for the channel model to cover a range of coupling loss considering current typical cell sizes, e.g. up to km-
range macro cells. Note: This is to enable investigation of the relevance of the 5G system using higher frequency
bands to existing deployments.

- Accommodate UT mobility

- Mobile speed up to 500 km/h.

- Develop a methodology considering that model extensions to D2D and V2V may be developed in future SI.

- Support large antenna arrays

7 Channel model(s) for 0.5-100 GHz

7.1 Coordinate system

7.1.1 Definition
A coordinate system is defined by the x, y, z axes, the spherical angles and the spherical unit vectors as shown in Figure
7.1.1. Figure 7.1.1 defines the zenith angle θ and the azimuth angle φ in a Cartesian coordinate system. Note that
θ = 0 points to the zenith and θ = 90 0 points to the horizon. The field component in the direction of θˆ is given by Fθ
and the field component in the direction of φˆ is given by Fφ .

θ
φˆ
n̂

θˆ
y
φ

Figure 7.1.1: Definition of spherical angles and spherical unit vectors in a Cartesian coordinate
system, where n̂ is the given direction, θˆ and φˆ are the spherical basis vectors

3GPP
Release 16 15 3GPP TR 38.901 V16.1.0 (2019-12)

7.1.2 Local and global coordinate systems

A Global Coordinate System (GCS) is defined for a system comprising multiple BSs and UTs. An array antenna for a
BS or a UT can be defined in a Local Coordinate System (LCS). An LCS is used as a reference to define the vector far-
field that is pattern and polarization, of each antenna element in an array. It is assumed that the far-field is known in the
LCS by formulae. The placement of an array within the GCS is defined by the translation between the GCS and a LCS.
The orientation of the array with respect to the GCS is defined in general by a sequence of rotations (described in clause
7.1.3). Since this orientation is in general different from the GCS orientation, it is necessary to map the vector fields of
the array elements from the LCS to the GCS. This mapping depends only on the orientation of the array and is given by
the equations in clause 7.1.3. Note that any arbitrary mechanical orientation of the array can be achieved by rotating the
LCS with respect to the GCS.

7.1.3 Transformation from a LCS to a GCS

A GCS with coordinates (x, y, z, θ , φ ) and unit vectors ( θˆ , φˆ ) and an LCS with "primed" coordinates (x', y', z', θ ' , φ ' )
and "primed" unit vectors ( θˆ' , φˆ' ) are defined with a common origins in Figures 7.1.3-1 and 7.1.3-2. Figure 7.1.3-1
illustrates the sequence of rotations that relate the GCS (gray) and the LCS (blue). Figure 7.1.3-2 shows the coordinate
direction and unit vectors of the GCS (gray) and the LCS (blue). Note that the vector fields of the array antenna
elements are defined in the LCS. In Figure 7.1.3-1 we consider an arbitrary 3D-rotation of the LCS with respect to the
GCS given by the angles α, β, γ. The set of angles α, β, γ can also be termed as the orientation of the array antenna with
respect to the GCS.

Note that the transformation from a LCS to a GCS depends only on the angles α, β, γ. The angle α is called the bearing
angle, β is called the downtilt angle and γ is called the slant angle.

Figure 7.1.3-1: Orienting the LCS (blue) with Figure 7.1.3-2: Definition of spherical
respect to the GCS (gray) by a sequence of 3 coordinates and unit vectors in both the GCS
rotations: α, β, γ. and LCS.

Let A' (θ ' , φ ' ) denote an antenna element pattern in the LCS and A(θ , φ ) denote the same antenna element pattern in the
GCS. Then the two are related simply by

A(θ ,φ ) = A' (θ ' ,φ ' ) (7.1-1)

with θ ' and φ ' given by (7.1-7) and (7.1-8).

3GPP
Release 16 16 3GPP TR 38.901 V16.1.0 (2019-12)

Let us denote the polarized field components in the LCS by Fθ′' (θ ' ,φ ' ) , Fφ′' (θ ' ,φ ' ) and in the GCS by Fθ (θ , φ ) ,
Fφ (θ , φ ) . Then they are related by equation (7.1-11).

Any arbitrary 3D rotation can be specified by at most 3 elemental rotations, and following the framework of
Figure 7.1.3-1, a series of rotations about the z, y and x axes are assumed here, in that order. The dotted and double-
dotted marks indicate that the rotations are intrinsic, which means that they are the result of one (⋅) or two (⋅⋅)
intermediate rotations. In other words, the y axis is the original y axis after the first rotation about z, and the x axis is
the original x axis after the first rotation about z and the second rotation about y . A first rotation of α about z sets the
antenna bearing angle (i.e. the sector pointing direction for a BS antenna element). The second rotation of β about y
sets the antenna downtilt angle. Finally, the third rotation of γ about x sets the antenna slant angle. The orientation of
the x, y and z axes after all three rotations can be denoted as x , y and z
 . These triple-dotted axes represents the final
orientation of the LCS, and for notational purposes denoted as the x', y' and z' axes (local or "primed" coordinate
system).

In order to establish the equations for transformation of the coordinate system and the polarized antenna field patterns
between the GCS and the LCS, it is necessary to determine the composite rotation matrix that describes the
transformation of point (x, y, z) in the GCS into point (x', y', z') in the LCS. This rotation matrix is computed as the
product of three elemental rotation matrices. The matrix to describe rotations about the z, y and x axes by the angles α,
β and γ respectively and in that order is defined as

 + cos α − sin α 0  + cos β 0 + sin β  1 0 0 

   
R = RZ (α )RY (β )RX (γ ) =  + sin α + cos α 0  0 1 0  0 + cos γ − sin γ 
 0 0 1  − sin β 0 + cos β  0 + sin γ + cos γ 
 (7.1-2)

The reverse transformation is given by the inverse of R, which is also equal to the transpose of R since it is orthogonal.

R −1 = R X (− γ )RY (− β )RZ (− α ) = R T (7.1-3)

The simplified forward and reverse composite rotation matrices are given by

 cos α cos β cos α sin β sin γ − sin α cos γ cos α sin β cos γ + sin α sin γ 
 
R =  sin α cos β sin α sin β sin γ + cos α cos γ sin α sin β cos γ − cos α sin γ 
 − sin β cos β sin γ cos β cos γ 
  (7.1-4)

and

 cos α cos β sin α cos β − sin β 

 
R =  cos α sin β sin γ − sin α cos γ
−1
sin α sin β sin γ + cos α cos γ cos β sin γ 
 cos α sin β cos γ + sin α sin γ sin α sin β cos γ − cos α sin γ cos β cos γ 
 (7.1-5)

These transformations can be used to derive the angular and polarization relationships between the two coordinate
systems.

In order to establish the angular relationships, consider a point (x, y, z) on the unit sphere defined by the spherical
coordinates (ρ=1, θ, φ), where ρ is the unit radius, θ is the zenith angle measured from the +z-axis, and φ is the azimuth
angle measured from the +x-axis in the x-y plane. The Cartesian representation of that point is given by

 x   sin θ cos φ 
   
ρˆ =  y  =  sin θ sin φ 
 z   cos θ 
    (7.1-6)

3GPP
Release 16 17 3GPP TR 38.901 V16.1.0 (2019-12)

The zenith angle is computed as arccos(ρˆ ⋅ ẑ ) and the azimuth angle as arg( xˆ ⋅ ρˆ + j yˆ ⋅ ρˆ ) , where x̂ , ŷ and ẑ
are the Cartesian unit vectors. If this point represents a location in the GCS defined by θ and φ, the corresponding
position in the LCS is given by R −1 ρ̂ , from which local angles θ' and φ' can be computed. The results are given in
equations (7.1-7) and (7.1-8).

 0  T 
 
 
θ ' (α , β , γ ; θ , φ ) = arccos 0 R ρˆ  = acos(cos β cos γ cos θ + (sin β cos γ cos(φ − α ) − sin γ sin (φ − α )) sin θ )
−1

 
 1  
  (7.1-7)

 1  T 
   (cos β sin θ cos(φ − α ) − sin β cos θ ) +  (7.1-8)
  
φ ' (α , β , γ ;θ , φ ) = arg  j  R ρˆ  = arg
−1


 0 

  j (cos β sin γ cos θ + (sin β sin γ cos(φ − α ) + cos γ sin (φ − α )) sin θ )
 

These formulae relate the spherical angles (θ, φ) of the GCS to the spherical angles (θ', φ') of the LCS given the rotation
operation defined by the angles (α, β, γ).

Let us denote the polarized field components Fθ (θ , φ ) , Fφ (θ , φ ) in the GCS and Fθ′' (θ ' ,φ ' ) , Fφ′' (θ ' ,φ ' ) in the LCS.
These are related by

 Fθ (θ , φ ) θˆ(θ , φ )T Rθˆ′(θ ′, φ ′) θˆ(θ , φ )T Rφˆ′(θ ′, φ ′) Fθ′′ (θ ′, φ ′)

    
 F (θ , φ ) =  φˆ(θ , φ )T Rθˆ′(θ ′, φ ′) φˆ(θ , φ )T Rφˆ′(θ ′, φ ′) F ′′ (θ ′, φ ′) (7.1-9)
 φ    φ 

In this equation, θˆ and φˆ represent the spherical unit vectors of the GCS, and θˆ′ and φˆ′ are the representations in
the LCS. The forward rotation matrix R transforms the LCS unit vectors into the GCS frame of reference. These pairs
of unit vectors are orthogonal and can be represented as shown in Figure 7.1.3-3.

Figure 7.1.3-3: Rotation of the spherical basis vectors by an angle ψ due to the orientation of the LCS
with respect to the GCS

Assuming an angular displacement of ψ between the two pairs of unit vectors, the rotation matrix of equation (7.1-9)
can be further simplified as:

θˆ(θ ,φ )T Rθˆ′(θ ′,φ ′) θˆ(θ ,φ )T Rφˆ′(θ ′,φ ′)  cosψ cos(π 2 +ψ )  + cosψ − sinψ 
 
 φˆ(θ ,φ )T Rθˆ′(θ ′,φ ′) φˆ(θ ,φ )T Rφˆ′(θ ′,φ ′) =  cos(π 2 −ψ ) cosψ
 =  
+ cosψ 
     + sinψ (7.1-10)

and equation (7.1-9) can be written as:

3GPP
Release 16 18 3GPP TR 38.901 V16.1.0 (2019-12)

 Fθ (θ , φ )  + cosψ − sinψ  Fθ′′ (θ ′, φ ′)

   
 F (θ , φ ) =  + sinψ + cosψ  Fφ′′ (θ ′, φ ′)
(7.1-11)
 φ  
The angle ψ can be computed in numerous ways from equation (7.1-10), with one such way approach being

(
ψ = arg θˆ(θ ,φ )T Rθˆ′(θ ′,φ ′) + j φˆ(θ ,φ )T Rθˆ′(θ ′,φ ′) ) (7.1-12)

The dot products are readily computed using the Cartesian representation of the spherical unit vectors. The general
expressions for these unit vectors are given by

 cos θ cos φ 
 
θˆ =  cos θ sin φ 
 
 − sin θ  (7.1-13)

and

 − sin φ 
 
ˆ
φ =  + cos φ 
 0 
  (7.1-14)

The angle ψ can be expressed as a function of mechanical orientation (α, β, γ) and spherical position (θ, φ), and is given
by

 (sin γ cos θ sin (φ − α ) + cos γ (cos β sin θ − sin β cos θ cos(φ − α ))) + 
ψ = arg 
 j (sin γ cos(φ − α ) + sin β cos γ sin (φ − α ))  (7.1-15)

It can be shown that cosψ and sin ψ can be expressed as:

cos β cos γ sin θ − (sin β cos γ cos(φ − α ) − sin γ sin (φ − α )) cos θ

cosψ =
1 − (cos β cos γ cos θ + (sin β cos γ cos(φ − α ) − sin γ sin (φ − α )) sin θ )
2
(7.1-16)

sin β cos γ sin (φ − α ) + sin γ cos(φ − α )

sinψ =
1 − (cos β cos γ cos θ + (sin β cos γ cos(φ − α ) − sin γ sin (φ − α )) sin θ )
2
(7.1-17)

7.1.4 Transformation from an LCS to a GCS for downtilt angle only

In this clause equations are provided for the transformation from LCS to GCS assuming that the orientation of the LCS
(with respect to the GCS) is such that the bearing angle α=0, the downtilt angle β is non-zero and the slant angle γ=0. In
other words the y'-axis of the LCS is parallel to the y-axis of the GCS. Considering a BS antenna element the x-axis of
the GCS is aligned with the pointing direction of the sector. Mechanical downtilt is modelled as a rotation of the LCS
around the y-axis. For zero mechanical downtilt the LCS coincides with the GCS.

This transformation relates the spherical angles ( θ , φ ) in the global coordinate system to spherical angles ( θ ' , φ ' ) in
the local (antenna-fixed) coordinate system and is defined as follows:

θ ' = arccos(cos φ sin θ sin β + cos θ cos β ) (7.1-18)

φ ' = arg(cos φ sin θ cos β − cosθ sin β + j sin φ sin θ ) (7.1-19)

3GPP
Release 16 19 3GPP TR 38.901 V16.1.0 (2019-12)

where β is the mechanical tilt angle around the y-axis as defined in Figure 7.1.4. Note that the equations (7.1-7), (7.1-
8) reduce to equations (7.1-18), (7.1-19) if both α and γ are zero.

The antenna element pattern A(θ , φ ) in the GCS is related to the antenna element pattern A' (θ ' , φ ' ) in the LCS by the
relation

A(θ , φ ) = A' (θ ' , φ ') (7.1-20)

with θ ' and φ ' given by (7.1-18) and (7.1-19).

z
z'
β

θ' n̂
φˆ' φˆ
φˆ' φˆ
n̂
θˆ'
θˆ'
y'
φ' θˆ
y

x
β
θˆ ψ
x'
Figure 7.1.4: Definition of angles and unit vectors when the LCS has been rotated an angle β around
the y-axis of the GCS

For a mechanical tilt angle β , the global coordinate system field components Fθ (θ , φ ) and Fφ (θ , φ ) , are calculated
from the field components Fθ′' (θ ' ,φ ' ) and Fφ′' (θ ' ,φ ' ) of the radiation pattern in the local (antenna-fixed) coordinate
system as:

Fθ (θ ,φ ) = Fθ′' (θ ' ,φ ' ) cosψ − Fφ′' (θ ' ,φ ' ) sinψ (7.1-21)

Fφ (θ ,φ ) = Fθ′' (θ ' ,φ ' ) sinψ + Fφ′' (θ ' ,φ ' ) cosψ (7.1-22)

where θ ' and φ ' are defined as in (7.1-18) and (7.1-19), and ψ is defined as:

ψ = arg(sin θ cos β − cos φ cos θ sin β + j sin φ sin β ) . (7.1-23)

Note that the equation (7.1-15) is reduced to equation (7.1-23) if both α and γ are zero.

As an example, in the horizontal cut, i.e., for θ = 90° , equations (7.1-18), (7.1-19) and (7.1-23) become
θ '= arccos(cos φ sin β ) (7.1-24)

φ ' = arg(cos φ cos β + j sin φ ) (7.1-25)

ψ = arg(cos β + j sin φ sin β ) (7.1-26)

7.2 Scenarios
The detailed scenario description in this clause can be used for channel model calibration.

UMi-street canyon and UMa

3GPP
Release 16 20 3GPP TR 38.901 V16.1.0 (2019-12)

Details on UMi-street canyon and UMa scenarios are listed in Table 7.2-1.

Table 7.2-1: Evaluation parameters for UMi-street canyon and UMa scenarios

Parameters UMi - street canyon UMa

Hexagonal grid, 19 micro sites, 3 sectors Hexagonal grid, 19 macro sites, 3
Cell layout
per site (ISD = 200m) sectors per site (ISD = 500m)
BS antenna height hBS 10m 25m
Outdoor/indoor Outdoor and indoor Outdoor and indoor
UT LOS/NLOS LOS and NLOS LOS and NLOS
location
Height hUT Same as 3D-UMi in TR36.873 Same as 3D-UMa in TR36.873
Indoor UT ratio 80% 80%
UT mobility (horizontal plane
3km/h 3km/h
only)
Min. BS - UT distance (2D) 10m 35m
UT distribution (horizontal) Uniform Uniform

Indoor-office

Details on indoor-office scenarios are listed in Table 7.2-2 and presented in Figure 7.2-1. More details, if necessary, can
be added to Figure 7.2-1.

Table 7.2-2: Evaluation parameters for indoor-office scenarios

Indoor - office Indoor - office

Parameters
open office mixed office

Room size
120mx50mx3m
Layout (WxLxH)

ISD 20m

BS antenna height hBS 3 m (ceiling)

LOS/NLOS LOS and NLOS

UT location
Height hUT 1m

UT mobility (horizontal plane only) 3 km/h

Min. BS - UT distance (2D) 0

UT distribution (horizontal) Uniform

Note: The only difference between the open office and mixed office models in this TR is the line of sight
probability.

3GPP
Release 16 21 3GPP TR 38.901 V16.1.0 (2019-12)

Figure 7.2-1: Layout of indoor office scenarios.

RMa

The rural deployment scenario focuses on larger and continuous coverage. The key characteristics of this scenario are
continuous wide area coverage supporting high speed vehicles. This scenario will be noise-limited and/or interference-
limited, using macro TRPs. Details of RMa scenario is described in Table 7.2-3.

Table 7.2-3: Evaluation parameters for RMa

Parameters RMa
Carrier Frequency Up to 7Ghz
BS height hBS 35m

Layout Hexagonal grid, 19 Macro sites, 3sectors per site, ISD = 1732m or 5000m
UT height hUT 1.5m

UT distribution Uniform
Indoor/Outdoor 50% indoor and 50% in car
LOS/NLOS LOS and NLOS
Min BS - UT 35m
distance(2D)

Indoor Factory (InF)

The indoor factory (InF) scenario focuses on factory halls of varying sizes and with varying levels of density of
"clutter", e.g. machinery, assembly lines, storage shelves, etc. Details of the InF scenario are listed in Table 7.2-4.

Table 7.2-4: Evaluation parameters for InF

InF

InF-SL InF-DL InF-SH InF-DH InF-HH

Parameters (sparse clutter, (dense clutter, (sparse clutter, (dense clutter, (high Tx, high
low BS) low BS) high BS) high BS) Rx)

Rectangular: 20-160000 m2
Room size
Layout
Ceiling
5-25 m 5-15 m 5-25 m 5-15 m 5-25 m
height

3GPP
Release 16 22 3GPP TR 38.901 V16.1.0 (2019-12)

Effective
clutter < Ceiling height, 0-10 m
height ℎ𝑐𝑐
External
wall and Concrete or metal walls and ceiling with metal-coated windows
ceiling type

Small to medium Small to medium

Big machineries metallic Big machineries metallic
composed of machinery and composed of machinery and
regular metallic objects with regular metallic objects with
surfaces. irregular surfaces. irregular
For example: structure. For example: structure.
Clutter type several mixed For example: several mixed For example: Any
production assembly and production areas assembly and
areas with open production lines with open spaces production lines
spaces and surrounded by and surrounded by
storage/commis mixed small- storage/commissio mixed small-
sioning areas sized ning areas sized
machineries. machineries.

Typical clutter size,

10 m 2m 10 m 2m Any
𝑑𝑑𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐

Clutter density 𝑟𝑟 Any

Low clutter High clutter High clutter
(percentage of surface Low clutter density
density density density
area occupied by (<40%)
(<40%) (≥40%) (≥40%)
clutter)
Clutter-embedded, i.e. the BS
BS antenna height antenna height is below the Above clutter Above clutter
average clutter height

LOS/NLOS LOS and NLOS 100% LOS

UT
location
Height Clutter-embedded Above clutter

7.3 Antenna modelling

This clause captures the antenna array structures considered in this SI for calibration.

The BS antenna is modelled by a uniform rectangular panel array, comprising MgNg panels, as illustrated in Figure 7.3-1
with Mg being the number of panels in a column and Ng being the number of panels in a row. Furthermore the following
properties apply:

- Antenna panels are uniformly spaced in the horizontal direction with a spacing of dg,H and in the vertical
direction with a spacing of dg,V.

- On each antenna panel, antenna elements are placed in the vertical and horizontal direction, where N is the
number of columns, M is the number of antenna elements with the same polarization in each column.

- Antenna numbering on the panel illustrated in Figure 7.3-1 assumes observation of the antenna array from
the front (with x-axis pointing towards broad-side and increasing y-coordinate for increasing column
number).

- The antenna elements are uniformly spaced in the horizontal direction with a spacing of dH and in the vertical
direction with a spacing of dV.

- The antenna panel is either single polarized (P =1) or dual polarized (P =2).

The rectangular panel array antenna can be described by the following tuple (M g , N g , M , N , P ).

3GPP
Release 16 23 3GPP TR 38.901 V16.1.0 (2019-12)

dg,H
……
(M-1,0) (M-1,1) (M-1,N-1)

dg,V ……

……

……

……
……

(1,0) (1,1) (1,N-1)

……

(0,0) (0,1) (0,N-1)

Figure 7.3-1: Cross-polarized panel array antenna model

The antenna radiation power pattern of each antenna element is generated according to Table 7.3-1.

Table 7.3-1: Radiation power pattern of a single antenna element

Parameter Values
  θ ′′ − 90°  2 
Vertical cut of the
′′ (θ ′′, φ ′′ = 0°) = − min 12
AdB  , SLAV 
radiation power pattern   θ 3dB  
with θ 3dB = 65°, SLAV = 30 dB and θ ′′ ∈ [0°, 180°]
(dB)

  φ ′′  2 
Horizontal cut of the
′′ (θ ′′ = 90°, φ ′′) = − min 12
AdB  , Amax 
radiation power pattern   φ3dB  
with φ3dB = 65°, Amax = 30 dB and φ ′′ ∈ [- 180°, 180°]
(dB)

3D radiation power ′′ (θ ′′, φ ′′) = − min{− ( AdB

AdB ′′ (θ ′′, φ ′′ = 0°) + AdB
′′ (θ ′′ = 90°, φ ′′)), Amax }
pattern (dB)
Maximum directional gain
of an antenna element 8 dBi
GE,max

7.3.1 Antenna port mapping

Legacy BS array antennas, i.e. uniform linear arrays with fix phase shifts between its M elements to obtain a beamtilt in
vertical direction are modelled using complex weights

1  2π
wm = exp − j (m − 1)dV cos θ etilt  (7.3-1)
M  λ 

where m=1, …, M, θ etilt is the electrical vertical steering angle defined between 0° and 180° (90° represents
perpendicular to the array). λ denotes the wavelength and dV the vertical element spacing.

7.3.2 Polarized antenna modelling

In general the relationship between radiation field and power pattern is given by:

A′′(θ ′′, φ ′′) = Fθ′′′′ (θ ′′, φ ′′) + Fφ′′′′ (θ ′′, φ ′′) .

2 2
(7.3-2)

3GPP
Release 16 24 3GPP TR 38.901 V16.1.0 (2019-12)

The following two models represent two options on how to determine the radiation field patterns based on a defined
radiation power pattern.

Model-1:

In case of polarized antenna elements assume ζ is the polarization slant angle where ζ = 0 degrees corresponds to a
purely vertically polarized antenna element and ζ = + /− 45 degrees correspond to a pair of cross-polarized antenna
elements. Then the antenna element field components in θ′ and φ′ direction are given by

 Fθ′′ (θ ′,φ ′)  + cosψ − sinψ  Fθ′′′′ (θ ′′,φ ′′)

   
 F ′′ (θ ′,φ ′) =  + sinψ + cosψ  Fφ′′′′ (θ ′′,φ ′′)
(7.3-3)
 φ  

cos ζ sin θ ′ + sin ζ sin φ ′ cos θ ′ sin ζ cos φ ′

where cosψ = , sinψ = .
1 − (cos ζ cos θ ′ − sin ζ sin φ ′ sin θ ′) 1 − (cos ζ cos θ ′ − sin ζ sin φ ′ sin θ ′)
2 2

Note that the zenith and the azimuth field components Fθ′′ (θ ′, φ ′) , Fφ′′ (θ ′, φ ′) , Fθ′′′′ (θ ′′, φ ′′) and Fφ′′′′ (θ ′′, φ ′′) are
defined in terms of the spherical basis vectors of an LCS as defined in Clause 7.1. The difference between the single-
primed and the double-primed components is that the single-primed field components account for the polarization slant
and the double-primed field components do not. For a single polarized antenna (purely vertically polarized antenna) we
A′′(θ ′′,φ ′′) and Fφ′′′′ (θ ′′,φ ′′) = 0 where A′′(θ ′′, φ ′′) is the 3D antenna radiation power
can write Fθ′′′′ (θ ′′, φ ′′) =
pattern as a function of azimuth angle φ ′′ and zenith angle θ ′′ in the LCS as defined in Table 7.3-1 converted into
linear scale.

Model-2:

In case of polarized antennas, the polarization is modelled as angle-independent in both azimuth and elevation, in an
LCS. For a linearly polarized antenna, the antenna element field pattern, in the vertical polarization and in the
horizontal polarization, are given by

Fθ′′ (θ ′,φ ′) = A′(θ ′,φ ′) cos(ζ ) (7.3-4)

and

Fφ′′ (θ ′,φ ′) = A′(θ ′,φ ′) sin (ζ ) , (7.3-5)

respectively, where ζ is the polarization slant angle and A′(θ ′, φ ′) is the 3D antenna element power pattern as a
function of azimuth angle, φ ′ and elevation angle, θ ′ in the LCS. Note that ζ = 0 degrees correspond to a purely
vertically polarized antenna element. The vertical and horizontal field directions are defined in terms of the spherical
basis vectors, θˆ′ and φˆ′ respectively in the LCS as defined in Clause 7.1.2. Also A′(θ ′, φ ′) = A′′(θ ′′, φ ′′) , θ ′ = θ ′′
and φ ′ = φ ′′ as defined in Table 7.1-1.

7.4 Pathloss, LOS probability and penetration modelling

7.4.1 Pathloss
The pathloss models are summarized in Table 7.4.1-1 and the distance definitions are indicated in Figure 7.4.1-1 and
Figure 7.4.1-2. Note that the distribution of the shadow fading is log-normal, and its standard deviation for each
scenario is given in Table 7.4.1-1.

3GPP
Release 16 25 3GPP TR 38.901 V16.1.0 (2019-12)

d3
D -o
d3 ut
D
hBS hBS
d3
D-i
n

hUT hUT

d2D d2D-out d2D-in

Figure7.4.1-1: Definition of d2D and d3D Figure 7.4.1-2: Definition of d2D-out, d2D-in
for outdoor UTs and d3D-out, d3D-in for indoor UTs.

