1 SCM Text Description SCM-083

3
4

5 Title: Spatial Channel Model Text Description

6

7 File: 431714279.doc
8

9 Source: Spatial Channel Model AHG (Combined ad-hoc from 3GPP & 3GPP2)
10
11

12 Date: December 19, 2002

13

14

15

16

17

18

Notice

©2002 Third Generation Partnership Project Two (3GPP2). All rights reserved. Permission is granted
for copying, reproducing, or duplicating this document only for the legitimate purposes of 3GPP2 and
its organizational partners. No other copying, reproduction, or distribution is permitted.

2
1SCM Text Description SCM-083

11 Introduction................................................................................................................ 4

2 1.1 Terminology (usage in 3GPP & 3GPP2)......................................................................4

3 1.2 Scope....................................................................................................................... 4

4 1.2.1 System level evaluation.................................................................................4

5 1.2.2 Link level evaluation......................................................................................5

62 Link Level Channel Models for Calibration Purposes....................................................5

7 2.1 Purpose.................................................................................................................... 5

8 2.2 Link Level Channel Model Parameter Summary........................................................5

9 2.3 Spatial Parameters per Path.....................................................................................7

10 2.4 BS and MS Array Topologies.....................................................................................7

11 2.5 Spatial Parameters for the BS...................................................................................7

12 2.5.1 BS Antenna Pattern......................................................................................7

13 2.5.2 Per Path BS Angle Spread (AS)....................................................................10

14 2.5.3 Per Path BS Angle of Arrival........................................................................10

15 2.5.4 Per Path BS Power Azimuth Spectrum.........................................................10

16 2.6 Spatial Parameters for the MS................................................................................11

17 2.6.1 MS Antenna Pattern....................................................................................11

18 2.6.2 Per Path MS Angle Spread (AS)....................................................................11

19 2.6.3 Per Path MS Angle of Arrival.......................................................................11

20 2.6.4 Per Path MS Power Azimuth Spectrum........................................................12

21 2.6.5 MS Direction of Travel.................................................................................12

22 2.6.6 Per Path Doppler Spectrum.........................................................................12

23 2.7 Generation of Channel Model..................................................................................12

24 2.8 Calibration and Reference Values...........................................................................13

3
25 System Level Spatial Channel Model..........................................................................13

26 3.1 System Wide Definitions and Parameters................................................................14

27 3.2 Scenarios............................................................................................................... 15

28 3.3 Channel Generation Steps......................................................................................16

29 3.4 BS and MS Array Topologies...................................................................................18

30 3.4.1 Polarized arrays........................................................................................... 18

31 3.4.2 Reference Antenna Configurations...............................................................19

32 3.4.3 General Modeling Assumptions...................................................................20

22
1SCM Text Description SCM-083

1 3.4.4 Micro-cell Assumptions...............................................................................20

2 3.5 Options for Urban Macro and Micro Environments.................................................20

3 3.5.1 Far scatterer clusters..................................................................................20

4 3.5.2 Line of sight................................................................................................ 21

5 3.5.3 Urban canyon............................................................................................. 22

6 3.6 Correlation Between Channel Parameters...............................................................23

7 3.6.1 Correlation between Delay Spread, Angle Spread, and Log Normal
8 Shadowing........................................................................................................... 23

9 3.7 System Level Calibration........................................................................................25

4
10 Appendix A: MMSE receiver description.....................................................................29

5
11 Reference.................................................................................................................. 29
12

23
1SCM Text Description SCM-083

21 INTRODUCTION

3This document details the current discussion of the combined 3GPP-3GPP2 Spatial Channel
4Ad-hoc group. A similar document1, developed independently in the 3GPP2 Spatial Channel
5Modeling Ad-hoc group, was used for reference.

61.1 Terminology (usage in 3GPP & 3GPP2)

7In this document the following are terms that are commonly used interchangeably and are
8equivalent. To promote consistency, the term on the left will be preferred in this document
9unless otherwise stated.

MS = Mobile Station = UE = User Equipment = Terminal = Subscriber Unit

10

BS = Base Station = Node-B = BTS

11

AS = Angle Spread = Azimuth Spread =

12 PAS
Path = Ray
13

Path Component = Sub-ray

14

PAS = Power Azimuth Spectrum

15

DoT = Direction of Travel

16

AoA = Angle of Arrival

17

PDP = Power Delay Profile

18

19

1.2
20 Scope

21The scope of the 3GPP-3GPP2 SCM AHG is to develop and specify parameters and methods
22associated with the spatial channel modeling that are common to the needs of the 3GPP and

233GPP2 organizations (harmonization). The scope includes development of specifications for:

1.2.1
24 System level evaluation.

Within this category, a list of four focus areas are identified, however the emphasis of the SCM
25

AHG work is on items a & b.

26

27 a. Physical parameters (e.g. power delay profiles, angle spreads, dependencies between
28 parameters)

29 b. System evaluation methodology.

30 c. Antenna arrangements, reference cases and definition of minimum requirements.

31 d. Some framework (air interface) dependent parameters.

24
1SCM Text Description SCM-083

11.2.2 Link level evaluation.

2The link level models are defined only for calibration purposes. It is a common view within the
3group that the link level simulation assumptions will not be used for evaluation and

4comparison of proposals.

62 LINK LEVEL CHANNEL MODELS FOR CALIBRATION PURPOSES

7(Editor’s Note: A harmonized link level model is being proposed, and further discussion is
8expected prior to agreement. This includes all aspects of the link level model described in
9Section 2.)

2.1
10 Purpose

11Link level simulations alone will not be used for algorithm comparison because they reflect only
12one snapshot of the channel behavior. Furthermore, they do not account for system attributes

13such as scheduling and HARQ. For these reasons, link level simulations do not allow any

14conclusions about the typical behavior of the system. Only system level simulations can

15achieve that. Therefore we require system level simulations for the final algorithm comparison.

16Link level simulations will not be used to compare performance of different algorithms. Rather,
17they will be used only for calibration, which is the comparison of performance results from

18different implementations of a given algorithm.

2.2
19 Link Level Channel Model Parameter Summary

(Editor’s Note: The following table is under discussion as a harmonized set of link level
20

parameters and values.)

21

22

25
SCM Text Description
1 SCM-083

1 Table Link Level Channel Models for Calibration Purposes-1. Summary of Suggested SCM Link Level
2 Parameters for Calibration Purposes

6
2
SCM Text Description
1 SCM-083

Model Case I Case II Case III Case IV

Corresponding Case B Case C Case D Case A

3GPP Designator*

Corresponding Model A, D, E Model C Model B Model F

3GPP2
Designator*

PDP Modified Pedestrian A Vehicular A Pedestrian B Single

Path

# of Paths 1) 4+1 (LOS on, K = 6dB) 6 6 1

2) 4 (LOS off)
Relative Path Power (dB)

Delay (ns)

1) 0.0
0 0,0 0 0.0 0 0 0
2) -Inf

1) -6.51
0 -1.0 310 -0.9 200
2) 0.0

1) -16.21
110 -9.0 710 -4.9 800
2) -9.7

1) -25.71
190 -10.0 1090 -8.0 1200
2) –19.2

1) -29.31
410 -15.0 1730 -7.8 2300
2) -22.8

-20.0 2510 -23.9 3700

Speed (km/h) 1) 3 3, 30, 120 3, 30, 120 3

2) 30, 120
UE/Mobile Station

Topology Reference 0.5λ Reference 0.5λ Reference 0.5λ N/A

PAS 1) LOS on: Fixed AoA for RMS angle RMS angle spread N/A
LOS component, spread of 35 of 35 degrees per
remaining power has degrees per path path with a
360 degree uniform PAS. with a Lapacian Lapacian

