Project Number: IST-4-027756

Project Title: Wireless World Initiative New Radio – WINNER II

Document Type: I (Internal)

Document Number:
Document Title: Matlab SW documentation of WIM2 model
Source Activity: <insert>
Editor: Milan Narandžić, Pekka Kyösti
Authors: Milan Narandžić, Pekka Kyösti, Lassi Hentilä, Martin Käske
Status / Issue: 1.0
Date Last changes: 19.08.2008
File: WIM2 Matlab documentation V1.0.doc

Abstract: This is intended for the abstract

Keywords:

Document History:
19.08.08 ver 0.1 (Pekka) Document created by modifying Phase I Matlab SW
Documentation of the WIM model.doc (Authors: Jari Salo, Pekka
Kyösti, Daniela Laselva, Giovanni DelGaldo)
9.11.07 ver 0.2 (Pekka) Document updated to Phase II final model
13.11.2007 ver 0.3 (Pekka) Input parameter tables refined.
09.04.2008 ver 0.4 (M. Narandzic) Integration of 3D-Antenna-Array description into this
document. Major changes.
19.08.2008 ver 1.0 (Lassi) Final version. Minor changes.

Document Number: IST-4-027756 Page 1 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Table of Contents

1. Introduction and scope ......................................................................... 3

2. Installation.............................................................................................. 4

3. Model features ....................................................................................... 5

3.1 Basic features .......................................................................................................................... 5
3.2 3D-Antenna-Array model ....................................................................................................... 6
3.2.1 Conceptual changes related to GCS and 3D-AA ............................................................. 6
3.3 CDL model option .................................................................................................................. 6
3.4 Supported propagation scenarios ............................................................................................ 7

4. Model usage........................................................................................... 9
4.1 Construction of antenna array model (Preprocessing phase) .................................................. 9
4.1.1.1 Array Geometry (AG) ............................................................................................. 9
4.1.1.2 Field pattern (FP) representation ........................................................................... 10
4.1.1.3 Examples ............................................................................................................... 11
4.2 Description of the network layout......................................................................................... 12
4.2.1 Construction of semi-random layout.............................................................................. 12
4.2.1.1 Example................................................................................................................. 13
4.2.2 Manual editing ............................................................................................................... 13
4.2.3 Layout setup visualization ............................................................................................. 13
4.3 Channel matrix generation .................................................................................................... 13
4.3.1 Simple initialization ....................................................................................................... 13
4.3.2 Initialization with arbitrary parameters.......................................................................... 14
4.3.3 Initialization of the structural model parameters ........................................................... 14

5. Technical description of the SW ........................................................ 15

5.1 SW high level structure......................................................................................................... 15
5.2 Model input/output interface................................................................................................. 16
5.2.1 Global simulation parameters (WIMPAR) .................................................................... 16
5.2.2 Parameters defining network layout (LAYOUTPAR)................................................... 18
5.2.2.1 Link-level parameters (LINKPAR) ....................................................................... 19
5.2.2.2 Stations (Array) structure ...................................................................................... 20
5.2.2.3 Antenna element parameters ................................................................................. 21
5.2.3 External initialization of structural model parameters (INITVALUES) - optional........ 21
5.2.4 Output parameters.......................................................................................................... 22
5.3 Implementation notes............................................................................................................ 22
5.3.1 Speed optimization ........................................................................................................ 22
5.3.1.1 Compatibility issues .............................................................................................. 23
5.3.2 Output interpolation ....................................................................................................... 23

6. References ........................................................................................... 23

Document Number: IST-4-027756 Page 2 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

1. Introduction and scope

The channel modeling work package WP1 provides a Matlab implementation of the WINNER Phase II
Model (referred as WIM this onward). The main purpose of the WIM channel models is to generate a radio
channel realisations for a link and system level simulations. The implementation is based on the earlier
implementations of WINNER II interim model [2], WINNER I (interim) model [4], 3GPP/3GPP2 Spatial
Channel Model (SCM) [5] and SCM(E) Extension models. For details concerning the WIM model, please
refer to [1]. This document describes the SW implementation structure of WIM, its input/output interface
and its functionality.

The channel model takes the user defined parameters, the MIMO radio link parameters and antenna
parameters described in [1] as an input. Channel matrices can be generated for multiple BS-MS links with
one function call. The output is a multi-dimensional array which contains the channel impulse responses for
the given radio links. In addition, the randomly drawn channel parameters for each link will be given as an
output.

The channel convolution and other related operations are beyond the scope of the implemented channel
model.

Document Number: IST-4-027756 Page 3 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

2. Installation
The WIM package installs as a MATLAB mini-toolbox. Unzip the files in their own directory, e.g.
‘winner‘. Add the directory to MATLAB path. Type ‘help winner’ at MATLAB command to get started.

The package includes the following modules:

% WINNER Phase II channel model
% Version 1.2, Apr. 09, 2008
%
%% Channel model functions
% wim - WINNER channel model (D1.1.2)
% wim_core - Channel coefficient computation for a geometric channel model
% scm_mex_core.m - SCM computations
% scenpartables - Set WIM parameters for WINNER scenarios
% pathloss - Pathloss models for 2GHz and 5GHz
%
%% Helper functions for model Initialization
% wimparset - Model parameter configuration for WIM
% layoutparset - Layout parameter configuration for WIM
%
%% Determination of link-level parameters
% layout2link - Computes and converts layout to link parameters. The following
helper functions are used:
% StationDistXY - Distance between stations in XY plane
% StationDirectionXY - Link angular direction in Global-Coordinate-System, w.r.t Y-axis
% StationVelocityXY - MS velocity
%
%% Utility functions
% StationNumElements - Retrieve number of array elements
% NTlayout - Visualisation of network layout
%
%% Generation of structural parameters
% generate_bulk_par - Generation of RANDOM WIM bulk parameters
% LScorrelation.m - Correlation of Large-Scale-Parameters (LSP)
% LOSprobability.m - Probability of LoS condition
% offset_matrix_generation.m - Generation of fixed angle offsets
% ScenarioMapping.m - Retrive WINNER scnario label
% struct_generation.m - Manipulation with bulk_parameters
% fixedAoas.m - FIXED AoAs for CDL model
% fixedAods.m - FIXED AoDs for CDL model
% fixedPdp.m - FIXED PDP for CDL model
%
%% 3D Antenna Array Model functions
% AntennaArray.m - 3D-AA model construction
% ArrayPreprocess.m - construction w. element field-pattern rotatation in ACS
% AntennaResponse.m - Computation of array response
%
% Aperture_Calc.p - EADF representation of radiation pattern
% BP2Aperture.m - Interface toward Aperture_Calc.p for 2D field pattern
% BP2Aperture1D.m - Interface toward Aperture_Calc.p for 1D field pattern
% G_Calc3D_simple.p - Calculation for 2D field pattern (3D-AA)
% G_Calc1D_simple.p - Calculation for 1D field pattern
% antenna_pol_vect.m - Computation of polarization vectors
%
% mycart2sph.m - Transfmation of carthesian into spherical coordinates
% mysph2cart.m - Transfmation of spherical into carthesian coordinates
% rotate_vector.m - 3D rotation
% unrotate_vector.m - 3D inverse rotation
%
%% Miscellaneous functions
% cas - Circular angle spread (3GPP TR 25.996)
% ds - RMS delay spread
% dipole - Field pattern of half wavelength dipole
%
%% Examples
% example_EADF_approx.m - Representation of radiation pattern
% example_syntetic_arrays.m - Array struct generation
% example_channel_matrix.m - Generation of channel matrix

Note! The code is created and tested on Matlab versions 6.5, 7.1 (R14) and 7.5 (R2007b). Some other
version might cause problems.