Note that

d 3D − out + d 3D − in = (d 2D − out + d 2D −in )2 + (hBS − hUT )2 (7.4-1)

3GPP
Release 16 26 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.4.1-1: Pathloss models

3GPP
Release 16 27 3GPP TR 38.901 V16.1.0 (2019-12)

LOS/NLOS
Scenario
Shadow Applicability range,
Pathloss [dB], fc is in GHz and d is in meters, see note 6 fading antenna height
std [dB] default values

 PL 10m ≤ d 2D ≤ d BP hBS = 35m

PLRMa −LOS =  1 , see note 5
 PL2 d BP ≤ d 2D ≤ 10km hUT = 1.5m
W = 20m
LOS

PL1 = 20 log10 (40πd 3D f c / 3) + min(0.03h1.72 ,10) log10 (d 3D ) σ SF = 4 h = 5m

− min(0.044h1.72 ,14.77) + 0.002 log10 (h)d 3D h = avg. building height

PL2 = PL1 (d BP ) + 40 log10 (d 3D / d BP ) σ SF = 6 W = avg. street width

The applicability ranges:
RMa

′ − NLOS )
PLRMa − NLOS = max( PLRMa −LOS , PLRMa 5m ≤ h ≤ 50m
for 10m ≤ d 2D ≤ 5km 5m ≤ W ≤ 50m
σ SF = 8
′ − NLOS = 161.04 − 7.1 log10 (W )+7.5 log10 (h)
PLRMa 10m ≤ hBS ≤ 150m
NLOS

− (24.37 − 3.7(h / hBS ) 2 ) log10 (hBS ) 1m ≤ hUT ≤ 10m

+ (43.42 − 3.1 log10 (hBS ))(log10 (d 3D ) − 3)
+ 20 log10 ( f c ) − (3.2(log10 (11.75hUT )) 2 − 4.97)
′
 PL 10m ≤ d 2D ≤ d BP
PLUMa −LOS =  1
′ ≤ d 2D ≤ 5km , see note 1
 PL2 d BP
1.5m ≤ hUT ≤ 22.5m
LOS

PL1 = 28.0 + 22 log10 (d 3D ) + 20 log10 ( f c ) σ SF = 4 hBS = 25m

PL2 = 28.0 + 40 log10 (d 3D ) + 20 log10 ( f c )
′ ) 2 + (hBS − hUT ) 2 )
− 9 log10 ((d BP
UMa

′ − NLOS )
PLUMa − NLOS = max( PLUMa −LOS , PLUMa
1.5m ≤ hUT ≤ 22.5m
for 10m ≤ d 2D ≤ 5km
hBS = 25m
NLOS

′ − NLOS = 13.54 + 39.08 log10 (d 3D ) +

PLUMa σ SF = 6 Explanations: see note
3
20 log10 ( f c ) − 0.6(hUT − 1.5)
Optional PL = 32.4 + 20 log10 ( f c ) + 30 log10 (d 3D ) σ SF = 7.8
′
 PL 10m ≤ d 2D ≤ d BP
PLUMi−LOS =  1
UMi - Street Canyon

′ ≤ d 2D ≤ 5km , see note 1

 PL2 d BP
1.5m ≤ hUT ≤ 22.5m
σ SF = 4
LOS

PL1 = 32.4 + 21log10 (d 3D ) + 20 log10 ( f c ) hBS = 10m

PL2 = 32.4 + 40 log10 (d 3D ) + 20 log10 ( f c )
′ ) 2 + (hBS − hUT ) 2 )
− 9.5 log10 ((d BP

3GPP
Release 16 28 3GPP TR 38.901 V16.1.0 (2019-12)

′ − NLOS )
PLUMi− NLOS = max( PLUMi−LOS , PLUMi
1.5m ≤ hUT ≤ 22.5m
NLOS for 10m ≤ d 2D ≤ 5km
σ SF = 7.82 hBS = 10m
′ − NLOS = 35.3 log10 (d 3D ) + 22.4
PLUMi Explanations: see note
4
+ 21.3 log10 ( f c ) − 0.3(hUT − 1.5)

3GPP
Release 16 29 3GPP TR 38.901 V16.1.0 (2019-12)

Optional PL = 32.4 + 20 log10 ( f c ) + 31.9 log10 (d 3D ) σ SF = 8.2

LOS
PLInH−LOS = 32.4 + 17.3 log10 (d 3D ) + 20 log10 ( f c ) σ SF = 3 1m ≤ d 3D ≤ 150m
InH - Office

′ − NLOS )
PLInH− NLOS = max( PLInH−LOS , PLInH
σ SF = 8.03 1m ≤ d 3D ≤ 150m
′ − NLOS = 38.3 log10 (d 3D ) + 17.30 + 24.9 log10 ( f c )
NLOS

PLInH
Optional ′ -NLOS = 32.4 + 20 log10 ( f c ) + 31.9 log10 (d 3D )
PLInH σ SF = 8.29 1m ≤ d 3D ≤ 150m
LOS

𝑃𝑃𝑃𝑃𝐿𝐿𝐿𝐿𝐿𝐿 = 31.84 + 21.50 log10 (𝑑𝑑3𝐷𝐷 ) + 19.00 log10 (𝑓𝑓𝑐𝑐 ) 𝜎𝜎𝑆𝑆𝑆𝑆 = 4.

InF-SL: 𝑃𝑃𝑃𝑃 = 33 + 25.5 log10 (𝑑𝑑3𝐷𝐷 ) + 20 log10 (𝑓𝑓𝑐𝑐 )

𝜎𝜎𝑆𝑆𝑆𝑆 = 5.7
𝑃𝑃𝑃𝑃𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = max (𝑃𝑃𝑃𝑃, 𝑃𝑃𝑃𝑃𝐿𝐿𝐿𝐿𝐿𝐿 )
InF

InF-DL: 𝑃𝑃𝑃𝑃 = 18.6 + 35.7 log10 (𝑑𝑑3𝐷𝐷 ) + 20 log10 (𝑓𝑓𝑐𝑐 ) 1 ≤ 𝑑𝑑3𝐷𝐷 ≤ 600 𝑚𝑚
𝜎𝜎𝑆𝑆𝑆𝑆 = 7.2
NLOS

𝑃𝑃𝑃𝑃𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = max (𝑃𝑃𝑃𝑃, 𝑃𝑃𝑃𝑃𝐿𝐿𝐿𝐿𝐿𝐿 , 𝑃𝑃𝑃𝑃𝐼𝐼𝐼𝐼𝐼𝐼−𝑆𝑆𝑆𝑆 )

InF-SH: 𝑃𝑃𝑃𝑃 = 32.4 + 23.0 log10 (𝑑𝑑3𝐷𝐷 ) + 20 log10 (𝑓𝑓𝑐𝑐 )

𝜎𝜎𝑆𝑆𝑆𝑆 = 5.9
𝑃𝑃𝑃𝑃𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = max (𝑃𝑃𝑃𝑃, 𝑃𝑃𝑃𝑃𝐿𝐿𝐿𝐿𝐿𝐿 )

InF-DH: 𝑃𝑃𝑃𝑃 = 33.63 + 21.9 log10 (𝑑𝑑3𝐷𝐷 ) + 20 log10 (𝑓𝑓𝑐𝑐 )

𝜎𝜎𝑆𝑆𝑆𝑆 = 4.0
𝑃𝑃𝑃𝑃𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = max (𝑃𝑃𝑃𝑃, 𝑃𝑃𝑃𝑃𝐿𝐿𝐿𝐿𝐿𝐿 )
Note 1: Breakpoint distance d'BP = 4 h'BS h'UT fc/c, where fc is the centre frequency in Hz, c = 3.0×108 m/s is the
propagation velocity in free space, and h'BS and h'UT are the effective antenna heights at the BS and the UT,
respectively. The effective antenna heights h'BS and h'UT are computed as follows: h'BS = hBS – hE, h'UT = hUT – hE,
where hBS and hUT are the actual antenna heights, and hE is the effective environment height. For UMi hE = 1.0m.
For UMa hE=1m with a probability equal to 1/(1+C(d2D, hUT)) and chosen from a discrete uniform distribution
uniform(12,15,…,(hUT-1.5)) otherwise. With C(d2D, hUT) given by
0 , hUT < 13m

C (d 2D , hUT ) =  hUT − 13 
1. 5
,

 10   g (d 2D ) ,13 m ≤ hUT ≤ 23m

where
0 , d 2D ≤ 18m

g (d 2D ) =  5  d 2D 
3
 − d 2D  .
 4  100  exp 150  ,18m < d 2D

Note that hE depends on d2D and hUT and thus needs to be independently determined for every link between BS
sites and UTs. A BS site may be a single BS or multiple co-located BSs.
Note 2: The applicable frequency range of the PL formula in this table is 0.5 < fc < fH GHz, where fH = 30 GHz for RMa
and fH = 100 GHz for all the other scenarios. It is noted that RMa pathloss model for >7 GHz is validated based
on a single measurement campaign conducted at 24 GHz.
Note 3: UMa NLOS pathloss is from TR36.873 with simplified format and PLUMa-LOS = Pathloss of UMa LOS outdoor
scenario.
Note 4: PLUMi-LOS = Pathloss of UMi-Street Canyon LOS outdoor scenario.
Note 5: Break point distance dBP = 2π hBS hUT fc/c, where fc is the centre frequency in Hz, c = 3.0 × 108 m/s is the
propagation velocity in free space, and hBS and hUT are the antenna heights at the BS and the UT, respectively.
Note 6: fc denotes the center frequency normalized by 1GHz, all distance related values are normalized by 1m, unless it
is stated otherwise.

3GPP
Release 16 30 3GPP TR 38.901 V16.1.0 (2019-12)

7.4.2 LOS probability

The Line-Of-Sight (LOS) probabilities are given in Table 7.4.2-1.

Table 7.4.2-1 LOS probability

Scenario LOS probability (distance is in meters)

RMa
 1 , d 2D -out ≤ 10m

PrLOS =   d 2D -out − 10 
exp −  ,10m < d 2D -out
  1000 
UMi - Street  1 , d 2D -out ≤ 18m
canyon 
PrLOS =  18  d  18 
+ exp − 2D -out 1 −  ,18m < d 2D -out
 d 2D -out  36  d
  2D - out 
UMa  1 , d 2D-out ≤ 18m

18  
3
PrLOS =  18  d  5d   d
+ exp − 2D-out 1 −  1 + C ′(hUT )  2D-out  exp − 2D-out   ,18m < d 2D-out
 d 2D-out  63  d 2D-out  4  100   150  

where
0 , hUT ≤ 13m
 1.5
′
C (hUT ) =   hUT − 13 
  10  ,13m < hUT ≤ 23m


Indoor - Mixed 
office 1
 , d 2D -in ≤ 1.2m
  d 2D -in − 1.2 
PrLOS = exp −  ,1.2m < d 2D -in < 6.5m
  4.7 
  d 2D -in − 6.5 
exp − 32.6 
 ⋅ 0.32 ,6.5m ≤ d 2D -in
 
Indoor - Open 
office 
1 , d 2D -in ≤ 5m
  d 2D -in − 5 
PrLOS = exp −  ,5m < d 2D -in ≤ 49m
  70.8 
  d 2D -in − 49 
exp − 211.7  ⋅ 0.54 ,49m < d 2D -in
  

InF-SL 𝑑𝑑2D
InF-SH PrLOS,subsce (𝑑𝑑2D ) = exp �− �
𝑘𝑘subsce
InF-DL
InF-DH where

𝑑𝑑𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
⎧ − for InF-SL and InF-DL
ln(1 − 𝑟𝑟)
𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 =
⎨− 𝑑𝑑𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ∙ ℎ𝐵𝐵𝐵𝐵 − ℎ𝑈𝑈𝑈𝑈 for InF-SH and InF-DH
⎩ ln(1 − 𝑟𝑟) ℎ𝑐𝑐 − ℎ𝑈𝑈𝑈𝑈

The parameters 𝑑𝑑𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 , 𝑟𝑟, and ℎ𝑐𝑐 are defined in Table 7.2-4
InF-HH PrLOS = 1

Note: The LOS probability is derived with assuming antenna heights of 3m for indoor, 10m for UMi, and 25m
for UMa

3GPP
Release 16 31 3GPP TR 38.901 V16.1.0 (2019-12)

7.4.3 O2I penetration loss

7.4.3.1 O2I building penetration loss

The pathloss incorporating O2I building penetration loss is modelled as in the following:

PL = PL b + PL tw + PL in + N 0, σ P2 ( ) (7.4-2)

where PL b is the basic outdoor path loss given in Clause 7.4.1, where d 3D is replaced by d 3D −out + d 3D −in . PL tw is
the building penetration loss through the external wall, PL in is the inside loss dependent on the depth into the building,
and σP is the standard deviation for the penetration loss.

PL tw is characterized as:

 Lmaterial _ i
N 
PL tw = PL npi − 10 log10 ∑ pi ×10 −10
 

i =1 

 (7.4-3)

PL npi is an additional loss is added to the external wall loss to account for non-perpendicular incidence;
Lmaterial _ i = amaterial _ i + bmaterial _ i ⋅ f , is the penetration loss of material i, example values of which can be found in
N
Table 7.4.3-1; pi is proportion of i-th materials, where ∑p i =1
i = 1 ; and N is the number of materials.

Table 7.4.3-1: Material penetration losses

Material Penetration loss [dB]

Standard multi-pane glass Lglass = 2 + 0.2 f
IRR glass LIIRglass = 23 + 0.3 f
Concrete Lconcrete = 5 + 4 f
Wood Lwood = 4.85 + 0.12 f
Note: f is in GHz

Table 7.4.3-2 gives PL tw , PL in and σP for two O2I penetration loss models. The O2I penetration is UT-specifically
generated, and is added to the SF realization in the log domain.

Table 7.4.3-2: O2I building penetration loss model

Path loss through external wall: Indoor loss: Standard deviation:

PL tw in [dB] PL in in [dB] σP in [dB]

 − Lglass − Lconcrete

Low-loss model 
5 − 10 log10 0.3 ⋅10 10
+ 0.7 ⋅10 10  0.5 d 2D−in 4.4
 
 
 − LIIRglass − Lconcrete

High-loss model 5 − 10 log10  0.7 ⋅ 10 10 + 0.3 ⋅ 10 10  0.5 d 2D−in 6.5
 
 

d 2D−in is minimum of two independently generated uniformly distributed variables between 0 and 25 m for UMa and
UMi-Street Canyon, and between 0 and 10 m for RMa. d 2D−in shall be UT-specifically generated.

3GPP
Release 16 32 3GPP TR 38.901 V16.1.0 (2019-12)

Both low-loss and high-loss models are applicable to UMa and UMi-Street Canyon.

Only the low-loss model is applicable to RMa.

Only the high-loss model is applicable to InF.

The composition of low and high loss is a simulation parameter that should be determined by the user of the channel
models, and is dependent on the use of metal-coated glass in buildings and the deployment scenarios. Such use is
expected to differ in different markets and regions of the world and also may increase over years to new regulations and
energy saving initiatives. Furthermore, the use of such high-loss glass currently appears to be more predominant in
commercial buildings than in residential buildings in some regions of the world (see note).

Note: One example survey for the US market can be found in [5]. The survey does not necessarily be
representative for all the scenarios. Other ratios outside of the survey should not be precluded.

For backwards compatibility with TR 36.873 [3], the following building penetration model should be used for UMa and
UMi single-frequency simulations at frequencies below 6 GHz.

Table 7.4.3-3. O2I building penetration loss model for single-frequency simulations <6 GHz

Parameter Value
PL tw 20 dB
0.5 d 2D−in
PL in with d 2D−in being a single, link-specific, uniformly distributed variable between 0
and 25 m
σP 0 dB

σ SF 7 dB (note: replacing the respective value in Table 7.4.1-1)

7.4.3.2 O2I car penetration loss

The pathloss incorporating O2I car penetration loss is modelled as in the following:

PL = PL b + Ν μ,σ P2 ( ) (7.4-4)

where PL b is the basic outdoor path loss given in Clause 7.4.1. μ = 9, and σP = 5. The car penetration loss shall be UT-
specifically generated. Optionally, for metallized car windows, μ = 20 can be used. The O2I car penetration loss models
are applicable for at least 0.6-60 GHz.

7.4.4 Autocorrelation of shadow fading

The long-term (log-normal) fading in the logarithmic scale around the mean path loss PL (dB) is characterized by a
Gaussian distribution with zero mean and standard deviation. Due to the slow fading process versus distance ∆x
(∆x is in the horizontal plane), adjacent fading values are correlated. Its normalized autocorrelation function R(∆x) can
be described with sufficient accuracy by the exponential function ITU-R Rec. P.1816 [18]
∆x
−
R(∆x )=e d cor
(7.4-5)

with the correlation length dcor being dependent on the environment, see the correlation parameters for shadowing and
other large scale parameters in Table 7.5-6 (Channel model parameters). In a spatial consistency procedure in Clause
7.6.3, the cluster specific random variables are also correlated following the exponential function with respect to
correlation distances in the two dimensional horizontal plane.

3GPP
Release 16 33 3GPP TR 38.901 V16.1.0 (2019-12)

7.5 Fast fading model

The radio channel realizations are created using the parameters listed in Table 7.5-1. The channel realizations are
obtained by a step-wise procedure illustrated in Figure 7.5-1 and described below. It has to be noted that the geometric
description covers arrival angles from the last bounce scatterers and respectively departure angles to the first scatterers
interacted from the transmitting side. The propagation between the first and the last interaction is not defined. Thus, this
approach can model also multiple interactions with the scattering media. This indicates also that e.g., the delay of a
multipath component cannot be determined by the geometry. In the following steps, downlink is assumed. For uplink,
arrival and departure parameters have to be swapped.

Note: the channel generation in this clause is enough for at least the following cases.

- Case 1: For low complexity evaluations

- Case 2: To compare with earlier simulation results,

- Case 3: When none of the additional modelling components are turned on.

For other advanced simulations, e.g., spatially consistency, large bandwidth and arrays, oxygen absorption, blockage,
absolute time of arrival, dual mobility, embedded devices, etc., some of the additional modelling components of Clause
7.6 should be considered.

General parameters:

Set scenario, Assign propagation Generate correlated

network layout and condition (NLOS/ large scale
Calculate pathloss
antenna parameters LOS) parameters (DS,
AS, SF, K)

Small scale parameters:

Perform random Generate arrival & Generate cluster

Generate XPRs Generate delays
coupling of rays departure angles powers

Coefficient generation:

Draw random initial Generate channel Apply pathloss and

phases coefficient shadowing

Figure 7.5-1 Channel coefficient generation procedure

3GPP
Release 16 34 3GPP TR 38.901 V16.1.0 (2019-12)

θ
φˆ
n̂

θˆ
y
φ

x
Figure 7.5-2: Definition of a global coordinate system showing the zenith angle θ and the azimuth
angle ϕ. θ=0° points to zenith and θ=+90° points to the horizon.
The spherical basis vectors θˆ and φˆ shown above are defined based on the direction of propagation
n̂ .

Table 7.5-1: Notations in the global coordinate system (GCS)

Parameter Notation
LOS AOD ϕLOS,AOD
LOS AOA ϕLOS,AOA
LOS ZOD θLOS,ZOD
LOS ZOA θLOS,ZOA
AOA for cluster n ϕn,AOA
AOD for cluster n ϕn,AOD
AOA for ray m in cluster n ϕn,m,AOA
AOD for ray m in cluster n ϕn,m,AOD
ZOA for cluster n θn,ZOA
ZOD for cluster n θn,ZOD
ZOA for ray m in cluster n θn,m,ZOA
ZOD for ray m in cluster n θn,m,ZOD

Receive antenna element u field pattern in the direction of the spherical basis vector θˆ Frx,u,θ

Receive antenna element u field pattern in the direction of the spherical basis vector φˆ Frx,u,ϕ

Transmit antenna element s field pattern in the direction of the spherical basis vector θˆ Ftx,s,θ

Transmit antenna element s field pattern in the direction of the spherical basis vector φˆ Frx,s,ϕ

Step 1: Set environment, network layout, and antenna array parameters

a) Choose one of the scenarios (e.g. UMa, UMi-Street Canyon, RMa, InH-Office or InF). Choose a global
coordinate system and define zenith angle θ, azimuth angle ϕ, and spherical basis vectors θˆ , φˆ as shown in
Figure 7.3-2. Note: Scenario RMa is for up to 7GHz while others are for up to 100GHz

b) Give number of BS and UT

c) Give 3D locations of BS and UT, and determine LOS AOD (ϕLOS,AOD), LOS ZOD (θLOS,ZOD), LOS AOA
(ϕLOS,AOA), and LOS ZOA (θLOS,ZOA) of each BS and UT in the global coordinate system

d) Give BS and UT antenna field patterns Frx and Ftx in the global coordinate system and array geometries
3GPP
Release 16 35 3GPP TR 38.901 V16.1.0 (2019-12)

e) Give BS and UT array orientations with respect to the global coordinate system. BS array orientation is defined
by three angles ΩBS,α (BS bearing angle), ΩBS,β (BS downtilt angle) and ΩBS,γ (BS slant angle). UT array
orientation is defined by three angles ΩUT,α (UT bearing angle), ΩUT,β (UT downtilt angle) and ΩUT,γ (UT slant
angle).

f) Give speed and direction of motion of UT in the global coordinate system

g) Specify system centre frequency fc and bandwidth B

Note: In case wrapping is used, each wrapping copy of a BS or site should be treated as a separate BS/site
considering channel generation.

Large scale parameters:

Step 2: Assign propagation condition (LOS/NLOS) according to Table 7.4.2-1. The propagation conditions for different
BS-UT links are uncorrelated.

Also, assign an indoor/outdoor state for each UT. It is noted that all the links from a UT have the same indoor/outdoor
state.

Step 3: Calculate pathloss with formulas in Table 7.4.1-1 for each BS-UT link to be modelled.

Step 4: 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.

These LSPs for different BS-UT links are uncorrelated, but the LSPs for links from co-sited sectors to a UT are the
same. In addition, these LSPs for the links of UTs on different floors are uncorrelated.

Limit random RMS azimuth arrival and azimuth departure spread values to 104 degrees, i.e., ASA= min(ASA, 104°),
ASD = min(ASD, 104°). Limit random RMS zenith arrival and zenith departure spread values to 52 degrees, i.e., ZSA =
min(ZSA, 52°), ZSD = min(ZSD, 52°).

Small scale parameters:

Step 5: Generate cluster delays τn .

Delays are drawn randomly from the delay distribution defined in Table 7.5-6. With exponential delay distribution
calculate

τ n′ = −rτ DS ln( X n ) , (7.5-1)

Where rτ is the delay distribution proportionality factor, Xn ~ uniform(0,1), and cluster index n = 1,…,N. With uniform
delay distribution the delay values τ n′ are drawn from the corresponding range. Normalise the delays by subtracting the
minimum delay and sort the normalised delays to ascending order:

τ n = sort (τ n′ − min (τ n′ )) . (7.5-2)

In the case of LOS condition, additional scaling of delays is required to compensate for the effect of LOS peak addition
to the delay spread. The heuristically determined Ricean K-factor dependent scaling constant is

Cτ = 0.7705 − 0.0433K + 0.0002 K 2 + 0.000017 K 3 , (7.5-3)

where K [dB] is the Ricean K-factor as generated in Step 4. The scaled delays

3GPP
Release 16 36 3GPP TR 38.901 V16.1.0 (2019-12)

τ nLOS = τ n / Cτ , (7.5-4)

are not to be used in cluster power generation.

Step 6: Generate cluster powers Pn .

Cluster powers are calculated assuming a single slope exponential power delay profile. Power assignment depends on
the delay distribution defined in Table 7.5-6. With exponential delay distribution the cluster powers are determined by
− Ζn
 rτ − 1 
Pn′ = exp − τ n  ⋅ 10 10 (7.5-5)
 rτ DS 

where ( )
Z n ~ N 0, ζ 2 is the per cluster shadowing term in [dB]. Normalize the cluster powers so that the sum of all
cluster powers is equal to one, i.e.,

Pn′
Pn = (7.5-6)
∑
N
P′
n =1 n

In the case of LOS condition an additional specular component is added to the first cluster. Power of the single LOS ray
is:

KR
P1, LOS =
KR + 1 (7.5-7)

and the cluster powers are not normalized as in equation (7.5-6) , but:

1 Pn′
Pn = N
+ δ(n − 1)P1, LOS
K R + 1 ∑ Pn′
n =1 (7.5-8)

where δ(.) is Dirac's delta function and KR is the Ricean K-factor as generated in Step 4 converted to linear scale. These
power values are used only in equations (7.5-9) and (7.5-14), but not in equation (7.5-22).

Assign the power of each ray within a cluster as Pn / M, where M is the number of rays per cluster.

Remove clusters with less than -25 dB power compared to the maximum cluster power. The scaling factors need not be
changed after cluster elimination.

Step 7: Generate arrival angles and departure angles for both azimuth and elevation.

The composite PAS in azimuth of all clusters is modelled as wrapped Gaussian. The AOAs are determined by applying
the inverse Gaussian function (7.5-9) with input parameters Pn and RMS angle spread ASA

2(ASA / 1.4) − ln (Pn max(Pn ))

φn′, AOA = , (7.5-9)
Cφ

with Cφ defined as

Cφ =  NLOS
(
CφNLOS ⋅ 1.1035 − 0.028K − 0.002 K 2 + 0.0001K 3 ) , for LOS
, (7.5-10)
Cφ , for NLOS

3GPP
Release 16 37 3GPP TR 38.901 V16.1.0 (2019-12)

NLOS
where Cφ is defined as a scaling factor related to the total number of clusters and is given in Table 7.5-2:

Table 7.5-2: Scaling factors for AOA, AOD generation

# clusters 4 5 8 10 11 12 14 15 16 19 20 25
NLOS
Cφ 0.779 0.860 1.018 1.090 1.123 1.146 1.190 1.211 1.226 1.273 1.289 1.358

In the LOS case, constant Cφ also depends on the Ricean K-factor K in [dB], as generated in Step 4. Additional scaling
of the angles is required to compensate for the effect of LOS peak addition to the angle spread.

Assign positive or negative sign to the angles by multiplying with a random variable Xn with uniform distribution to the
(
discrete set of {1,–1}, and add component Yn ~ N 0, (ASA 7 )
2
) to introduce random variation
φn , AOA = X nφn′, AOA + Yn + φLOS , AOA , (7.5-11)

where ϕLOS,AOA is the LOS direction defined in the network layout description, see Step1c.

In the LOS case, substitute (7.5-11) by (7.5-12) to enforce the first cluster to the LOS direction ϕLOS, AOA

φn , AOA = (X nφn′, AOA + Yn ) − (X 1φ1′, AOA + Y1 − φLOS , AOA ) (7.5-12)

Finally add offset angles αm from Table 7.5-3 to the cluster angles

φn ,m, AOA = φn , AOA + c ASAα m , (7.5-13)

where cASA is the cluster-wise rms azimuth spread of arrival angles (cluster ASA) in Table 7.5-6.

Table 7.5-3: Ray offset angles within a cluster, given for rms angle spread normalized to 1

Ray number m Basis vector of offset angles αm

1,2 ± 0.0447
3,4 ± 0.1413
5,6 ± 0.2492
7,8 ± 0.3715
9,10 ± 0.5129
11,12 ± 0.6797
13,14 ± 0.8844
15,16 ± 1.1481
17,18 ± 1.5195
19,20 ± 2.1551

The generation of AOD ( φn , m , AOD ) follows a procedure similar to AOA as described above.

The generation of ZOA assumes that the composite PAS in the zenith dimension of all clusters is Laplacian (see Table
7.5-6). The ZOAs are determined by applying the inverse Laplacian function (7.5-14) with input parameters Pn and
RMS angle spread ZSA

ZSA ln (Pn max(Pn ))

θ n′,ZOA = − , (7.5-14)
Cθ

with Cθ defined as

3GPP
Release 16 38 3GPP TR 38.901 V16.1.0 (2019-12)

Cθ =  NLOS
(
CθNLOS ⋅ 1.3086 + 0.0339 K − 0.0077 K 2 + 0.0002 K 3 ) , for LOS
, (7.5-15)
Cθ , for NLOS

Where CθNLOS is a scaling factor related to the total number of clusters and is given in Table 7.5-4:

Table 7.5-4: Scaling factors for ZOA, ZOD generation

# clusters 8 10 11 12 15 19 20 25
NLOS
Cθ 0.889 0.957 1.031 1.104 1.1088 1.184 1.178 1.282

In the LOS case, constant Cθ also depends on the Ricean K-factor K in [dB], as generated in Step 4. Additional scaling
of the angles is required to compensate for the effect of LOS peak addition to the angle spread.