2) LOS off: PAS with a distribution distribution

Lapacian distribution, Or 360 degree

RMS angle spread of 35 uniform PAS.
degrees per path

DoT 0 22.5 -22.5 N/A

(degrees)

AoA 22.5 (LOS component) 67.5 (all paths) 22.5 (odd N/A
(degrees) 67.5 (all other paths) numbered paths),
-67.5 (even
numbered paths)

7
2
1SCM Text Description SCM-083

Node B/ Base Station Model Case I Case II Case III Case IV

Topology Reference: ULA with N/A

0.5λ-spacing or 4λ-spacing or 10λ-spacing

PAS Lapacian distribution with RMS angle spread of N/A

2 degrees or 5 degrees,

per path depending on AoA/AoD

AoD/AoA 50 for 2 RMS angle spread per path N/A

 
(degrees) 20 for 5 RMS angle spread per path

1*Designators correspond to channel models previously proposed in 3GPP and 3GPP2 ad-hoc
2groups.

32.3 Spatial Parameters per Path

4Each resolvable path is characterized by its own spatial channel parameters (angle spread,
5angle of arrival, power azimuth spectrum). All paths are assumed independent. These
6assumptions apply to both the BS and the MS specific spatial parameters. The above
7assumptions are in effect only for the Link Level channel model.

82.4 BS and MS Array Topologies

9 The spatial channel model should allow any type of antenna configuration to be selected,
10although details of a given configuration must be shared to allow others to reproduce the model
11and verify the results.

12Calibrating simulators at the link level requires a common set of assumptions including a
13specific set of antenna topologies to define a baseline case. At the MS, the reference element
14spacing is 0.5. At the BS, three values for reference element spacing are defined: 0.5, 4, and

1510.

16

2.5
17 Spatial Parameters for the BS

2.5.1
18 BS Antenna Pattern

19The 3-sector antenna pattern used for each sector, Reverse Link and Forward Link, is plotted
20in Figure Link Level Channel Models for Calibration Purposes-1 and is specified by

� �q � 2
�
21 A ( q ) = - min �
12 � �, Am � where - 180 �q �180 (1)
� q �
� �3dB � �
22 q is defined as the angle between the direction of interest and the boresight of the antenna,
23 q 3 dB is the 3dB beamwidth in degrees, and A m is the maximum attenuation. For a 3 sector

24scenario q 3 dB is 70 , Am = 20dB ,and the antenna boresight pointing direction is given by

o

25Figure Link Level Channel Models for Calibration Purposes-2. For a 6 sector scenario q 3 dB is

2635 , Am = 23dB , which results in the pattern shown in Figure Link Level Channel Models for
o

28
1SCM Text Description SCM-083

1Calibration Purposes-3, and the boresight pointing direction defined by Figure Link Level
2Channel Models for Calibration Purposes-4. The boresight is defined to be the direction to
3which the antenna shows the maximum gain. The gain specified by previous 3GPP
2 o o
4documents[ ] for the 3-sector 70 antenna is 14dBi. By reducing the beamwidth by half to 35 ,

5the corresponding gain will be 3dB higher resulting in 17dBi. The antenna pattern shown is

6targeted for diversity oriented implementations (i.e. large inter-element spacings). For

7beamforming applications that require small spacings, alternative antenna designs may have to

8be considered leading to a different antenna pattern.

3 Sector Antenna Pattern

-5
Gain in dB.

-10

-15

-20

-25
-120 -100 -80 -60 -40 -20 0 20 40 60 80 100 120
Azim uth in Degrees

10 Figure Link Level Channel Models for Calibration Purposes-1 Antenna Pattern for 3-Sector Cells

3-Sector Scenario

BS

Antenna Boresight in
direction of arrow
11

12 Figure Link Level Channel Models for Calibration Purposes-2 Boresight pointing direction for 3-sector cells

13

29
1SCM Text Description SCM-083

6 Sector Antenna Pattern

-5

Gain in dB
-10

-15

-20

-25
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
Azim uth in Degrees

2 Figure Link Level Channel Models for Calibration Purposes-3 Antenna Pattern for 6-Sector Cells

6-Sector Boundaries

BS

Antenna Boresight in
direction of arrow
4

5 Figure Link Level Channel Models for Calibration Purposes-4 Boresight Pointing direction for 6-Sector Cells

72.5.2 Per Path BS Angle Spread (AS)

8 The base station per path angle spread is defined as the root mean square (RMS) of angles with
9which an arriving path’s power is received by the base station array. The individual path

10powers are defined in the temporal channel model described in Table Link Level Channel

11Models for Calibration Purposes-1. Two values of BS angle spread (each associated with a

12corresponding mean angle of arrival, AoA) are considered:

13 - AS: 2 degrees at AoA 50 degrees

14 - AS: 5 degrees at AoA 20 degrees

210
1SCM Text Description SCM-083

1It should be noted that attention should be paid when comparing the link level performance
2between the two angle spread values since the BS antenna gain for the two corresponding AoAs

3will be different. The BS antenna gain is applied to the path powers specified in Table Link

4Level Channel Models for Calibration Purposes-1.

52.5.3 Per Path BS Angle of Arrival

6The Angle of Arrival (AoA) or Angle of Departure (AoD) is defined to be the mean angle with
7which an arriving or departing path’s power is received or transmitted by the BS array with

8respect to the boresite. The two values considered are:

9 - AoA: 50 degrees (associated with the RMS Angle Spread of 2 degrees)

10 - AoA: 20 degrees (associated with the RMS Angle Spread of 5 degrees)

2.5.4
11 Per Path BS Power Azimuth Spectrum
12The Power Azimuth Spectrum (PAS) of a path arriving at the base station is assumed to have a
13Laplacian distribution. For an incoming AOA q and RMS angle-spread , the BS per path PAS
14value at an angle q is given by expression below:

15
- 2 q -q 
16 P (q ,  ,q ) = N o exp   G (q )
  
 

17where both angles q and q are given with respect to the boresight of the antenna elements. It
18is assumed that all antenna elements’ orientations are aligned. Also, P is the average received

19power and G is the numeric base station antenna gain described in Section 2.5.1 by

20 G (q ) = 10 0.1 A(q )

Finally, No is the normalization constant:

21

 -q - 2 (q - q ) 
22 N o -1 = 
- -q
exp 
 
  G (q ) dq


In the above equation, q represents path components (sub-rays) of the path power arriving at
23

an incoming AoA q . The distribution of these path components is TBD.

24

2.6
25 Spatial Parameters for the MS

2.6.1
26 MS Antenna Pattern

27For each and every antenna element at the MS, the antenna pattern will be assumed omni
28directional with an antenna gain of -1 dBi.

2.6.2
29 Per Path MS Angle Spread (AS)

30The MS per path AS is defined as the root mean square (RMS) of angles of an incident path’s
31power at the MS array. Two values of the path’s angle spread are considered:

32 - AS: 104 degrees (results from a uniform over 360 degree PAS),

33 - AS: 35 degrees for a Laplacian PAS with a certain path specific Angle of Arrival (AoA).