Document Number: IST-4-027756 Page 4 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

3. Model features

3.1 Basic features

WINNER MIMO radio channel model enables system level simulations and testing. This means that
multiple links are to be simulated (evolved) simultaneously. System level simulation may include multiple
base stations, multiple relay stations, and multiple mobile terminals as in Figure 3.1. Link level simulation
is done for one link, which is shown by blue dashed ellipse. The short blue lines represent channel
segments where large scale parameters are fixed. System level simulation consists of multiple links. Both
link level and system level simulations can be done by modelling multiple segments, or by only one (CDL
model).

MS

BS
link

segments FRS v
σϕ
MS τ1
σφ
φ1 ϕ1
φ2
τ2 ϕ2 MS

BS

Figure 3.1. System level approach, several segments (drops).

A single link model is shown in Figure 3.2. The parameters used in the models are also shown in the figure.
Each circle with several dots represents scattering region causing one cluster. The number of clusters varies
from scenario to another.

Figure 3.2. Single link.

The number of paths is scenario dependent and varies from 8 to 24 (see [1]). The number of rays (subpaths)
is fixed to 20 for all scenarios.

Document Number: IST-4-027756 Page 5 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

3.2 3D-Antenna-Array model

3D-Antenna-Array model allows complete 3D description of polarization, directional filtering and spatial
displacement. It supports arbitrary array geometries (not limited to Uniform Linear Arrays). Each mobile or
base station can use different antenna array, where radiation pattern of each element can be independently
defined.
Detailed reasoning for chosen antenna array representation is given in [1]. In order to fully describe
positions and directivity of antenna elements in 3D space, 2 coordinate systems will be used (Figure 3.3):
• GCS – Global Coordinate System (used to define radio-network system layout, and as a reference
system for polarization)
• ACS – Array Coordinate System - describes array geometry and rotated radiation patterns of
antenna elements.
Relative position of CS are described by orientation and translation in 3D space.

Figure 3.3 Coordinate system definitions: GCS, ACS and ECS

Usage of the third co called Element-Coordinate-System (ECS) to represent radiation pattern of each
antenna element is not suitable since it increases simulation complexity. In [1] it was shown that
representation of field patterns (of all elements in array) in ACS requires lower complexity.
Additionally, if field patterns are represented with its Effective-Aperture-Density-Functions (EADF, 2D-
Fourier transform of the field pattern) that will reduce memory storage requirements and provide effective
way of interpolating the field pattern for arbitrary (θ, φ).
It was therefore concluded that most suitable representation for element field patterns is EADF
defined for all elements in the array in respect to common ACS.

3.2.1 Conceptual changes related to GCS and 3D-AA

• All angles used in simulations (AoA/AoD) are now defined relative to the GCS (Figure 3.2)
and not to the arrays broadside. This change is following evolution of link-level to system-
level model: single station can be considered in multiple links – therefore same reference is
necessary.
• All field patterns are polarized patterns. In the case where only single polarization of field
pattern is known, the other polarization dimension has to be filled with zeros. This is necessary
to support arbitrary array rotation.

3.3 CDL model option

Deliverable D1.1.2 [1] contains fixed channel models called Clustered Delay Line (CDL) models with
tabulated spatio-temporal channel parameters. They are fully deterministic despite random initial phases of
the rays. Doppler is not explicitly defined, because it results from the angular properties. These models
might be useful e.g. in the simulation system calibration simulations. With CDL models there is no
randomness in the large scale parameters.

Document Number: IST-4-027756 Page 6 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

CDL models can be selected by setting parameters:

wimpar.FixedPdpUsed =‘yes’ and wimpar.FixedAnglesUsed =‘yes’.

3.4 Supported propagation scenarios

For details about the scenarios definition see [1], sect 2.3. Spatial channel model parameters, derived
mostly based on the WINNER measurement campaigns [1], characterize each channel scenario. The
scenarios-dependent parameters are currently supported at center frequency of 2-6 GHz and at bandwidth
of 100 MHz.
Fixed parameters, defining every parameterized WINNER scenario are provided in the function
ScenParTables.m.

Document Number: IST-4-027756 Page 7 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Table 3.1. Supported propagation scenarios.

Scenario Definition LOS/ Mob. Freque CG Note

NLOS km/h ncy
(GHz)
A1 Indoor small office / LOS/ 0–5 2-6 LA
In building residential NLOS
A2 Indoor to outdoor NLOS 0–5 2-6 LA AP inside UT outside.
Outdoor environment urban
B1 Typical urban micro-cell LOS 0–70 2-6 LA, Typical distance ranges.
Hotspot MA Actual ranges depend on
NLOS frequency and antenna
heights
B2 Bad Urban LOS/ 0–70 2-6 MA Same as B1 +
micro-cell NLOS long delays
B3 Large indoor hall LOS 0–5 2-6 LA
Hotspot
B4 Outdoor to indoor. NLOS 0–5 2-6 MA B1 or C2 to the wall/window.
-Outdoor typical urban.
-Indoor A1
B5a LOS stat. feeder, LOS 0 2-6 MA Same channel model for hot
Hotspot rooftop to rooftop spot and metropol.
Metropol
B5b LOS stat. feeder, street- LOS 0 2-6 MA Typical distance ranges.
Hotspot level to street-level Actual ranges depend on
Metropol frequency and antenna
heights
B5c LOS stat. feeder, below- LOS 0 2-6 MA Typical distance range.
Hotspot rooftop to street-level Actual range depends on
Metropol frequency and antenna
heights
B5d NLOS stat. feeder, NLOS 0 2-6 MA Extended C2
Hotspot above rooftop to street-
Metropol level
B5f Feeder link BS -> FRS. LOS/ 0 2-6 WA Desired link: LOS or OLOS,
Approximately RT to RT OLOS/ Interfering links:
level. NLOS LOS/(OLOS) /NLOS
FRS -> MS = B1*
C1 Suburban LOS/ 0–120 2-6 WA
Metropol NLOS