Assign positive or negative sign to the angles by multiplying with a random variable Xn with uniform distribution to the
(
discrete set of {1,–1}, and add component Yn ~ N 0, (ZSA 7 )
2
) to introduce random variation
θ n ,ZOA = X nθ n′,ZOA + Yn + θ ZOA , (7.5-16)

where θ ZOA = 90° if the BS-UT link is O2I and θ ZOA = θ LOS ,ZOA otherwise. The LOS direction is defined in the
network layout description, see Step1c.

In the LOS case, substitute (7.5-16) by (7.5-17) to enforce the first cluster to the LOS direction θLOS,ZOA

θ n ,ZOA = (X nθ n′,ZOA + Yn ) − (X 1θ1′,ZOA + Y1 − θ LOS ,ZOA ). (7.5-17)

Finally add offset angles αm from Table 7.5-3 to the cluster angles

θ n,m,ZOA = θ n,ZOA + cZSAα m , (7.5-18)

where cZSA is the cluster-wise rms spread of ZOA (cluster ZSA) in Table 7.5-6. Assuming that θ n ,m, ZOA is wrapped
within [0, 360°], if θ n ,m,ZOA ∈ [180°,360°] , then θ n,m,ZOA is set to (360° − θ n ,m,ZOA ) .
The generation of ZOD follows the same procedure as ZOA described above except equation (7.5-16) is replaced by

θ n,ZOD = X nθ n′,ZOD + Yn + θ LOS ,ZOD + µ offset ,ZOD , (7.5-19)

where variable Xn is with uniform distribution to the discrete set of {1,–1}, Yn ~ N 0, (ZSD 7 ) , ( 2
) µ offset , ZOD is given
in Tables 7.5-6/7/8 and equation (7.5-18) is replaced by
µlgZSD
θ n ,m,ZOD = θ n ,ZOD + (3 / 8)(10 )α m (7.5-20)

where µlgZSD is the mean of the ZSD log-normal distribution.

In the LOS case, the generation of ZOD follows the same procedure as ZOA described above using equation (7.5-17).

Step 8: Coupling of rays within a cluster for both azimuth and elevation

Couple randomly AOD angles φn,m,AOD to AOA angles φn,m,AOA within a cluster n, or within a sub-cluster in the case of
two strongest clusters (see Step 11 and Table 7.5-3). Couple randomly ZOD angles θ n ,m , ZOD with ZOA angles

3GPP
Release 16 39 3GPP TR 38.901 V16.1.0 (2019-12)

θ n ,m,ZOA using the same procedure. Couple randomly AOD angles φn,m,AOD with ZOD angles θ n ,m,ZOD within a cluster n
or within a sub-cluster in the case of two strongest clusters.

Step 9: Generate the cross polarization power ratios

Generate the cross polarization power ratios (XPR) κ for each ray m of each cluster n. XPR is log-Normal distributed.
Draw XPR values as
X n ,m / 10
κ n , m = 10 , (7.5-21)

where
2
X n , m ~ N ( µ XPR , σ XPR ) is Gaussian distributed with σ XPR and µXPR from Table 7.5-6.

Note: X n , m is independently drawn for each ray and each cluster.

The outcome of Steps 1-9 shall be identical for all the links from co-sited sectors to a UT.

Coefficient generation:

Step 10: Draw initial random phases

{ θθ θφ φθ φφ
}
Draw random initial phase Φ n ,m , Φ n ,m , Φ n ,m , Φ n ,m for each ray m of each cluster n and for four different polarisation
combinations (θθ, θϕ, ϕθ, ϕϕ). The distribution for initial phases is uniform within (-π,π).

Step 11: Generate channel coefficients for each cluster n and each receiver and transmitter element pair u, s.

The method described below is used at least for drop-based evaluations irrespective of UT speeds. Relevant cases for
drop-based evaluations are:

- Case 1: For low complexity evaluations

- Case 2: To compare with earlier simulation results,

- Case 3: When none of the additional modelling components are turned on.

- Case 4: When spatial consistency and/or blockage is modeled for MU-MIMO simulations

- Other cases are not precluded

For the N – 2 weakest clusters, say n = 3, 4,…, N, the channel coefficients are given by:

Pn M  Frx ,u ,θ (θ n ,m ,ZOA ,φn ,m , AOA )

T
 exp jΦ θθ ( ) κ n ,m −1 exp( jΦ θφ
n ,m )

H NLOS
(t ) = ∑
m =1  Frx ,u ,φ (θ n , m , ZOA , φ n , m , AOA )
   n ,m

u , s ,n
M  κ n ,m exp jΦ φθ

−1
n ,m ( ) (
exp jΦ φφ
n ,m ) 

 Ftx ,s ,θ (θ n ,m ,ZOD ,φn ,m , AOD )
exp
(
 j 2π rˆrxT ,n ,m .d rx ,u )

 exp
(
 j 2π rˆtxT,n ,m .d tx ,s )   rˆ T .v 
 exp j 2π rx ,n ,m t 
 F (θ
 tx ,s ,φ n ,m ,ZOD ,φn ,m , AOD )
      
 λ0   λ0   λ0  (7.5-22)

where Frx,u,θ and Frx,u,ϕ are the field patterns of receive antenna element u according to (7.1-11) and in the direction of
the spherical basis vectors, θˆ and φˆ respectively, Ftx,s,θ and Ftx,s,ϕ are the field patterns of transmit antenna element s in
the direction of the spherical basis vectors, θˆ and φˆ respectively. Note that the patterns are given in the GCS and
therefore include transformations with respect to antenna orientation as described in Clause 7.1. rˆrx , n , m is the spherical
unit vector with azimuth arrival angle ϕn,m,AOA and elevation arrival angle θn,m,ZOA, given by
3GPP
Release 16 40 3GPP TR 38.901 V16.1.0 (2019-12)

sin θ n ,m ,ZOA cos φn ,m , AOA 

 
rˆrx ,n ,m =  sin θ n ,m ,ZOA sin φn ,m , AOA 
 cos θ n ,m ,ZOA 
, (7.5-23)

where n denotes a cluster and m denotes a ray within cluster n. rˆtx , n , m is the spherical unit vector with azimuth departure
angle ϕn,m,AOD and elevation departure angle θn,m,ZOD, given by

sin θ n ,m ,ZOD cos φn ,m , AOD 

 
rˆtx ,n ,m =  sin θ n ,m ,ZOD sin φn ,m , AOD 
 cos θ n ,m ,ZOD  , (7.5-24)

where n denotes a cluster and m denotes a ray within cluster n. Also, d rx ,u is the location vector of receive antenna
element u and d tx ,s is the location vector of transmit antenna element s, κn,m is the cross polarisation power ratio in
linear scale, and λ0 is the wavelength of the carrier frequency. If polarisation is not considered, the 2x2 polarisation
( )
matrix can be replaced by the scalar exp jΦ n ,m and only vertically polarised field patterns are applied.

The Doppler frequency component depends on the arrival angles (AOA, ZOA), and the UT velocity vector v with
speed v, travel azimuth angle ϕv, elevation angle θv and is given by

rˆrxT , n , m .v
, where v = v.[sin θ v cos φv cos θ v ] .
T
vn , m = sin θ v sin φv (7.5-25)
λ0
For the two strongest clusters, say n = 1 and 2, rays are spread in delay to three sub-clusters (per cluster), with fixed
delay offset. The delays of the sub-clusters are

τ n ,1 = τ n
τ n , 2 = τ n + 1.28 cDS (7.5-26)
τ n ,3 = τ n + 2.56 cDS

where cDS is cluster delay spread specified in Table 7.5-6. When intra-cluster delay spread is unspecified (i.e., N/A)
the value 3.91 ns is used; it is noted that this value results in the legacy behaviour with 5 and 10 ns sub-cluster delays.

Twenty rays of a cluster are mapped to sub-clusters as presented in Table 7.5-5 below. The corresponding offset angles
are taken from Table 7.5-3 with mapping of Table 7.5-5.

Table 7.5-5: Sub-cluster information for intra cluster delay spread clusters

sub-cluster # mapping to rays Power delay offset

i Ri Ri M τ n ,i − τ n
i =1 R1 = {1,2,3,4,5,6,7,8,19,20} 10/20 0

i=2 R2 = {9,10,11,12,17,18} 6/20 1.28 cDS

i=3 R3 = {13,14,15,16} 4/20 2.56 cDS

Then, the channel impulse response is given by:

2 3 N
H uNLOS
,s (τ , t ) = ∑∑ ∑ H uNLOS
, s , n , m (t )δ (τ − τ n ,i ) + ∑ H u , s , n (t )δ (τ − τ n )
NLOS
(7.5-27)
n =1 i =1 m∈Ri n =3

3GPP
Release 16 41 3GPP TR 38.901 V16.1.0 (2019-12)

where H uNLOS NLOS

, s , n (t ) is given in (7.5-22) and H u , s ,n ,m ( t ) defined as:

H NLOS
(t ) =
Pn  Frx ,u ,θ (θ n , m, ZOA , φn , m, AOA ) 
T


exp jΦ θθ (n,m ) κ n, m −1 exp( jΦθφ
n,m )


 rx ,u ,φ (θ n , m, ZOA , φn , m, AOA )
F 
u , s ,n,m
M  −1
 κ n , m exp jΦ n , m ( φθ
) (
exp jΦ φφ
n,m ) 
 (7.5-28)
 Ftx , s ,θ (θ n , m, ZOD , φn , m, AOD )  rˆrxT , n , m .d rx ,u   rˆtxT, n , m .d tx , s   rˆT
.v 
 F (θ exp j 2π  exp  j 2π  exp j 2π t
rx , n , m

 tx , s ,φ n , m, ZOD , φn , m, AOD )
      
 λ0   λ0   λ 0 
In the LOS case, determine the LOS channel coefficient by:

 Frx ,u ,θ (θ LOS ,ZOA , φLOS , AOA ) 1 0   Ftx ,s ,θ (θ LOS ,ZOD , φLOS , AOD )
T
LOS
H (t ) = 
 Frx ,u ,φ (θ LOS ,ZOA , φLOS , AOA ) 0 − 1  Ftx ,s ,φ (θ LOS ,ZOD , φLOS , AOD )
u ,s ,1    
(7.5-29)
 d   rˆT .d   rˆT .d   rˆT .v 
⋅ exp − j 2π 3D  exp j 2π rx ,LOS rx ,u  exp j 2π tx ,LOS tx ,s  exp j 2π rx ,LOS t 
 λ0   λ0   λ0   λ0 
where δ(.) is the Dirac's delta function and KR is the Ricean K-factor as generated in Step 4 converted to linear scale.

Then, the channel impulse response is given by adding the LOS channel coefficient to the NLOS channel impulse
response and scaling both terms according to the desired K-factor K R as

1 KR
, s (τ , t ) =
H uLOS H uNLOS
,s (τ , t ) + , s ,1 (t )δ (τ − τ 1 ) .
H uLOS (7.5-30)
KR +1 KR +1

Step 12: Apply pathloss and shadowing for the channel coefficients.

3GPP
Release 16 42 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.5-6 Part-1: Channel model parameters for UMi-Street Canyon and UMa

3GPP
Release 16 43 3GPP TR 38.901 V16.1.0 (2019-12)

UMi - Street Canyon UMa

Scenarios
LOS NLOS O2I LOS NLOS O2I
Delay spread (DS) µlgDS -0.24 log10(1+ fc) - 7.14 -0.24 log10(1+ fc) - 6.83 -6.62 -6.955 - 0.0963 log10(fc) -6.28 - 0.204 log10(fc) -6.62
lgDS=log10(DS/1s) σlgDS 0.38 0.16 log10(1+ fc) + 0.28 0.32 0.66 0.39 0.32
AOD spread (ASD) µlgASD -0.05 log10(1+ fc) + 1.21 -0.23 log10(1+ fc) + 1.53 1.25 1.06 + 0.1114 log10(fc) 1.5 - 0.1144 log10(fc) 1.25
lgASD=log10(ASD/1°) σlgASD 0.41 0.11 log10(1+ fc) + 0.33 0.42 0.28 0.28 0.42
AOA spread (ASA) µlgASA -0.08 log10(1+ fc) + 1.73 -0.08 log10(1+ fc) + 1.81 1.76 1.81 2.08 - 0.27 log10(fc) 1.76
lgASA=log10(ASA/1°) σlgASA 0.014 log10(1+ fc) + 0.28 0.05 log10(1+ fc) + 0.3 0.16 0.20 0.11 0.16
ZOA spread (ZSA) µlgZSA -0.1 log10(1+ fc) + 0.73 -0.04 log10(1+ fc) + 0.92 1.01 0.95 -0.3236 log10(fc) + 1.512 1.01
lgZSA=log10(ZSA/1°) σlgZSA -0.04 log10(1+ fc) + 0.34 -0.07 log10(1+ fc) + 0.41 0.43 0.16 0.16 0.43
Shadow fading (SF) [dB] σSF See Table 7.4.1-1 See Table 7.4.1-1 7 See Table 7.4.1-1 See Table 7.4.1-1 7
µK 9 N/A N/A 9 N/A N/A
K-factor (K) [dB]
σK 5 N/A N/A 3.5 N/A N/A
ASD vs DS 0.5 0 0.4 0.4 0.4 0.4
ASA vs DS 0.8 0.4 0.4 0.8 0.6 0.4
ASA vs SF -0.4 -0.4 0 -0.5 0 0
ASD vs SF -0.5 0 0.2 -0.5 -0.6 0.2
DS vs SF -0.4 -0.7 -0.5 -0.4 -0.4 -0.5
Cross-Correlations ASD vs ASA 0.4 0 0 0 0.4 0
ASD vs Κ -0.2 N/A N/A 0 N/A N/A
ASA vs Κ -0.3 N/A N/A -0.2 N/A N/A
DS vs Κ -0.7 N/A N/A -0.4 N/A N/A
SF vs Κ 0.5 N/A N/A 0 N/A N/A
ZSD vs SF 0 0 0 0 0 0
ZSA vs SF 0 0 0 -0.8 -0.4 0
ZSD vs K 0 N/A N/A 0 N/A N/A
ZSA vs K 0 N/A N/A 0 N/A N/A
ZSD vs DS 0 -0.5 -0.6 -0.2 -0.5 -0.6
Cross-Correlations 1) ZSA vs DS 0.2 0 -0.2 0 0 -0.2
ZSD vs ASD 0.5 0.5 -0.2 0.5 0.5 -0.2
ZSA vs ASD 0.3 0.5 0 0 -0.1 0
ZSD vs ASA 0 0 0 -0.3 0 0
ZSA vs ASA 0 0.2 0.5 0.4 0 0.5
ZSD vs ZSA 0 0 0.5 0 0 0.5
Delay scaling parameter rτ 3 2.1 2.2 2.5 2.3 2.2
µXPR 9 8.0 9 8 7 9
XPR [dB]
σXPR 3 3 5 4 3 5
Number of clusters N 12 19 12 12 20 12
Number of rays per cluster M 20 20 20 20 20 20
max(0.25, 6.5622 max(0.25, 6.5622
Cluster DS ( cDS ) in [ns] 5 11 11 11
-3.4084 log10(fc)) -3.4084 log10(fc))
3GPP
Release 16 44 3GPP TR 38.901 V16.1.0 (2019-12)

Cluster ASD ( c ASD ) in [deg] 3 10 5 5 2 5

Cluster ASA ( c ASA ) in [deg] 17 22 8 11 15 8

Cluster ZSA ( cZSA ) in [deg] 7 7 3 7 7 3

Per cluster shadowing std ζ [dB] 3 3 4 3 3 4
DS 7 10 10 30 40 10
ASD 8 10 11 18 50 11
ASA 8 9 17 15 50 17
Correlation distance in
SF 10 13 7 37 50 7
the horizontal plane [m]
Κ 15 N/A N/A 12 N/A N/A
ZSA 12 10 25 15 50 25
ZSD 12 10 25 15 50 25
fc is carrier frequency in GHz; d2D is BS-UT distance in km.
NOTE 1: DS = rms delay spread, ASD = rms azimuth spread of departure angles, ASA = rms azimuth spread of arrival angles, ZSD = rms zenith spread of departure angles,
ZSA = rms zenith spread of arrival angles, SF = shadow fading, and K = Ricean K-factor.
NOTE 2: The sign of the shadow fading is defined so that positive SF means more received power at UT than predicted by the path loss model.
NOTE 3: All large scale parameters are assumed to have no correlation between different floors.
NOTE 4: The following notation for mean (μlgX=mean{log10(X) }) and standard deviation (σlgX=std{log10(X) }) is used for logarithmized parameters X.
NOTE 5: For all considered scenarios the AOD/AOA distributions are modelled by a wrapped Gaussian distribution, the ZOD/ZOA distributions are modelled by a Laplacian
distribution and the delay distribution is modelled by an exponential distribution.
NOTE 6: For UMa and frequencies below 6 GHz, use fc = 6 when determining the values of the frequency-dependent LSP values
NOTE 7: For UMi and frequencies below 2 GHz, use fc = 2 when determining the values of the frequency-dependent LSP values

3GPP
Release 16 45 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.5-6 Part-2: Channel model parameters for RMa (up to 7GHz) and Indoor-Office

3GPP
Release 16 46 3GPP TR 38.901 V16.1.0 (2019-12)

RMa Indoor-Office
Scenarios
LOS NLOS O2I LOS NLOS
Delay spread (DS) µlgDS -7.49 -7.43 -7.47 -0.01 log10(1+fc) - 7.692 -0.28 log10(1+fc) - 7.173
lgDS=log10(DS/1s) σlgDS 0.55 0.48 0.24 0.18 0.10 log10(1+fc) + 0.055
AOD spread (ASD) µlgASD 0.90 0.95 0.67 1.60 1.62
lgASD=log10(ASD/1°) σlgASD 0.38 0.45 0.18 0.18 0.25
AOA spread (ASA) µlgASA 1.52 1.52 1.66 -0.19 log10(1+fc) + 1.781 -0.11 log10(1+fc) + 1.863
lgASA=log10(ASA/1°) σlgASA 0.24 0.13 0.21 0.12 log10(1+fc) + 0.119 0.12 log10(1+fc) + 0.059
ZOA spread (ZSA) µlgZSA 0.47 0.58 0.93 -0.26 log10(1+fc) + 1.44 -0.15 log10(1+fc) + 1.387
lgZSA=log10(ZSA/1°) σlgZSA 0.40 0.37 0.22 -0.04 log10(1+fc) + 0.264 -0.09 log10(1+fc) + 0.746
Shadow fading (SF) [dB] σSF See Table 7.4.1-1 8 See Table 7.4.1-1
µK 7 N/A N/A 7 N/A
K-factor (K) [dB]
σK 4 N/A N/A 4 N/A
ASD vs DS 0 -0.4 0 0.6 0.4
ASA vs DS 0 0 0 0.8 0
ASA vs SF 0 0 0 –0.5 –0.4
ASD vs SF 0 0.6 0 –0.4 0
DS vs SF -0.5 -0.5 0 –0.8 –0.5
Cross-Correlations ASD vs ASA 0 0 -0.7 0.4 0
ASD vs Κ 0 N/A N/A 0 N/A
ASA vs Κ 0 N/A N/A 0 N/A
DS vs Κ 0 N/A N/A -0.5 N/A
SF vs Κ 0 N/A N/A 0.5 N/A
ZSD vs SF 0.01 -0.04 0 0.2 0
ZSA vs SF -0.17 -0.25 0 0.3 0
ZSD vs K 0 N/A N/A 0 N/A
ZSA vs K -0.02 N/A N/A 0.1 N/A
ZSD vs DS -0.05 -0.10 0 0.1 -0.27
Cross-Correlations 1) ZSA vs DS 0.27 -0.40 0 0.2 -0.06
ZSD vs ASD 0.73 0.42 0.66 0.5 0.35
ZSA vs ASD -0.14 -0.27 0.47 0 0.23
ZSD vs ASA -0.20 -0.18 -0.55 0 -0.08
ZSA vs ASA 0.24 0.26 -0.22 0.5 0.43
ZSD vs ZSA -0.07 -0.27 0 0 0.42
Delay scaling parameter rτ 3.8 1.7 1.7 3.6 3
µXPR 12 7 7 11 10
XPR [dB]
σXPR 4 3 3 4 4
Number of clusters N 11 10 10 15 19
Number of rays per cluster M 20 20 20 20 20
Cluster DS ( cDS ) in [ns] N/A N/A N/A N/A N/A

3GPP
 Release 16 47 3GPP TR 38.901 V16.1.0 (2019-12)

Cluster ASD ( c ASD ) in [deg] 2 2 2 5 5

Cluster ASA ( c ASA ) in [deg] 3 3 3 8 11

Cluster ZSA ( cZSA ) in [deg] 3 3 3 9 9

Per cluster shadowing std ζ [dB] 3 3 3 6 3
DS 50 36 36 8 5
ASD 25 30 30 7 3
ASA 35 40 40 5 3
Correlation distance in the
SF 37 120 120 10 6
horizontal plane [m]
Κ 40 N/A N/A 4 N/A
ZSA 15 50 50 4 4
ZSD 15 50 50 4 4
fc is carrier frequency in GHz; d2D is BS-UT distance in km.
NOTE 1: DS = rms delay spread, ASD = rms azimuth spread of departure angles, ASA = rms azimuth spread of arrival angles, ZSD = rms zenith spread of departure angles, ZSA =
rms zenith spread of arrival angles, SF = shadow fading, and K = Ricean K-factor.
NOTE 2: The sign of the shadow fading is defined so that positive SF means more received power at UT than predicted by the path loss model.
NOTE 3: The following notation for mean (μlgX=mean{log10(X) }) and standard deviation (σlgX=std{log10(X) }) is used for logarithmized parameters X.
NOTE 4: Void.
NOTE 5: For all considered scenarios the AOD/AOA distributions are modelled by a wrapped Gaussian distribution, the ZOD/ZOA distributions are modelled by a Laplacian
distribution and the delay distribution is modelled by an exponential distribution.
NOTE 6: For InH and frequencies below 6 GHz, use fc = 6 when determining the values of the frequency-dependent LSP values

3GPP
Release 16 48 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.5-6 Part-3: Channel model parameters for InF

3GPP
Release 16 49 3GPP TR 38.901 V16.1.0 (2019-12)

InF
Scenarios
LOS NLOS
log10(26(V/S)+14)-9.35 log10(30(V/S)+32)-9.44
Delay spread (DS) µlgDS
lgDS=log10(DS/1s) 4)
σlgDS 0.15 0.19
AOD spread (ASD) µlgASD 1.56 1.57
lgASD=log10(ASD/1°) σlgASD 0.25 0.2
AOA spread (ASA) µlgASA -0.18*log10(1+fc) + 1.78 1.72
lgASA=log10(ASA/1°) σlgASA 0.12*log10(1+fc) + 0.2 0.3
ZOA spread (ZSA) µlgZSA -0.2*log10(1+fc) + 1.5 -0.13*log10(1+fc) + 1.45
lgZSA=log10(ZSA/1°) σlgZSA 0.35 0.45
Shadow fading (SF)
σSF Specified as part of path loss models
[dB]
µK 7 N/A
K-factor (K) [dB]
σK 8 N/A
ASD vs DS 0 0
ASA vs DS 0 0
ASA vs SF 0 0
ASD vs SF 0 0
DS vs SF 0 0
Cross-Correlations ASD vs ASA 0 0
ASD vs Κ -0.5 N/A
ASA vs Κ 0 N/A
DS vs Κ -0.7 N/A
SF vs Κ 0 N/A
ZSD vs SF 0 0
ZSA vs SF 0 0
ZSD vs K 0 N/A
ZSA vs K 0 N/A
ZSD vs DS 0 0
Cross-Correlations 1) ZSA vs DS 0 0
ZSD vs ASD 0 0
ZSA vs ASD 0 0
ZSD vs ASA 0 0
ZSA vs ASA 0 0
ZSD vs ZSA 0 0
Delay scaling parameter rτ 2.7 3
µXPR 12 11
XPR [dB]
σXPR 6 6
Number of clusters N 25 25
Number of rays per cluster M 20 20

c
Cluster DS ( DS ) in [ns] N/A N/A

c
Cluster ASD ( ASD ) in [deg] 5 5

c
Cluster ASA ( ASA ) in [deg] 8 8

c
Cluster ZSA ( ZSA ) in [deg] 9 9
Per cluster shadowing std ζ [dB] 4 3
DS 10 10
ASD 10 10
Correlation distance ASA 10 10
in the horizontal SF 10 10
plane [m] 5) Κ 10 N/A
ZSA 10 10
ZSD 10 10

3GPP
Release 16 50 3GPP TR 38.901 V16.1.0 (2019-12)

fc is carrier frequency in GHz; d2D is BS-UT distance in km.

NOTE 1: DS = rms delay spread, ASD = rms azimuth spread of departure angles, ASA = rms azimuth spread of arrival
angles, ZSD = rms zenith spread of departure angles, ZSA = rms zenith spread of arrival angles, SF = shadow
fading, and K = Ricean K-factor.
NOTE 2: The sign of the shadow fading is defined so that positive SF means more received power at UT than predicted
by the path loss model.
NOTE 3: The following notation for mean (μlgX=mean{log10(X) }) and standard deviation (σlgX=std{log10(X) }) is used for
logarithmized parameters X.
NOTE 4: V = hall volume in m3, S = total surface area of hall in m2 (walls+floor+ceiling)
NOTE 5: When the simulation scenario is small it is recommended to consider methods to improve the statistical
confidence, such as using multiple random seeds in simulations.

Table 7.5-7: ZSD and ZOD offset parameters for UMa

Scenarios LOS NLOS
max[-0.5, -2.1(d2D/1000) max[-0.5, -2.1(d2D/1000)
ZOD spread (ZSD) µlgZSD
-0.01 (hUT - 1.5)+0.75] -0.01(hUT - 1.5)+0.9]
lgZSD=log10(ZSD/1°)
σlgZSD 0.40 0.49
e(fc)-10^{a(fc) log10(max(b(fc), d2D))
ZOD offset µoffset,ZOD 0
+c(fc) -0.07(hUT-1.5)}
Note: For NLOS ZOD offset:
a(fc) = 0.208log10(fc)- 0.782;
b(fc) = 25;
c(fc) = -0.13log10(fc)+2.03;
e(fc) = 7.66log10(fc)-5.96.

Table 7.5-8: ZSD and ZOD offset parameters for UMi – Street Canyon

Scenarios LOS NLOS

max[-0.21, -14.8(d2D/1000) max[-0.5, -3.1(d2D/1000)
ZOD spread (ZSD) µlgZSD
+ 0.01|hUT-hBS| + 0.83] + 0.01 max(hUT-hBS,0) +0.2]
lgZSD=log10(ZSD/1°)
σlgZSD 0.35 0.35
ZOD offset µoffset,ZOD 0 -10^{-1.5log10(max(10, d2D))+3.3}

Table 7.5-9: ZSD and ZOD offset parameters for RMa

Scenarios LOS NLOS O2I

max[-1, -0.17(d2D/1000) max[-1, -0.19(d2D/1000) max[-1, -0.19(d2D/1000)
ZOD spread (ZSD) µlgZSD
-0.01(hUT - 1.5) + 0.22] -0.01(hUT - 1.5) + 0.28] -0.01(hUT - 1.5) + 0.28]
lgZSD=log10(ZSD/1°)
σlgZSD 0.34 0.30 0.30
arctan((35 - 3.5)/d2D) arctan((35 - 3.5)/d2D)
ZOD offset µoffset,ZOD 0
-arctan((35 - 1.5)/d2D) -arctan((35 - 1.5)/d2D)

Table 7.5-10: ZSD and ZOD offset parameters for Indoor-Office

Scenarios LOS NLOS

ZOD spread (ZSD) µlgZSD -1.43 log10(1+ fc)+2.228 1.08
lgZSD=log10(ZSD/1°) σlgZSD 0.13 log10(1+fc)+0.30 0.36
ZOD offset µoffset,ZOD 0 0

Table 7.5-11: ZSD and ZOD offset parameters for InF

Scenarios LOS NLOS

ZOD spread (ZSD) µlgZSD 1.35 1.2
lgZSD=log10(ZSD/1°) σlgZSD 0.35 0.55
ZOD offset µoffset,ZOD 0 0

Notes for Tables 7.5-7, 7.5-8, 7.5-9, 7.5-10, 7.5-11:

3GPP
Release 16 51 3GPP TR 38.901 V16.1.0 (2019-12)

NOTE 1: fc is center frequency in GHz; d2D is BS-UT distance in m.