211
1SCM Text Description SCM-083

12.6.3 Per Path MS Angle of Arrival

2The per path Angle of Arrival (AOA) is defined as the mean of angles of an incident path’s power
3at the UE/Mobile Station array with respect to the broadside as shown Figure Link Level

4Channel Models for Calibration Purposes-5.

AOA = 0

AOA < 0 AOA > 0

6 Figure Link Level Channel Models for Calibration Purposes-5. Angle of Arrival orientation at the MS.

8Three different per path AoA values at the MS are suggested for the cases of a non-uniform
9PAS, see Table Link Level Channel Models for Calibration Purposes-1 for details:

10 - AoA: -67.5 degrees (associated with an RMS Angle Spread of 35 degrees)

11 - AoA: +67.5 degrees (associated with an RMS Angle Spread of 35 degrees)

12 - AoA: +22.5 degrees (associated with an RMS Angle Spread of 35 degrees or

13 with an LOS component)

2.6.4
14 Per Path MS Power Azimuth Spectrum

The Laplacian distribution and the Uniform distribution are used to model the per path Power
15

Azimuth Spectrum (PAS) at the MS.

16

17The Power Azimuth Spectrum (PAS) of a path arriving at the MS is modeled as either a
18Laplacian distribution or a uniform over 360 degree distribution. Since an omni directional MS

19antenna gain is assumed, the received per path PAS will remain either Laplacian or uniform.

20For an incoming AOA q and RMS angle-spread , the MS per path Laplacian PAS value at an
21angle q is given by expression below:
- 2 q -q 
22 P (q , ,q ) = N o exp   ,
  
 

23where both angles q and q are given with respect to the boresight of the antenna elements. It
24is assumed that all antenna elements’ orientations are aligned. Also, P is the average received

25power and No is the normalization constant:


=  P (q ,  ,q )dq .
-1
26 No
-

In the above equation, q represents path components (sub-rays) of the path power arriving at
27

an incoming AoA q . The distribution of these path components is TBD.

28

212
1SCM Text Description SCM-083

12.6.5 MS Direction of Travel

2The mobile station direction of travel is defined with respect to the broadside of the mobile
3antenna array as shown in Figure Link Level Channel Models for Calibration Purposes-5.

42.6.6 Per Path Doppler Spectrum

5The per path Doppler Spectrum is defined as a function of the direction of travel and the per
6path PAS and AoA at the MS. This should correspond to the per path fading behavior for either

7the correlation-based or ray-based method.

82.7 Generation of Channel Model

9The proponent can determine the model implementation. Examples of implementations include
correlation or ray-based techniques.
10

Further details on examples of the model implementation are TBD.

11

Outline of methodology, including doppler spectrum filter required for correlation method.
12

2.8
13 Calibration and Reference Values

14For the purpose of link level simulations, reference values of the average correlation are given
15below in Table Link Level Channel Models for Calibration Purposes-2. The reference values are
16provided for the calibration of the simulation software and to assist in the resolution of possible

17errors in the simulation methods implemented. Specifically, the average complex correlation

18and magnitude of the complex correlation is reported between BS antennas and between MS

19antennas. The spatial parameter values used are those defined already throughout Section 2.

20Nokia to provide formulas for computing correlation matrices

21

22 Table Link Level Channel Models for Calibration Purposes-2. Reference Correlation Values.

Antenna AS (degrees) AOA (degrees) Correlation Complex

Spacing (magnitude) Correlation

BS 0.5  5 20 0.9688 0.4743+0.8448i

0.5  2 50 0.9975 -0.7367+0.6725i
4 5 20 0.3224 -0.2144+0.2408i
4 2 50 0.8624 0.8025+0.3158i
10  5 20 0.0704 -0.0617+i0.034
10  2 50 0.5018 -0.2762-i0.4190

MS  /2 104 0 0.3042 -0.3042

 /2 35 -67.5 0.7744 -0.6948-i0.342

 /2 35 22.5 0.4399 0.0861+0.431i

 /2 35 67.5 0.7744 -0.6948+i0.342

213
1SCM Text Description SCM-083

23 SYSTEM LEVEL SPATIAL CHANNEL MODEL

3A Spatial Channel Model for use at the system level is described below. Parameter values and
4their ranges are specified to enable the generation of a Spatio-Temporal channel model for use
5in evaluating the system-level performance of multiple antenna schemes. The goal of this

6section is to define the methodology for generating the spatial channel coefficients between a

7given base and mobile for use in system level simulations. For a given base and mobile pair,

8this procedure consists of three basic steps

9 1. Specify an environment, either suburban macro, urban macro, or urban micro.

10 2. Obtain the parameters associated with that environment.

11 3. Generate the channel coefficients based on the parameters.

12The remainder of this section gives the details for these steps. Figure X below provides a
13roadmap for generating the channel coefficients. (This diagram should be greatly expanded and

14should show which section numbers each of the items is discussed.)

15

Suburban Macro Urban Macro Urban Micro

Options: Options: Options:

Line of sight (TBD) Urban canyon Urban canyon
Far scattering cluster Far scattering cluster
Line of sight Line of sight

Pathloss model: (Hata) Pathloss model:

(Feurestein)

Generate coefficients
16

17

3.1
18 System Wide Definitions and Parameters

19The received signal at the MS consists of N time-delayed multipath replicas of the transmitted
20signal. These N paths are defined by powers and delays and are chosen randomly according to

21the channel generation procedure. Each path consists of M subpaths which will be described

22later.

23
24Figure System Level Spatial Channel Model-6 shows the angular parameters used in the
25proposed model. The following definitions are used:
26  BS Base Station orientation. It is defined as the difference between the broadside of the BS

27 array and a North (N) reference direction.

214
1SCM Text Description SCM-083

1 q0 Angle of the LOS direction between the BS and MS, referenced to the broadside of the
2 BS array.
3 i mean angle of arrival difference for the nth (n = 1 … N) path relative to the LOS AOA q 0 .
4 q l ,i Absolute AOA for the mth (m = 1 … M) subpath of the nth path at the BS with respect to

5 the BS broadside.
6q 0 , MS Angle between the BS-MS LOS and the MS broadside.
7  MS MS orientation. It is defined as the difference between the broadside of the MS array
8 and the North reference direction.
9 q v , MS The angle reference of the MS velocity vector.
10
The angles shown in Figure 1 that are measured in a clockwise direction are assumed to be
11
negative in value.
12

Cluster i N
Single Path MS
Case qvMS
i MS
q o MS
N

i
 BS
q l ,i (singlewave)
qi
qo
BS
BS Array broadside

14 Figure System Level Spatial Channel Model-6 BS and MS angle parameters

15

16For system level simulation purposes, the fast fading per-path will be evolved in time, although
17bulk parameters including Angle Spread, Delay Spread, Log Normal Shadowing, and MS
18location will remain fixed during the evaluation of the given MS.

3.2
19 Scenarios

20To limit the number of cases for consideration, the following represent a definition of the
21unique environments for simulation.

22 a. Suburban Macro (approximately 3Km distance BS to BS)

23 b. Urban Macro (approximately 3Km distance BS to BS)

24 c. Urban Micro (less than 1Km distance BS to BS)

25The macro cell definition applies when the base station antennas are above rooftop height. The
26micro cell definition applies otherwise. Table System Level Spatial Channel Model-3 Scenario
27parameters describes the parameters used in each of the scenarios.