C2 Typical urban NLOS 0–120 2-6 MA

Metropol macro-cell WA

C3 Bad Urban macro-cell LOS/ 0–70 2-6 - Same as C2 + long delays

NLOS

D1 Rural macro-cell LOS/ 0–200 2-6 WA

Rural NLOS

D2 a) Moving networks: LOS 0 –350 2-6 WA Very large Doppler

BS – MRS, rural variability.

b) Moving networks: LOS / 0–5 2-6 LA

MRS – MS, rural OLOS/
NLOS

Document Number: IST-4-027756 Page 8 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

4. Model usage
The model is used for generation of multidimensional channel matrix H, containing time-variant Channel-
Impulse-Responses (CIR) between all transmitter and receiver antenna combinations of MIMO system.
Generation of matrix is performed by MATLAB function wim.m. Apart from the matrix H full set of
parameters characterizing physical propagation channel can be retrieved as well. Detailed description of
input and output arguments of wim.m function is given in section 5.2. In this section preparation of network
layout description will be discussed, and particularly construction of antenna array simulation model.
Additionally, some examples will be given for both usage of model in simulation, and construction of the
antenna array models.

4.1 Construction of antenna array model (Preprocessing phase)

Although WINNER channel model is constructed as antenna independent model, antenna array model is
necessary to obtain signals at the output of the radio-channel. This model is deterministic and permanent -
from the viewpoint of the simulation, and can be created independently from channel model simulations.
Therefore certain type of array requires only single construction, which should be performed independently
from WIM simulations - in pre-processing phase. It is not good strategy to construct arrays each time when
WIM is used, instead defined antenna arrays should be stored and retrieved when needed. This becomes
particularly important when large number of operations is performed in this phase, i.e. when 3D field
patterns are rotated from ECS to ACS.
WINNER phase II channel model uses structure Array (subsection 5.2.2.2) for representation 3D-AA
model. For construction of Array structure MATLAB function AntennaArray.m is provided. If one would
like to verify created 3D-AA model, this can be performed by AntennaResponse.m function.
In order to make Array it is necessary to define its geometry (positions and rotation of elements), and to
provide element field patterns. The arguments provided to AntennaArray .m are always processed in
predefined way: first array geometry is created and after that field patterns are assigned.

Figure 4.1 Construction of antenna array

Supported options for a creation of Array structure are illustrated in the following examples:

ArrayGeometryExample1=AntennaArray('Pos',Position,’Rot’,Rotation, ’FP-ECS’,
FieldPattern);
ArrayGeometryExample2=AntennaArray('UCA',N,r, ’FP-ECS’, FieldPattern);
ArrayGeometryExample3=AntennaArray('ULA',N,d, ’FP-ACS’, FieldPattern);
4.1.1.1 Array Geometry (AG)
Geometry can be defined explicitly using ‘Pos’ and ‘Rot’ arguments followed by ELNUMx3 matrix, where
ELNUM is the number of elements. Alternatively, common array types ‘UCA’ and ‘ULA’ could be defined

Document Number: IST-4-027756 Page 9 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

2π
with few parameters only. For UCA elements are placed starting from x-axis (phi=0) every ∆ϕ = ,
N
and nth element is rotated for (n − 1) ∆ϕ in positive mathematical direction. ULA elements are placed
along x-axis is such a way that the center of the array is at [0;0;0] (for even N there is no antenna element at
[0;0;0]).
Default geometry:
If there are no parameters defining geometry default is single antenna positioned at centre of ACS, without
rotation.

Table 4.1. Input parameters for AntennaArray.m function.

Parameter Default
Group Definition value Unit Note
Name
Position ELNUMx3 matrix, where n-th row contains [x,
AG - m
y, z] position of n-th antenna element in ACS.
Rotation ELNUMx3 matrix, where n-th row contains
AG [Rot ;Rot ;Rot ] rotation of n-th antenna 0Nx3 rad
x y z
element afound axes of ACS.
ELNUM Number of physical antenna elements in array.
Used as 1st argument for ‘UCA’ and ‘ULA’
AG - -
options. Implicitly defines the first dimension
of Position, Rotation and FieldPattern.
r AG Radius of UCA. 1 m Optional
d Distance between antenna elements in 1/ELNU Optional
AG m
ULA. M
FieldPattern 4D array containing field patterns of antenna Same format is
elements. The dimensions of FieldPattern are used for both
FP [ELNUM POL EL AZ] = SIZE(FieldPattern) ‘FP-ECS’ and
‘FP-ACS’
options
POL Number of polarizations used to characterize
FieldPattern. The first dimension in
FP FieldPattern is used to store vertical
2
polarization, the second for horizontal. Missing
polarization dimensions of FieldPattern are
substituted with zeros.
EL Number of equidistant FieldPattern samples
FP - -
taken over elevation angle.
AZ Number of equidistant FieldPattern samples
FP - -
taken over azimuth angle.
Elevation Vector of elevation angles corresponding to 3rd
FP - deg
dimension of FieldPattern.
Azimuth Vector of azimuth angles corresponding to 4rd
FP - deg
dimension of FieldPattern.

4.1.1.2 Field pattern (FP) representation

The field patterns of individual array elements are described using the EADF1 defined in ACS. This was
done because EADF has proven to be superior in terms of memory requirements and interpolation errors.
EADF representation is created from a sampled field pattern (FP) that is provided as an input to
AntennaArray. The number of provided field patterns (#FP) for ‘FP-ACS’ and ‘FP-ECS’ should be equal to
number of elements in array (ELNUM), however:
a) if #FP is equal to 1, same FieldPattern will be apply to all antennas.
b) if 1<#FP< ELNUM the error will be issued
c) if #FP> ELNUM take only first N.
NOTE: When single FP defined in ECS is shared by all elements, EADF calculation will be performed
twice: i) calculation of CommonAperture in ECS and ii) during FP rotation to ACS.
Reference coordinating system

1
Effective Aperture Distribution Function

Document Number: IST-4-027756 Page 10 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

In order to distinguish between FPs that are expressed in ECS and ACS it was necessary to define different
argument types: ‘FP-ECS’ and ‘FP-ACS’. However, in both cases input argument FP has the same
structure.
Sampling grid
In order to calculate 2D Fourier transformation of FP equidistant sampling points are required. If non-
uniform samples are provided, additional interpolation (e.g. spline) of FP, before calculating its EADF, is
necessary. This may introduce significant errors in the FP representation and therefore we would assume
that uniform sampling is always used. This would mean that the azimuth angles must be equidistant from -
180 to (but not including) 180 degrees.
Polarization
Since 3D-Antenna-Array-Model generally allows arbitrary rotation of antenna elements, it is necessary to
calculate projections to referent polarization vectors in GCS. Using of wimpar.PolarizedArrays=’no’ option
in wimparset.m (WIM1) with 3D-Antenna Array is not directly applicable since the representation of the
polarization for single-polarized but rotated antenna still requires two orthogonal reference polarization
vectors – what is equivalent to previous polarized option (wimpar.PolarizedArrays=’yes’). Otherwise, only
single (vertical or horizontal) projection of polarization vector will be handled, what is obviously wrong.
Therefore control parameter wimpar.PolarizedArrays can not be used (in general) to reduce complexity –
channel coefficients will be always based on polarized FP (after rotation).