NOTE 2: hBS and hUT are antenna heights in m for BS and UT respectively.

NOTE 3: The following notation for mean (μlgX=mean{log10(X) }) and standard deviation (σlgX=std{log10(X) }) is
used for logarithmized parameters X.

NOTE 4: For frequencies below 6 GHz, use fc = 6 when determining the values of the frequency-dependent ZSD
and ZOD offset parameters in Table 7.5-7 and 7.5-10

NOTE 5: The ZSD parameters for O2I links are the same parameters that are used for outdoor links, depending on
the LOS condition of the outdoor link part in Tables 7.5-7 and 7.5-8.

7.6 Additional modelling components

The additional modelling components in this clause are to support advanced simulations, such as simulations with very
large arrays and large bandwidth, simulations affected by oxygen absorption (frequencies between 53 and 67 GHz),
simulations in which spatial consistency is important (e.g. high number of closely located user), simulations of mobility,
simulations of blockage effects, simulations of absolute time of arrival, simulations of dual mobility, simulations that
include sources of EM interference, and simulations of devices embedded in machinery or enclosures. These modelling
components affect some of the steps between Step 1 and Step 12 in Clause 7.5.

These extensions are computationally more expensive and might not be required in all evaluation cases.

7.6.1 Oxygen absorption

Oxygen absorption loss is applied to the cluster responses generated in Step 11 in Clause 7.5. The additional loss,
OLn(fc) for cluster n at centre frequency fc is modelled as:

α ( fc )
OLn ( f c ) = ⋅ (d 3 D + c ⋅ (τ n + τ ∆ )) [dB] (7.6-1)
1000
where:

- α(fc) is frequency dependent oxygen loss [dB/km] characterized in Table 7.6.1-1;

- c is the speed of light [m/s]; and d3D is the distance [m];

- τn is the n-th cluster delay [s] in Step 11 in Clause 7.5;

- τΔ is 0 in the LOS case and min( τ n′ ) otherwise, where min( τ n′ ) is the minimum delay in Step 5.

For centre frequencies not specified in this table, the frequency dependent oxygen loss α(fc) is obtained from a linear
interpolation between two loss values corresponding to the two adjacent centre frequencies of the centre frequency fc.

Table 7.6.1-1 Frequency dependent oxygen loss α(f) [dB/km]

f in [GHz] 0-52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68-100

α(f) in [dB/km] 0 1 2.2 4 6.6 9.7 12.6 14.6 15 14.6 14.3 10.5 6.8 3.9 1.9 1 0

For large channel bandwidth, first transform the time-domain channel response of each cluster (all rays within one
cluster share common oxygen absorption loss for simplicity) into frequency-domain channel response, and apply the
oxygen absorption loss to the cluster's frequency-domain channel response for frequency fc + Δf within the considered
bandwidth. The oxygen loss, OLn(fc+ Δf) for cluster n at frequency fc+ Δf is modelled as:
α ( f c + ∆f )
( f c + ∆f )
OLn= ⋅ (d3 D + c ⋅ (τ n + τ ∆ ))
1000 [dB] (7.6-2)

3GPP
Release 16 52 3GPP TR 38.901 V16.1.0 (2019-12)

where:
- α(fc+ Δf) is the oxygen loss (dB/km) at frequency fc+ Δf characterized in Table 7.6.1-1. Note that Δf is in [-B/2,
B/2], where B is the bandwidth. Linear interpolation is applied for frequencies not provided in Table 7.6.1-1.

The final frequency-domain channel response is obtained by the summation of frequency-domain channel responses of
all clusters.
Time-domain channel response is obtained by the reverse transform from the obtained frequency-domain channel
response.

7.6.2 Large bandwidth and large antenna array

7.6.2.1 Modelling of the propagation delay

The modelling in this clause applies only when the bandwidth B is greater than c/D Hz, where

- D is the maximum antenna aperture in either azimuth or elevation (m)

- c is the speed of light (m/s)

Each ray within a cluster for a given u (Rx) and s (Tx) has unique time of arrival (TOA). The channel coefficient
generation step (Step 11 in Clause 7.5) is updated to model individual rays. In this case, the channel response of ray m
in cluster n for a link between Rx antenna u and Tx antenna s at delay τ at time t is given by:

H NLOS  Frx ,u ,θ (θ n ,m,ZOA ,φ n ,m, AOA )

(t ;τ ) = Pn ,m 
T


(
exp jΦ θθ n ,m ) (
κ n ,m −1 exp jΦ θφ n ,m )
 Frx ,u ,φ (θ n ,m ,ZOA ,φn ,m , AOA )

u , s ,n ,m
 κ n ,m exp jΦ n ,m

−1
( φθ
) (
exp jΦ n ,m φφ
) 

 Ftx ,s ,θ (θ n ,m ,ZOD ,φn ,m , AOD ) 
exp
( )
j 2π rˆrxT ,n ,m .d rx ,u 
 exp
( )
 j 2π rˆtxT,n ,m .d tx ,s 
 (7.6-3)
 F (θ
 tx ,s ,φ n ,m ,ZOD ,φn ,m , AOD )

 λ( f ) 


 λ ( f ) 

 rˆ T .v 
⋅ exp j 2π rx ,n ,m t δ (τ − τ n ,m )
 λ0 

 B B
with λ ( f ) is the wavelength on frequency f ∈  f c − , f c +  , which can be implemented by user's own
 2 2
method. The delay (TOA) for ray m in cluster n for a link between Rx antenna u and Tx antenna s is given by:

τ u ,s ,n ,m = τ n ,m − 1c rˆrxT ,n ,m .d rx ,u − 1c rˆtxT,n ,m .d tx ,s (7.6-4)

Note that Equation (7.6-3) only considers the delays τ n, m intentionally. If unequal ray powers are considered, Pn ,m are
generated according to Clause 7.6.2.2. Otherwise, ray powers are equal within a cluster, i.e., Pn , m = Pn M for all m.
Note: this model is developed assuming plane wave propagation.

7.6.2.2 Modelling of intra-cluster angular and delay spreads

With large antenna arrays or large bandwidths, the angle and/or delay resolution can be larger than what the fast fading
model in Clause 7.5 is designed to support. To model this effect, the following modifications to Step 7 in Clause 7.5 can
be optionally used.

1. The offset angles αm in (7.5-13), (7.5-18) and (7.5-20) are generated independently per cluster and ray using:

α n , m,{AOA,AOD,ZOA,ZOD} ~ unif (− 2,2) (7.6-5)

3GPP
Release 16 53 3GPP TR 38.901 V16.1.0 (2019-12)

[ ]
where unif (a, b ) denotes the continuous uniform distribution in the interval a, b . These random variables may
further be modelled as spatially consistent with correlation distance equal to the cluster-specific random variable
correlation distance of Table 7.6.3.1-2.

2. The relative delay of m-th ray is given by (

τ n′ ,m = τ n′′,m − min {τ n′′,i }iM=1 ) , where n is the cluster index,
τ n′′,m ~ unif (0, 2cDS ) , the cluster DS cDS is given in Table 7.5-6. These random variables may further be
modelled as spatially consistent with correlation distance equal to the cluster-specific random variable
correlation distance of Table 7.6.3.1-2. In this case, the sub-cluster mapping according to (7.5-26) and Table 7.5-
5 shall not be applied. The delays to be used in Equation (7.6-3) are given by τ n ,m = τ n + τ n′ ,m .

3. Ray powers are determined unequally by the following process:

Pn′,m
The power of m-th ray in n-th cluster is given by Pn ,m = Pn ⋅ M
for m = 1,…,M, where
∑ P′
m =1
n ,m

 τ′   2 α n ,m,AOA   2 α n ,m,AOD 
Pn′,m = exp − n ,m  exp −  exp − 
 cDS   cASA   cASD 
   
(7.6-6)
 2 α n ,m,ZOA   2 α n ,m,ZOD 
⋅ exp −  exp − 
 cZSA   cZSD 
   

and cDS , cASA , cASD , and cZSA are respectively the intra-cluster delay spread and the corresponding intra-cluster
angular spread that are given in Table 7.5-6. The cluster zenith spread of departure is given by

3 µ
cZSD = 10 lgZSD , (7.6-7)
8
with µlgZSD being defined in Tables 7.5-7, 7.5-8, 7.5-9, 7.5-10 and 7.5-11.

4. The number of rays per cluster shall be calculated as follows:

M = min{max(M t M AOD M ZOD ,20), M max } (7.6-8)

where:

- M t = 4kcDS B 

 π ⋅ Dh 
- M AOD = 4kc ASD
 180 ⋅ λ 

 π ⋅ Dv 
- M ZOD = 4kc ZSD
 180 ⋅ λ 

- M max is the upper limit of M , and it should be selected by the user of channel model based on the trade-off
between simulation complexity and accuracy.

- Dh and Dv are the array size in m in horizontal and vertical dimension, B is bandwidth in Hz, c ASD and c ZSD
are the cluster spreads in degrees, and λ is the wavelength.

- k is a "sparseness" parameter with value 0.5.

3GPP
Release 16 54 3GPP TR 38.901 V16.1.0 (2019-12)

It is noted that each MPC may have different AOD, ZOD, and delay.

7.6.3 Spatial consistency

7.6.3.1 Spatial consistency procedure

A new procedure, namely a spatial consistency procedure, can be used for both cluster-specific and ray-specific
random variables to be generated in various channel generation steps in Clause 7.5, so that they are spatially consistent
for drop-based simulations. Alternatively, this can be used together with Procedure B described in Clause 7.6.3.2 for
spatially consistent mobility simulations.

The procedure can be considered as a 2D random process (in the horizontal plane) given the UT locations based on the
parameter-specific correlation distance values for spatial consistency, specified in Table 7.6.3.1-2. The cluster specific
random variables include:

- Cluster specific random delay in Step 5;

- Cluster specific shadowing in Step 6; and

- Cluster specific offset for AOD/AOA/ZOD/ZOA in Step 7.

- Cluster specific sign for AOD/AOA/ZOD/ZOA in Step 7.

- Optionally in case of large bandwidth as described in Clause 7.6.2.2 the procedure may apply as well for the
parameters of rays within a cluster.

The procedure shall apply to each cluster before sorting the delay. Cluster specific sign for AOD/AOA/ZOD/ZOA in
Step 7 shall be kept unchanged per simulation drop even if UT position changes during simulation. The ray specific
random variables include:

- Random coupling of rays in Step 8;

- XPR in Step 9; and

- Random phase in Step 10.

The random coupling of rays in Step 8 and the intra-cluster delays in Step 11 shall be kept unchanged per simulation
drop even if UT position changes during simulation.

Table 7.6.3.1-2 Correlation distance for spatial consistency

Correlation RMa UMi UMa Indoor InF

distance in [m]
LOS NLOS O2I LOS NLOS O2I LOS NLOS O2I

Cluster and ray 50 60 15 12 15 15 40 50 15 10 10

specific random
variables

LOS/NLOS state 60 50 50 10 𝑑𝑑𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ⁄2

Indoor/outdoor 50 50 50 N/A N/A

state

7.6.3.2 Spatially-consistent UT/BS mobility modelling

For mobility simulation enhancement, two alternative spatial consistency procedures – Procedure A and Procedure B –
are described as follows. The procedures presented below consider the downlink direction same as in Clause 7.5.

Procedure A:

For t0 = 0 when a UT/BS is dropped into the network, spatially consistent powers/delays/angles of clusters are
generated according to Clause 7.6.3.1.

3GPP
Release 16 55 3GPP TR 38.901 V16.1.0 (2019-12)

The updated distance of UT/BS should be limited within 1 meter, i.e. when v ⋅ ∆t < 1 m , and the updated procedure in
the following should take the closest realization instead of t0 = 0 . In case the absolute time of arrival modeling of
clause 7.6.9 is simultaneously used, the update distance should be the minimum of 1 meter and 𝑑𝑑3D ⁄10.

Att k = t k −1 + ∆t , update channel cluster power/delay/angles based on UT/BS channel cluster power/delay/angles,
moving speed moving direction and UT position at tk −1 .

Cluster delay is updated as:

𝑟𝑟̂rx,𝑛𝑛 (𝑡𝑡𝑘𝑘−1 )𝑇𝑇 𝑣𝑣�rx (𝑡𝑡𝑘𝑘−1 )+𝑟𝑟̂tx,𝑛𝑛(𝑡𝑡𝑘𝑘−1 )𝑇𝑇 𝑣𝑣�tx (𝑡𝑡𝑘𝑘−1 )
𝜏𝜏̃𝑛𝑛 (𝑡𝑡𝑘𝑘−1 ) − ∆𝑡𝑡 for 𝑘𝑘 > 0
𝜏𝜏̃𝑛𝑛 (𝑡𝑡𝑘𝑘 ) = � 𝑐𝑐
𝑑𝑑3D (𝑡𝑡0 )
(7.6-9)
𝜏𝜏𝑛𝑛 (𝑡𝑡0 ) + 𝜏𝜏∆ (𝑡𝑡0 ) + for 𝑘𝑘 = 0
𝑐𝑐

where

c is the speed of light in [m/s];

𝑣𝑣̅tx (𝑡𝑡𝑘𝑘 ) and 𝑣𝑣̅rx (𝑡𝑡𝑘𝑘 ) are the BS and UTvelocity vectors respectively, with ‖𝑣𝑣𝑡𝑡𝑡𝑡 (𝑡𝑡𝑘𝑘 )‖2 = 𝑣𝑣𝑡𝑡𝑡𝑡 and ‖𝑣𝑣𝑟𝑟𝑟𝑟 (𝑡𝑡𝑘𝑘 )‖2 = 𝑣𝑣𝑟𝑟𝑟𝑟 ;

d3D (t0 ) is the 3D distance between Rx antenna and Tx antenna at t0 in [m];

τ n (t0 ) are the cluster delays in [s] as in Step 11 in Clause 7.5;

 0 for LOS
τ ∆ (t0 ) =  τ n′
(
min {τ ′ }
N
n n =1 ) for NLOS
, where are the delays from Equation (7.5-1). In case the absolute time

0 for LOS
of arrival modelling of clause 7.6.9 is simultaneously used, τ ∆ (t0 ) =  where Δ𝜏𝜏 is
∆τ for NLOS
determined according to clause 7.6.9.

Also,

sin (θ n ,ZOA (tk −1 ) ) cos (φn ,AOA (tk −1 ) ) 

 
rˆrx ,n ( tk −1 ) =  sin (θ n ,ZOA (tk −1 ) ) sin (φn ,AOA (tk −1 ) )  (7.6-10)
 
 cos (θ n ,ZOA (tk −1 ) ) 

where θ n,ZOA and φn, AOA are cluster specific zenith and azimuth angles of arrival.

⎡sin �𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘−1 )� cos �𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘−1 )�⎤

⎢ ⎥
𝑟𝑟̂tx,𝑛𝑛 (𝑡𝑡𝑘𝑘−1 ) = ⎢ sin �𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘−1 )� sin �𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘−1 )� ⎥ (7.6-10aa)
⎢ ⎥
⎣ cos �𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘−1 )� ⎦

where 𝜃𝜃𝑛𝑛,ZOD and 𝜙𝜙𝑛𝑛,AOD are the cluster specific zenith and azimuth angles of departure.

After updating the delays according to equation (7.6-9), the delays over the mobility range are normalized. Equation
(7.5-2) of the fast fading model is replaced by

τ n (tk ) = τ~n (tk ) − min {τ~n (tk )}nN=1 ( ) (7.6-10a)

in which tk covers the entire duration of the mobility model.

3GPP
Release 16 56 3GPP TR 38.901 V16.1.0 (2019-12)

Cluster powers are updated as in Step 6 using the cluster delays from Equation (7.6-10a).
Cluster departure angles ( θ n , ZOD and φn , AOD ) and arrival angles ( θ n , ZOA and φn , AOA ) are updated using a cluster-wise
transformation of the UT/BS velocity vector given by
𝑣𝑣̅rx (𝑡𝑡𝑘𝑘 )−𝑣𝑣̅tx (𝑡𝑡𝑘𝑘 ) for LOS
′
𝑣𝑣̅𝑛𝑛,rx (𝑡𝑡𝑘𝑘 ) = � (7.6-10b)
𝑅𝑅𝑛𝑛,rx ∙ 𝑣𝑣̅rx (𝑡𝑡𝑘𝑘 ) − 𝑣𝑣̅tx (𝑡𝑡𝑘𝑘 ) for NLOS

𝑣𝑣̅tx (𝑡𝑡𝑘𝑘 )−𝑣𝑣̅rx (𝑡𝑡𝑘𝑘 ) for LOS

′
𝑣𝑣̅𝑛𝑛,tx (𝑡𝑡𝑘𝑘 ) = � (7.6-10c)
𝑅𝑅𝑛𝑛,tx ∙ 𝑣𝑣̅tx (𝑡𝑡𝑘𝑘 ) − 𝑣𝑣̅rx (𝑡𝑡𝑘𝑘 ) for NLOS

with
1 0 0
𝜋𝜋 𝜋𝜋
𝑅𝑅𝑛𝑛,rx = 𝑅𝑅𝑍𝑍 �𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘 ) + 𝜋𝜋� ∙ 𝑅𝑅𝑌𝑌 � −𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘 )� ∙ �0 𝑋𝑋𝑛𝑛 0� ∙ 𝑅𝑅𝑌𝑌 � 2 − 𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘 )� ∙ 𝑅𝑅𝑍𝑍 �−𝜙𝜙𝑛𝑛,AOA (𝑡𝑡𝑘𝑘 )�,
2
0 0 1
and
1 0 0
𝜋𝜋 𝜋𝜋
𝑅𝑅𝑛𝑛,tx = 𝑅𝑅𝑍𝑍 �−𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘 )� ∙ 𝑅𝑅𝑌𝑌 � −𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘 )� ∙ �0 𝑋𝑋𝑛𝑛 0� ∙ 𝑅𝑅𝑌𝑌 � 2 −𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘 )� ∙ 𝑅𝑅𝑍𝑍 �𝜙𝜙𝑛𝑛,AOA (𝑡𝑡𝑘𝑘 ) + 𝜋𝜋�,
2
0 0 1

where RY (β ) and RZ (α ) being the rotation matrices around y-axis and z-axis as defined in Equation (7.1-2). Random
variable X n ∈ {1,−1} is sampled from a uniform distribution on a NLOS cluster basis and is not changed during the
UT mobility within a drop. The cluster specific decorrelation distances are 60m, 15m, 50m and 10m for RMa, UMi,
UMa and Indoor scenarios, respectively.

Now, the departure and arrival angles in radians are updated as:
′
𝑣𝑣�𝑛𝑛,rx (𝑡𝑡𝑘𝑘−1 )𝑇𝑇 𝜙𝜙
� �𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘−1 ),𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘−1 )�
𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘 ) = 𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘−1 ) + ∆𝑡𝑡 (7.6-11)
𝑐𝑐∙𝜏𝜏�𝑛𝑛 (𝑡𝑡𝑘𝑘−1 ) sin�𝜃𝜃𝑛𝑛,𝑍𝑍𝑍𝑍𝑍𝑍 (𝑡𝑡𝑘𝑘−1 )�

′
𝑣𝑣�𝑛𝑛,rx (𝑡𝑡𝑘𝑘−1 )𝑇𝑇 𝜃𝜃
� �𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘−1 ),𝜙𝜙𝑛𝑛,AOD (𝑡𝑡𝑘𝑘−1 )�
𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘 ) = 𝜃𝜃𝑛𝑛,ZOD (𝑡𝑡𝑘𝑘−1 ) + ∆𝑡𝑡 (7.6-12)
𝑐𝑐∙𝜏𝜏�𝑛𝑛(𝑡𝑡𝑘𝑘−1 )

and
′ (𝑡𝑡 𝑇𝑇 �
𝑣𝑣�𝑛𝑛,tx 𝑘𝑘−1 ) 𝜙𝜙 �𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘−1 ),𝜙𝜙𝑛𝑛,AOA (𝑡𝑡𝑘𝑘−1 )�
𝜙𝜙𝑛𝑛,AOA (𝑡𝑡𝑘𝑘 ) = 𝜙𝜙𝑛𝑛,AOA (𝑡𝑡𝑘𝑘−1 ) + ∆𝑡𝑡 (7.6-13)
𝑐𝑐∙𝜏𝜏�𝑛𝑛 (𝑡𝑡𝑘𝑘−1 ) sin�𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘−1 )�

′ (𝑡𝑡 𝑇𝑇 �
𝑣𝑣�𝑛𝑛,tx 𝑘𝑘−1 ) 𝜃𝜃 �𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘−1 ),𝜙𝜙𝑛𝑛,AOA (𝑡𝑡𝑘𝑘−1 )�
𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘 ) = 𝜃𝜃𝑛𝑛,ZOA (𝑡𝑡𝑘𝑘−1 ) + ∆𝑡𝑡, (7.6-14)
𝑐𝑐∙𝜏𝜏�𝑛𝑛(𝑡𝑡𝑘𝑘−1 )

with θˆ(θ ,φ ) and φˆ(θ ,φ ) being the spherical unit vectors defined in Equations (7.1-13) and (7.1-14).

Procedure B:

In procedure B, spatial or time evolution of the channel is obtained by generating channel realizations separately for all
links to different Rx positions using Steps 1-12 of Clause 7.5 together with the spatially consistent procedure of Clause
7.6.3.1. In the case of mobility these positions may be a function of time along one or more Rx trajectories.
Furthermore, to ensure that the spatial or time evolution of delays and angles are within reasonable limits, Steps 5, 6,
and 7 in Clause 7.5 should be replaced by the below procedure.

Note: For implementation purposes, LSPs and SSPs may be interpolated within the coherence length or time of the
respective parameter.

Step 5: Generate cluster delays τ n , with n ∈ [1, N ] .

N delays are drawn randomly from a uniform distribution.

3GPP
Release 16 57 3GPP TR 38.901 V16.1.0 (2019-12)

(
τ n′ ~ unif 0, 2 ⋅10
µ lgDS +σ lgDS
) (7.6-15)

Normalise the delays by subtracting the minimum delay: τ n = τ n′ − min (τ n′ ) . The autocorrelation distance for τ n′ is
µ lgDS +σ lgDS
2c ⋅ 10 .

In the case of LOS, set the delay of the first cluster τ 1 to 0.

Step 6: Generate N arrival angles and departure angles for both azimuth and elevation using (7.6-16) instead of (7.5-9)
and (7.5-14).

unif (− 1,1)
µ lgASA +σ lgASA
φn′, AOA ~ 2 ⋅10 (7.6-16)

[ ]
with n ∈ 1, N . This step is repeated independently for AOD, AOA, ZOD, and ZOA with corresponding maximum
angles for the uniform distribution. In case of LOS, set the angles of the first cluster ( φ1′, AOA , etc) to 0.

µ lgDS +σ lgDS
The autocorrelation distances are 2c ⋅ 10 for AOD, and ZOD, while they are fixed to 50m for AOA, and
ZOA.

Offset angles etc are applied in the modified Step 7b below after cluster powers have been calculated.

Step 7: Generate cluster powers Pn

Cluster powers are calculated assuming a single slope exponential power profile and Laplacian angular power profiles.
The cluster powers are determined by

 −τ ′   − 2 φn′,AOA   − 2 φn′,AOD 
Pn′ = exp n  exp  exp 
 DS   ASA   ASD 
   
(7.6-17)
 − 2 θ n′,ZOA   − 2 θ n′,ZOD  −Zn
⋅ exp  exp  ⋅ 10 10
 ZSA   ZSD 
   

where Z n ~ N (0, ζ 2 ) (autocorrelation distance same as for shadow fading) is the per cluster shadowing term in [dB].
Delay spread DS and angular spreads ASA , ASD , ZSA , ZSD are generated in Step 4 of Clause 7.5. Normalize
the cluster powers so that the sum of all cluster powers is equal to one, i.e.,

Pn′
Pn = (7.6-17a)
∑
N
P′
n =1 n

In the case of LOS condition, substitute DS with 1 + K R 2 ⋅ DS and { ASA , ASD , ZSA , ZSD } with
{ 1 + K R ⋅ ASA , 1 + K R ⋅ ASD , 1 + K R ⋅ ZSA , 1 + K R ⋅ ZSD } respectively to preserve the delay and
angular spreads. K R is the Ricean K-factor as generated in Step 4 converted to linear scale. Furthermore, an additional
specular component is added to the first cluster. Power of the single LOS ray is:

KR
P1, LOS =
KR + 1 (7.6-17b)

and the cluster powers are not normalized as in equation (7.6-17a), but:

1 Pn′
Pn = N
+ δ(n − 1)P1, LOS
K R + 1 ∑ Pn′
n =1 (7.6-17c)

3GPP
Release 16 58 3GPP TR 38.901 V16.1.0 (2019-12)

where δ(.) is Dirac's delta function.

Assign the power of each ray within a cluster as Pn / M, where M is the number of rays per cluster.

Step 7b: Apply offset angles

The ray AOA angles are determined by

φ n ,m,AOA = φ n′,AOA + φ LOS ,AOA + c ASAα m (7.6-17d)

where φ LOS,AOA is the LOS direction defined in the network layout description, see Step1c.

The generation of AOD ( φn , m , AOD ) follows a procedure similar to AOA as described above.

The ZOA angles are determined by

θ n ,m , ZOA = φ n′, ZOA + θ ZOA + c ZSAα m (7.6-17e)

where θ ZOA = 90 0 if the BS-UT link is O2I and θ ZOA = θ LOS ,ZOA otherwise. The LOS direction is defined in the
network layout description, see Step1c.

The ZOA angles are determined by

µlgZSD
θ n,m ,ZOD = φn′,ZOD + θ LOS ,ZOD + µ offset ,ZOD + (3 8)(10 )α m (7.6-17f)

where µlgZSD is the mean of the ZSD log-normal distribution.

Some of the delay and angle spreads and standard deviations used in equations (7.6-15) and (7.6-16) may be frequency-
dependent. In the case of multi-frequency simulations according to Clause 7.6.5, the largest value among all the
simulated frequencies should be used in (7.6-15) and (7.6-16) so that the cluster and ray delays and angles (but not the
powers or the resulting delay or angular spreads) are the same for all frequencies.

7.6.3.3 LOS/NLOS, indoor states and O2I parameters

The LOS state can be determined according to the spatial consistency procedure for random variables as mentioned in
Clause 7.6.3.1, by comparing a realization of a random variable generated with distance-dependent LOS probability. If
the realization is less than the LOS probability, the state is LOS; otherwise NLOS. Decision of LOS and NLOS status
should be used in Step 2 in Clause 7.5 if this advanced simulation is performed.

The same procedure can be applied for determining the indoor state, with the indoor probability instead of the LOS
probability.

The correlation distance for LOS state and indoor/outdoor is specified in Table 7.6.3.1-2.

The indoor distance can be modeled as the minimum of two spatially consistent uniform random variables within (0,
25) meters with correlation distance 25m.

Note in case the UT is in an indoor state, the pathloss model changes and a penetration loss is considered. For details on
the model, see Clause 7.4.3. Here, the focus is on modelling aspects with respect to spatial consistency. As described in
Clause 7.4.3, the penetration loss deviation σp represents variations within and between buildings of the same type. For
spatial consistency this can be modeled as a spatially consistent random variable with correlation distance 10m, see
Clause 7.6.3.1. The "building type" is determined using a spatially consistent uniform random variable with correlation
distance 50 m. The building type is determined by comparing the random variable with P1, where P1 is the probability of
the building type with low loss penetration. If the realization of the random variable is less than P1, the building type is
low loss; otherwise the building type is high loss.

