Sie sind auf Seite 1von 27

Channel models for MIMO

Adaptive Antenna Systems


Persa Kyritsi
December 16, 2004
What you are going to learn today:
z Fundamentals of MIMO systems
z Transmission techniques for MIMO
z Channel models for MIMO
(details on 802.11n)
z Real MIMO measurements
Power
Capacity
Rate
Why all the hype?
z MIMO advantages
Capacity
Range extension
Diversity
z Original results derived with iid Rayleigh h
ij
z More realistically, we need to characterize
Each link h
ij
Interdependence of links
Why do we need channel models?
z Prediction models for site planning
Site specific
Antenna dependent
Accurate
z Models for system design and algorithm
testing
Site and antenna independent
Tolerance for lower accuracy
Classification of MIMO channel models
MIMO channel models
Deterministic Stochastic
Recorded
impulse
responses
Ray-tracing
technique
Geometrically
based
Parametric
stochastic
Correlation
based
Non-
physical
Physical
Correlation based models
( ) ( ) ( ) k i l j h h
RX TX kl ij
, , , =
Kronecker assumption:
Separability of transmit and receive correlations
| |
( ) | |
( )
N N
M M
vec
RX
TX
iid RX TX vec
T
T
M
T T
vec
M

=
= =
=
:
:
H H
H H H H
H H H
2 1
2 1
2 1
R
R
R R
H
H
L
L
How to calculate the correlations
( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )

=
=
+ =

d PAS d k d R
d PAS d k d R
d jR d R d
l XY l
l XX l
XY l XX l
IQ
l
sin
, cos
,
,
, ,
z PAS( ): Power Azimuth Spectrum
(how much energy is arriving from where)
z How do the correlations look?
Discussion
z Simple & elegant
z Experimental validation in some environments and
discrediting in others
z Original expression cannot capture the pinhole
effect
z Generalization:
( ) ( ) ( )
2 1 2 1 2 1
TX t s r RX
R H R H R H =
MIMO channels: Wideband
z Narrowband z Wideband
( ) ( ) ( )
( ) ( )
( ) ( ) ( )
( ) ( )
l
l
l
l
t t t
t n t t x t y MIMO
t t h t h
t n t t x h t y SISO
=
+ =
=
+ =

H H
H :
:
( ) ( ) ( ) ( )
( ) ( )
( ) ( ) ( ) ( )
( ) ( )

=
=
+ =
=
+ =
1
0
1
0
:
:
L
l
l l
L
l
l l
t t t
t n t x t t y MIMO
t t h t h
t n t x t h t y SISO

H H
H
A specific example: 802.11n
Develop a MIMO channel model for 802.11
channels
z Applicable to
Environments where 802.11 systems are to be
used
Both 2.4GHz and 5GHz bands
z Approach
From narrowband to broadband
From SISO to MIMO
MIMO channels: Narrowband
TRANSMITTERS RECEIVERS
1
2
M
1
2
N
y = Hx + n
x
j
: transmitted signal from tx j
y
i
: received signal on rx i
n
i
: noise on rx i
h
ij
: From tx j to rx i (h
ij
C)
h
11
h
12
h
1M
h
21
h
22
h
2M
h
N1
h
N2
h
NM
I R
2
=
n n
802.11n Channel model
z SISO channel models (Medbo 98):
Tap delay line model for various envts
z MIMO channel models (Erceg et al 03):
Correlation-based model
Clustering in
Time (Saleh-Valenzuela)
Angle (AoA and AoD)
Additional parameters in 802.11n MIMO
channel models
z Local signal statistics (Ricean/ Rayleigh)
z Polarization
z Doppler spectrum
z Power roll-off law
Interdependence of parameters
z ds AS
The distribution is always Laplacian
Value selected to match experimental results
z d More Rayleigh than Ricean
(LOS for d< d
BP
, NLOS for d>d
BP
)
z ds K
Value selected to match experimental results
SISO Channel Models
250ns No
Large open
space
E
140ns YES
Large open
space
D
150ns No
Large open
space
C
100ns No
Large open
space/ Office
B
50ns No Office A
Delay spread LOS Environment Model
MIMO Channel Models
Large open space
Large open space/ Office
Office
Residential/ Small office
Residential
Narrowband
Environment
150ns
LOS(K=6)/
NLOS
F (C)
100ns
LOS(K=6)/
NLOS
E (B)
50ns
LOS(K=3)/
NLOS
D (A)
30ns
LOS(K=0)/
NLOS
C
15ns
LOS(K=0)/
NLOS
B
0ns LOS/NLOS A
Delay spread
Scattering
situation
Model
From SISO to MIMO
SISO channel
(Clustering in time)
0
1 , l

0
3 , l

0
2 , l

1 , l

2 , l

3 , l

MIMO channel
(Clustering in time & angle)
From SISO to MIMO
From SISO to MIMO
From SISO to MIMO
From SISO to MIMO
1 , l

0
1 , l

0
2 , l

2 , l

Tap 4
0
1 , l

0
2 , l

1 , l

2 , l

Tap 5
Additional parameters (I)
d
BP
z Local signal statistics
(LOS for d< d
BP
, NLOS for d>d
BP
)
z Polarization
LOS: 3dB X-pol discrimination
NLOS: 10dB X-pol discrimination
Additional parameters (II)
z Doppler spectrum
Bell shaped
Possibly:
Spike due to moving vehicle
Effect of fluorescent ligths
2
1
1
) (
|
|
.
|

\
|
+
=
d
f
f
A
f S
Additional parameters (III)
z Power roll-off law
Exponential power roll-off
Log normal distribution
Log normal variance depends on
The distance from the TX
The environment
( )
( )
( )

>
|
|
.
|

\
|
+
<
=
BP
BP
BP FS
BP FS
d d
d
d
d L
d d d L
d L
, log 5 . 3
,
10
Simulation methodology
z Define parameters
M, N
Type of environment
Distance from source
z For each tap
Calculate R
Tx
and R
RX
Generate independent samples & filter through R
Tx
and R
RX
Add LOS component if there is such
Filter through Doppler filter
Limitations of the model
z The model can be used
for any array geometry
z Simulation software
free
z Angular parameters are
hard-wired (both for
LOS and scattered
components)
z Change from LOS to
NLOS is more gradual
The reality

Das könnte Ihnen auch gefallen