Default field pattern:

If neither ‘FP-ACS’ nor ‘FP-ECS’ are defined use isotropic, vertically polarized antenna with XPD=∞.

4.1.1.3 Examples
Helper function arrayparset.m is provided to illustrate construction of Arrays. It shows
construction of ULA 2/4/8 and UCA 4/8 with synthetic field pattern defined in ECS. Slanted dipole and
isotropic radiation patterns are used for that purpose.
This section provides some additional examples and explanations about antenna array model creation.

Example 1:
Geometry: Unifom-Circular-Array (UCA) with 16 elements, and radius of 5 cm
Field pattern for each element is defined in its own ECS. This happens if each array element is measured
separately or if analytical description (i.e. mathematical expression/function) is used to define of the
patterns.

Array=AntennaArray('UCA',16,0.05,’FP-ECS’,FP);

FP should have dimensions  ELNUM POL EL AZ  :

 16 1 # AS 
 2
• First dimension should fit to number of elements required for UCA geometry, alternatively can be
1 what would mean that same FP will be applied to all array elements.
• Both polarizations must be provided for FP. It only single polarization is available other dimension
must be filled with zeros.
• Since model is still 2D only single (zero) elevation is used, and third dimension of FP is equal to 1.
• Typicaly FP is sampled once per angular degree, giving total number of azimuth samples #AS=
360. It is assumed that FP samples are taken from uniform grid. If optional ‘Azimuth’ argument is
provided it should have #AS elements.

Example 2:
Custom Geometry is defined with Position matrix.
Field pattern for all elements are defined in the ACS. When field patterns of array elements are measured
after array creation it is preferable to define same measurement grid in ACS for all elements.

Array=AntennaArray('Pos',Position,’FP-ACS’,FP,’Azimuth’,Az);

Comments:
• Number of rows in Position matrix will determine number of elements in array, and each row
contains x, y and z position (in [m]) of antenna element inside array.
• Since ‘Rot’ argument is not defined, there will be no rotation: ECS axis will be align with ACS
axis.

Document Number: IST-4-027756 Page 11 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

• FP argument to AntennaArray function has equivalent form like in previous example. Difference
is that in this case internal rotation of FRs from ECSs to ACS is not longer necessary.
• Az vector must contain equidistant azimuth angles from -180 to (but not including) 180 degrees.
Length of this vector must correspond to 4th dimension of FP.

4.2 Description of the network layout

WIM implementation supports a multi-base station and multi-mobile station network layout. The network
layout includes information about: the number and locations of MSs and BSs in the GCS; the number of
sectors in a BS (in case of a multi-cell network); the array broad side orientations at both MS and BS; the
coupling of an active radio link from a MS to a certain sector of a BS (or vice-versa); and the directions of
the MSs movement.
More detailed guidelines for system-level simulations could be found in [1]. That document describes
handover, multi-cell, multi-user, multi-hop and relaying simulations.

4.2.1 Construction of semi-random layout

Supported version of layoutparset.m is used to define type of the antenna array that is assigned to each
station, and therefore layout is not completely random. Desired number of K links is formed by random
BS-MS pairing (number of BSs and MSs is given implicitly during antenna selection). Function
layoutparset.m. generates random positions for all stations, and assigns random scenario and propagation
conditions to all links. MSs and BSs locations are randomly generated within e.g. the 500x500m2 cell area.
Default height of 32 m is used for BSs and 1.5 m for MSs.

layoutpar=layoutparset(MsAAIdx, BsAAIdxCell, K, Arrays)

Table 4.2. Input parameters for layoutparset.m function.

Parameter Default
Group Definition value Unit Note
Name
Arrays Vector of Array definitions (see subsection
- 5.2.2.2) containing all array types that should - -
be used in the simulation
MsAAIdx Vector of Arrays indices; single Array The total
index defines type of AA used by that MS. number of MS
- - - MsNum =
length(MsAA
Idx);
BsAAIdxCell Cell array of Arrays indices, where single cell The total
contains vector that describe multi-sector-BS. number of
Vector entries are Arrays indices, showing multi-sector-BS
wich of available AA is assigned to particular BsNum =
sectors. Multi-sector-BSs could have arbitrary length(BsAA
number of sectors. Indexing of sectors will be IdxCell);
taken according to the order of their occurrence The total
- in vector. - - number of
sectors (one-
sector-BS) is:
BsSectNum =
length([BsA
AIdxCell{:}
]);

K Number of links formed by random BS-MS

- - -
pairing

WIM model assumes that different sectors of multi-sector-BS are closely located and therefore links
between MS and different sectors exhibit full correlation: the same LoS/NLoS conditions and LSPs (AS,
DS) values. Links from mobile to different sectors are still not identical due to the specific array orientation
and directional filtering, and because they use different low-level parameters (MPC delays, angles etc.).
The second assumption is that sectors of the other BSs are located “very far away”, so that there is no
considerable correlation between links from single MS toward sectors belonging to different BSs.

Document Number: IST-4-027756 Page 12 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

4.2.1.1 Example
If e.g. MsAAIdx and BsAAIdxCell is defined like this:

>> MsAAIdx = [1 1 2 3];

>> BsAAIdxCell = {[1 3]; [2]; [1 1 2]};

It means that a total of 4 MS are present: The first two will use array type defined in Arrays(1) the third
MS will use Arrays(2) and the fourth Arrays(3). Created layout will have 3 multi-sector-BSs. The
first of them has two sectors, that are using Arrays(1) and Arrays(3), the second sector is one-sector-
BS with Arrays(2), while the third has three sectors: two of them are using Arrays(1) and one is using
Arrays(2). Please check the file ArraysExample.m for a more complex example.

4.2.2 Manual editing

It is also possible to define the network layout manually by directly editing all or part of the layout
parameters in the layoutpar structure. Starting from the previous semi-random layout wanted links (i.e.
(BS,Ms) pairs) could be defined by modifying LAYOUTPAR.Pairing. Position and orientation of each
station could be manually adjusted using LAYOUTPAR.Station.Pos/Rot parameters. Change of per-link
scenario and propagation conditions can be modified through
LAYOUTPAR.ScenarioVector/PropagConditionVector. For more details please refer to section 5.2.2
where these parameters are explained.