3GPP
Release 16 59 3GPP TR 38.901 V16.1.0 (2019-12)

The cluster-specific and ray-specific random variables as defined in Clause 7.6.3.1 on the same floor are generated in
the spatial consistency modelling; otherwise, these variables across different floors are uncorrelated.

In case there is a transition from LOS to NLOS due to UT mobility, there will be a hard transition in the channel
response. This is because pathloss and LS parameters are different for these states, leading automatically to very
different channel realizations. To circumvent such hard transitions the optional soft LOS state can be considered to
determine the PL and the channel impulse responses containing characteristics of both LOS and NLOS. Soft LOS state
LOSsoft is generated by floating numbers between 0 (NLOS) and 1 (LOS) in the spatial consistency modelling. The
value of LOSsoft is determined by

1 1  20 
LOSsoft = + arctan (G + F (d )) , (7.6-18)
2 π  λ 
where:

- G is a spatially consistent Gaussian random variable with correlation distance according to Table 7.6.3.1-2;

- F (d ) = 2erf -1 (2 PrLOS (d ) − 1) ; and

- PrLOS (d ) is the distance dependent LOS probability function.

After LOSsoft is obtained, Steps 2-12 of the channel coefficient generation described in Clause 7.5 are performed
twice, once with the propagation condition in Step 2 set as LOS and once with the propagation condition in Step 2 set as
H LOS and H NLOS respectively, where H LOS is generated
NLOS. The resulting channel coefficients are denoted as
NLOS
with the LOS path loss formula and channel model parameters while is H generated using the NLOS path loss
formula and channel model parameters. The channel matrix H with soft LOS state is determined from a linear
combination of H LOS and H NLOS as:

H (LOSsoft ) = H LOS LOSsoft + H NLOS 1 − LOSsoft

2
(7.6-19)

It is noted that soft indoor/outdoor states are not modelled in this TR. Thus the model does not support transitions
between indoor/outdoor states in mobility simulations.

7.6.3.4 Applicability of spatial consistency

According to the characteristics of different cluster specific parameters and ray specific parameters, the following types
on spatial consistency are defined:

- Site-specific: parameters for different BS-UT links are uncorrelated, but the parameters for links from co-sited
sectors to a UT are correlated.

- All-correlated: BS-UT links are correlated.

In Table 7.6.3.4-1, correlation type for each large scale parameter, cluster specific parameter and ray specific parameter
is clarified.

Table 7.6.3.4-1: Correlation type among TRPs

Parameters Correlation type

Delays Site-specific

Cluster powers Site-specific

AOA/ZOA/AOD/ZOD offset Site-specific

AOA/ZOA/AOD/ZOD sign Site-specific

3GPP
Release 16 60 3GPP TR 38.901 V16.1.0 (2019-12)

Random coupling Site-specific

XPR Site-specific

Initial random phase Site-specific

LOS/NLOS states Site-specific

Blockage (Model A) All-correlated

O2I penetration loss All-correlated

Indoor distance All-correlated

Indoor states All-correlated

Spatial consistency is not modelled in the following situations:

- Different link types, e.g., outdoor LOS, outdoor NLOS or O2I

- UE locates on different floors

Table 7.6.3.4-2, lists conditions in which each large scale parameter, cluster specific parameter and ray specific parameter
is assumed not spatially correlated.

Table 7.6.3.4-2: Conditions when parameters are not spatially correlated.

Parameters Uncorrelated states

Delays Outdoor LOS/outdoor NLOS/O2I (different floors)

Cluster powers Outdoor LOS/outdoor NLOS/O2I (different floors)

AOA/ZOA/AOD/ZOD offset Outdoor LOS/outdoor NLOS/O2I (different floors)

AOA/ZOA/AOD/ZOD sign Outdoor LOS/outdoor NLOS/O2I (different floors)

Random coupling Outdoor LOS/outdoor NLOS/O2I (different floors)

XPR Outdoor LOS/outdoor NLOS/O2I (different floors)

Initial random phase Outdoor LOS/outdoor NLOS/O2I (different floors)

Blockage Outdoor/O2I (different floors)

Standard deviation for O2I penetration loss Different building types, i.e., high/low loss

7.6.4 Blockage
Blockage modelling is an add-on feature to the channel model. The method described in the following applies only
when this feature is turned on. In addition, the temporal variability of the blockage modelling parameters is on-demand
basis. It is also noted that the modelling of the blockage does not change LOS/NLOS state of each link.

When blockage model is applied, the channel generation in Clause 7.5 should have several additional steps between
Step 9 and 10 as illustrated in Figure 7.6.4-1.

3GPP
Release 16 61 3GPP TR 38.901 V16.1.0 (2019-12)

Figure.7.6.4-1 Channel generation procedure with blockage model

Two alternative models (Model A and Model B) are provided for the blockage modelling. Both approaches have their
own use cases. Model A is applicable when a generic and computationally efficient blockage modelling is desired. Model
B is applicable when a specific and more realistic blocking modelling is desired.

7.6.4.1 Blockage model A

Model A adopts a stochastic method for capturing human and vehicular blocking.

Step a: Determine the number of blockers.

Multiple 2-dimensional (2D) angular blocking regions, in terms of centre angle, azimuth and elevation angular span are
generated around the UT. There is one self-blocking region, and K = 4 non-self-blocking regions, where K may be
changed for certain scenarios (e.g., higher blocker density). Note that the self-blocking component of the model is
important in capturing the effects of human body blocking.

Step b: Generate the size and location of each blocker.

For self-blocking, the blocking region in UT LCS is defined in terms of elevation and azimuth angles, ( θ 'sb , φ 'sb ) and
azimuth and elevation angular span ( xsb , ysb ).

  ysb y x x 
(θ ' , φ ') θ 'sb − ≤ θ ' ≤ θ 'sb + sb , φ 'sb − sb ≤ φ ' ≤ φ 'sb + sb  (7.6-20)
  2 2 2 2 

where the parameters are described in Table 7.6.4.1-1.

3GPP
Release 16 62 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.6.4.1-1: Self-blocking region parameters.

φsb′ xsb θ 'sb ysb

Portrait mode 260o 120o 100o 80o
Landscape mode 40o 160o 110o 75o

For non-self-blocking k = 1, …, 4, the blocking region in GCS is defined by

  yk y x x 
(θ , φ ) θ k − ≤ θ ≤ θ k + k , φk − k ≤ φ ≤ φk + k   (7.6-21)
  2 2 2 2 

where the parameters are described in Table 7.6.4.1-2, as well as the distance r between the UT and the blocker.

Table 7.6.4.1-2: Blocking region parameters.

Blocker index (k = 1, …, 4) φk xk θk yk r
InH scenario Uniform in Uniform in 90o Uniform in 2m
[0o, 360o] [15o, 45o] [5o, 15o]
UMi, UMa, RMa scenarios Uniform in Uniform in 90o 5o 10 m
[0o, 360o] [5o, 15o]

Step c: Determine the attenuation of each cluster due to blockers.

The attenuation of each cluster due to self-blocking corresponding to the centre angle pair ( θ 'sb , φ 'sb ), is 30 dB provided
xsb y
that ′ − φsb′ <
φAOA ′ − θ sb′ < sb . Otherwise, the attenuation is 0 dB.
and θ ZOA
2 2
The attenuation of each cluster due to the non-self-blocking regions (k=1, …, 4) is given by

( (
LdB = −20 log10 1 − FA1 + FA2 FZ1 + FZ 2 )( )) (7.6-22)

provided that φAOA − φk < xk and θ ZOA − θ k < yk . Otherwise, the attenuation is 0 dB. The terms in the above
equation are given as

 π π  1 
tan −1  ± r  − 1 
 2 λ  cos( A1 | A2 | Z1 | Z 2 )  
FA1 | A2 | Z1 | Z 2 =   (7.6-23)
π
where

 x 
A1 = φAOA −  φk + k  , (7.6-24)
 2

 x 
A2 = φAOA −  φk − k  , (7.6-25)
 2

 y 
Z1 = θ ZOA − θ k + k  , (7.6-26)
 2

 y 
Z 2 = θ ZOA − θ k − k . (7.6-27)
 2

3GPP
Release 16 63 3GPP TR 38.901 V16.1.0 (2019-12)

In the above formula for FA1| A2 |Z1|Z 2 , λ is the wavelength. The appropriate signs ( ± )within the tan −1 term are
described in Table 7.6.4.1-3.

Table 7.6.4.1-3: Description of signs

yk yk y yk
− y k < θ ZOA − θ k ≤ − − < θ ZOA − θ k ≤ k < θ ZOA − θ k ≤ y k
2 2 2 2
xk (-, +) for ( A1 , A2 ) (-, +) for ( A1 , A2 ) (-, +) for ( A1 , A2 )
< φAOA − φk ≤ xk (+, -) for (Z1 , Z 2 ) (+, +) for (Z1 , Z 2 ) (-, +) for (Z1 , Z 2 )
2
x x (+, +) for ( A1 , A2 ) (+, +) for ( A1 , A2 ) (+, +) for ( A1 , A2 )
− k < φAOA − φk ≤ k (+, -) for (Z1 , Z 2 ) (+, +) for (Z1 , Z 2 ) (-, +) for (Z1 , Z 2 )
2 2
xk (+, -) for ( A1 , A2 ) (+, -) for ( A1 , A2 ) (+, -) for ( A1 , A2 )
− xk < φAOA − φk ≤ − (+, -) for (Z1 , Z 2 ) (+, +) for (Z1 , Z 2 ) (-, +) for (Z1 , Z 2 )
2

Step d: Spatial and temporal consistency of each blocker.

The centre of the blocker is generated based on a uniformly distributed random variable, which is temporally and spatially
consistent. The two-dimensional autocorrelation function R(∆ x , ∆ t ) can be described with sufficient accuracy by the
exponential function

  ∆ ∆ 
R(∆ x , ∆ t ) = exp −  x + t   (7.6-28)
 d 
  corr tcorr  

The spatial correlation distance d corr for the random variable determining the centre of the blocker is given in Table
7.6.4.1-4 for different scenarios.

Table 7.6.4.1-4: Spatial correlation distance for different scenarios.

Spatial correlation UMi UMa RMa InH

distance d corr in [m] LOS NLOS O2I LOS NLOS O2I LOS NLOS O2I LOS NLOS
for the random
variable determining
the centre of the 10 10 5 10 10 5 10 10 5 5 5
blocker

The correlation time is defined by tcorr = dcorr v , where v is the speed of the moving blocker.

Note: The rectangular blocker description is chosen for self-blocking region with the specific choices of ( θ 'sb , φ 'sb )
assumed here. Generalization of this description to other choices should be done with care as the rectangular description
may not be accurate.

7.6.4.2 Blockage model B

Model B adopts a geometric method for capturing e.g., human and vehicular blocking.

Step a: Determine blockers

A number, K , of blockers are modelled as rectangular screens that are physically placed on the map. Each screen has
the dimension by height ( hk ) and width ( wk ), with the screen centre at coordinate (xk , yk , zk ) .
Note:

3GPP
Release 16 64 3GPP TR 38.901 V16.1.0 (2019-12)

- The number of blockers ( K ), their vertical and horizontal extensions ( hk and wk ), locations (xk , yk , zk ) ,
density, and movement pattern (if non-stationary) are all simulation assumptions, to allow different blocking
scenarios to be constructed depending on the need of the particular simulation study.

Recommended parameters for typical blockers are provided in Table 7.6.4.2-5.

- The blocking effect diminishes with increasing distance to the blocker. For implementation purposes it may be
sufficient to consider only the K nearest blockers or the blockers closer than some distance from a specific UT.

Table 7.6.4.2-5: Recommended blocker parameters

Typical set of blockers Blocker dimensions Mobility pattern

Indoor; Outdoor; InF Human Cartesian: w=0.3m; h=1.7m Stationary or up to 3 km/h
Outdoor Vehicle Cartesian: w=4.8m; h=1.4m Stationary or up to 100 km/h
InF AGV Cartesian: w=3m; h=1.5m Up to 30 km/h
InF Industrial robot Cartesian: w=2m; h=0.2m Up to 3 m/s

Step b: Determine the blockage attenuation per sub-path

Attenuation caused by each blocker to each sub-path is modelled using a simple knife edge diffraction model and is given
by

( ( )(
LdB = −20 log10 1 − Fh1 + Fh2 Fw1 + Fw2 )) (7.6-29)

where Fh1 , Fh2 and Fw1 , Fw2 account for knife edge diffraction at the four edges, and are given by

 −1  π π
 tan  ± (
D1h1|h2|w1|w2 + D2 h1|h2|w1|w2 − r ) 
  2 λ  , for direct path in LOS
 π
Fh1|h2|w1|w2 = (7.6-30)
 tan −1  ± π π D1 
  2 λ
( ) 
h1|h2 |w1|w2 − r ' 

   , for all other paths

 π

where λ is the wave length. As shown in Figure 7.6.4.2-2, D1h1|h2 |w1|w2 are the projected (onto the side and top view

planes) distances between the receiver and four edges of the corresponding blocker, and D2 h1|h2 |w1|w2 are the projected
(onto the side and top view planes) distances between the transmitter and four edges of the corresponding blocker. The
side view plane is perpendicular to the horizontal ground plane. The top view is perpendicular to the side view. For each
cluster, the blocker screen is rotated around its centre such that the arrival direction of the corresponding path is always
perpendicular to the screen. It should be noted that different rotations are required for each individual sub-path. Meanwhile,
the base and top edges of the screens are always parallel to the horizontal plane. As the screen is perpendicular to each
sub-path, r is the distance between the transmitter and receiver for the direct path in LOS, and r ' is the distance between
the blocker screen and receiver, projected onto the incoming sub-path direction, for all the other (NLOS) paths. In the
equation of Fh1|h2 |w1|w2 , the plus and minus signs are determined in such a way that, as shown in Figure 7.6.4.2-2,
- if the sub-path (terminated at the receiver or transmitter) does not intersect the screen in side view, minus sign is
applied for the shortest path among D1h1 and D1h2 in the NLOS case ( D1h1 + D2h1 and D1h2 + D2h2 in the
LOS case) and plus sign is applied for the other edge.

- if the sub-path (terminated at the receiver or transmitter) does not intersect the screen in top view, minus sign is
applied for the shortest path among D1w1 and D1w2 in the NLOS case ( D1w1 + D2w1 and D1w2 + D2w2 for the
LOS case) and plus sign is applied for the other edge.

3GPP
Release 16 65 3GPP TR 38.901 V16.1.0 (2019-12)

- if the sub-path intersects the screen plus signs are applied for both edges.

For the case of multiple screens the total loss is given by summing the losses of each contributing screen in dB units.

The model according to option B is consistent in time, frequency and space, and is more appropriate to be used for
simulations with arbitrarily designated blocker density.

. Top View
+ D1w1 Rx
.
Rx D2w1 + .
Tx r .
. (xk, yk, zk) w
D1w2
Tx
.
+
D2w2 +

Side View
+
D2h1 D2h1 +
+
h . Tx
h D1h1
D1h1 r
. .
D2h2 −
Tx + D2h2
. D1h2
+
Rx
− + D1h2
r

Rx

Figure 7.6.4.2-2(a): Illustration of the geometric relation among blocker, receiver and transmitter for
the LOS path

. Top View
+ D1w1 Rx
.
Rx .
r'
.
. (xk, yk, zk) w
D1w2

.
+

Side View

+
h .
h D1h1
D1h1
. .
+ r'
. D1h2 Rx
− + D1h2
r'
Rx

Figure 7.6.4.2-2(b): Illustration of the geometric relation between blocker and receiver for NLOS paths

7.6.5 Correlation modelling for multi-frequency simulations

This clause describes how to generate parameters to reflect correlation across different frequencies for a BS-UT link, for
simulations involved with multiple frequencies.

For those simulations, the steps in Clause 7.5 should be revised according to the following:

- The parameters generated in Step 1 are the same for all the frequencies, except for antenna patterns, array
geometries, system center frequency and bandwidth.

- Propagation conditions generated in Step 2 are the same for all the frequencies. It is noted that soft LOS states
may be different due to frequency dependent function.

3GPP
Release 16 66 3GPP TR 38.901 V16.1.0 (2019-12)

- The parameters generated in Step 4 are the same for all the frequencies, except for possibly frequency-dependent
scaling of e.g. delay spread and angular spreads according to the LSP tables. I.e. let x be a random variable
drawn from a Gaussian distribution: x ~ N(0,1). Then the delay spread at frequency f is DS(f)= 10^(µlgDS(f) +
σlgDS(f)⋅x), where the same value of x is used for all frequencies. A corresponding procedure applies to each of
the angular spreads.

- The cluster delays and angles resulting from Steps 5-7 are the same for all frequency bands

- Per-cluster shadowing Ζn in Step 6 are independently generated for the frequency bands.

- Cluster powers in Step 6 may be frequency-dependent.

- Steps 8-11 are independently applied for the frequency bands.

In addition, when blockage is modeled according to Clause 7.6.4, the positions of blockers are the same across all the
frequencies.

Note: The requirements above may not be fully aligned with the behavior of the model according to Clause 7.5,
since cluster delays and angles will be frequency-dependent in scenarios where the DS or AS is
frequency-dependent. The procedure below may alternatively be used to ensure that cluster delays and
angles are frequency-independently generated.

7.6.5.1 Alternative channel generation method

The alternative method replaces Steps 5-7 in Clause 7.5 with the below Steps 5'-7'. The inputs to the alternative method
are the delay and angular spreads determined according to Step 4 at an anchor frequency, e.g. 2 GHz: DS0, ASD0,
ASA0, ZSD0, ZSA0, the delay and angular spreads determined according to Step 4 at a frequency of interest: DS, ASD,
ASA, ZSD, ZSA, and the number of clusters N from Table 7.5-6.

Step 5': Generate nominal delays τ n′ ′ ,n , φAOA

and angles φAOD ′ ,n , ′ ,n , θ ZOA
θ ZOD ′ ,n .

Generate N delays from a single-sided exponential distribution with zero mean and standard deviation of r DS0 ,
according to τ n′ = −r DS0 ln( X n ) with X n ~ unif (0, 1) .

Generate N AODs from a wrapped Gaussian distribution with zero mean and standard deviation of r ASD0 , according
to ′ ,n = arg(exp(− j r ASD 0Yn )) with Yn ~ N (0, 1) .
φAOD

Generate N AOAs from a wrapped Gaussian distribution with zero mean and standard deviation of r ASA 0 , according
to ′ ,n = arg(exp(− j r ASA 0 Z n )) with Z n ~ N (0, 1) .
φ AOA

Generate N ZODs from a wrapped Laplacian distribution with zero mean and standard deviation of r ZSD0 , according
to ( (
′ ,n = arg exp − j r ZSD 0 sgn (Vn − 0.5) ln (1 − 2 Vn − 0.5 )
θ ZOD ))
2 with Vn ~ unif (0, 1) .

Generate N ZOAs from a wrapped Laplacian distribution with zero mean and standard deviation of r ZSA 0 , according
to ( (
′ ,n = arg exp − j r ZSA 0 sgn (Wn − 0.5) ln (1 − 2 Wn − 0.5 )
θ ZOA ))
2 with Wn ~ unif (0, 1) .

r is a proportionality factor, r=1.5. The principal value of the arg function should be used, e.g. (-180,180).

In case of LOS, set τ 1′ = 0 , ′ ,1 = 0 , φ AOA

φAOD ′ ,1 = 0 , θ ZOD
′ ,1 = 0 , and θ ZOA
′ ,1 = 0 .

Step 6': Generate cluster powers Pn .

3GPP
Release 16 67 3GPP TR 38.901 V16.1.0 (2019-12)

Generate cluster powers as

  φ AOD
′ ,n g ASD
2
  φ AOA
′ ,n g ASA 
2

Pn′ = exp − τ n′ g DS −   −  ′ ,n g ZSA  ⋅ 10 −Qn 10
′ ,n g ZSD − 2 θ ZOA
 − 2 θ ZOD
  2   2  
 
(7.6.30a)

where (
Qn ~ N 0, ζ 2 ) is the per cluster shadowing term in [dB] and
max(r ⋅ DS 0 − DS, 0 )
g DS = (7.6.30b)
DS ⋅ r ⋅ DS 0

g ASD =
(
max (r ⋅ ASD 0 ) − ASD 2 , 0
2
) (7.6.30c)
ASD ⋅ r ⋅ ASD 0

g ASA =
(
max (r ⋅ ASA 0 ) − ASA 2 , 0
2
) (7.6.30d)
ASA ⋅ r ⋅ ASA 0

max(r ⋅ ZSD 0 − ZSD, 0 )

g ZSD = (7.6.30e)
ZSD ⋅ r ⋅ ZSD 0

max(r ⋅ ZSA 0 − ZSA, 0 )

g ZSA = (7.6.30f)
ZSA ⋅ r ⋅ ZSA 0
Normalize the cluster powers so that the sum of all cluster powers is equal to one, i.e.,

Pn′
Pn = (7.6.30g)
∑n=1 Pn′
N

or, in the case of LOS, so that

1 Pn′ KR
Pn = + δ (n − 1) (7.6.30h)
1+ KR ∑n=1 Pn′ 1+ KR
N

where K R is the K-factor converted to linear scale.

Step 7': Generate delays τn and angles φ AOD,n , φ AOA,n , θ ZOD,n , θ ZOA,n .

 τ n′
 , for NLOS
τn =  KR (7.6.30i)
 1 + 2 τ n′ , for LOS

 φ n′,AOA + φ LOS,AOA + cASAα m , for NLOS

φ n ,m,AOA =  (7.6.30j)
 1 + K R φ n′,AOA + φ LOS,AOA + cASAα m , for LOS

3GPP
Release 16 68 3GPP TR 38.901 V16.1.0 (2019-12)

 φ n′,AOD + φ LOS,AOD + cASDα m , for NLOS

φ n ,m,AOD =  (7.6.30k)
 1 + K R φ n′,AOD + φ LOS,AOD + cASDα m , for LOS

 θ n′,ZOA + θ LOS,ZOA + c ZSAα m , for NLOS

θ n,m,ZOA =  (7.6.30l)
 1 + K R θ n′,ZOA + θ LOS,ZOA + c ZSAα m , for LOS

 θ n′,ZOD + θ LOS,ZOD + c ZSDα m , for NLOS

θ n ,m,ZOD =  (7.6.30m)
 1 + K R θ n′,ZOD + θ LOS,ZOD + c ZSDα m , for LOS

Repeat Steps 6'-7' for each frequency of interest, reusing the delays and angles from Step 5' for all frequencies.

Note: The resulting delay and angular spreads of channels generated with this alternative method will be similar but not
identical to when using Steps 5-7 in Clause 7.5.

7.6.6 Time-varying Doppler shift

The Doppler shift generally depends on the time evolution of the channel as it is defined as the derivative of the
channels phase over time. It can be the result from Tx, Rx, or scatterer movement. The more general form of the
exponential Doppler term as used in Equation (7.5-22) is given by

~ ~


t T
rˆrx ,n ,m ( t ) ⋅ v (t ) ~ 
exp j 2π ∫ dt . (7.6-31)
 λ 
 t 0 0 

Here, rˆrx , n , m (t ) is the normalized vector that points into the direction of the incoming wave as seen from the Rx at

time t . v (t ) denotes the velocity vector of the Rx at time t , while t 0 denotes a reference point in time that defines


the initial phase, e.g. t 0 = 0.

 
Note that Equation (7.5-22) only holds for time-invariant Doppler shift, i.e. rˆrxT ,n ,m (t ) ⋅ v (t ) = rˆrxT ,n ,m ⋅ v .

7.6.7 UT rotation
UT rotation modelling is an add-on feature. When modelled, Step 1 in Clause 7.5 shall consider UT rotational motion.

Step 1:

Add: h) Give rotational motion of UT in terms of its bearing angle, downtilt angle and slant angle.

7.6.8 Explicit ground reflection model

In case the ground reflection shall be modelled explicitly, Equation (7.5-30) has to be replaced by

1
, s (τ , t ) =
H uLOS H uNLOS
,s (τ − τ LOS , t )
KR + 1
(7.6-32)
K R  LOS d 
+  H u , s ,1 (t )δ (τ − τ LOS ) + 3D H uGR
, s (t )δ (τ − τ GR )

KR + 1 d GR 

3GPP
Release 16 69 3GPP TR 38.901 V16.1.0 (2019-12)

with the delays for the ground reflected and the LOS paths being defined by their lengths, according to the Tx-Rx
separation d 2D and the Tx and Rx heights htx and hrx respectively, as

d (htx + hrx )2 + d 2D2

τ GR = GR = , (7.6-33)
c c
and

d 3D (htx − hrx )2 + d 2D2

τ LOS = = . (7.6-34)
c c
The channel coefficient for the ground reflected path is given by

 Frx ,u ,θ (θ GR, ZOA ,φGR, AOA )  R||GR 0   Ftx , s ,θ (θ GR, ZOD ,φGR, AOD )
T
GR
H (t ) = 
 Frx ,u ,φ (θ GR, ZOA ,φGR, AOA )  0 − R⊥GR   Ftx , s ,φ (θ GR, ZOD ,φGR, AOD )
u,s    
(7.6-35)
 d   rˆT .d   rˆT .d   rˆT .v 
⋅ exp − j 2π GR  exp j 2π rx, GR rx ,u  exp j 2π tx, GR tx , s  exp j 2π rx, GR t 
λ0  λ0   λ0   λ0 
       
with the normalized vectors pointing towards the ground reflection point from the Tx

sin θ GR, ZOD cos φGR, AOD 

 
rˆtx, GR = er (θ GR, ZOD , φGR, AOD ) =  sin θ GR, ZOD sin φGR, AOD  (7.6-36)
 cos θ GR, ZOD 

and from the Rx side

sin θ GR, ZOA cos φGR, AOA 

 
rˆrx, GR = er (θ GR, ZOA ,φGR, AOA ) =  sin θ GR, ZOA sin φGR, AOA  . (7.6-37)
 cosθ GR, ZOA 

The angles for the ground reflected path are given by the geometry assuming a flat surface with its normal pointing into
z-direction. The angles at the Tx side can be determined by

 d 2D 
θ GR, ZOD = 180° - atan  (7.6-37a)
 hrx + htx 

φGR,AOD = φLOS,AOD (7.6-37b)

and at the Rx side by

θGR, ZOA = θGR, ZOD (7.6-37c)

φGR, AOA = φGR, AOD + 180° . (7.6-37d)

The reflection coefficients for parallel and perpendicular polarization on the ground, cf. [21], are given by

3GPP
Release 16 70 3GPP TR 38.901 V16.1.0 (2019-12)

ε GR ε GR
cos(θ GR,ZOD ) + − sin 2 (θ GR,ZOD )
ε0 ε0
R||GR = , (7.6-38)
ε GR ε GR
cos(θ GR,ZOD ) − − sin 2 (θ GR,ZOD )
ε0 ε0

and

ε GR
cos(θ GR,ZOD ) + − sin 2 (θ GR,ZOD )
ε0
R⊥GR = , (7.6-39)
ε GR
cos(θ GR,ZOD ) − − sin (θ GR,ZOD )
2

ε0

with the complex relative permittivity of the ground material given by

ε GR σ
= εr − j . (7.6-40)
ε0 2π f cε 0

The electric constant ε0 is given by 8.854187817... × 10−12 F·m−1.

For applicable frequency ranges, the real relative permittivity can be modelled by
b
 fc 
ε

ε r = aε ⋅  9  , (7.6-41)
 10 
while the conductivity in [S/m] may be modelled by
d
 fc 
σ

σ = cσ ⋅  9  , (7.6-42)
 10 

with fc being the center frequency in Hz.