215
 1SCM Text Description SCM-083

2 Table System Level Spatial Channel Model-3 Scenario parameters

Channel Scenario Suburban Macro Urban Macro Urban Micro

Mean composite AS at BS E(AS)=50 E(AS)=80, 150 N/A

rds (delays/DS) Input 1.4 TBD 1.7 TBD

rds Output 1.29 TBD 1.54 TBD

ras (AoDs/PAS) Output 1.22 TBD 1.37 TBD

Composite AS at BS as a lognormal RV A= 0.69 A= 0.87 N/A

when simulating with 6 paths
A= 0.17 A= 0.36
AS=10Ax+A, x~N(0,1)

Per path AS at BS (Fixed) 2 deg 2 deg 5 deg

2 2 2 2
Per path AoD Distribution st dev N (0, r AOD ,CS �
s AS ) N (0,r AOD ,CS �
s AS ) U(-30deg, 30deg)
rAOD , CS =1.07 rAOD , CS =1.3
TBD TBD

Mean of RMS composite AS at UE E(AS, comp,UE)=720 E(AS, comp, UE)=720 E(AS, comp, UE)=720

Per path AS at UE (fixed) 350 350 350

Per path AoA Distribution N(0,AoA2(Pr)) N(0,AoA2(Pr)) N(0,AoA2(Pr))

Note 1 Note 1 Note 1

Mean total RMS Delay Spread E(DS)=0.17 s E(DS)=0.65 s U(0,0.8) s

Narrowband composite delay spread as a D = - 0.80 D = -0.175

lognormal RV when simulating with 6 paths
D = 0.288 D = 0.17
DS=10Dx+D, x~N(0,1)

Lognormal shadowing standard deviation 8dB 8dB 9-10dB TBD

3Note 1, Per path AoA is chosen from Normal distribution whose sigma is a function of the
4relative of path power, N(0,AoA2(Pr)), where 0 = LOS, and AoA = 104.12(1-exp(-0.3125*|Pr|),
5with Pr equal to the fraction of power in the given path in dBr.

63.3 Channel Generation Steps

7From these parameters, realizations of the user parameters such as the path delays, powers,
8and subpath angles of departure and arrival can be derived using the following 12-step

9procedure.

-
10 Step 1: Choose a single channel scenario common to all drops (i.e. to be applied to
11 the entire simulation).

216
1SCM Text Description SCM-083

1- Step 2: Generate Drops. Assign geometry (LOS direction and distance of UE from
2 NodeB), UE antenna structure orientation, UE speed vector direction and magnitude.

3- Step 3: Select lognormal random draws for DS, AS (at NodeB:  ASk ), LN as
4 described in Section 3.4.2 below.
5- Step 4: Assign N = 6 paths (for macro channels). Assign 6 random delays. Delays
6 are ordered and the minimum delay is subtracted from all so that the first delay is always
7 zero. k is the drop and UE index.
8 Ratio values rds are rds =1.17 suburban macro, rds =1.41 urban macro. Realization of random
9 delays are made according to the model below:

10  n = - rDS DSk log zn z n : U (0,1)

11 Note that while there is some evidence that delay spread may depend on distance between
12 the transmitter and receiver, the effect is considered to be minor (compared to other
13 dependencies: DS-AS, DS-LN.) Various inputs based on multiple data sets indicate that the
14 trend of DS can be either slightly positive or negative, and may sometimes be relatively flat
15 with distance. For these reasons and also for simplicity, a distance dependence on DS is not
16 modeled.
17
18 In the development of the Spatial Channel Model, care was taken to include the statistical
19 relationships between Angles and Powers, as well as Delays and Powers. This was done
20 using the proportionality factors r ds = delays/DS and ras = AoDs/PAS that were based on
21 measurements. (the r value is calculated by a ratio of statistically biased estimators).
22

-
23 Step 5: Assign a power to each path n:
(1-rDS ) n
24
Pn = e rDS  DSk
10 -n where  n : N (0,  2 RND )
25 where  RND is a shadowing randomization effect on the per-path powers. The value for RND
26 = 3 dB. Powers are normalized so that total power (for all six paths is equal to one).
27

28 The equations presented here for the power of the nth path are based on an power-delay
29 envelope which is the average behavior of the power-delay profile. Defining the powers to
30 reproduce the average behavior limits the dynamic range of the result and does not
31 reproduce the expected randomness from trial to trial. The randomizing noise n is used to
32 vary the powers with respect to the average envelope to reproduce the variations experienced
33 in the actual channel. This parameter is also necessary to produce a dynamic range
34 comparable to measurements.
35
36
37

-
38 Step 6: Generation of AODs per path and ordering at NodeB. Random draws of
39 AODs from N (0, r 2AOD,CS �
s 2AS ) . The rAOD , CS =1.07 (0.3 in dB) (Suburban macro), rAOD , CS =1.3
40 (1.0 in dB) (urban macro).
-
41 Step 7: The 6 AODs generated in Step 6 are ordered in increasing absolute value
42 and each of the six delays (Step 4) is assigned to each AOD. Increasing delays are matched to
43 increasing relative AOD angle  n is defined in Figure System Level Spatial Channel Model-6
44 based on their absolute value and deterministically.

217
1SCM Text Description SCM-083

1- Step 8: 20 sub-rays are used to generate a 2 o Laplacian spread for each path (ray)
2 at the NodeB. All 20 sub-rays have identical powers (1/20 of the path power) but random
3 phase. Sub-components have a predefined fixed angle distribution as shown in Table System
4 Level Spatial Channel Model-4.
5- Step 9: At the UE, assign per-path (ray) AOA variance as a function of the path (ray)
6 relative power. Draw the relative AOA (with respect to LOS) from a distribution: N (0, 2 AOA ) .
7 Where 0 = LOS, andAoA = 104.12(1-exp(-0.3125*|Pr|), with Pr equal to the fraction of power
8 in the given path in dBr.
9- Step 10: 20 sub-rays are used to generate a 35 o Laplacian spread for each path
10 (ray) at the UE. All 20 sub-rays have identical powers (1/20 of the path power) but random
11 phase. Sub-components have a predefined fixed angle distribution as shown in Table System
12 Level Spatial Channel Model-4.
-
13 Step 11: Pairing of each NodeB sub-ray with a corresponding UE sub-ray is made
14 to create the actual channel gain. Random pairing is used.
-
15 Step 12: Assign antenna gains to NodeB paths (rays). Assign the antenna gain to
16 each UE sub-ray.
-
17

18 Table System Level Spatial Channel Model-4. Fixed Sub-path components to produce per-path spreads

Sub-path # (+/-) degrees for a (+/-) degrees for a (+/-) degrees for a
2 deg Laplacian 5 deg Laplacian 35 deg Laplacian
(Node-B Macro-cell) (Node-B Micro-cell) (UE)

1, 2 0.08716 0.2179 1.525

3, 4 0.2754 0.6884 4.819

5, 6 0.4858 1.215 8.502

7, 8 0.7243 1.811 12.68

9, 10 0.9997 2.499 17.49

11, 12 1.325 3.313 23.19

13, 14 1.724 4.310 30.17

15, 16 2.238 5.595 39.16

17, 18 2.962 7.405 51.84

19, 20 4.201 10.50 73.51

19

3.4
20 BS and MS Array Topologies

21The spatial channel model should allow any type of antenna configuration to be selected,
22although details of a given configuration must be shared to allow others to reproduce the model

23and verify the results. It is intended that the spatial channel model be capable of operating on

24any given antenna array configuration.