4.2.3 Layout setup visualization

A rudimental visualization of the network layout is implemented as helper utility in order to visualize the
layout of the generated channel; the network layout with 5 BSs and MSs, and 7 active links looks like as in
the following figures. In the figures red arrows denote the BS sector array orientation, black arrows denote
the MS direction of motion, and blue arrows denote the active (modelled) links.

Figure 4.2: Examples of the network layout: a) 5 BSs (three sectors in any BS), 5 MSs and 7 active
links; b) 5 BSs (number of sectors varies from 1-3 in different BSs), 5 MSs and 7 active links

The utility function NTlayout.m is used the following way:

>> NTlayout(layoutpar)

By setting the MSs and BSs coordinates and selecting the Pairing matrix properly it is possible to simulate
many system-level cases (see next paragraph).

4.3 Channel matrix generation

4.3.1 Simple initialization

%% Matrix generation for 10 MS-BS links (semi-random layout)
Arrays=arrayparset;

Document Number: IST-4-027756 Page 13 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

MsAAIdx = [1 1 2 3];
BsAAIdxCell = {[1 3]; [2]; [1 1 2]};
layoutpar=layoutparset(MsAAIdx, BsAAIdxCell,10,Arrays);

wimpar= wimparset;

H=wim(wimpar,layoutpar);

4.3.2 Initialization with arbitrary parameters

Instructions for the generation of arbitrary network layout and antenna array models are given in previous
sections. Control of the general model features (for all links), simulation parameters and output formats
could be adjusted in wimpar structure. In general, wimpar structure can be created from the scratch, or one
can initialize parameters to default values (by calling wimparset.m) and then modify selected parameters.

%% Matrix generation with modified parameter settings

% Setting and modifying wimpar and layoutpar input stuctures
Arrays=arrayparset;

MsAAIdx = [1 1 2 3];
BsAAIdxCell = {[1 3]; [2]; [1 1 2]};
layoutpar=layoutparset(MsAAIdx, BsAAIdxCell,10,Arrays);
layoutpar.PropagConditionVector=zeros(1,10); % (NLOS=0/LOS=1)
layoutpar.ScenarioVector=3*ones(1,10); % B1 scenario

wimpar=wimparset;
wimpar.NumTimeSamples=1000; % 100 time samples per link

% Generate channel realisations

[H1,delays,out]=wim(wimpar,layoutpar);

4.3.3 Initialization of the structural model parameters

This kind of initialization assumes complete knowledge of model structure. This option is provided to
enable consequtive calls of wim.m functions, without (default) random initialization of structural
parameters. This means that structural parameters obtained after one simulation run could be used to
initialize new run, preserving in that way previous channel conditions – what somehow means continuation
of the previous simulation run. This enables performing of seamless channel simulation is several
simuation runs. In example bellow it is assumed that input argument out is provided during some previous
call of wim.m, e.g. by example from the previous section.

% using final conditions as initial conditions in next function call

[H2,delays,out]=wim(wimpar,layoutpar,out);

Document Number: IST-4-027756 Page 14 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

5. Technical description of the SW

5.1 SW high level structure

The high-level WIM implementation structure is shown in the block diagram given in Figure 5.1. It is
assumed that the user mobility model, which is not specified in [1] is external to the channel matrix
generation routine. The path loss model is also implemented as a separate user-supplied function. The
default path loss function, complying with [1], is [pathloss.m]. Interpolation of antenna field patterns is
also required since AoD/AoAs can be any values over (-180,180) degrees. In WIM, field pattern
interpolation is based on EADF representation - function [ArrayResponse.m].

The WIM computation consists of two main parts:

• the random user parameter generation and
• the actual channel matrix computation.

Input and output arguments are defined in more detail in the next section.

generate location and

pre-processing phase:
array orientation helper functions for
generation of Array
parameters for all links setting model input
structure containing
(network layout parameters
3D-AA models
dependent)

external functions

ScenParTables.m wim.m

INPUT OUTPUT INPUT MATLAB structs:

wimpar - WIM configuration parameters (same for all links) [wimparset.m]
wimpar.Scenario fixpar layoutpar - location and orientation parameters for all simulated [layoutparset.m]
wimpar.Propagation struct stations, including antenna array parameters
initvalues - (opt.) initial path parameters for all links

[ScenParTables.m]
[generate_bulk_par.m]
Setting WIM
scenario specific [ArrayResponse.m]
parameters Generation of
random bulk Interpolation of
antenna field
parameters
Generation of patterns
for all links
path loss
ArrayResponse.m
parameters
INPUT OUTPUT
[pathloss.m]
AR
Generation of Array (complex polar. field
pathloss.m structure pattern interpolated
channel at azimuth angles)
INPUT OUTPUT coefficients
loss
for all links
wimpar wim_core.m
(Path losses [dB] for
linkpar INPUT OUTPUT
all the links) [wim_core.m]
WIM wimpar H
linkpar delta_t
[wim.m] bulkpar output_SubPathPhases
wim.m ... output_Phi_LOS
OUTPUT MATLAB structs:
H - channel matrix (cell array of size K (number of links);
each element of cell array contains UxSxNxT matrix)
fulloutput - (MATLAB struct) delays, AoD, AoA, phases,
path_loss, shadowing for all links
delays - multipath delays given in seconds

WIM implementation

data format conversion

model output
functions to link and etc
visualization
system simulators

external functions

Figure 5.2. High-level description of the WIM computation. The actual WIM model is in the box
labeled ‘WIM’.

Document Number: IST-4-027756 Page 15 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

5.2 Model input/output interface

The full syntax for the WIM function is ([ . ] indicate optional arguments):

[H, [DELAYS], [FULL_OUTPUT]] = WIM ( WIMPAR, LAYOUTPAR, [INITVALUES] ).

Table 5.1 Short overview of input and output aruments for WIM function

Helper Note
functions for
Argument name Type Description
structure
generation:
WIMPAR General simulation parameters. wimparset.m
The function linkparset should
not be used anymore. Instead
WIM should always be
Defines position of terminal stations, initialized with a layoutpar
their assigned antenna arrays and layoutparset.
LAYOUTPAR structure. This is necessary to
gives links of interest for simulation. m
keep consistency of data
defined in layoutpar structure.
input
Internally layout2link is used to
create missing link parameters.
Parameters of the propagation Optional
channel. When this parameter is
given WIM does not generate the
INITVALUES channel parameters randomly, but -
uses the supplied initial channel
values.
MIMO channel 5D-matrix is Calculated for a specified
collection of time-variant CIRs number of time samples
H -
(=f(t,τ)) between all (Tx,Rx) pairs, for
all links defined in LINKPAR.
Multipath delays for all links, Optional
DELAYS -
given in [s].
output Stores the randomly generated Optional
link parameters and the final
phases of the complex sinusoids.
This MATLAB struct can be used
FULL_OUTPUT as INITVALUES in subsequent -
function calls to generate time
continuous channel realizations
with separate function calls.
NOTE: All input and output arguments of WIM function are MATLAB structures/cells.