A selection of material properties from [19] is given below. Since the simplified models for conductivity and relative
permittivity are only applicable for frequencies between 1 and 10 GHz, Figure 7.6.8-1 presents curves up to 100 GHz at
least for very, medium dry and wet ground, cf. [20].

Table 7.6.8-1 Material properties [19]

Relative permittivity Conductivity

Material class Frequency range in [GHz]
aε bε cσ dσ
Concrete 5.31 0 0.0326 0.8095 1-100
Brick 3.75 0 0.038 0 1-10
Plasterboard 2.94 0 0.0116 0.7076 1-100
Wood 1.99 0 0.0047 1.0718 0.001-100
Floorboard 3.66 0 0.0044 1.3515 50-100
Metal 1 0 107 0 1-100
Very dry ground 3 0 0.00015 2.52 1-10
Medium dry ground 15 −0.1 0.035 1.63 1-10
Wet ground 30 −0.4 0.15 1.30 1-10

3GPP
Release 16 71 3GPP TR 38.901 V16.1.0 (2019-12)

Figure 7.6.8-1: Relative permittivity and conductivity as a function of frequency [20]

7.6.9 Absolute time of arrival

To support simulations in which absolute time of arrival is important, the propagation time delay due to the total path
length is considered in step 11 of the fast fading model as follows. The impulse response in NLOS is determined using
equation (7.6-43) instead of (7.5-27) and the impulse response in LOS is determined using equation (7.6-44) instead of
(7.5-30), where 𝑐𝑐 is the speed of light. Δ𝜏𝜏 is generated from a lognormal distribution with parameters according to Table
7.6.9-1. Δ𝜏𝜏 is generated independently for links between the same UT and different BS sites. The excess delay in
NLOS, ∆τ, should further be upper bounded by 2𝐿𝐿/𝑐𝑐, where 𝐿𝐿 is the largest dimension of the factory hall, i.e. 𝐿𝐿 =
max(length, width, height).

3GPP
Release 16 72 3GPP TR 38.901 V16.1.0 (2019-12)

𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 (𝜏𝜏,
𝐻𝐻𝑢𝑢,𝑠𝑠 𝑡𝑡) = ∑2𝑛𝑛=1 ∑3𝑖𝑖=1 ∑𝑚𝑚∈𝑅𝑅𝑖𝑖 𝐻𝐻𝑢𝑢,𝑠𝑠,𝑛𝑛,𝑚𝑚
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 (𝑡𝑡)𝛿𝛿�𝜏𝜏
𝑛𝑛=3 𝐻𝐻𝑢𝑢,𝑠𝑠,𝑛𝑛 (𝑡𝑡)𝛿𝛿(𝜏𝜏 − 𝜏𝜏𝑛𝑛 − 𝑑𝑑3𝐷𝐷 ⁄𝑐𝑐 − Δ𝜏𝜏) (7.6-
− 𝜏𝜏𝑛𝑛,𝑖𝑖 − 𝑑𝑑3𝐷𝐷 ⁄𝑐𝑐 − Δ𝜏𝜏� + ∑𝑁𝑁 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁

43)

1 𝐾𝐾𝑅𝑅 𝐿𝐿𝐿𝐿𝐿𝐿 (𝑡𝑡)𝛿𝛿(𝜏𝜏

𝐿𝐿𝐿𝐿𝐿𝐿 (𝜏𝜏,
𝐻𝐻𝑢𝑢,𝑠𝑠 𝑡𝑡) = � 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 (𝜏𝜏,
𝐻𝐻𝑢𝑢,𝑠𝑠 𝑡𝑡) + � 𝐻𝐻𝑢𝑢,𝑠𝑠,1 − 𝜏𝜏1 − 𝑑𝑑3𝐷𝐷 ⁄𝑐𝑐 ). (7.6-44)
𝐾𝐾𝑅𝑅 +1 𝐾𝐾𝑅𝑅 +1

Table 7.6.9-1: Parameters for the absolute time of arrival model

Scenarios InF-SL, InF-DL InF-SH, InF-DH

𝜇𝜇𝑙𝑙𝑙𝑙Δ𝜏𝜏 -7.5 -7.5
𝑙𝑙𝑙𝑙Δ𝜏𝜏 = 𝑙𝑙𝑙𝑙𝑙𝑙10 (Δ𝜏𝜏⁄1𝑠𝑠)
𝜎𝜎𝑙𝑙𝑙𝑙Δ𝜏𝜏 0.4 0.4
Correlation distance in the horizontal plane 6 11
[m]

7.6.10 Dual mobility

To support simulations that involve dual Tx and Rx mobility or scatterer mobility, the Doppler frequency component in
the channel coefficient generation in step 11 in clause 7.5 should be updated as follows.

For the LOS path, the Doppler frequency is given by

𝒓𝒓�𝑻𝑻 �𝒓𝒓𝒓𝒓 +𝒓𝒓�𝑻𝑻
𝒓𝒓𝒓𝒓,𝒏𝒏,𝒎𝒎 ∙𝒗𝒗 �𝒕𝒕𝒕𝒕
𝒕𝒕𝒕𝒕,𝒏𝒏,𝒎𝒎 ∙𝒗𝒗
𝒗𝒗𝒏𝒏,𝒎𝒎 = (7.6-45)
𝝀𝝀𝟎𝟎

where 𝒓𝒓�𝒓𝒓𝒓𝒓,𝒏𝒏,𝒎𝒎 and 𝒓𝒓�𝒕𝒕𝒕𝒕,𝒏𝒏,𝒎𝒎 are defined in (7.5-23) and (7.5-24) respectively, and

vrx = vrx [sin θ v ,rx cos φv ,rx sin θ v ,rx sin φv ,rx cosθ v ,rx ]
T

vtx = vtx [sin θ v ,tx cos φv ,tx sin θ v ,tx sin φv ,tx cos θ v ,tx ]
T

For all other paths, the Doppler frequency component is given by:
𝒓𝒓�𝑻𝑻 �𝒓𝒓𝒓𝒓 +𝒓𝒓�𝑻𝑻
𝒓𝒓𝒓𝒓,𝒏𝒏,𝒎𝒎 ∙𝒗𝒗 �𝒕𝒕𝒕𝒕 +𝟐𝟐𝜶𝜶𝒏𝒏,𝒎𝒎 𝑫𝑫𝒏𝒏,𝒎𝒎
𝒕𝒕𝒕𝒕,𝒏𝒏,𝒎𝒎 ∙𝒗𝒗
𝒗𝒗𝒏𝒏,𝒎𝒎 = (7.6-46)
𝝀𝝀𝟎𝟎

where 𝐷𝐷𝑛𝑛,𝑚𝑚 is a random variable from −𝑣𝑣𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 to 𝑣𝑣𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 , 𝛼𝛼𝑛𝑛,𝑚𝑚 is a random variable of Bernoulli distribution with mean
p, and 𝑣𝑣𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 is the maximum speed of the clutter. Parameter p determines the proportion of mobile scatterers and can
thus be selected to appropriately model statistically larger number of mobile scatterers (higher p) or statistically smaller
number of mobile scatterers (e.g. in case of a completely static environment: p=0 results in all scatteres having zero
speed). A typical value of p is 0.2.

7.6.11 Sources of EM interference

When the simulation includes sources of EM interference in industrial scenarios, the channel(s) between the
interferer(s) and the BS or UT should be modeled using the appropriate sub-scenario(s), e.g. clutter-embedded or
elevated Tx or Rx and low or high clutter density. Characterization of EM interference sources is out of the scope of
this TR.

7.6.12 Embedded devices

To simulate devices that are embedded inside machinery or enclosures, an additional loss parameter Lembedded is added to
the path loss in step 12. The additional path loss parameter Lembedded to embedded devices is left as a simulation
assumption, where it may be formulated e.g. as a value or range, or derived using a methodology similar to the building
penetration loss in clause 7.4.3.1.

3GPP
Release 16 73 3GPP TR 38.901 V16.1.0 (2019-12)

7.7 Channel models for link-level evaluations

7.7.1 Clustered Delay Line (CDL) models
The CDL models are defined for the full frequency range from 0.5 GHz to 100 GHz with a maximum bandwidth of 2
GHz. CDL models can be implemented by e.g., coefficient generation Step 10 and Step 11 in Clause 7.5 or generating
TDL model using spatial filter from Clause 7.7.4. Three CDL models, namely CDL-A, CDL-B and CDL-C are
constructed to represent three different channel profiles for NLOS while CDL-D and CDL-E are constructed for LOS,
the parameters of which can be found respectively in Table 7.7.1-1, Table 7.7.1-2 , Table 7.7.1-3, Table 7.7.1-4 and
Table 7.7.1-5.

Each CDL model can be scaled in delay so that the model achieves a desired RMS delay spread, according to the
procedure described in Clause 7.7.3. Each CDL model can also be scaled in angles so that the model achieves desired
angle spreads, according to the procedure described in Clause 7.7.5.1.

For LOS channel models, the K-factor of CDL-D and CDL-E can be set to a desired value following the procedure
described in Clause 7.7.6.

For modelling effect of beamforming in a simplified way, a brick-wall window can be applied to a delay-scaled CDL
model. The power shall be normalized after applying the window. A TDL model for simplified evaluations can be
obtained from the CDL model, according to this method.

The following step by step procedure should be used to generate channel coefficients using the CDL models.

Step 1: Generate departure and arrival angles

Generate arrival angles of azimuth using the following equation

φ n,m,AOA = φ n,AOA + c ASAα m , (7.7-0a)

Where φn,AOA is the cluster AOA and cASA is the cluster-wise rms azimuth spread of arrival angles (cluster ASA) in
Tables 7.7.1.1 – 7.7.1.5 below, and αm denotes the ray offset angles within a cluster given by Table 7.5-3. If angular
scaling according to Clause 7.7.5.1 is used, this is applied to the ray angles φ n ,m ,AOA , The generation of AOD

( φ n ,m ,AOD ), ZSA ( θ n ,m , ZOA ), and ZSD ( θ n ,m , ZOD ) follows a procedure similar to AOA as described above.

Step 2: Coupling of rays within a cluster for both azimuth and elevation

Couple randomly AOD angles φ n ,m ,AOD to AOA angles φ n ,m ,AOA within a cluster n. Couple randomly ZOD angles

θ n,m,ZOD with ZOA angles θ n,m,ZOA using the same procedure. Couple randomly AOD angles φ n,m,AOD with ZOD
angles θ n,m,ZOD within a cluster n.

Step 3: Generate the cross polarization power ratios

Generate the cross polarization power ratios (XPR) κ for each ray m of each cluster n as

κ n ,m = 10 X /10 , (7.7-0b)

where X is the per-cluster XPR in dB from Tables 7.7.1.1 – 7.7.1.5.

Step 4: Coefficient generation

3GPP
Release 16 74 3GPP TR 38.901 V16.1.0 (2019-12)

Follow the same procedure as in Steps 10 and 11 in Clause 7.5, with the exception that all clusters are treated as
"weaker cluster", i.e. no further sub-clusters in delay should be generated. Additional clusters representing delay spread
of the stronger clusters are already provided in Tables 7.7.1.1 – 7.7.1.5.

Table 7.7.1-1. CDL-A

Cluster # Normalized delay Power in [dB] AOD in [°] AOA in [°] ZOD in [°] ZOA in [°]
1 0.0000 -13.4 -178.1 51.3 50.2 125.4
2 0.3819 0 -4.2 -152.7 93.2 91.3
3 0.4025 -2.2 -4.2 -152.7 93.2 91.3
4 0.5868 -4 -4.2 -152.7 93.2 91.3
5 0.4610 -6 90.2 76.6 122 94
6 0.5375 -8.2 90.2 76.6 122 94
7 0.6708 -9.9 90.2 76.6 122 94
8 0.5750 -10.5 121.5 -1.8 150.2 47.1
9 0.7618 -7.5 -81.7 -41.9 55.2 56
10 1.5375 -15.9 158.4 94.2 26.4 30.1
11 1.8978 -6.6 -83 51.9 126.4 58.8
12 2.2242 -16.7 134.8 -115.9 171.6 26
13 2.1718 -12.4 -153 26.6 151.4 49.2
14 2.4942 -15.2 -172 76.6 157.2 143.1
15 2.5119 -10.8 -129.9 -7 47.2 117.4
16 3.0582 -11.3 -136 -23 40.4 122.7
17 4.0810 -12.7 165.4 -47.2 43.3 123.2
18 4.4579 -16.2 148.4 110.4 161.8 32.6
19 4.5695 -18.3 132.7 144.5 10.8 27.2
20 4.7966 -18.9 -118.6 155.3 16.7 15.2
21 5.0066 -16.6 -154.1 102 171.7 146
22 5.3043 -19.9 126.5 -151.8 22.7 150.7
23 9.6586 -29.7 -56.2 55.2 144.9 156.1
Per-Cluster Parameters
Parameter cASD in [°] cASA in [°] cZSD in [°] cZSA in [°] XPR in [dB]
Value 5 11 3 3 10

3GPP
Release 16 75 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.7.1-2. CDL-B

Cluster # Normalized delay Power in [dB] AOD in [°] AOA in [°] ZOD in [°] ZOA in [°]
1 0.0000 0 9.3 -173.3 105.8 78.9
2 0.1072 -2.2 9.3 -173.3 105.8 78.9
3 0.2155 -4 9.3 -173.3 105.8 78.9
4 0.2095 -3.2 -34.1 125.5 115.3 63.3
5 0.2870 -9.8 -65.4 -88.0 119.3 59.9
6 0.2986 -1.2 -11.4 155.1 103.2 67.5
7 0.3752 -3.4 -11.4 155.1 103.2 67.5
8 0.5055 -5.2 -11.4 155.1 103.2 67.5
9 0.3681 -7.6 -67.2 -89.8 118.2 82.6
10 0.3697 -3 52.5 132.1 102.0 66.3
11 0.5700 -8.9 -72 -83.6 100.4 61.6
12 0.5283 -9 74.3 95.3 98.3 58.0
13 1.1021 -4.8 -52.2 103.7 103.4 78.2
14 1.2756 -5.7 -50.5 -87.8 102.5 82.0
15 1.5474 -7.5 61.4 -92.5 101.4 62.4
16 1.7842 -1.9 30.6 -139.1 103.0 78.0
17 2.0169 -7.6 -72.5 -90.6 100.0 60.9
18 2.8294 -12.2 -90.6 58.6 115.2 82.9
19 3.0219 -9.8 -77.6 -79.0 100.5 60.8
20 3.6187 -11.4 -82.6 65.8 119.6 57.3
21 4.1067 -14.9 -103.6 52.7 118.7 59.9
22 4.2790 -9.2 75.6 88.7 117.8 60.1
23 4.7834 -11.3 -77.6 -60.4 115.7 62.3
Per-Cluster Parameters
Parameter cASD in [°] cASA in [°] cZSD in [°] cZSA in [°] XPR in [dB]
Value 10 22 3 7 8

Table 7.7.1-3. CDL-C

Cluster # Normalized delay Power in [dB] AOD in [°] AOA in [°] ZOD in [°] ZOA in [°]
1 0 -4.4 -46.6 -101 97.2 87.6
2 0.2099 -1.2 -22.8 120 98.6 72.1
3 0.2219 -3.5 -22.8 120 98.6 72.1
4 0.2329 -5.2 -22.8 120 98.6 72.1
5 0.2176 -2.5 -40.7 -127.5 100.6 70.1
6 0.6366 0 0.3 170.4 99.2 75.3
7 0.6448 -2.2 0.3 170.4 99.2 75.3
8 0.6560 -3.9 0.3 170.4 99.2 75.3
9 0.6584 -7.4 73.1 55.4 105.2 67.4
10 0.7935 -7.1 -64.5 66.5 95.3 63.8
11 0.8213 -10.7 80.2 -48.1 106.1 71.4
12 0.9336 -11.1 -97.1 46.9 93.5 60.5
13 1.2285 -5.1 -55.3 68.1 103.7 90.6
14 1.3083 -6.8 -64.3 -68.7 104.2 60.1
15 2.1704 -8.7 -78.5 81.5 93.0 61.0
16 2.7105 -13.2 102.7 30.7 104.2 100.7
17 4.2589 -13.9 99.2 -16.4 94.9 62.3
18 4.6003 -13.9 88.8 3.8 93.1 66.7
19 5.4902 -15.8 -101.9 -13.7 92.2 52.9
20 5.6077 -17.1 92.2 9.7 106.7 61.8
21 6.3065 -16 93.3 5.6 93.0 51.9
22 6.6374 -15.7 106.6 0.7 92.9 61.7
23 7.0427 -21.6 119.5 -21.9 105.2 58
24 8.6523 -22.8 -123.8 33.6 107.8 57
Per-Cluster Parameters
Parameter cASD in [°] cASA in [°] cZSD in [°] cZSA in [°] XPR in [dB]
Value 2 15 3 7 7

3GPP
Release 16 76 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.7.1-4. CDL-D.

Cluster # Cluster PAS Normalized Delay Power in [dB] AOD in [°] AOA in [°] ZOD in [°] ZOA in [°]
Specular(LOS path) 0 -0.2 0 -180 98.5 81.5
1
Laplacian 0 -13.5 0 -180 98.5 81.5
2 Laplacian 0.035 -18.8 89.2 89.2 85.5 86.9
3 Laplacian 0.612 -21 89.2 89.2 85.5 86.9
4 Laplacian 1.363 -22.8 89.2 89.2 85.5 86.9
5 Laplacian 1.405 -17.9 13 163 97.5 79.4
6 Laplacian 1.804 -20.1 13 163 97.5 79.4
7 Laplacian 2.596 -21.9 13 163 97.5 79.4
8 Laplacian 1.775 -22.9 34.6 -137 98.5 78.2
9 Laplacian 4.042 -27.8 -64.5 74.5 88.4 73.6
10 Laplacian 7.937 -23.6 -32.9 127.7 91.3 78.3
11 Laplacian 9.424 -24.8 52.6 -119.6 103.8 87
12 Laplacian 9.708 -30.0 -132.1 -9.1 80.3 70.6
13 Laplacian 12.525 -27.7 77.2 -83.8 86.5 72.9
Per-Cluster Parameters
Parameter cASD in [°] cASA in [°] cZSD in [°] cZSA in [°] XPR in [dB]
Value 5 8 3 3 11

Table 7.7.1-5. CDL-E.

Cluster # Cluster PAS Normalized Delay Power in [dB] AOD in [°] AOA in [°] ZOD in [°] ZOA in [°]
Specular (LOS path) 0.000 -0.03 0 -180 99.6 80.4
1
Laplacian 0.000 -22.03 0 -180 99.6 80.4
2 Laplacian 0.5133 -15.8 57.5 18.2 104.2 80.4
3 Laplacian 0.5440 -18.1 57.5 18.2 104.2 80.4
4 Laplacian 0.5630 -19.8 57.5 18.2 104.2 80.4
5 Laplacian 0.5440 -22.9 -20.1 101.8 99.4 80.8
6 Laplacian 0.7112 -22.4 16.2 112.9 100.8 86.3
7 Laplacian 1.9092 -18.6 9.3 -155.5 98.8 82.7
8 Laplacian 1.9293 -20.8 9.3 -155.5 98.8 82.7
9 Laplacian 1.9589 -22.6 9.3 -155.5 98.8 82.7
10 Laplacian 2.6426 -22.3 19 -143.3 100.8 82.9
11 Laplacian 3.7136 -25.6 32.7 -94.7 96.4 88
12 Laplacian 5.4524 -20.2 0.5 147 98.9 81
13 Laplacian 12.0034 -29.8 55.9 -36.2 95.6 88.6
14 Laplacian 20.6419 -29.2 57.6 -26 104.6 78.3
Per-Cluster Parameters
Parameter cASD in [°] cASA in [°] cZSD in [°] cZSA in [°] XPR in [dB]
Value 5 11 3 7 8

7.7.2 Tapped Delay Line (TDL) models

The TDL models for simplified evaluations, e.g., for non-MIMO evaluations, are defined for the full frequency range
from 0.5 GHz to 100 GHz with a maximum bandwidth of 2 GHz.

Three TDL models, namely TDL-A, TDL-B and TDL-C, are constructed to represent three different channel profiles
for NLOS while TDL-D and TDL-E are constructed for LOS, the parameters of which can be found respectively in
Table 7.7.2-1, Table 7.7.2-2 , Table 7.7.2-3, Table 7.7.2-4 and Table 7.7.2-5.

The Doppler spectrum for each tap is characterized by a classical (Jakes) spectrum shape and a maximum Doppler shift
fD where f D = v λ0 . Due to the presence of a LOS path, the first tap in TDL-D and TDL-E follows a Ricean fading
distribution. For those taps the Doppler spectrum additionally contains a peak at the Doppler shift fS = 0.7 fD with an
amplitude such that the resulting fading distribution has the specified K-factor.

3GPP
Release 16 77 3GPP TR 38.901 V16.1.0 (2019-12)

Each TDL model can be scaled in delay so that the model achieves a desired RMS delay spread, according to the
procedure described in Clause 7.7.3.

For LOS channel models, the K-factor of TDL-D and TDL-E can be set to a desired value following the procedure
described in Clause 7.7.6.

Table 7.7.2-1. TDL-A

Tap # Normalized delay Power in [dB] Fading distribution

1 0.0000 -13.4 Rayleigh
2 0.3819 0 Rayleigh
3 0.4025 -2.2 Rayleigh
4 0.5868 -4 Rayleigh
5 0.4610 -6 Rayleigh
6 0.5375 -8.2 Rayleigh
7 0.6708 -9.9 Rayleigh
8 0.5750 -10.5 Rayleigh
9 0.7618 -7.5 Rayleigh
10 1.5375 -15.9 Rayleigh
11 1.8978 -6.6 Rayleigh
12 2.2242 -16.7 Rayleigh
13 2.1718 -12.4 Rayleigh
14 2.4942 -15.2 Rayleigh
15 2.5119 -10.8 Rayleigh
16 3.0582 -11.3 Rayleigh
17 4.0810 -12.7 Rayleigh
18 4.4579 -16.2 Rayleigh
19 4.5695 -18.3 Rayleigh
20 4.7966 -18.9 Rayleigh
21 5.0066 -16.6 Rayleigh
22 5.3043 -19.9 Rayleigh
23 9.6586 -29.7 Rayleigh

Table 7.7.2-2. TDL-B

Tap # Normalized delay Power in [dB] Fading distribution

1 0.0000 0 Rayleigh
2 0.1072 -2.2 Rayleigh
3 0.2155 -4 Rayleigh
4 0.2095 -3.2 Rayleigh
5 0.2870 -9.8 Rayleigh
6 0.2986 -1.2 Rayleigh
7 0.3752 -3.4 Rayleigh
8 0.5055 -5.2 Rayleigh
9 0.3681 -7.6 Rayleigh
10 0.3697 -3 Rayleigh
11 0.5700 -8.9 Rayleigh
12 0.5283 -9 Rayleigh
13 1.1021 -4.8 Rayleigh
14 1.2756 -5.7 Rayleigh
15 1.5474 -7.5 Rayleigh
16 1.7842 -1.9 Rayleigh
17 2.0169 -7.6 Rayleigh
18 2.8294 -12.2 Rayleigh
19 3.0219 -9.8 Rayleigh
20 3.6187 -11.4 Rayleigh
21 4.1067 -14.9 Rayleigh
22 4.2790 -9.2 Rayleigh
23 4.7834 -11.3 Rayleigh

3GPP
Release 16 78 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.7.2-3. TDL-C

Tap # Normalized delays Power in [dB] Fading distribution

1 0 -4.4 Rayleigh
2 0.2099 -1.2 Rayleigh
3 0.2219 -3.5 Rayleigh
4 0.2329 -5.2 Rayleigh
5 0.2176 -2.5 Rayleigh
6 0.6366 0 Rayleigh
7 0.6448 -2.2 Rayleigh
8 0.6560 -3.9 Rayleigh
9 0.6584 -7.4 Rayleigh
10 0.7935 -7.1 Rayleigh
11 0.8213 -10.7 Rayleigh
12 0.9336 -11.1 Rayleigh
13 1.2285 -5.1 Rayleigh
14 1.3083 -6.8 Rayleigh
15 2.1704 -8.7 Rayleigh
16 2.7105 -13.2 Rayleigh
17 4.2589 -13.9 Rayleigh
18 4.6003 -13.9 Rayleigh
19 5.4902 -15.8 Rayleigh
20 5.6077 -17.1 Rayleigh
21 6.3065 -16 Rayleigh
22 6.6374 -15.7 Rayleigh
23 7.0427 -21.6 Rayleigh
24 8.6523 -22.8 Rayleigh

Table 7.7.2-4. TDL-D.

Tap # Normalized delay Power in [dB] Fading distribution

0 -0.2 LOS path
1
0 -13.5 Rayleigh
2 0.035 -18.8 Rayleigh
3 0.612 -21 Rayleigh
4 1.363 -22.8 Rayleigh
5 1.405 -17.9 Rayleigh
6 1.804 -20.1 Rayleigh
7 2.596 -21.9 Rayleigh
8 1.775 -22.9 Rayleigh
9 4.042 -27.8 Rayleigh
10 7.937 -23.6 Rayleigh
11 9.424 -24.8 Rayleigh
12 9.708 -30.0 Rayleigh
13 12.525 -27.7 Rayleigh
NOTE: The first tap follows a Ricean distribution with a K-factor of K1 = 13.3 dB and a mean power of 0dB.

3GPP
Release 16 79 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.7.2-5. TDL-E.

Tap # Normalized delay Power in [dB] Fading distribution

0 -0.03 LOS path
1
0 -22.03 Rayleigh
2 0.5133 -15.8 Rayleigh
3 0.5440 -18.1 Rayleigh
4 0.5630 -19.8 Rayleigh
5 0.5440 -22.9 Rayleigh
6 0.7112 -22.4 Rayleigh
7 1.9092 -18.6 Rayleigh
8 1.9293 -20.8 Rayleigh
9 1.9589 -22.6 Rayleigh
10 2.6426 -22.3 Rayleigh
11 3.7136 -25.6 Rayleigh
12 5.4524 -20.2 Rayleigh
13 12.0034 -29.8 Rayleigh
14 20.6519 -29.2 Rayleigh
NOTE: The first tap follows a Ricean distribution with a K-factor of K1 = 22 dB and a mean power of 0dB.

7.7.3 Scaling of delays

The RMS delay spread values of both CDL and TDL models are normalized and they can be scaled in delay so that a
desired RMS delay spread can be achieved. The scaled delays can be obtained according to the following equation:

τ n,scaled = τ n,model ⋅ DSdesired (7.7-1)

in which

τ n ,model is the normalized delay value of the nth cluster in a CDL or a TDL model

τ n ,scaled is the new delay value (in [ns]) of the nth cluster

DSdesired is the wanted delay spread (in [ns])

The example scaling parameters are selected according to Table 7.7.3-1 where the values have been chosen such that
the RMS delay spreads span the range observed in measurements corresponding to the typical 5G evaluation scenarios.
For information purposes, examples of such RMS delay spreads for the different scenarios are given in Table 7.7.3-2
where the "short-delay profile" corresponds to the median RMS delay spread for LOS scenarios, while "normal-delay
profile" and "long-delay profile" correspond to the median and the 90th percentile RMS delay spread for NLOS
scenarios according to the channel parameters in Table 7.5-6. It can therefore be understood that a particular RMS delay
spread in Table 7.7.3-1 may occur in any scenario; however certain values may be more likely in some scenarios than in
others.

The example parameters given in Table 7.7.3-1 does not preclude the use of other scaling values if this is found
appropriate, for instance if additional scenarios are introduced or if e.g. the effect of beamforming needs to be captured
in a TDL. Both of these examples can potentially result in an increased range of experienced RMS delay spreads.