218
1SCM Text Description SCM-083

13.4.1 Polarized arrays

2Practical antennas on handheld devices require spacings much less than /2. Polarized
3antennas are likely to be the primary way to implement multiple antennas. A cross-polarized

4model is therefore included here.

5A method of describing polarized antennas is presented, which is compatible with the 12 step
6procedure given in section 3.3. The following steps extend the original 12 to add the additional

7polarized components necessary.

8- Step 13: For each of the 6 paths of step 4, generate 20 sub-rays at the MS and 20
9 sub-rays at the BS, to represent the portion of each signal that leaks into the quadrature
10 antenna orientation due to scattering. Randomly pair each of the new sub-rays with a
11 corresponding sub-ray that was generated in step 8 for the BS and step 10 for the MS.
-
12 Step 14:. Set the AoD and AoA of each sub-ray in step 13 equal to that of the
13 corresponding sub-ray of the inline antenna orientation. (Orthogonal sub-rays arrive/depart
14 at common angles.)
-
15 Step 15: Set the phase angle of each sub-ray in step 13 to a random angle drawn
16 from U(0,2pi).
-
17 Step 16: The power P2 of each ray in the quadrature orientation is set relative to
18 the power P1 of each ray in the inline orientation according to an XPD ratio, defined as XPD=
19 P1/P2. For urban macrocells: P2 = P1 - A - B*N(0,1), where A=0.34*(mean relative path
20 power)+7.2 dB, and B=5.5dB is the standard deviation of the XPD variation.

21 For urban microcells: P2 = P1 - A - B*N(0,1), where A=8 dB, and B=8dB is the standard
22 deviation of the XPD variation.
-
23 Step 17: Decompose each of the inline and quadrature sub-rays into vertical and
24 horizontal components based on the in-line and quadrature orientations.
-
25 Step 18: At the receive antennas, decompose each of the vertical and horizontal
26 components into components that are in-line and quadrature with the receive antennas and
27 sum the in-line components.

28The fading behavior between the cross pol elements will be a function of the per-ray spreads
29and the Doppler. The fading between orthogonal polarizations has been observed to be
30independent and therefore the sub-rays phases are chosen randomly.

31

219
1SCM Text Description SCM-083

Y Y
T1 T2 R1 R2
UE1 UE2 BS1 BS2

X X

o
A. XY Plane at node UE (T=0 ) B. XY Plane at node B (R=0 )
o

Z Z
X X

Ti Ri
subray i subray i
C. XZ Plane at node UE D. XZ Plane at node B
1

2 Figure System Level Spatial Channel Model-7 Dual Polarization Example

3
4Additional description of the polarization model will be included in this section with equations
5describing the various components.
6

7Expressions describing the per antenna resulting complex waveforms can be described as
8follows (variables defined as in Figure System Level Spatial Channel Model-6).

9
hs ,u (t ) = i =1
T Pi
L
( L
l =1
G BS (q l ,i )  e
j k d s sin(q l , i )
 G MS (q lMS
,i )  e
j k d uMS sin(q lMS
,i )
e
j l , i
) (1)

j k  v cos( θl,iMS -q vMS ) t
e
10
11The polarization model can be illustrated by a matrix describing the propagation of and mixing
12between horizontal and vertical amplitude of each sub-path. The resulting channel realization
13is:
14
T
 e j l , i   G MS uV (q MS ) 
1
j l3, i
Pi L  G sV (q l ,i )  rx ,i  e
hs ,u (t ) = i =1  l =1  
T
   
l ,i

 GsH (q l ,i )   rx ,i  e l ,i
j 2 j l4, i
15 L e   G uH (q lMS
MS
,i )  (2)

j k d s sin(q l , i ) j k d uMS sin(q lMS
,i ) j k  v cos(θl,iMS -q vMS ) t
e e e
16
where:
17

18 s is the antenna index at the BS (NodeB) where s = 1...S

19 u is the antenna index at the MS (UE) where u = 1...U
20 T is the number of paths as defined in the temporal channel model

21 Pi is the relative power of the i -th path

22 L s the number of sub-paths used to model each path (L=20).

220
1SCM Text Description SCM-083

 GsV (q l , i ) 
1   is the antenna complex response in the vertical and horizontal polarizations at
 GsH (q l , i ) 
2 the BS (NodeB) as functions of the AOA.

 G MS uV (q MS ) 
3 
l ,i
is the antenna gain at the MS (UE) as a function of the AOA for the two
 G MS uH (q lMS )
,i 

4 polarizations.
5

6 q l ,i is the angle, with respect to the BS broadside, of the l -th sub-path in the i -th path.
7 Note that q l , i = q o   i  q l ,i _ sub - path , where q l ,i _ sub - path is the sub-path’s relative angle
8 within the path as defined in Table 4 in the SCM-Text.

9 q lMS
,i is the angle, with respect to the UE broadside, of the l -th sub-path in the i -th path.
10 l , i is the uniformly distributed random phase of the l -th sub-path in the i -th path.
2
11 k= is the wave number.

12 ds is the distance in wavelengths of the antenna s from the reference (s=0 ) antenna. If s is
13 the reference antenna then the distance is zero.

14 v is the MS (UE) speed vector in meters/sec. The norm of the vector is the magnitude of
15 the speed.

16 q vMS is the angle of the speed vector with respect to the MS array’s broadsise
17
18The 2x2 matrix represents the scattering phases and amplitudes of a plane wave leaving the
19 UE with a given angle and polarization and arriving Node B with another direction and
20 polarization. rx , i is the average power ratio of waves leaving the UE in the vertical
21 direction and arriving at Node B in the horizontal direction (v-h) to those arriving at
22 Node B in the vertical direction (v-v). By symmetry the power ratio of the opposite
23 process (h-v over v-v) is chosen to be the same. Note that: rx , i =1/XPD; for the
24 macrocell model, the XPD is dependent on the path index; for the microcell model, the
25 XPD is independent of path index.

26  lx,i where x=1,2,3,4 are the uniform and i.i.d random phases of the l -th sub-path in the i -
27 th path for each of the four mixing directions.
28Expression (2) assumes a random pairing of the of the sub-paths from the MS and BS. The
29random orientation of the MS (UE) array affects the value of the angle q l , i
MS
of each sub-path.

30If for example, vertically polarized antennas are used only at both NodeB and UE then the
1 
31antenna responses become   and expression (2) becomes identical to (1). For an ideal dipole
0 

221
1SCM Text Description SCM-083

1antenna at the NodeB tilted with respect to the z-axis at  degrees the above vector becomes
 cos( ) 
2 sin( )  cos(q ) .
 l ,i 

3The elevation spectrum is not modeled.

53.4.2 Reference Antenna Configurations

6In order to compare algorithms, reference antenna configurations based on uniform linear
7array configurations with 0.5, 4, and 10 wavelength inter-element spacing will be used.

83.4.3 General Modeling Assumptions

91. Mapping of paths to resolvable paths. TBD
2.
10 Fractional Unrecovered Power (FURP) is TBD.
3.
11 The macrocell pathloss from 3GPP2 evaluation methodology will be used for the SCM
12 model as shown below.
Propagation Model 28.6+ 35log10(d) dB, Modified Hata Urban Prop.
(BTS Ant Ht=32m, MS=1.5m) d in meters Model @1.9GHz (COST 231).
Minimum of 35 meters
separation between MS and
BS.1

13

4.
14 Uplink-Downlink Reciprocity: The AOD/AOA values are identical between uplink and
15 downlink.

