0. Spatial Channel Model for Link Level Simulations
Younglok Kim
July 14, 2004

1. Outline Importance of spatial channel model (SCM)
Spatial model implementation
Old spatial correlation model
Time-invariant correlation
Time-varying correlation
MIMO spatial channel model by 3GPP-3GPP2
More spatial parameters for BS and UE
Discussions

2. Channel Models for Multiple Antennas
The spatial properties of channels are extremely important in determining the performance of multiple antenna systems
Each diversity scheme has own channel model description
Need to consider
Time-varying angular spread
Angular distribution with plane wave assumptions
Adaptive array antenna geometries
Various estimations of spatial correlation based on various assumptions
Must verify models by field measurements [R ]
If we adopt more than two antenna elements for TxD scheme, SCM description is important to evaluate the performances of various different schemes.

3. Spatial Correlated Channel Model
Assumptions
Long-term properties of the channel remain unchanged over time
Independence between different taps
Spatial covariance matrices are equal for each tap
Spatially correlated multiple channels can be generated by the linear transformation of the same number of uncorrelated channels

4. Block Diagram for SCM Generation

5. Alternative Modification for Rx Diversity Simulations
Assumption: spatial covariance matrices are equal for each tap
Equivalent simulation method without change the channel impulse responses by transformation of
Received signal before AWGN
Antenna weight

6. Spatial Correlation Coefficients
Spatial correlation matrix for 4 antenna case
Spatially uncorrelated channels: a=b=c=0
Correlation coefficients are given in TSGR1#17(00)1358

7. Recommended Correlation Parameters
Environment
Parameters
Recommended path model
Angular spread
Macro cell
(Rural area)
a=0.97 exp(-0.8 j)
b=0.94 exp(-1.6 j)
c=0.88 exp(-2.4 j)
1-path Rayleigh
Vehicular A
10° AS
70° AoA
Micro cell
(Urban area)
a=0.7 exp(-2.2 j)
b=0.1 exp(-1.2 j)
c=0.2 exp(-3.0 j)
Pedestrian A
45° AS
60° AoA
Pico cell or uncorrelated
a, b, c = 0
Large angular spread

8. PSD of Macro Cell Model: 70 AoA, 10 AS
AoA is around 15° but not 70 °
Long-term average power profile according to angle
Beam pattern of the dominant eigenvector
Dominant eigenvector is associated with 99.9% power

9. Eigenbeam Patterns of Macro Model
0.1%
99.9%
0 %
0 %

10. PSD of Micro Cell Model: 60 AoA, 45 AS
Dominant eigenvector is associated with 96% of total power

11. Eigenbeam Patterns of Micro Cell
38.7%
57.2%
3.7 %
0.4 %

12. Time Varying Correlation Model
Discrete uniform distribution model

13. Time Varying Correlation Model…
Angle of sub-paths:
Angle of center:
Steering vector of ULA:

14. Steering Vector of ULA Phase Difference between two adjacent antenna

15. Time Varying Model Parameters
Propagation Environment
Macro Cell
Micro Cell
Distance from BS to UE
2000 m
500 m
Angular spread
10 °
45 °
Channel models
&
UE velocities (km/h)
1-path Rayleigth: 3,10,40,120
Vehicular A: 10,40,120
Pedestrian A: 3, 10, 40
Number of sub-paths (Q) 10
100
Closed loop feedback error rate
4 %

16. Time Varying Coeffs (Macro Cell)
Power of dominant eigenvector is about 99.9%

17. Time Varying Coeffs (Micro Cell)
Power of dominant eigenvector ranges from 75.4 to 99.7%

18. Comments on Previous Channel Model
Only eigenbeamformer scheme has adopted this spatial channel model
Uniformly distributed power on an angular region
Very limited simulation environments can be considered
Constant time-invariant AoA can be assumed
Since it is a slow varying effect similar to shadow fading
Time varying AoA will not included in link level simulations
Common & more reliable simulation spatial channel model is required
Need specific descriptions of more channel parameters, AOA, PAS, AS, etc

19. Spatial Channel Model by AHG
Text description: SCM /2002
Recommended to use MIMO physical layer channel model proposed
SCM AHG (AH-62) from 3GPP & 3GPP2
Develop and specify parameters and methods associated with
Link level spatial channel model
For calibration only
System level spatial channel model
Define physical parameters and system evaluation methodology

20. Link level channel model
Only for calibration purposes
Reflect only one snapshot of the channel behavior
Do not account for system attributes such as scheduling and HARQ
Only for comparison of performance results from different implementation of a given algorithm
Status: 95% completed

21. System level channel model
Required for the final algorithm comparison
Define the methodology for generating the spatial channel coefficients between BS and MS
95% completed

22. Terminologies MS = UE = terminal = subscriber unit BS = Node-B = BTS
AS = angle spread = azimuth spread = PAS
Path = Ray
Path component = Sub-ray
PAS = power azimuth spectrum
DoT = direction of travel
AoA = angle of arrival
AoD = angle of departure
PDP = power delay profile