Table 7.7.3-1. Example scaling parameters for CDL and TDL models.

Model DSdesired
Very short delay spread 10 ns
Short delay spread 30 ns
Nominal delay spread 100 ns
Long delay spread 300 ns
Very long delay spread 1000 ns

3GPP
Release 16 80 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.7.3-2. Scenario specific scaling factors - for information only

Frequency [GHz]
Proposed Scaling Factor DSdesired in [ns] 2 6 15 28 39 60 70
Short-delay profile 20 16 16 16 16 16 16
Indoor office Normal-delay profile 39 30 24 20 18 16 16
Long-delay profile 59 53 47 43 41 38 37
Short-delay profile 65 45 37 32 30 27 26
UMi Street-canyon Normal-delay profile 129 93 76 66 61 55 53
Long-delay profile 634 316 307 301 297 293 291
Short-delay profile 93 93 85 80 78 75 74
UMa Normal-delay profile 363 363 302 266 249 228 221
Long-delay profile 1148 1148 955 841 786 720 698
Short-delay profile 32 32 N/A N/A N/A N/A N/A
RMa & RMa O2I Normal-delay profile 37 37 N/A N/A N/A N/A N/A
Long-delay profile 153 153 N/A N/A N/A N/A N/A
Normal-delay profile 240
UMi / UMa O2I
Long-delay profile 616

7.7.4 Spatial filter for generating TDL channel model

The TDL models described in Clause 7.7.2 are generated from the CDL models assuming ideal isotropic antennas at
both Tx and Rx. It is also possible to generate TDL models by assuming non-isotropic antennas like directive horn
antennas or array antennas.

The basic idea to generate a TDL model based on a filtered CDL model is shown in Figure 7.7.4-1 below.

Figure 7.7.4-1 The basic idea for filtering the CDL model to TDL model.

7.7.4.1 Exemplary filters/antenna patterns

Note that any filter/pattern can be applied on a CDL to derive a TDL for evaluating directional algorithms.

Example 1: Isotropic pattern

A{′tx ,rx} (θ ′, φ ′) = 1 (7.7-2)

Example 2: Rectangular mask

BW BW
1, θ ′ − 90° ≤ & φ′ ≤
A{tx ,rx} (θ ,φ , BW ) = 
′ ′ ′ 2 2 (7.7-3)
0, otherwise.

with BW denotes beamwidth.

Example 3: Simplified antenna pattern given in [ITU-R M.2135]

3GPP
Release 16 81 3GPP TR 38.901 V16.1.0 (2019-12)

Figure 7.7.4-2 Simplified antenna pattern [ITU-R M.2135].

7.7.4.2 Generation procedure

1. The following steps are needed to generate tapped delay line (TDL) models:Choose a CDL model (e.g. CDL-A).
Note that the models may be scaled according to Clause 7.7.5.1 prior to the filtering in order to represent
different angular spreads.

2. Choose spatial filters Atx′ and Arx′ defined in LCS

3. Transform the spatial filter into GCS to obtain Atx and Arx such that the pointing direction (θ ,φ ) is
p p
centered within the filter. The pointing direction may be defined

a. by the dominant path (θ P ,φP ) = (θi ,φi ) with (

i = arg max n PnCDL ) , where P n
CDL
denotes the CDL
cluster power values

b. Or an arbitrary direction

4. Calculate TDL cluster power values PnTDL given the following equation

PnTDL = PnCDL Arx (θ n ,ZOA , φn ,AOA )Atx (θ n ,ZOD , φn ,AOD ) (7.7-4)

7.7.5 Extension for MIMO simulations

Extended MIMO link-level channel models can be constructed according to two alternative methods described in the
following.

7.7.5.1 CDL extension: Scaling of angles

The angle values of CDL models are fixed, which is not very suitable for MIMO simulations for several reasons; The
PMI statistics can become biased, and a fixed precoder may perform better than open-loop and on par with closed-loop
or reciprocity beamforming. Furthermore, a CDL only represents a single channel realization. The predefined angle values
in the CDL models can be generalized by introducing angular translation and scaling. By translation, mean angle can be
changed to µφ , desired and angular spread can be changed by scaling. The translated and scaled ray angles can be obtained
according to the following equation:

ASdesired
φn,scaled = (φn,model − µφ ,model ) + µφ ,desired (7.7-5)
ASmodel
in which:

3GPP
Release 16 82 3GPP TR 38.901 V16.1.0 (2019-12)

φn ,model is the tabulated CDL ray angle

ASmodel is the rms angular spread of the tabulated CDL including the offset ray angles, calculated using the
angular spread definition in Annex A

µφ ,model is the mean angle of the tabulated CDL, calculated using the definition in Annex A

µφ ,desired is the desired mean angle

ASdesired is the desired rms angular spread

φn ,scaled is the resulting scaled ray angle.

The angular scaling is applied on the ray angles including offsets from the tabulated cluster angles. Typical angular
spreads for different scenarios can be obtained from the system-level model.
Example scaling values are:
- AOD spread (ASD) for each CDL model: {5, 10, 15, 25} degrees.

- AOA spread (ASA) for each CDL model: {30, 45, 60} degrees.

- ZOA spread (ZSA) for each CDL model: {5, 10, 15} degrees.

- ZOD spread (ZSD) for each CDL model: {1, 3, 5} degrees.

The angular scaling and translation can be applied to some or all of the azimuth and zenith angles of departure and arrival.
Note: The azimuth angles may need to be wrapped around to be within [0, 360] degrees, while the zenith angles may need
to be clipped to be within [0, 180] degrees.

7.7.5.2 TDL extension: Applying a correlation matrix

The TDLs and the spatial-filtered TDLs can be used with the correlation matrices for MIMO link-level simulations.
Typical correlation parameters can be derived from 1) delay & angular scaled CDLs with antenna array assumptions, 2)
system-level model with antenna array assumptions, or 3) by selecting extreme cases, e.g. uncorrelated, highly
correlated etc. For example, these following options can be considered:
1) Zero correlation (IID channel coefficients) can be used for any number of antenna elements

2) The correlation matrix construction method from 3GPP TS36.101/104 [15][16] can be used for linear and planar
(single- or dual-polarized) arrays.

- Other correlation parameters α, β, γ than those specified in 3GPP TS36.101/104 [15][16] and extensions to
larger antenna arrays can be considered. For typical scenarios, α and β will be in the range 0-1

- A representative set of values is {0,0.7,0.9,0.99}

Note: This approach can be applied to TDLs derived from spatially filtered CDLs to emulate hybrid BF system

Note: Other methodologies could also be developed, e.g.

- extending the 3GPP TS36.101/104 [15][16] procedure to planar arrays or more elements

- using CDLs in combination with array assumptions to derive per-tap correlation matrices as in [17].

- using the system-level model in combination with array assumptions to derive per-tap or per-channel
correlation matrices

3GPP
Release 16 83 3GPP TR 38.901 V16.1.0 (2019-12)

7.7.6 K-factor for LOS channel models

For the LOS channel models of CDL/TDL-D and CDL/TDL-E, the K-factor values may be changed by the user. Mean
and standard deviation of K-factor values can be found in Table 7.5-6, although other values may also be used.

If the K-factor of a model shall be changed to K desired [dB], the cluster powers for the Laplacian clusters (in case of
CDL) or the tap powers for the Rayleigh fading taps (in case of TDL) are determined by

Pn,scaled = Pn, model − K desired + K model , (7.7.6-1)

where Pn ,scaled and Pn ,model denote the scaled and the model path power (as given in the tables) of tap/cluster n . The

model's K-factor K model is defined as

 N Pn,model 10 
K model = P1,LOS
model − 10 log10  ∑ 10 . (7.7.6-2)
 n=1 
After scaling the powers, the delay spread needs to be re-normalized. This is done through the two steps below.

1) Calculate the actual RMS delay spread after the K-factor adjustment.

2) Divide the delays by that value to obtain DS = 1.

7.8 Channel model calibration

7.8.1 Large scale calibration
For large scale calibration, fast fading is not modeled. The calibration parameters can be found in Table 7.8-1. The
calibration results based on TR 38.900 V14.0.0 can be found in R1-165974.

3GPP
Release 16 84 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.8-1: Simulation assumptions for large scale calibration

Parameter Values
Scenarios UMa, UMi-Street Canyon, Indoor-office (open office)
Sectorization 3 sectors per cell site: 30, 150 and 270 degrees

BS antenna configurations Mg = Ng = 1; (M,N,P) = (10, 1, 1), dV = 0.5λ

BS port mapping The 10 elements are mapped to a single CRS port
102 degrees for UMa and UMi Street Canyon
BS antenna electrical downtilting
110 degrees for indoor
DFT precoding according to TR 36.897 with application of panning and tilting
Antenna virtualization
angles
44 dBm for UMi-Street Canyon, 49 for UMa at 6GHz
BS Tx power 35 dBm at 30GHz and 70 GHz for UMa and UMi-Street canyon
24 dBm for Indoor for all carrier frequencies
Bandwidth 20MHz for 6GHz, and 100MHz for 30GHz and 70 GHz
UT antenna configurations 1 element (vertically polarized), Isotropic antenna gain pattern
Handover margin (for calibration) 0dB
Following TR36.873 for UMa and UMi, (3D dropping)
UT distribution
uniform dropping for indoor with minimum distance (2D) of 0 m
UT attachment Based on pathloss considering LOS angle
UT noise figure 9 dB
Fast fading channel Fast fading channel is not modelled
O2I penetration loss 50% low loss and 50% high loss
Carrier Frequency 6 GHz, 30 GHz, 70GHz
Wrapping method for UMa and geographical distance based wrapping (mandatory)
UMi radio distance (optional)
Metrics 1) Coupling loss – serving cell (based on LOS pathloss)
2) Geometry (based on LOS pathloss) with and without white noise

7.8.2 Full calibration

The calibration parameters for full calibration including the fast fading modelling can be found in Table 7.8-2.
Unspecified parameters in Table 7.8-2 are the same as those in Table 7.8-1. When P=2, X-pol (+/-45 degree) is used for
BS antenna configuration 1 and X-pol (0/+90 degree) is used for UT antenna configuration. The calibration results
based on TR 38.900 V14.0.0 can be found in R1-165975.

3GPP
Release 16 85 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.8-2: Simulation assumptions for full calibration

Parameter Values
Scenarios UMa, UMi-street Canyon, Indoor-office (open office)
Carrier Frequency 6 GHz, 30 GHz, 60GHz, 70GHz
Bandwidth 20MHz for 6GHz, and 100MHz for 30GHz, 60 GHz and 70 GHz
44 dBm for UMi-Street Canyon, 49 for UMa at 6GHz
BS Tx power 35 dBm at 30GHz, 60 GHz and 70 GHz for UMa and UMi-Street canyon
24 dBm for Indoor for all carrier frequencies
Config 1: M = 4,N = 4,P = 2, Mg = 1, Ng = 2, dH = dV = 0.5λ, dH,g = dV,g = 2.5λ … calibration
BS antenna
metrics 1), 2), 3) are calibrated
configurations
Config 2: Mg = Ng = 1, M = N = 2, P = 1 … calibration metrics 1), 2), 4) are calibrated
Config 1: all 16 elements for each polarization on each panel are mapped to a single CRS port;
panning angles of the two subarrays: (0,0) degs; same downtilt angles as used for the large-
BS port mapping
scale calibrations
Config 2: each antenna element is mapped to one CRS port
UT antenna Mg = Ng = 1, M = N = 1, P = 2
configurations
Following TR36.873 for UMa and UMi, (3D dropping)
UT distribution
uniform dropping for indoor with minimum distance (2D) of 0 m
UT attachment Based on RSRP (formula) from CRS port 0
Polarized antenna
Model-2 in TR36.873
modelling
UT array orientation ΩUT,α uniformly distributed on [0,360] degree, ΩUT,β = 90 degree, ΩUT,γ = 0 degree
UT antenna pattern Isotropic
1) Coupling loss – serving cell
2) Wideband SIR before receiver without noise
3) CDF of Delay Spread and Angle Spread (ASD, ZSD, ASA, ZSA) from the serving cell
(according to circular angle spread definition of TR 25.996)
4)
Metrics CDF of largest (1st) PRB singular values (serving cell) at t=0 plotted in 10*log10 scale.
CDF of smallest (2nd) PRB singular values (serving cell) at t=0 plotted in 10*log10 scale.
CDF of the ratio between the largest PRB singular value and the smallest PRB singular value
(serving cell) at t=0 plotted in 10*log10 scale.

Note: The PRB singular values of a PRB are the eigenvalues of the mean covariance matrix in
the PRB.

7.8.3 Calibration of additional features

The calibration parameters for the calibration of oxygen absorption, large bandwidth and large antenna array, spatial
consistency, and blockage can be respectively found in Table 7.8-3, 7.8-4, 7.8-5, and 7.8-6. Unspecified parameters in
these tables are the same as those in Tables 7.8-1 and 7.8-2. When P=2, X-pol (+/-45 degree) is used for BS antenna
configuration 1 and X-pol (0/+90 degree) is used for UT antenna configuration. The calibration results based on TR
38.900 V14.0.0 can be found in R1-1700990.

Table 7.8-3: Simulation assumptions for calibration for oxygen absorption

Parameter Values
Scenarios UMi-street Canyon
Carrier Frequency 60 GHz
BS antenna
M = 4, N = 4, P = 2, Mg = 1, Ng = 2, dH = dV = 0.5λ, dH,g = dV,g = 2.5λ
configurations
all 16 elements for each polarization on each panel are mapped to a single CRS port; panning
BS port mapping angles of the two subarrays: (0,0) degs; same downtilt angles as used for the large-scale
calibrations
Drop multiple users in the multiple cells randomly, and collect the following metrics for each
Calibration method
user after attachment.

1) CDF of coupling loss (serving cell)

Metrics 2) Wideband SINR before receiver – determined from RSRP (formula) from CRS port 0
3) CDF of Delay Spread from the serving cell

3GPP
Release 16 86 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.8-4: Simulation assumptions for calibration for large bandwidth and large antenna array

Parameter Values
Scenarios UMi-street Canyon
Carrier Frequency 30GHz
Bandwidth 2GHz
BS antenna
M = 8, N = 8, P = 2, Mg = 1, Ng = 4, dH = dV = 0.5λ, dH,g = dV,g = 4λ
configurations
all 64 elements for each polarization on each panel are mapped to a single CRS port; panning
BS port mapping angles of the two subarrays: (0,0) degs; same downtilt angles as used for the large-scale
calibrations
The number of rays
NMPC = 40
per cluster
Drop multiple users in the multiple cells randomly, and collect the following metrics for each
Calibration method
user after attachment.

1) CDF of coupling loss (serving cell)

Metrics 2) Wideband SINR before receiver – determined from RSRP (formula) from CRS port 0
3) CDF of largest (1st) PRB singular values (serving cell) at t=0 plotted in 10*log10 scale

Table 7.8-5: Simulation assumptions for calibration for spatial consistency

Parameter Values
Scenarios UMi-street Canyon
Carrier Frequency 30 GHz
BS antenna
M = 4, N = 4, P = 2, Mg = 1, Ng = 2, dH = dV = 0.5λ, dH,g = dV,g = 2.5λ
configurations
all 16 elements for each polarization on each panel are mapped to a single CRS port; panning
BS port mapping angles of the two subarrays: (0,0) degs; same downtilt angles as used for the large-scale
calibrations
Following TR36.873, 3D dropping
uniform dropping for indoor with minimum distance of 0 m
UE distribution
For Config1: 100% UE indoor and in the 1st floor
For Config2: 100% UE outdoor
Config1: UE is stationary
Mobility
Config2: UE is moving with random direction and fixed speed, e.g., 30 km/h
For Config1:
Drop multiple UEs in a single cell, determine all permutations of pairs of UEs, collect the
variables for each pair and bin them into certain distance ranges, e.g., 1m/2m/10m, to get
Calibration method enough samples. Collect the following metrics 1) –6).
For Config2:
Drop multiple users in the single cell, and collect metric 1)-2) and 7)-9) for each user after
attachment.
1) CDF of coupling loss (serving cell)
2) Wideband SINR before receiver – determined from RSRP (formula) from CRS port 0
3) Cross-correlation coefficient of delay for the third cluster between paired UEs, see note
1
4) Cross-correlation coefficient of AOA for the third cluster between paired UEs
5) Cross-correlation coefficient of LOS/NLOS status between paired UEs
Metrics
6) Cross correlation coefficient of the channel response in Step 11 on the first non-DC
subcarrier in an OFDM symbol on antenna port 0 received on the first UE antenna over
multiple realizations between paired UEs
7) CDF of average varying rate of power for the third cluster, see note 2
8) CDF of average varying rate of delay for the third cluster (ns)
9) CDF of average varying rate of AOA for the third cluster (degree)
NOTE 1 For the UT pair at a certain distance range, the variables collected by two UEs can be denoted as X and Y,
respectively, then the cross-correlation coefficient for real number can be written as [E(XY)-E(X)E(Y)]
/sqrt([E(X^2)-E(X)^2])/ sqrt([E(Y^2)-E(Y)^2]), the cross-correlation coefficient for complex number can be
written as |[E(XY*)-E(X)E(Y)*]/sqrt([E(XX*)-E(X)E(X)*])/sqrt([E(YY*)-E(Y)E(Y)*])|.
NOTE 2 For the average varying rate, we assume the collecting interval, e.g., 100ms, and then get the samples for
a certain UE, the varying rate can be written as the standard variance of the samples per 100ms.

3GPP
Release 16 87 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.8-6: Simulation assumptions for calibration for blockage

Parameter Values
Scenarios UMi-street Canyon
Carrier Frequency 30 GHz
BS antenna
M = 4, N = 4, P = 2, Mg = 1, Ng = 2, dH = dV = 0.5λ, dH,g = dV,g = 2.5λ
configurations
all 16 elements for each polarization on each panel are mapped to a single CRS port; panning
BS port mapping angles of the two subarrays: (0,0) degs; same downtilt angles as used for the large-scale
calibrations
Calibration method For Model A:
Drop multiple users in the multiple cells, and collect the following metrics 1) – 3) for each user
after attachment. Optional self-blocking feature is made mandatory in the Landscape mode only
for calibration purposes.
For Model B:
Drop a BS in (0,0,30) and a UT in (100,0,1.5),
Jump directly to Step 11 and replace the channel with CDL-E.
Drop a blocking screen of size h=10m, w = 2m in (80,10,1.5)
Move the UT from (100,0,1.5) to (100,20,1.5) in small increments. For each UT position,
translate all the AODs and AOAs of CDL-E such that the specular (LOS path) of CDL-E is
pointing along the direct path between the BS and UT.
Collect metric 4)
1) CDF of coupling loss (serving cell)
2) Wideband SINR before receiver – determined from RSRP (formula) from CRS port 0
Metrics
3) CDF of ASA from the serving cell
4) RSRP as a function of UT position

7.8.4 Calibration of the indoor factory scenario

For the InF, the calibration parameters can be found in Table 7.8-7. The calibration results can be found in R1-1909704.

It should be noted absolute delay model had not been agreed by the deadline, so companies were not able to submit
CDF of first path excess delay for serving cell.

3GPP
Release 16 88 3GPP TR 38.901 V16.1.0 (2019-12)

Table 7.8-7: Simulation assumptions for large scale calibration for the indoor factory scenario

Parameter Values
Scenario InF-SL, InF-DL, InF-SH, InF-DH
Hall size InF-SL: 120x60 m
InF-DL: 300x150 m
InF-SH: 300x150 m
InF-DH: 120x60 m
Room height 10 m
Sectorization None
BS antenna
1 element (vertically polarized), Isotropic antenna gain pattern
configurations
UT antenna
1 element (vertically polarized), Isotropic antenna gain pattern
configurations
Handover margin (for 0dB
calibration)
18 BSs on a square lattice with spacing D, located D/2 from the walls.

- for the small hall (L=120m x W=60m): D=20m

- for the big hall (L=300m x W=150m): D=50m

BS deployment

BS height = 1.5 m for InF-SL and InF-DL

BS-height = 8 m for for InF-SH and InF-DH
uniform dropping for indoor with minimum 2D distance of 1 m
UT distribution
UT height = 1.5 m
UT attachment Based on pathloss
UT noise figure 9 dB
Carrier frequency 3.5 GHz, 28 GHz
Bandwidth 100 MHz
Clutter density: 𝑟𝑟 Low clutter density: 20%
High clutter density: 60%
Clutter height: ℎ𝑐𝑐 Low clutter density: 2 m
High clutter density: 6 m
Clutter size: 𝑑𝑑𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 Low clutter density: 10 m
High clutter density: 2 m
Metrics 1) Coupling loss – serving cell
2) Geometry with and without noise
3) CDF of delay spread and angle spread (ASD, ZSD, ASA, ZSA) according to the
definition in Annex A.1
4) CDF of first path excess delay for serving cell

8 Map-based hybrid channel model (Alternative

channel model methodology)
Map-based hybrid model is composed of a deterministic component following, e.g., METIS work [6] and a stochastic
component following mainly the model described in clause 7. The channel model methodology described in this clause
is an alternative to the methodology specified in clause 7, and can be used if:

3GPP
Release 16 89 3GPP TR 38.901 V16.1.0 (2019-12)

- The system performance is desired to be evaluated or predicted with the use of digital map to take into account
the impacts from environmental structures and materials.

The map-based hybrid model defined in this clause is not calibrated and can be used per company basis.

8.1 Coordinate system

The same coordinate system as defined in Clause 7.1 is applied.

8.2 Scenarios
The same scenarios as in Clause 7.2 can be applied.

8.3 Antenna modelling

The same antenna modelling as defined in Clause 7.3 can be applied.

8.4 Channel generation

The radio channels are created using the deterministic ray-tracing upon a digitized map and emulating certain stochastic
components according to the statistic parameters listed in Tables 7.5-6 to 7.5-11 [Note: Not all parameters listed in
these tables are used in hybrid model]. The channel realizations are obtained by a step-wise procedure illustrated in
Figure 8.4-1 and described below. In the following steps, downlink is assumed. For uplink, arrival and departure
parameters have to be swapped.

Configurable number of random clusters

Set network Generate Generate
Set scenario layout correlated large- cluster
Import digital map Antenna scale parameters delays
parameters (DS, AS, K)

LoS/NLoS
Ray-tracing
Deterministic
clusters

Perform
Generate Generate Generate
Generate random Merge
ray angle cluster cluster
XPRs coupling of clusters
offsets angles power
rays

Generate
Draw random
channel
initial phases
coefficient

Figure 8.4-1: Channel coefficient generation procedure

Step-wise procedure:

Step 1: Set environment and import digitized map accordingly

a) Choose scenario. Choose a global coordinate system and define zenith angle θ, azimuth angle ϕ, and spherical
basis vectors θˆ , φˆ as shown in Figure 7.5-2.

3GPP
Release 16 90 3GPP TR 38.901 V16.1.0 (2019-12)

b) Import digitized map according to the chosen scenario. The digitized map should at least contain the following
information:
- The 3D geometric information for each of major structures involving with buildings or rooms. The external
building walls and internal room walls are represented by surfaces and identified by the coordinates of the
vertices on each wall.

- The material and thickness of each wall as well as the corresponding electromagnetic properties including
permittivity and conductivity.

- Random small objects in certain scenarios (e.g, UMi outdoor)

The format of digitized map, including additional information besides above-mentioned, is per implementation wise
and out of scope of this description.

Step 2: Set network layout, and antenna array parameters

a) Give number of BS and UT.

b) Give 3D locations of BS and UT, and calculate LOS AOD (ϕLOS,AOD), LOS ZOD (θLOS,ZOD), LOS AOA (ϕLOS,AOA),
LOS ZOA (θLOS,ZOA) of each BS and UT in the global coordinate system

c) Give BS and UT antenna field patterns Frx and Ftx in the global coordinate system and array geometries

d) Give BS and UT array orientations with respect to the global coordinate system. BS array orientation is defined
by three angles ΩBS,α (BS bearing angle), ΩBS,β (BS downtilt angle) and ΩBS,γ (BS slant angle). UT array
orientation is defined by three angles ΩUT,α (UT bearing angle), ΩUT,β (UT downtilt angle) and ΩUT,γ (UT slant
angle). Give rotational motion of UT in terms of its bearing angle, downtilt angle and slant angle if UT rotation
is modelled.

e) Give speed and direction of motion of UT in the global coordinate system for virtual motion.

f) Give system centre frequency/frequencies and bandwidth(s) for each of BS-UT links

If the bandwidth (denoted as B ) is greater than c/D Hz, where c is the speed of light and D is the maximum antenna
aperture in either azimuth or elevation, the whole bandwidth is split into K B equal-sized frequency bins, where
 B 
KB ≥   is a per-implementation parameter taking into account the channel constancy as well as other
 c D 
B
potential evaluation needs, and the bandwidth of each frequency bin is ∆B = . Within k-th frequency bin,
KB
the channel power attenuation, phase rotation, Doppler are assumed constant, whose corresponding values are
K B − 2k + 1
calculated based on the centre frequency of k-th frequency bin fk = fc − ∆B for 1 ≤ k ≤ K B ,
2
where fc is the centre frequency of the corresponding BS-UT link.

Step 3: Apply ray-tracing to each pair of link ends (i.e., end-to-end propagation between pair of Tx/Rx arrays).

a) Perform geometric calculations in ray-tracing to identify propagation interaction types, including LOS,
reflections, diffractions, penetrations and scattering (in case the digitized map contains random small objects),
for each propagation path. In general, some maximum orders of different interaction types can be set.

- The theoretical principles and procedures of geometric tracing calculations can be found in [6]~[10][12]. This
description does not intend to mandate new concepts and/or procedures to the conventional ray-tracing
algorithms; on the other hand, the implementation-based variations aiming to reduce computation complexity
are allowed within limits of acceptable calibration tolerances.

- The same geometric calculation is shared among all K B frequency bins.

b) Perform electric field calculations over propagation path, based on identified propagation interaction types (LOS,
reflection, diffraction, penetration and scattering) and centre frequencies of frequency bins.

3GPP
Release 16 91 3GPP TR 38.901 V16.1.0 (2019-12)

The details of electric field calculation can be found in [6]~[13].

The modelling algorithms in geometry and electric field calculations for different propagation interactions are
summarized in the table below.

Table 8.4-1 Principles applied in ray-tracing

Geometry calculation Electric field calculation

LOS Free space LOS Friis equation [11]
Reflection Snell's law with image-based method [7] Fresnel equation [7]
Diffraction Fermat's principle [12] UTD [13]
Penetration Snell's law for transmission through slab [8] Fresnel equation [7]
Scattering (upon small Isotropic scattering [6] RCS-based scattering coefficient [6]
objects)
Note: For reasons of simplicity and simulation speed, the maximum order of reflection on a path without
diffraction is configurable from {1,2,3}; the maximum order of diffraction on a path without reflection is
configurable from {1,2}; the path containing both reflection and diffraction has 1-order reflection and 1-
order diffraction, besides any potential penetrations; and the maximum order of penetration on a path is
configurable, with the recommended value equal to 5.