5.
16 Random path phases between UL, DL are uncorrelated.

6.
17 mobile-to-mobile shadowing is uncorrelated.

18

3.4.4
19 Micro-cell Assumptions
(a)
20 The BS antenna is always located at rooftop.
(b)
21 The pathloss model is TBD
(c)
22 Antenna patterns at BS/NodeB will be those already defined in the SCM-
23 Text. Question about backlobes remain since micro-cells typically have poor backlobes.
24 TBD.
(d)
25 Site-to-site correlation. Shadowing between different paths (delay) are iid.
26 Site-to-site correlation for the same delay follows the macro model (50% correlation). Some
27 questions remain on this method.
(e)
28 Log Normal St.Dev of 4-5dB per path should be used prior to normalization.
29 The log normal per-path sets the distribution and the dynamic range for the path powers,

2
1

322
1SCM Text Description SCM-083

1 and the distribution for the composite base angle spread. A separate bulk log normal
2 shadowing is needed for the path losse03
3(f) . The value of 10dB was suggested. The final value is TBD.
4(g) The hexagonal cell repeats will be the assumed layout.

63.5 Options for Urban Macro and Micro Environments

73.5.1 Far scatterer clusters

8 The Far scatterer cluster model is switch selectable. It represents the bad-urban case where
9 additional clusters are seen in the environment. This model is limited to use with the urban
10macro-cell where the first cluster will be the primary cluster and the second will be the far

11scattering cluster (FSC). When the model is active, it will have the following characteristics:

121. There is a reduction in the number of paths in the primary cluster from N = 6 to N = 4, with
13 the far scattering cluster then having N = 2. Thus the total number of paths will stay the
14 same, now N = 4 + 2. This is a modification to the SCM channel generation procedure in
15 section 3.3.

2. FSCs will lie only outside a 1000m radius from the BS/NodeB. (the 500-600m radius is
16

17 under consideration also but not finalized).

183. The model statistics of the two clusters are identical (cluster DS, AS, PDP) but
19 independently drawn. The FSC also has independent shadowing.

204. The FCS is attenuated by 1dB/microsec delay with respect to the 1 st cluster with a 10dB
21 maximum. The excess delay will be defined as the difference in propagation time between
22 the BS-MS LOS distance, and the BS-FSC-MS distance. The delay of the FSCs of the other
23 cells (Ioc components) remains an open issue.

245. There will be one FSC modeled within each cell and dropped following a uniform
25 distribution. FSCs will be modeled for all the cells in the simulation in order to capture the
26 Ioc spatial characteristics.

27An exact method of setting the powers is TBD. The following approach is a possible method:
28Draw the N=4 path powers from the channel generation procedure in section 3.3, then draw a

29separate set of N=2 path powers from the same procedure. The two groups are kept separate

30and un-normalized. Now the delay based attenuation is applied to the group of N=2 paths, and

31the N=6 total paths are normalized to unity power.

3.5.2
32 Line of sight

33The Line-of-sight model an option that is switch selectable. It can be selected for the urban
34macro and micro cases. Suburban is TBD. It uses the following description when this
35function is selected.

For the NLOS case, the Rice factor is set to 0, thus the fading is determined by the combination
36

of sub-rays as described in section 3.3 of the model.

37

223
1SCM Text Description SCM-083

1For the LOS case, the Rice factor is based on a simplified version of [Foster 1994], and is
2applied based on a probability factor:

3K = 13-0.03*d (dB)

4where d is the distance between MS and BS in meters.

5The probability for LOS or NLOS depends on various environmental factors, including clutter,
6street canyons, and distance. For simplicity, the probability of LOS is defined to be unity at
7zero distance, and decreases linearly until a cutoff point at d=300m, where the LOS probability

8is zero.

9P(LOS) = (300 – d)/300, for 0< d < 300m.

10 0, for d > 300m.

11The K-factor will be formed by adding a direct component (sine wave) at the average AoD and
12AoA of the path such that the ratio of the power assigned to the direct component to the power

13assigned to the 6 paths is equal to the K-factor measured in dB. After the power of the direct

14component is added, the total power in the channel is normalized to unity power. The K-factor

15is defined as the ratio of power in the LOS component to the total power in the diffused-NLOS

16component. The LOS path will coincide in time with the first (earliest) diffused path. When
17pairing sub-rays between transmitter and receiver, the direct components are paired

18representing the LOS path.

193.5.3 Urban canyon

20The urban canyon model is switch selectable. When switched on, the model modifies the AoAs
21of the paths arriving at the subscriber unit. It is for use in both the urban macro and urban
22micro scenarios.

23Urban-canyons exist in dense urban areas served by macro-cells, and below-rooftop micro-
24cells. When this model is used, the spatial channel for all subscribers in the simulated
25universe will be defined by the statistical model given below. Thus for the SCM channel
26generation steps given in Section 3.3, Step 9 is replaced with steps 9a-d given below, which
27describe the AoAs of the paths arriving at the subscriber in the urban canyon scenario.

28The following procedure is used to determine the subscriber mean AoAs of the six paths. This
29model does not use a building grid, but assigns angles based on statistical data presented in
30the figures below. The procedures is defined in terms of the subscriber terminal:

31 Step 9a. Pick a random direction of travel from a uniform random variable U(-180,180).
32 This is going to be also the assumed street orientation for this subscriber, and their
33 direction of travel.

34 Step 9b. Pick the AoA corresponding to the strongest ray in the channel impulse response
35 according to the CDF depicted in Figure System Level Spatial Channel Model-8. Note that
36 +/- angles are equally likely.

37 Step 9c. For the remaining 5 weaker paths, pick the AoAs according to the CDF shown in
38 Figure System Level Spatial Channel Model-9, where zero degrees is defined to be AoA of
39 the strongest path from step 9b.

224
1SCM Text Description SCM-083

1 Step 9d. For interferers, steps 9b and 9c are repeated, assuming the street orientation and
2 direction of travel did not change from step 9a.
3

4The fixed 35 degree per-path Azimuth Spread is applied after the six ray AoA are defined.

0.9

0.8

0.7
Pr(Angle<=abscissa)

0.6

0.5

0.4

0.3

0.2

0.1

0
0 20 40 60 80 100 120 140 160 180
Angle wrt street orientation degrees
5

6 Figure System Level Spatial Channel Model-8, AoA of Strongest Path (+ & - angles have equal probability)

8The CDF for the function given in Figure System Level Spatial Channel Model-8, is defined by:

 x 0o  x  15o
 50
x  0.2723 15o  x  80o
 x541 . 66
Fx ( x) =  - 0.22 80o  x  100o
125
9

x
 541.66  0.3954 100o  x  165o
 x - 2.6 165o  x  180o
 50
10

225
1SCM Text Description SCM-083

0.9

0.8

0.7

Pr(Angle<=abscissa) 0.6

0.5

0.4

0.3

0.2

0.1

0
-150 -100 -50 0 50 100 150
Angle between strongest and the rest degrees
1

2 Figure System Level Spatial Channel Model-9, Angle difference between Strongest Path and remaining paths
3

4The CDF of the function given in Figure System Level Spatial Channel Model-9, is defined by:

x  0.15 - 180o  x  0 o
 1200
5 Fx ( x) =  0.7 x=0
x  0 .85 0 o
 x  180 o
 1200
6

73.6 Correlation Between Channel Parameters

83.6.1 Correlation between Delay Spread, Angle Spread, and Log Normal Shadowing

9 In [3], Greenstein presents a model for correlating delay spread (DS) with log normal (LN)
10shadow fading. Since both are shown to be log-normal distributed, the correlation between the

11DS and LN are correlated by the coefficient . The best value for suburban and urban data was

12shown to be  = -0.75, presented in [Error: Reference source not found] from data measured by
4
13[ ].