23. Parameters of Link Level SCM
Multipath fading propagation conditions
Model
Case I
Case II
Case III
Case IV
3GPP
Case B
Case C
Case D
Case A
3GPP2
Model A, D, E
Model C
Model B
Model F
PDP
Mod. Pedestrian A
Vehicular A
Pedestrian B
Single path
Speed (Km/h)
1) 3
2) 30, 120
3, 30, 120 3
# of paths
1) (LOS on, K=6 dB)
2) 4 (LOS off) 6 1
Relative path power (dB)
& Delay (ns)
LOS on
0.0
-6.51
-16.21
-25.71
-29.31
LOS off
–Inf
-9.7
-19.2
-22.8
110
190
410
-1.0
-9.0
-10.0
-15.0
-20.0
310
710
1090
1730
2510
-0.9
-4.9
-8.0
-7.8
-23.9
200
800
1200
2300
3700

24. Parameters of Link Level SCM…
Spatial parameters for NodeB
Model
Case I
Case II
Case III
Case IV
Topology
Reference: ULA with 0.5, 4, 10 spacing
N/A
PAS
Laplacian distribution with
RMS angle spread of 2 or 5 degrees per path
depending on AoA/AoD
AoD/AoA
(degrees)
50 for 2 RMS AS per path
20 for 5 RMS AS per path
Antenna gain pattern
3 or 6 sector antenna pattern
(For diversity oriented applications rather than beamforming applications)

25. Parameters of Link Level SCM…
Spatial parameters for UE
Model
Case I
Case II
Case III
Case IV
Topology
Reference 0.5
N/A
PAS
1) LOS on: Fixed AoA for LOS component, remaining power has 360 degree uniform PAS. ( RMS angle spread of 104 degrees)
2) LOS off: Laplacian distribution with RMS angle spread of 35 degrees per path
Laplacian distribution with RMS angle spread of 35 degrees per path OR
360 degree uniform PAS ( RMS angle spread of 104 degrees)
DoT
(degrees)
22.5
-22.5
AoA
22.5 (LOS component)
67.5 (all other paths)
67.5 (all paths)
22.5 (odd paths)
-67.5 (even paths)
Antenna gain pattern
Omni directional with -1 dBi gain

26. Assumptions of Link Level SCM
Spatial channel parameters per path
Each resolvable path is characterized by spatial channel parameters: AS, AoA, PAS
All paths are assumed independent
Array Topologies
Allow any type of antenna configuration, but must be shared to reproduce and verify the results
ULA with element spacing of 0.5, 4, 10 wavelengths

27. Antenna Gain Patterns UE: -1 dBi gain omni-direction
Node B uplink/downlink
Only for diversity oriented implementations (large spacing)
Need different antenna patterns for beamforming applications
3 sector cell
Bandwidth 70°, Maximum attenuation 20 dB, 14 dBi gain
6 sector cell
Bandwidth 35°, Maximum attenuation 23 dB, 17 dBi gain
Sector antenna formula in dB scale

28. Sector Antenna Patterns at BS…

29. Sector Antenna Patterns at BS…

30. Average Received Power
Laplacian distributed PAS
Angle of Arrival
RMS angle spread
Normalization factor
Antenna gain in linear scale

31. Average Received Power at BS
3 sectored antenna with AoA = 20 and RMS AS = 5

32. Average Received Power at MS
Laplacian distributed PAS with omni-directional gain

33. Average Received Power at MS…
22.5 AoA & 35 RMS AS

34. Average Received Power at MS…
22.5 AoA & 104 RMS AS (uniform over 360 degree PAS)

35. Doppler Spectrum at UE
Dependent on
DoT
PAS
AoA

36. Generation of link level channel model
Average received power
Angle theta represents path components
Its distribution is TBD
Uniformly distribution over [ ] degrees
Channel implementation techniques
Correlation based
Ray based
Details are TBD
Reference correlation values are provided
Nokia to provide formulas for computing correlation matrices

37. Future Plan & Discussions
SCM generation and verification by using MatLab UE BS
Study on system level spatial channel model

38. References Old 3GPP Model: MIMO Model: Tx Div Model: TSGR1#14(00)0867
0765, 770
TSGR1#29(02)
1139,1419,1440
1441 (TR25.869)
MIMO Model:
TSGR1#22(01)1132
TSGR1#22(01)1136
TSGR1#23(01)1179
TSGR1(02)0141
SCM-077, November 2002
by 3GPP-3GPP2

39. Standard for Space Channel Model
Common & more reliable simulation spatial channel model is required
Need specific descriptions of more channel parameters, AOA, PAS, AS, etc
Spatial channel model by 3GPP-3GPP2
More spatial parameters for BS and UE
SCM Text V7.0, 8/2003
ftp://ftp.3gpp2.org/TSGC/Working/2003/3GPP_3GPP2_SCM_(Spatial_Modeling)/ConfCall /