5.2.1 Global simulation parameters (WIMPAR)

Global simulation parameters could be classified into two major groups:
• MDL – parameters defining model structure (also includes system dependant parameters like
CenterFrequency). Change of parameters from this group causes different model behavior.
• SIM –simulation control parameters. These parameters control sampling in time and delay,
parameter initialization mode (random vs. manual), and format of output parameters (e.g.
inclusion/exclusion of path-loss in channel matrix H).

Table 5.2: MATLAB struct WIMPAR (set in wimparset.m). General channel model parameters,
common for all links.

Parameter Default
Group Definition value Unit Note
Name
The carrier center frequency. Center
CenterFrequency MDL frequency affects path loss and time sampling 5.25e9 Hz
interval.

Document Number: IST-4-027756 Page 16 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Parameter Default
Group Definition value Unit Note
Name
NumTimeSamples SIM Number of time samples 100 -
Oversampling factor, number of time samples
per half wavelength. For successful Doppler
analysis, one should select SampleDensity >
1. The time sample interval is calculated from
CenterFrequency and MsVelocity (see
SampleDensity SIM LAYOUTPARSET and LAYOUT2LINK) 2 -
according to
wavelength/(MsVelocity*SampleDensity).
The calculated time sample interval for each
link is included in the optional output
argument of WIM.
If UniformTimeSampling is 'yes' all links will ‘Yes’ setting could be
be sampled at simultaneous time instants. In useful in some
this case, the time sample interval is the same system-level
for all links it is calculated by replacing simulations where all
UniformTimeSamp SIM
MsVelocity with MAX(MsVelocity), where ‘no’ - simulated links need
ling
the maximum is over all links. If ‘no’ all the to be sampled at equal
links are time sampled with different rate time intervals,
depending on MsVelocity. regardless of MS
speeds.
If ‘yes’ the two strongest clusters in power
are divided in delay into three subclusters.
Fixed delays are [0 5 10] ns, fixed powers are
IntraClusterDsUsed MDL ‘yes’ -
[10 6 4]/20. For details see [1], section 4.2.
Number of delay tap grows by four. If ‘no’,
the clusters are not spread in delay.
Number of rays (i.e. complex sinusoids, plane This is a fixed value.
NumSubPathsPerPa waves) per cluster. It is not possible to change
MDL 20 -
th this value from 20 without modifying the
code.
If ‘yes’ the power and delay parameters are
not drawn randomly, but taken from the CDL
FixedPdpUsed MDL parameter tables [1], table 6-1..26. In the ‘no’ -
default mode ‘no’, the parameters are random
variables.
If ‘yes’ the angle parameters are not drawn
randomly, but taken from the CDL parameter
tables [1], table 6-1..26 and the random
FixedAnglesUsed MDL ‘no’ -
pairing of AoDs and AoAs is not used. In the
default mode ‘no’, the parameters are random
variables.
If TimeEvolution='yes', the transition Implem-entation is no
between adjacent channel segments is complete. This option
enabled. Transition from segment to segment is not supported.
is carried out by replacing clusters of the
"old" segment by the clusters of the "new"
segment, one by one. The route between
adjacent channel segments is divided to
number of sub-intervals equal to maximum
number of clusters within the channel
TimeEvolution MDL ‘no’ -
segments. During each sub-interval the power
of one old cluster ramps down and one new
cluster ramps up. Power ramps are linear.
Clusters from the old and new segments are
coupled based on their power. If number of
clusters is different in the channel segments
the weakest clusters are ramped up or down
without a pair from other cluster. See [1], Sec
3.4.
DelaySamplingInterval determines the
DelaySamplingInte sampling grid in delay domain. All path
SIM 5e-9 sec
rval delays are rounded to the nearest grid point. It
can also be set to zero.

Document Number: IST-4-027756 Page 17 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Parameter Default
Group Definition value Unit Note
Name
When PathLossModelUsed is 'no' the path
losses are still computed for each link but
PathLossModelUse they are not multiplied into the channel
SIM ‘no’ -
d matrices. Path loss is given only as an output
parameter. If ‘yes’, path loss is multiplied to
channel matrices.
If ‘yes’ the propagation condition (los/nlos)
setting is defined manually in
UseManualPropCondi LAYOUTPARSET in
SIM ‘yes’ -
tion PropagConditionVector. If ‘no’, the
propagation condition is drawn from LOS
probabilities in [1], table 4-7.
When ShadowingModelUsed is 'no' the
shadowing coefficients are still computed for
ShadowingModelU each link but they are not multiplied into the
SIM ‘no’ -
sed channel matrices. Shadowing is given only as
an output parameter. If ‘yes’, shadowing is
multiplied to channel matrices.
The path loss model function name. Path loss
model is implemented in a separate function,
whose name is defined in PathLossModel. ‘pathl
PathLossModel MDL -
For syntax, see PATHLOSS. The default oss’
function is PATHLOSS, which complies with
[1].
A wall material option for A1 NLOS path
loss calculation with default formulas from
[1]. Options are ['CR_light', ‘CR_heavy',
'RR_light', 'RR_heavy']. Here CR denotes
Corridor-Room condition and RR denotes 'CR_li
PathLossOption MDL -
Room-Room nlos condition. Light wall loss ght'
is 5dB and heavy wall loss is 12dB. Light
wall material is e.g. plaster board, heavy wall
material is e.g. brick or concrete. Note, in [1],
table 4-4 default A1 NLOS case is CR.
Sets random seed for Matlab random number see Matlab help
generators. The default value is empty. Even
RandomSeed SIM fixing the random seed may not result in fully [] -
repeatable simulations due to differences in
e.g. MATLAB versions.
Path loss parameter for scenario B5b. In B5b B5 specific
range MDL the path-loss ranges 1, 2 and 3 are defined, 1 -
see [1], table 6-20.
Observation end time for B5 scenarios time B5 specific
points are taken as:
end_time SIM 1 sec
wimpar.TimeVector=linspace(0,wimpar.end_
time,T);

5.2.2 Parameters defining network layout (LAYOUTPAR)

A network layout is defined with positions of terminal stations, where each station has assigned antenna
array. The coordinate system where placement of stations takes place is called Global-Coordinate-System
(GCS). Given 3D-AA representation allows rotation of the antenna array (or equivalently station) around
any axis of the GCS.
The monitored links are given as a collection of terminal station pairs. Propagation condition and radio
environment are defined on link-level.
It is therefore possible to identify several groups of parameters:
• ENV – parameters for selection of propagation conditions and radio environment description,
• POS – parameters defining positions of terminals (also includes mobility model – in this case
“MsVelocity” and “MsDirection”).
• ANT – antenna array parameters.
Table 5.3. MATLAB struct LAYOUTPAR (set in layoutparset.m): network layout parameters.