The outputs from Step 3 should at least contain following for each pair of link ends:

- the LOS/NLOS flag to indicate whether a LOS propagation mechanism exists;

- the number of deterministic propagation paths LRT (also referred as deterministic clusters in Step 8. To avoid the
unnecessary computation complexity, these LRT deterministic paths only include those paths whose powers are
RT , real
higher than 25dB below the maximum deterministic path power, where the path power is denoted as Pl RT
and defined below);

- for each deterministic path ( lRT -th path sorted in ascending order of path delay):

- the flag indicating whether the deterministic path is generated with scattering upon random small objects;

-
RT RT RT
l RT
( )
the normalized path delay τ lRT = τ l′ − min τ l′ and the first arrival absolute delay min τ l′
RT l RT
( ) (with τ ′ lRT to

be the real absolute propagation delay of the path);

angles of arrival and departure [ φlRT , AOA , θ lRT ,ZOA , φlRT , AOD , θ lRT ,ZOD ];
RT RT RT RT
-

1 K B RT , real
- the power P
RT , real
l RT , k for k-th frequency bin, and the path power
RT , real
Pl RT = ∑ Pl , k
K B k =1 RT

1 K B RT
- the XPR κ lRT of the path, where κ lRT
RT RT
= ∑ κ l , k with κ lRTRT ,k being the XPR for k-th frequency bin.
K B k =1 RT

- to support for true motion, i.e. the case when a trajectory is specified for UT, a path ID is associated for each
deterministic path. The same ID is associated for a path across a number of UT locations as far as 1) it has
same interaction types in the same order and 2) its interactions occur in same walls or other surfaces.

The LRT deterministic paths are sorted by normalized path delay ( τ lRT ) in ascending order. That is to say, τ 1 =
RT RT

0.

If LRT =0 for a pair of link ends, the channel gain for this pair of link ends is assumed to be zero and the
remaining steps are skipped with none of random cluster.

Step 4: Generate large scale parameters e.g. delay spread, angular spreads and Ricean K factor for random clusters.

The generation of large scale parameters takes into account cross correlation according to Table 7.5-6 and uses
the procedure described in Clause 3.3.1 of [14] with the square root matrix C MxM (0) being generated using the

3GPP
Release 16 92 3GPP TR 38.901 V16.1.0 (2019-12)

Cholesky decomposition and the following order of the large scale parameter vector: sM = [sK, sDS, sASD, sASA, sZSD,
sZSA]T. Limit random RMS azimuth arrival and azimuth departure spread values to 104 degrees, i.e., ASA=
min(ASA ,104°), ASD = min(ASD ,104°). Limit random RMS zenith arrival and zenith departure spread values
to 52 degrees, i.e., ZSA = min(ZSA,52°), ZSD = min(ZSD,52°). For the parameter selection from Table 7.5-6,
the LOS/NLOS condition determined in Step 3 is applied.

Step 5: Generate delays (denoted as { τ

RC
}) for random clusters.

Delays are drawn randomly according to the exponential delay distribution

τ n′ = − µτRC ln( X n ) (8.4-1)

 1  LRT
where µτRC = max µτ ,
LRT
∑τ
 , Xn ~ uniform(0,1), and cluster index n = 0,…, LRC
RT
lRT
′ to be
′ with LRC
 lRT =1 
configurable. A recommended value for LRC′ is the number of clusters given in Table 7.5-6.

LRT  1 LRT

µτ = rτ DS +  rτ DS −
′ + 1 
LRC LRT
∑τ RT
l RT
 , where rτ is the delay distribution proportionality factor given

l RT =1 
in Table 7.5-6.

Normalise the delays by subtracting the minimum delay and sort the normalised delays to ascending order:

τ n = sort (τ n′ − min(τ n′ )) Cτ (8.4-2)

where Cτ is the additional scaling of delays to compensate for the effect of LOS peak addition to the delay
spread, and is depending on the heuristically determined Ricean K-factor [dB] as generated in Step 4:

0.7705 − 0.0433K + 0.0002 K 2 + 0.000017 K 3 LOS condition

Cτ =  (8.4-3)
 1 NLOS condition

For the delay used in cluster power generation in Step 6, the scaling factor Cτ is always 1.

The n-th random cluster is removed if n=0 or τ n − τ lRT < τ th

RT
for any of 1≤lRT≤LRT, where τth is given by

 1 
τ th = µτRC ⋅ ln  , and p0 is the configurable probability for cluster inter-arrival interval to be less than
 1 − p0 
τth. For example, set p0 =0.2 to obtain τth=0.223 µτRC .

Denote τ n
RC
for 1≤n≤LRC as the delays of the LRC random clusters that remain after the cluster removal.

As an add-on feature, the absolute time of arrival of a cluster is derived as:

- τ l′RT
, if the cluster is the lRT -th deterministic path obtained in Step 3 for 1≤lRT≤LRT;

min(τ l′RT ) + τ n , if the cluster is the n-th random cluster for 1≤n≤LRC.
RC
-
l RT

Step 6: Generate powers (denoted as Pi RC ,real for 1≤i≤LRC) for random clusters.

3GPP
Release 16 93 3GPP TR 38.901 V16.1.0 (2019-12)

Cluster powers for the random clusters are calculated assuming a single slope exponential power delay profile.
First, the virtual powers (denoted as Pi RC ,virtual for 1≤i≤LRC) of random clusters and virtual powers (denoted as
PjRT ,virtual for 1≤j≤LRT) of deterministic clusters are calculated as following.

Denote:
− Ζi , RC
 r − 1
Vi RC = exp − τ iRC τ  ⋅ 10 10
(8.4-4)
 rτ DS 

− Ζ j , RT
 rτ − 1 
V jRT = exp − τ RT  ⋅ 10 10
(8.4-5)
rτ DS 
j

where Ζi,RC and Ζj,RT are the per cluster shadowing terms in [dB] and meet distribution of N(0, ζ2 ). Then,

1 Vi RC
Pi RC ,virtual
= ⋅ (8.4-6)
A + 1 LRC LRT

i
=i 1 =j 1
∑V RC
+ ∑V j
RT

1 V jRT A
P j
RT ,virtual
= ⋅ LRC LRT
+ ⋅ δ ( j − 1) (8.4-7)
A +1 A +1
∑V
i =1
i
RC
+ ∑V
j =1
j
RT

In the case of LOS condition, A=KR with KR being the Ricean K-factor obtained in Step 4 and converted to linear
scale; otherwise, A=0. The real power (including effects of pathloss) per random cluster in k-th frequency bin is
given by
LRT

∑P
j =1
RT , real
j ,k
, real
Pi ,RC
k = LRT
⋅ Pi RC ,virtual (8.4-8)

∑P j =1
j
RT ,virtual

for 1≤i≤LRC and 1 ≤ k ≤ K B . Similar to path power of deterministic cluster, the path power of i-th random
cluster is calculated as

1 K B RC , real
Pi RC , real = ∑ Pi, k .
K B k =1
(8.4-9)

Step 7: Generate arrival angles and departure angles for both azimuth and elevation, for each random cluster.

For azimuth angles of the n-th random cluster:

The composite PAS in azimuth of all random clusters is modelled as wrapped Gaussian (see Table 7.5-6). The
AOAs are determined by applying the inverse Gaussian function with input parameters PnRC ,real and RMS angle
spread ASA

2(ASA / 1.4) − ln PnRC ,real max Pi RC ,real , PjRT ,real 

 i, j 
( )
φ n′, AOA = (8.4-10)
Cφ

with constant Cφ defined as

3GPP
Release 16 94 3GPP TR 38.901 V16.1.0 (2019-12)

Cφ
NLOS

Cφ =  NLOS
(
⋅ 1.1035 − 0.028 K − 0.002 K 2 + 0.0001K 3 ) , for LOS
(8.4-11)
Cφ , for NLOS

NLOS
where Cφ is defined as a scaling factor related to the total number of clusters and is given in Table 7.5-2.

In the LOS case, constant Cφ also depends on the Ricean K-factor K in [dB], as generated in Step 4. Additional
scaling of the angles is required to compensate for the effect of LOS peak addition to the angle spread.
Assign positive or negative sign to the angles by multiplying with a random variable Xn with uniform
distribution to the discrete set of {1,–1}, and add component Yn ~ N 0, (ASA 7 ) ( 2
) to introduce random
variation

φn , AOA = X nφn′, AOA + Yn + φcenter , AOA (8.4-12)

where φcenter, AOA is calculated as

 LRT 
φcenter , AOA = arg ∑ Pl RT ,real ⋅ exp( jφlRT
, AOA )
 (8.4-13)
 l =1 

Note that φlRT

, AOA shall be given in radians here.

The generation of AOD (φn,AOD) follows a procedure similar to AOA as described above.

For zenith angles of the n-th random cluster:

The generation of ZOA assumes that the composite PAS in the zenith dimension of all random clusters is
Laplacian (see Table 7.5-6). The ZOAs are determined by applying the inverse Laplacian function with input
parameters PnRC ,real and RMS angle spread ZSA

 i, j
(
ZSA ln PnRC ,real max Pi RC ,real , PjRT ,real 

)
θ n′, ZOA =− (8.4-14)
Cθ

with Cθ defined as

Cθ =  θNLOS
(
C NLOS ⋅ 1.3086 + 0.0339 K − 0.0077 K 2 + 0.0002 K 3 ) , for LOS
, (8.4-15)
Cθ , for NLOS

where CθNLOS is a scaling factor related to the total number of clusters and is given in Table 7.5-4.

In the LOS case, constant Cθ also depends on the Ricean K-factor K in [dB], as generated in Step 4. Additional
scaling of the angles is required to compensate for the effect of LOS peak addition to the angle spread.

Assign positive or negative sign to the angles by multiplying with a random variable Xn with uniform
distribution to the discrete set of {1,–1}, and add component Yn ~ N 0, (ZSA 7 ) ( 2
) to introduce random
variation
(8.4-16)
θ n , ZOA = X nθ n′, ZOA + Yn + θ ZOA

3GPP
Release 16 95 3GPP TR 38.901 V16.1.0 (2019-12)

where θ ZOA = 90 0 if the UT is located indoors and θ ZOA = θ center ,ZOA if the UT is located outdoors, where
θ center, ZOA is calculated as

 LRT 
θ center ,ZOA = arg ∑ Pl RT ,real ⋅ exp( jθ lRT
, ZOA )
 (8.4-17)
 l =1 

Note that θ lRT

, ZOA shall be given in radians here.

The generation of ZOD follows the same procedure as ZOA described above except equation (8.4-16) is
replaced by

θ n , ZOD = X nθ n′, ZOD + Yn + θ center , ZOD + µoffset , ZOD (8.4-18)

where variable Xn is with uniform distribution to the discrete set of {1,–1}, Yn ~ N 0, (ZSD 7 ) , ( 2
) µ offset , ZOD
is given in Tables 7.5-7 to 7.5-11.

Step 8: Merge deterministic clusters and random clusters.

RT , real
First, remove any deterministic or random cluster with less than -25 dB power compared to max{ Pj ,

Pi RC ,real } for all 1≤j≤LRT and 1≤i≤LRC. Then, simply put the remaining deterministic clusters and random
clusters into single set of clusters, and meanwhile maintain an attribute for each cluster to indicate whether the
cluster is a deterministic cluster or a random cluster.

Step 9: Generate ray delays and ray angle offsets inside each cluster, where the cluster can be either random or
deterministic.

Denote M as the number of rays per cluster, where M=1 if the cluster corresponds to n=1 in the LOS case,
otherwise the value of M is given in Table 7.5-6.
When K B = 1 :

The relative delay of m-th ray within n-th cluster is given by τ n′ , m = 0 for m = 1,…,M.
The azimuth angle of arrival (AOA) for the m-th ray in n-th cluster is given by

φn ,m, AOA = φn , AOA + c ASAα m (8.4-19)

where c ASA is the cluster-wise rms azimuth spread of arrival angles (cluster ASA) in Table 7.5-6, and offset
angle αm is given in Table 7.5-3. φn, AOA equals to the AOA angle output from Step 3 if n-th cluster is
deterministic cluster, and equals to the AOA angle (8.4-12) in Step 7 if n-th cluster is random cluster.

The generation of AOD (φn,m,AOD) follows a procedure similar to AOA as described above.

The zenith angle of arrival (ZOA) for the m-th ray in n-th cluster is given by

θ n ,m,ZOA = θ n ,ZOA + cZSAα m (8.4-20)

where cZSA is the cluster-wise rms spread of ZOA (cluster ZOA) in Table 7.5-6, and offset angle αm is given in
Table 7.5-3. Assuming that θ n ,m, ZOA is wrapped within [0, 360°], if θ n ,m,ZOA ∈ [180°,360°] , then θ n,m,ZOA is
set to (360° − θ n ,m ,ZOA ) . θ n,ZOA equals to the ZOA angle output from Step 3 if n-th cluster is deterministic
cluster, and equals to the ZOA angle (8.4-16) in Step 7 if n-th cluster is random cluster.
The zenith angle of departure (ZOD) for the m-th ray in n-th cluster is given by

3GPP
Release 16 96 3GPP TR 38.901 V16.1.0 (2019-12)

µlgZSD (8.4-21)
θ n ,m ,ZOD = θ n ,ZOD + (3 / 8)(10 )α m

where µ lgZSD is the mean of the ZSD log-normal distribution. θ n,ZOD equals to the ZOD angle output from
Step 3 if n-th cluster is deterministic cluster, and equals to the ZOD angle (8.4-18) in Step 7 if n-th cluster is
random cluster.

When K B > 1 :

(
The relative delay of m-th ray within n-th cluster is given by τ n′ , m = sort τ n′′, m − min
1≤ m ≤ M
{τ ′′ }) that are sorted in
n, m

ascending order, where τ n′′,m ~ unif (0,2cDS ) , with the cluster delay spread as given in Table 7.5-6.
unif (a, b ) denotes the continuous uniform distribution on the interval [a, b] . Note that τ n,′′ m shall be the
independently generated.
The azimuth angles (AOA and AOD) and zenith angles (ZOA and ZOD) for the m-th ray in n-th cluster in each
frequency bin is given by
φ n ,m , AOA = φ n , AOA + φ n′,m , AOA
φ n ,m , AOD = φ n , AOD + φ n′,m , AOD
(8.4-22)
θ n ,m , ZOA = θ n , ZOA + θ n′,m , ZOA
θ n ,m , ZOD = θ n , ZOD + θ n′,m , ZOD

for m = 1,…,M, where φn,{ AOA| AOD} and θ n ,{ZOA| ZOD} equal to the {AOA,AOD} and {ZOA, ZOD} angle
outputs from Step 3 if n-th cluster is deterministic cluster, and equal to the {AOA,AOD} and {ZOA, ZOD}
angle in Step 7 if n-th cluster is random cluster; and
φn′, m, AOA ~ 2cASA unif (-1,1)
φn′, m, AOD ~ 2cASD unif (-1,1)
(8.4-23)
θ n′, m, ZOA ~ 2cZSA unif (-1,1)
θ n′, m, ZOD ~ 2cZSD unif (-1,1)

with the respective cluster angular spreads as given in Tables 7.5-6 to 7.5-11.

Assuming that θ n ,m, ZOA is wrapped within [0, 360°], if θ n ,m,ZOA ∈ [180°,360°] , then θ n,m,ZOA is set to
(360° − θ n ,m ,ZOA ) .
Step 10: Generate power of rays in each cluster, where coupling of rays within a cluster for both azimuth and elevation
could be needed.

Given Pn ,k as the real power in k-th frequency bin for the n-th cluster (either deterministic or random) obtained
from Step 8,
When K B = 1 :

Couple randomly AOD angles φn,m,AOD to AOA angles φn,m,AOA within a cluster n. Couple randomly ZOD angles
θ n ,m,ZOD with ZOA angles θ n ,m,ZOA using the same procedure. Couple randomly AOD angles φn,m,AOD with
ZOD angles θ n ,m,ZOD within a cluster n.
The power of m-th ray in n-th cluster and in k-th frequency bin is given by Pn ,m ,k = Pn ,k M for m = 1,…,M.

3GPP
Release 16 97 3GPP TR 38.901 V16.1.0 (2019-12)

When K B > 1 :

Pn′,m
The power of m-th ray in n-th cluster and in k-th frequency bin is given by Pn ,m ,k = Pn ,k ⋅ M
for
∑ P′
m =1
n ,m

m = 1,…,M, where

 τ′   2 φn′,m , AOA   2 φn′,m , AOD 

Pn′,m = exp − n ,m  exp −  exp − 
 cDS   cASA   cASD 
   
(8.4-24)
 2 θ n′,m ,ZOA   2 θ n′,m ,ZOD 
⋅ exp −  exp − 
 cZSA   cZSD 
   

and cDS , cASA , cASD , and cZSA are respectively the intra-cluster delay spread and the corresponding intra-
cluster angular spreads that are given in Table 7.5-6. The cluster zenith spread of departure is given by
3 µ
cZSD = 10 lgZSD , (8.4-25)
8

with µ lgZSD being defined in Tables 7.5-7, 7.5-8, 7.5-9, 7.5-10 and 7.5-11.

Step 11: Generate XPRs

Generate the cross polarization power ratios (XPR) κ for each ray m of each cluster n. XPR is log-Normal
distributed. Draw XPR values as
X n ,m / 10
κ n , m = 10 (8.4-26)

where
2
X n , m ~ N ( µ XPR , σ XPR ) is Gaussian distributed with σ XPR given from Table 7.5-6. If n-th cluster is a
deterministic cluster, µ = 10 log10 κ lRT ; otherwise, µ = µ XPR is given in Table 7.5-6.
RT

Note: X n , m is independently drawn for each ray and each cluster.

Step 12: Draw initial random phases

{ θθ θφ φθ φφ
}
Draw random initial phase Φ n ,m , Φ n ,m , Φ n ,m , Φ n ,m for each ray m of each cluster n and for four different
polarisation combinations (θθ, θϕ, ϕθ, ϕϕ). The distribution for initial phases is uniform within (-π,π).

In the LOS case, calculate an initial phase Φ LOS = −2π d 3 D λ0 for both θθ and ϕϕ polarisations, where d3D
is the 3D distance between transmitter and receiver and λ0=c/fc is the wavelength of the modelled propagation
link.
Step 13: Generate channel coefficients for each cluster n and each receiver and transmitter element pair u, s.

In case of NLOS, the channel coefficients of ray m in cluster n for a link between Rx antenna u and Tx antenna s
at time t in k-th frequency bin can be calculated as

3GPP
Release 16 98 3GPP TR 38.901 V16.1.0 (2019-12)

 Frx ,u ,θ (θ n ,m,ZOA , φn ,m, AOA )

H u ,s ,n ,m,k (t ) = 
T
 exp jΦθθ

( n ,m ) κ n−,1m exp( jΦθφ
n ,m )


 Frx ,u ,φ (θ n ,m,ZOA , φn ,m, AOA )
 −1
(
 κ n ,m exp jΦ φθn ,m ) (
exp jΦ φφ
n ,m ) 
 Ftx ,s ,θ (θ n ,m ,ZOD , φn ,m , AOD ) f (8.4-27)
⋅
F (θ , φ ) 

( c
) 
⋅ exp j 2π k rˆrxT ,n ,m .d rx ,u + rˆtxT,n ,m .d tx ,s 
 tx ,s ,φ n ,m ,ZOD n ,m , AOD   
−(OLn , m ( f k )+ BLn , m ( f k ,t ))
 
⋅  Pn ,m ,k ⋅ 10 20  exp j 2π f k rˆrxT ,n ,m .v t 
   c 
 
where Frx,u,θ and Frx,u,ϕ are the receive antenna element u field patterns in the direction of the spherical basis
vectors, θˆ and φˆ respectively, Ftx,s,θ and Ftx,s,ϕ are the transmit antenna element s field patterns in the direction
of the spherical basis vectors, θˆ and φˆ respectively. The delay (TOA) for ray m in cluster n for a link between
Rx antenna u and Tx antenna s is given by:

τ u , s ,n ,m = τ n + τ n′ ,m − 1c rˆrxT ,n ,m .d rx ,u − 1c rˆtxT,n ,m .d tx , s (8.4-28)

For the m-th ray within n-th cluster, rˆrx , n , m is the spherical unit vector with azimuth arrival angle φn ,m, AOA and
elevation arrival angle θ n ,m, ZOA , given by

sin θ n ,m ,ZOA cos φn ,m , AOA 

 
rˆrx ,n ,m =  sin θ n ,m ,ZOA sin φn ,m , AOA  (8.4-29)
 cos θ n ,m ,ZOA 

rˆtx , n , m is the spherical unit vector with azimuth departure angle φn ,m , AOD and elevation departure angle
θ n ,m, ZOD , given by

sin θ n ,m ,ZOD cos φn ,m , AOD 

 
rˆtx ,n ,m =  sin θ n ,m ,ZOD sin φn ,m , AOD  (8.4-30)

 cos θ n ,m ,ZOD 

Also, d rx ,u is the location vector of receive antenna element u and d tx ,s is the location vector of transmit antenna
element s, κn,m is the cross polarisation power ratio in linear scale. If polarisation is not considered, the 2x2
( )
polarisation matrix can be replaced by the scalar exp jΦ n ,m and only vertically polarised field patterns are
applied.
The Doppler frequency component is calculated from the arrival angles (AOA, ZOA), and the UT velocity
vector v with speed v, travel azimuth angle ϕv, elevation angle θv and is given by

v = v.[sin θ v cos φv cos θ v ] ,

T
sin θ v sin φv (8.4-31)

In case of LOS, the channel coefficient is calculated in the same way as in (8.4-27) except for n=1:

 Frx ,u ,θ (θ LOS ,ZOA , φ LOS , AOA ) exp jΦ LOS

( )   Ftx ,s ,θ (θ LOS ,ZOD , φ LOS , AOD )
T
0
H u ,s ,n=1,k (t ) = 
 Frx ,u ,φ (θ LOS ,ZOA , φ LOS , AOA )  − exp( jΦ LOS )  Ftx ,s ,φ (θ LOS ,ZOD , φ LOS , AOD )
   
0 (8.4-32)
−(OLn , m =1 ( f k )+ BLn , m =1 ( f k ,t ))
f  
 exp j 2π f k rˆrxT ,LOS .v t 

c
( 
)
⋅ exp j 2π k rˆrxT ,LOS .d rx ,u + rˆtxT,LOS .d tx ,s . P1,k ⋅ 10 20
 c
    

where the corresponding delay (TOA) for cluster n=1 for a link between Rx antenna u and Tx antenna s is
given by τ u , s , n =1 = τ n − 1c rˆrx , LOS .d rx ,u − 1c rˆtx , LOS .d tx , s .
T T

3GPP
Release 16 99 3GPP TR 38.901 V16.1.0 (2019-12)

In (8.4-27) and (8.4-32), the oxygen absorption loss, OLn,m(f), for each ray m in cluster n at carrier frequency f is
modelled as

OLn,m(f) = α(f)/1000 · c ·
 n n, m
l RT
( )
τ + τ ′ + min τ ′  [dB]
l RT

(8.4-33)

where:
- α(f) is the frequency dependent oxygen loss per distance (dB/km) characterized in Clause 7.6.1;

- c is speed of light (m/s); and

- τn is the delay (s) obtained from Step 3 for deterministic clusters and from Step 5 for random clusters.
min(τ l′RT ) is from the output of Step 3.
l RT

In (8.4-27) and (8.4-32), blockage modelling is an add-on feature. If the blockage model is applied, the blockage
loss, BLn,m(f,t) in unit of dB, for each ray m in cluster n at carrier frequency f and time t is modelled in the same
way as given in Clause 7.6.4; otherwise BLn,m(f,t)=0dB for all f and t.

3GPP
Release 16 100 3GPP TR 38.901 V16.1.0 (2019-12)

Annex A:
Further parameter definitions

A.1 Calculation of angular spread

Based on the circular standard deviation in directional statistics, the following expression for the angular spread AS in
radians is proposed

 N M 
 ∑∑ exp ( jφn ,m ) Pn ,m 
AS= −2 ln  =n 1 =m 1 N M  (A-1)
 
 ∑∑ Pn ,m 
 =n 1=
m 1 

where Pn ,m is the power for the mth subpath of the nth path and φ n,m is the subpaths angle (either AOA, AOD, ZOA,
ZOD) given in radians.

A.2 Calculation of mean angle

The power weighted mean angle is given by

 N M

µφ = arg ∑∑ exp( jφn , m )Pn , m  (A-2)
 n =1 m =1 
where Pn ,m is the power for the mth subpath of the nth path and φ n,m is the subpaths angle (either AOA, AOD, ZOA,
ZOD) given in radians.

3GPP
Release 16 101 3GPP TR 38.901 V16.1.0 (2019-12)

Annex B: Change history

Change history
Date Meeting TDoc CR Rev Cat Subject/Comment New
version
2017-02 RAN1#88 R1-1703641 Skeleton TR based on TR 38.900 v14.2.0, adding new changes 0.0.1
(from R1-1701231, R1-1701406, R1-1701410, R1-1701411, R1-
1701412, R1-1701413, R1-1701414, R1-1701416, R1-1701417 and
R1-1701425) to have a TR applicable for the frequency range 0.5 to
100 GHz
2017-02 RAN1#88 R1-1704145 Adding new agreements (from R1-1702701, R1-1703635, R1- 0.1.0
1703873, R1-1703358, R1-1702703, R1-1703637, R1-1703354, R1-
1703458, R1-1703647, R1-1701936) and reflecting comments from
RAN1 e-mail discussions [88-17].
2017-02 RAN#75 RP-170394 Submitted to RAN for information 1.0.0
2017-03 RAN#75 RP-170737 Add TR number further to the approval of revised SID in RP-170379. 1.0.1
Submitted to RAN for one step approval
2017-03 RAN#75 MCC clean-up – Further to RAN#75 decision, TR goes under 14.0.0
change control as Rel-14
2017-06 RAN#76 RP-171208 0001 - F Correction on site specific effective height for TR38.901 14.1.0
2017-06 RAN#76 RP-171208 0003 - F Correction for LOS probability 14.1.0
2017-06 RAN#76 RP-171208 0004 - F Correction for scaling factor for ZOD generation 14.1.0
2017-06 RAN#76 RP-171208 0005 - F Correction for modelling of intra-cluster angular and delay spreads 14.1.0
2017-06 RAN#76 RP-171208 0007 - F TR38.901_CR_LCS_Radiation_Field_Notation 14.1.0
2017-06 RAN#76 RP-171208 0008 - F TR38.901_CR_Mean_angle_definition 14.1.0
2017-06 RAN#76 RP-171208 0009 - F TR38.901_CR_O2I_penetration_loss 14.1.0
2017-06 RAN#76 RP-171208 0010 1 F TR38.901_CR_Correction_on_Rural_LSP 14.1.0
2017-06 RAN#76 RP-171208 0011 1 F Clarification for spatial consistency procedure 14.1.0
2017-06 RAN#76 RP-171208 0013 - F TR38.901_CR_Scenarios_in_Simulation_Assumptions 14.1.0
2017-06 RAN#76 RP-171208 0014 - F TR38.901_CR_TDL_Spatial_Filter 14.1.0
2017-06 RAN#76 RP-171208 0015 - F Correction of spatial-consistency UE mobility modelling Procedure A 14.1.0
2017-07 MCC: UMa LOS probability in Table 7.4-2 is made fully visible 14.1.1
2017-09 RAN#77 RP-171648 0016 - F Correction on the formula of Annex A 14.2.0
2017-09 RAN#77 RP-171648 0018 2 F Correction on cross polarization power ratios in TR 38.901 14.2.0
2017-09 RAN#77 RP-171648 0019 1 F TR38.901_CR_Spatially-consistent_UT_modelling 14.2.0
2017-09 RAN#77 RP-171648 0020 1 F TR38.901_CR_Calibraiton_Results 14.2.0
2017-12 RAN#78 RP-172687 0022 - F Correction on 3D-InH channel model in TR 38.901 14.3.0
2018-06 SA#80 - - - - Update to Rel-15 version (MCC) 15.0.0
2019-09 RAN#85 RP-191944 0023 - B Addition of indoor industrial channel model – version created in error 15.1.0
– withdrawn
2019-10 RAN#85 RP-191944 0023 - B Addition of indoor industrial channel model – creation of rel-16 report 16.0.0
2019-12 RAN#86 RP-192629 0024 - F CR to TR 38.901 for remaining open issues in IIOT channel 16.1.0
modelling

3GPP