14The result of the correlation between log normal shadowing and delay spread is significant
15because it indicates that for a strong signal (positive LN), the DS is reduced, and for a weak
16signal condition (negative LN), the DS is increased.

17Cost 259[5] presents the azimuth spread (AS) as also being log-normal distributed, and likewise
18being correlated to the DS and LN. Since the correlation of these parameters is quite high, a

19spatial channel model needs to be specified that can reproduce this correlation behavior along

226
1SCM Text Description SCM-083

1with the expected probability and range of each parameter. For a macro-cell environment, the
2following values are given in [Error: Reference source not found]:

4 = Correlation between DS & AS = +0.5

5 = Correlation between LN & AS = -0.75

6 = Correlation between LN & DS = -0.75

7The random variables for correlating DS, AS, and LN shadowing are generated using

 n   c11 c12 c13   wn1 

   = c c22 c23   wn 2  , (1)
8
 n   21  
  n  c31 c32 c33   wn 3 
9 where n refers to the nth base station (BS). and are the correlated zero-mean unit
10variance Gaussian distributed random variables associated with the DS, AS, and LN,

11respectively, and wn1, wn2, and wn3 are unit-variance, independent Gaussian noise samples.

12Equation 1 is performed once for each BS per drop.

1
 c11 c12 c13          2

c  
13
 21 c 22 c 23  =         , (2)
c31 c32 c33         

Equation 2 incorporates the matrix square root function of correlation coefficients.

14

15The LN shadow fading component is typically correlated between Base Stations (BS). To
16produce correlated shadow fading components () between BS1 and BS2, a simplified approach

17is shown where the independent unit-variance Gaussian noise samples w 13 and w23 are

18replaced with correlated, unit-variance Gaussian noise samples that are generated using:

cc cc
19 w13 =  c  1 - 2 1 (3)
c33 c33

and
20

cc cc
21 w23 =  c  1 - 2 2 , (4)
c33 c33

22where c, 1, 2 are independent, unit-variance Gaussian noise samples and cc is the desired
23site-to-site shadow fading correlation between the MS and nodes BS 1 and BS2. A value for the
24common component cc=0.5 or 50% site-to-site correlation will be assumed. The cumulative
25distribution functions of DS, LN, and AS are described in terms of the correlated Gaussian
26random variables that characterize their distributions [Error: Reference source not found]
6
27[Error: Reference source not found][ ].

Using nomenclature from[Error: Reference source not found] the distribution of DS is given by:

28

227
1SCM Text Description SCM-083

1 DS = 10^(Dn +D) (5)

2where  n  is   the correlated zero-mean unit variance Gaussian distributed random variable
3associated with the DS for the nth BS, D  =   E{log10(DS)} is the logarithmic mean of the

4distribution of DS, and  D =   Std{log10(DS)} is the logarithmic standard deviation of the

5distribution of DS.

6Similarly the distribution of AS is given by:

7 AS = 10^(An +A) (6)

8 where n is the correlated zero-mean unit variance Gaussian distributed random variable
9 associated with the AS, A = E{log10(AS)} is the logarithmic mean of the distribution of AS, and
10A = Std{log10(AS)} is the logarithmic standard deviation of the distribution of AS.

Finally, the distribution for the LN is given by:

11

12 LN = 10^(SFn/10), (7)

13where n is the correlated zero-mean unit variance Gaussian distributed random variable
14associated with the LN, SF (shadow fading) is obtained from analysis of the standard deviation
15from the regression line of the path loss versus distance, and the value is given in dB. Since

16the LN is normally used as a dB value, there is no need to convert it to linear, but rather

17simply use the quantity SFn.

Correlated channel parameters may be drawn within a system simulation based on the
18

equations described above.

19

3.7
20 System Level Calibration
21The following examples are given for calibration purposes. A resolvable path at the receiver is
22assumed to be the energy from one (or more) paths falling within one chip interval. The Chip
23rate in UMTS is 3.84Mcps. The PDF of the number of resulting resolvable paths is recorded.

24

25

26 Table System Level Spatial Channel Model-5, SCM Parameter Summary with Simulated Outputs

Urban 8 Urban 15 Suburban 5

Parameter
RND = 3dB RND = 3dB RND = 3dB

Input Output Input Output Input Output

rDS
1.7 1.54 1.7 1.54 1.4 1.29

Input Ideal Input Ideal Input Ideal

DS
-6.195 -6.26 -6.195 -6.26 -6.80 -6.92

Input Ideal Input Ideal Input Ideal

DS
0.18 0.25 0.18 0.25 0.288 0.363

Input Output Input Output Input Output

rAS
1.3 1.37 1.3 1.37 1.2 1.22

228
1SCM Text Description SCM-083

Input Ideal Input Ideal Input Ideal

AS
0.810 0.75* 1.18 1.0938 0.66 0.66

Input Ideal Input Ideal Input Ideal

AS
0.3295 0.37 0.210 0.2669 0.12 0.18

Ideal Simulated Ideal Simulated Ideal Simulated

E[DS]
0.65s 0.63s 0.65s 0.63s 0.17s 0.172s

Ideal Simulated Ideal Simulated Ideal Simulated

E[ AS Node B]
8 7.97 15 14.94 5 5.01

Ideal Simulated Ideal Simulated Ideal Simulated

E[ AS UE]
72 72.05 72 72.69 72 74.07

1*This value was 0.77, however this will produce a value of E[ AS Node B]=8.5 instead of 8.
2

3The following figures: Figure System Level Spatial Channel Model-10, Figure System Level
4Spatial Channel Model-11, Figure System Level Spatial Channel Model-12, Figure System Level
5Spatial Channel Model-13, Figure System Level Spatial Channel Model-14, represent
6calibration cases for the current SCM model. These curves correspond to the parameters
7presented in Table System Level Spatial Channel Model-5, and include the 3dB randomizing

8factor for the generation of path powers.

3840000 chips/second
0.4
Urban Macro
Suburban Macro
0.35

0.3

0.25
Probability

0.2

0.15

0.1

0.05

0
1 2 3 4 5 6
Number of resolvable paths

10 Figure System Level Spatial Channel Model-10, Probability of Urban and Suburban Time Resolvable Paths

229
SCM Text Description
1 SCM-083

0.9

0.8

0.7

Pr(RMS DS<=abscissa)
0.6 Urban Macro
Urban Macro Ideal
Suburban Macro
0.5 Suburban Macro Ideal

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
RMS Delay Spread seconds -6
x 10
1

2 Figure System Level Spatial Channel Model-11, RMS Delay Spread, Simulated versus Ideal

CDF of Composite Angle Spread at Node B

1

0.9

0.8

0.7
Pr(RMS AS<=Abscissa)

0.6

0.5

0.4

Suburban Macro 5o
0.3
Suburban Macro Ideal 5o
0.2 Urban Macro 8o
Urban Macro Ideal 8o
0.1 Urban Macro 15o
Urban Macro Ideal 15o
0
0 2 4 6 8 10 12 14 16 18 20
RMS Angle Spread, in degrees
5