Parameter Group Definition Default Unit Note

Document Number: IST-4-027756 Page 18 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Name value
POS MATLAB
Parameters of the stations being included see Table structure
Stations + -
into system simulations. 5.5
ANT
To define
pairing Stations
A 2xK matrix whose k-th colum contains see are ordered: first
Pairing - layoutparse - all BS Sectors,
indices of Stations constituting k-th link.
t.m than MS (1 AA
is assumed per
MS)
A parameter defining the number of see
NofSect - - layoutparset.m
sectors in each of the BSs.
A 1xK vector mapping scenarios to links.
Scenarios are [1=A1, 2=A2, 3=B1, 4=B2, {1,2,
ScenarioVector ENV ones(1,K)
5=B3, 6=B4, 7=B5a, 8=B5c, 9=B5f, 10=C1, …,15}
11=C2, 12=C3, 13=C4, 14=D1, 15=D2a].
A 1xK vector mapping propagation condition Possible values
(NLOS/LOS) to links. If WIMPAR 0=NLOS and
PropagConditionVe ENV UseManualPropCondition = ‘yes’, link zeros(1,K) {0,1} 1=LOS.
ctor propagation conditions (NLOS=0/LOS=1)
are defined by this vector.
A parameter for B1 and B2 path loss model.
StreetWidth ENV 20 m
Average width of the streets, same for all users.
A parameter for A2/B4 path loss model.
NumFloor is the floor number in which the
NumFloors ENV indoor MS/BS is located. E.g. in A2 scenario 1 -
NumFloors is 5 if BS is located on the 5th floor.
On ground floor (=street level) NumFloor = 0.
A parameter for A1 NLOS path loss model [1],
NumPenetratedFloo ENV table 4-4. Number of penetrated floors between 0 -
rs BS and MS.
Distance definition for B1 and B2 path loss
model. Dist1 is a distance from BS to the “last
line-of-sight point”, typically street crossing,
Dist1 POS NaN -
see [1], fig 4-3. Default value is NaN, which
denotes random distance determination in
PATHLOSS function.

5.2.2.1 Link-level parameters (LINKPAR)

Based on parameters in LAYOUTPAR structure necessary link parameters are calculated. They should not
be modified directly in system level simulations to prevent inconsistencies with chosen network layout.
Please note that some of the link parameters are kept only to minimize changes in respect to previous
versions of the source code.

Both the distance and line of sight (LOS) direction information of the radio links are calculated w.r.t. XY
plane of GCS, i.e. without consideration of z dimension. The distance between the BSi and MSk
(MsBsDistance) is
d BS i , MS k = ( xBS i − xMS k ) 2 + ( yBS i − yMS k ) 2
.
In order to express link orientation and array broadside direction, Y axes of GCS, being aligned with North
direction, is used as a reference the zero angle (Figure 5.3). The positive direction of the angles is the
clockwise direction.

Document Number: IST-4-027756 Page 19 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

N (Y)
Y

BSi , MSk

Z GCS
X N (Y)
BSi array BS i
MS array

BS array broadside
MSk , BSi
v
MS k

MS array broadside

Figure 5.3: Link and BS and MS antenna array orientations.

The LOS direction from BSi to MSk with respect to BS antenna array broad side is
 yMS k − y BS i 
θ BS , MS = − arctan  + 90o
i k  xMS − xBS 
 k i 

Table 5.4: Internally used MATLAB struct LINKPAR (calculated by layout2link.m).

Parameter
Group Definition Unit Note
Name
calculated
pairwise from
Stations (as
A 1xK vector defining distance between BS and MS for K
MsBsDistance POS m defined in
links.
Pairing)
StationDistXY.
m
BsHeight POS A 1xK vector defining BS height from ground level. m Stations.Pos(3)
MsHeight POS A 1xK vector defining MS height from ground level. m Stations.Pos(3)
ThetaBs POS θBS (see Figure 5.3) deg calculated
pairwise from
Stations (as
defined in
ThetaMs POS θMS (see Figure 5.3) deg Pairing)
StationDirection
.m
MsVelocity POS MS velocity m/s calculated from
Stations.Velocit
MsDirection POS θϖ (see Figure 5.3) deg
y

5.2.2.2 Stations (Array) structure

Properties of the “Stations” are defined by its position and antenna array characteristics. The antenna array
description is created by making a copy of the Array structure2. The desired collection of Array structures,
intended for use in simulation, should be created only once - in preprocessing phase, independently from
channel simulations (section 4.1). The Array structure contains definitions of array geometry and radiation
patterns. Remaining parameters for position, orientation and velocity, should be adjusted according to
targeted network layout.
Table 5.5. Stations (Array) parameters.

Parameter Default
Group Definition value Unit Note
Name

2
Array structure is not provided as input argument to WIM function since necessary Array definitions are embedded
into Station definition, being part of layoutpar.

Document Number: IST-4-027756 Page 20 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Pos POS [x;y;z] position in GCS m

Rot [Rot ;Rot ;Rot ] rotation of array by respective
POS x y z rad
axes of GCS
Element ANT vector of array elements
CommonAperture EADF common to all elements (if empty each Temporary
element has to provide its own pattern). This parameter used
parameter is used only during Array only in
ANT construction when provided field patterns are preprocessing
defined in ECS. In this case rotation of ECS 3D phase
field patterns is performed to calculate their
EADF representation in ACS.
Aperture if preprocessing is used, this field contains structure
ANT
EADF for the whole array
Velocity POS [v ;v ;v ] velocity of this station
x y z

Since velocity vector is assigned to this structure, in the new WIM implementation MS and one-sector-
BS can be treated in the same way. The fundamental difference between MS/BS is that BS has zero
velocity. In the new model this is achieved by setting Station.Velocity to zero.

5.2.2.3 Antenna element parameters

Each antenna element in characterized with its position in resprect to ACS:
Table 5.6. Element parameters.

Parameter Default
Group Definition value Unit Note
Name
Pos POS [x;y;z] position in ACS m
Rot [Rot ;Rot ;Rot ] rotation of element by
POS x y z
respective axes of ACS
Aperture EADF for this element. This parameter is used Temporary
only during Array construction when provided parameter used
ANT field patterns are defined in ECS. In this case only in
rotation of ECS 3D field patterns is performed preprocessing
to calculate their EADF representation in ACS. phase

The geometry of the array is defined in meters (as well as station positions in GCS), not in normalized
distance (number of wavelengths) as in the previous model versions.

Note that the mean power of narrowband channel matrix elements (i.e. summed over delay domain)
depends on the antenna gains
5.2.3 External initialization of structural model parameters (INITVALUES) - optional
The fourth input argument, which is also a MATLAB struct, is optional. It can be used to specify the initial
AoDs, AoAs, cisoid phases, path losses and shadowing values when WIM is called recursively, or for
testing purposes. If this argument is given, the random parameter generation as defined in [1] is not needed.
Only the antenna gain values will be interpolated for the supplied AoAs and AoDs.
The fields of the MATLAB struct are given in the following table. Note! The fourth input argument can be
directly the output structure FULLOUTPUT defined in Table 5.8 or the structure defined in Table 5.7.

Table 5.7. MATLAB struct INITVALUES: initial values, fourth optional input argument.

Parameter name Definition Unit

InitDelays A K x N matrix of path delays. Sec
InitSubPathPowers A K x N x M array of powers of the subpaths. -
InitAods A K x N x M array Degrees
InitAoas A K x N x M array Degrees
InitSubPathPhases A complex-valued K x N x M array. When polarization option Degrees

Document Number: IST-4-027756 Page 21 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

is used, this is a K x P x N x M array, where P=4. In this case

the second dimension includes the phases for [VV VH HV
HH] polarized components.
InitPathLosses A K x 1 vector Decibel
InitShadowLosses A K x 1 vector Decibel
A K x 2 matrix. Index to two strongest clusters. These clusters
InitIndOfSpreadClust are spread to three delay positions if parameter
IntraClusterDsUsed = ‘yes’

5.2.4 Output parameters

There are three output arguments: H, DELAYS, FULLOUTPUT. The last two are optional.

Table 5.8. MATLAB output parameters.

Parameter name Definition Unit Note

cell array of size K (number of links). Each element of See Appendix 1 for
H
this cell array contains a U x S x N x T matrix. terminology
A K x N vector of path delay values. Note that delays
DELAYS are, for compatibility with the INITVALUES, also sec
included in FULLOUTPUT.
FULLOUTPUT A MATLAB struct with the following elements:
A K x N matrix of path delays. This is identical to the
delays sec
second output argument.
path_powers A K x N array of path powers. linear
aods A K x N x M array of subpath angles of departure degrees
aoas A K x N x M array of subpath angles of arrival degrees
path_losses A K x 1 vector linear scale
MsBsDistance 1 x K vector of MS-BS distances m
shadow_fading A K x 1 vector linear scale
A K x 4 vector of per link large scale parameters (ASD
sigmas ASA DS SF)
-
A K x 1 vector indicating LOS / NLOS condition
propag_condition (0=NLOS, 1=LOS)
-
Kcluster A K x 1 vector defining narrowband K-factors of links. linear scale Only with LOS
Phi_LOS Final phases for LOS paths, K x 1 array. deg Only with LOS
scatterer_freq A K x N x M array of subpath Doppler frequecies Hz Only with B5 CDL
A complex-valued K x N x M array giving the final
phases of all subpaths. When polarization option is
subpath_phases used, a K x P x N x M array, where P=4. In this case degrees
the second dimension includes the phases for [VV VH
HV HH] polarized components.
A K x 1 vector defining time sampling interval for all
delta_t sec
links.
A K x 2 matrix. Index to two strongest clusters. These
IndexOfDividedClust clusters are spread to three delay positions if parameter -
IntraClusterDsUsed = ‘yes’
Only with
A K x N x M array of cross-polarization coupling
xpr linear scale PolarisedArrays
power ratios.
case.

5.3 Implementation notes

5.3.1 Speed optimization

A performance improvement may be achieved by setting wimpar. LookUpTable=-1. This activates the
lookup table for computing the complex exponential, in the core equation of the channel model.
Alternatively, one can set the number of points in the look-up table by e.g. setting wimpar.
LookUpTable=1024. The default lookup table size (with LookUpTable = -1) is 2^14=16384.

Document Number: IST-4-027756 Page 22 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

5.3.1.1 Compatibility issues

In the previous implementation of WIM implementation ANSI-C support was provided for some critical
functions: scm_mex_core.dll and interp_gain_c.dll. The new WIM version uses EADF-based antenna
pattern interpolation, and modified version3 of scm_mex_core.m, and therefore *.dll files from previous
implementations can not be used for speed optimization.

5.3.2 Output interpolation

Channel sampling frequency has to be finally equal to the simulation system sampling frequency. To have
feasible computing time it is not reasonable to generate channel realisations on the sampling frequency of
the system to be simulated. The channel realisations can be generated on some lower sampling frequency
and then interpolated to the desired frequency. A practical solution is e.g. to generate channel samples with
sample density (i.e. over-sampling factor) two, interpolate them accurately to sample density 64 and to
apply zero-order-hold interpolation to the system sampling frequency. Channel impulse responses can be
generated during the simulation or stored on a file before the simulation on low sample density.
Interpolation can be done during the system simulation.

6. References
[1] IST-WINNER II, D1.1.2 “WINNER II Channel Models”, ver 1.0, Sep 2007, https://www.ist-
winner.org/WINNER2-Deliverables/.
[2] IST-WINNER II, D1.1.1 “WINNER II Interim Channel Models”, ver 1.2, Feb 2007,
https://www.ist-winner.org/WINNER2-Deliverables/.
[3] D5.4, “Final Report on Link Level and System Level Channel Models”, ver 1.4 , November
2005.
[4] D5.3, “Interim Channel Models”, April 2005.
[5] “Spatial channel model for Multiple Input Multiple Output (MIMO) simulations” 3GPP TR
25.996 V6.1.0
[6] M. Narandžić, M. Käske, C. Schneider, M. Milojević, M. Landmann, G. Sommerkorn, and
R.S. Thomä, „3D-Antenna Array Model for IST-WINNER Channel Simulations,” Proc. of
IEEE VTC2007-Spring, Dublin, Ireland, April 23 - 25, 2007.

3
scm_mex_core.dll created for previous version does not support intra-cluster delay spread option.

Document Number: IST-4-027756 Page 23 (24)

File: WIM2 Matlab documentation V1.0.doc 22.09.08

Appendix A: Terminology and notation

‘Link’ = one-directional (downlink or uplink) BS-MS connection. The term ‘link’ is sometimes, but not
always, interchangeable with ‘user’, i.e. MS.

‘Channel realisation’ = a sequence of channel matrices over a pre-defined number of time samples (in
WIM, each element of the channel matrix has six channel clusters in delay domain).

Cluster ≡ tap ≡ path

Ray ≡ subpath
K number of links
N number of paths
M number of subpaths within a path
U number of receiver elements
S number of transmitter elements
T number of time samples

Document Number: IST-4-027756 Page 24 (24)