6 Figure System Level Spatial Channel Model-12, Node-B Composite Angle Spread, Simulated versus Ideal

30
2
SCM Text Description
1 SCM-083

Noise power3 dB
1

0.9

0.8

0.7

Pr(Dynamic Range<=abscissa) 0.6

0.5

0.4

0.3
Urban Macro w/o Quantization
Urban Macro Resolvable
0.2 Suburban Macro w/o Quantization
Suburban Macro Resolvable

0.1

0
0 5 10 15 20 25
Dynamic Range dB

2 Figure System Level Spatial Channel Model-13, Dynamic Range (dB) for each channel model

0
Noise power3 dB
10

-1
10
Pr(Path Power<=abscissa)

-2
10

Urban Macro w/o Quantization

Urban Macro Resolvable
Suburban Macro w/o Quantization
Suburban Macro Resolvable

-3
10
-30 -25 -20 -15 -10 -5 0
Path Power dB

5 Figure System Level Spatial Channel Model-14, CDF of all Path Powers

31
2
 SCM Text Description
1 SCM-083

Channel Scenario: Urban Microcellular

1

Composite Delay Spread (DS) for Microcellular Scenario Number of resolvable paths for Microcellular scenario
1 7000

0.9
6000
0.8

0.7 5000

Occurence per 104 drops

0.6
4000
CDF

0.5
3000
0.4

0.3 2000

0.2
1000
0.1

0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1 2 3 4
sec N
4

Composite Angle Spread (AS) for Microcellular Scenario

1

0.9

0.8

0.7

0.6
CDF

0.5

0.4

0.3

0.2

0.1

0
5 10 15 20 25 30
Degrees
5

Figure
6 4 Statistics for Urban Microcellular Environment with chip-rate frequency =
3.84MHz
7 (clockwise from top left): (a) CDF of total rms DS, (b) Probabilities for number of
resolvable
8 paths present (in occurrences per 10,000 drops). (c) CDF of Composite AS
9

4
10 APPENDIX A: MMSE RECEIVER DESCRIPTION

(The following text is a preliminary description of the MMSE receiver).

11

12

13This procedure generates SINR values at the output of a linear MMSE receiver for a single
14instant in time.

15Step 1: Given the space-time propagation model and transmitter state, form a channel
16 (expressed here as one or more convolution matrices) relating all transmitting sources and
17 receive antennas from every sector in the system.

18 At the UE, the received samples are represented as a column vector,

32
2
1SCM Text Description SCM-083

r = [r1T , r2T , , rM
T T
]
1
= [ r1 (1), r1 (2), , r1 ( N ), r2 (1), r2 (2), , r2 ( N ), , rM (1), rM (2), , rM ( N )]T ,

2 where M is the number of receive antennas at the UE, and N is the number of received
3 symbols per antenna . This received time-space vector is related to the transmitted symbols as
2

4 follows:

G1(1)  G1( j )   n1 
 (1)     
J
 G 2  (1) J G (2 j )  ( j )  n 2 
r = G x  G x  n =
(1) (1) ( j) ( j)
x  x 
      
5
j =2 j =2
 (1)   ( j)   
G M  G M  n M 

( j)
6 where G i , 1 ≤ i, j ≤ M are Toeplitz convolution matrices defining the channel between the
7 i-th receive antenna and the j-th transmitted data stream, x ( j ) is the j-th transmitted data
8 stream, J is the total number of data streams in the system, and n is the vector of noise
9 samples. The j = 1 data stream is the primary data stream intended for the user. The j-th
10 data stream can be a transmission from an interfering base station, another sector of the
11 desired base station, or another data stream intended for the desired user (which is
12 considered interference to the primary data stream). If the composite channel response is
13 limited to K samples, then each of the convolution matrices has N rows by (N+K-1) columns,
 g i( j ) ( K ) g i( j ) ( K - 1)  g i( j ) (1) 0 0  0 
 
0 ( j)
gi (K ) g i ( K - 1)
( j)
 ( j)
g i (1) 0  0 
G i( j ) = ,
  0 
14
     
 
 0 0  0 g i( j ) ( K ) g i( j ) ( K - 1)  g i( j ) (1)

and  g i  is the vector of discrete channel samples of length K.
( j)
15

16 Note that in the above formulation, the vector x has M(N+K-1) rows, and thus, it is longer
17 than the received vector, r. Also, the vector x will be interleaved with zero values if a
18 fractionally-spaced approach with more than one received sample per symbol is used.

Step 2: Using the above channel, produce an estimate of the channel.

19

20 gˆ i( j ) = g i( j )  Δg i( j ) ,

( j)
21 where Δg i is a vector representing the channel estimation error for the i-th receive antenna
22 and the j-th transmitted data stream. The estimation error is due to noise and interference in

2
2 Actually, this is the number of received samples per antenna, if more than one sample per symbol is

3collected.

433
1SCM Text Description SCM-083

1 the pilot channel and can also be due to the channel estimator’s inability to track a fast fading
2 channel.

3Step 3: Using the estimated channel, compute the SINR per data stream at the output of the
4 MMSE filters.
2
ˆ -1fˆ
f jH Ω j j
5 SINR j = ,
fˆ jH Ω
ˆ -1Ω Ω
j j
ˆ -1fˆ
j j

6 where

       ,
J
Ω j = G( j)E x( j)x( j)H G( j)H - f ( j)E x( j)(d)x( j)(d) f ( j)H G(m)E x(m)x(m)H G(m)HE nnH
7
m=1
m j

8 Ω̂ j is an estimate of Ω j , d = max(  ( N - K ) / 2  K , K ) , f j is the d-th column of G ( j ) ,

9 x ( j ) ( d ) is the d-th element (desired symbol) of the x ( j ) data stream vector, and SINRj
10represents the SINR for the j-th transmitted data stream in the system. In this example, the
11primary data stream sent to a user will be j = 1. In a MIMO system where multiple data
12streams are sent to a single user, the second stream could be j = 2, etc.

13

5
14 REFERENCE
15

234
11[] WG5 Evaluation Methodology – Addendum (V6), WG5 Evaluation AHG, July 25, 2001
22[] Nokia, Ericsson, Motorola. Common HSDPA system simulation assumptions. TSG-R1 document,
3TSGR#15(00)1094, 22-25th, August, 2000, Berlin, Germany, 12 pp.
43[] L. Greenstein, V. Erceg, Y. S. Yeh, M. V. Clark, “A New Path-Gain/Delay-Spread Propagation
5Model for Digital Cellular Channels,” IEEE Transactions on Vehicular Technology, VOL. 46, NO.2,
6May 1997, pp.477-485.

74[] E. Sousa, V. Jovanovic, C. Daigneault, “Delay Spread Measurements for the Digital Cellular
8Channel in Toronto,” IEEE Transactions on Vehicular Technology, VOL. 43, NO.4, Nov 1994, pp.837-
9847.

5[] L. M. Correia, Wireless Flexible Personalized Communications, COST 259: European Co-operation
10
in Mobile Radio Research, Chichester: John Wiley & Sons, 2001.
11

126[] A. Algans, K. I. Pedersen, P. Mogensen, “Experimental Analysis of the Joint Statistical Properties
13of Azimuth Spread, Delay Spread, and Shadow Fading,” IEEE Journal on Selected Areas in
14Communications, Vol. 20, No. 3, April 2002, pp. 523-531.