Upload
others
View
7
Download
1
Embed Size (px)
Citation preview
Ghassan Dahman / Fredrik Tufvesson
Department of Electrical and Information Technology
Lund University, Sweden
Channel Modelling – ETIN10
Lecture no:
2014-02-14 Fredrik Tufvesson - ETIN10 1
6
Channel models
Content
• Modelling methods
• Okumura-Hata path loss model
• COST 231 model
• Indoor models
• Wideband models
• COST 207 (GSM model)
• ITU-R model for 3G
• Directional channel models
• Multiantenna (MIMO) models
• Ray tracing & Ray launching
2014-02-14 Fredrik Tufvesson - ETIN10 2
2014-02-14 Fredrik Tufvesson - ETIN10 3
Channel measures
2014-02-14 Fredrik Tufvesson - ETIN10 4
Modeling methods
• Stored channel impulse responses
– realistic
– reproducible
– hard to cover all scenarios
• Deterministic channel models
– based on Maxwell’s equations (or approximations: ray tracing)
– site specific
– computationally demanding
• Stochastic channel models
– describes the distribution of the field strength etc
– mainly used for design and system comparisons
– example: the Rayleigh-fading model
2014-02-14 Fredrik Tufvesson - ETIN10 5
Narrowband models
Review of properties
Narrowband models contain ”only one” attenuation, which is modeled
as a propagation loss, plus large- and small-scale fading.
Large-scale fading: Log-normal distribution (normal distr. in dB scale)
Small-scale fading: Rayleigh, Rice, Nakagami distributions ...
(of amplitudes and not in dB-scale)
Path loss: Often proportional to 1/dn, where n is the propagation
exponent (n may be different at different distances).
2014-02-14 Fredrik Tufvesson - ETIN10 6
Okumura’s measurements
Extensive measurement campaign in Japan in the 1960’s.
Parameters varied during measurements:
Frequency
Distance
Mobile station height
Base station height
Environment
100 – 3000 MHz
1 – 100 km
1 – 10 m
20 – 1000 m
medium-size city, large city, etc.
Propagation loss is given as median values (50% of the
time and 50% of the area).
Results from these measurements are displayed in figures
7.12 – 7.14 in the appendix.
2014-02-14 Fredrik Tufvesson - ETIN10 7
Okumura’s measurements
excess loss
Excess loss [dB
]
Frequency [MHz]
Dis
tan
ce
[km
]
These curves
are only for
hb=200 m and
hm=3 m
900 MHz and
30 km distance
FIGURE 7.12 in appendix
2014-02-14 Fredrik Tufvesson - ETIN10 8
The Okumura-Hata model
Background
In 1980 Hata published a parameterized model, based on Okumura’s
measurements.
The parameterized model has a smaller range of validity than
the measurements by Okumura:
Frequency
Distance
Mobile station height
Base station height
150 – 1500 MHz
1 – 20 km
1 – 10 m
30 – 200 m
It doesn’t encompass the 1800 MHz frequency range. This problem
was solved by the COST 231-Hata model.
2014-02-14 Fredrik Tufvesson - ETIN10 9
The Okumura-Hata model
How to calculate prop. loss
|logO H kmL A B d C
Small/medium-
size cities
Metropolitan
areas
Suburban
environments
Rural areas
0|
0|
1.1log 0.7
1.56log 0.8
MHz m
MHz
f h
f
2
0
2
0
8.29 log 1.54 1.1 for 200 MHz
3.2 log 11.75 4.97 for 400 MHz
m
m
h f
h f
ma h
0
0
2
0|2 log / 28 5.4MHzf
2
0| 0|4.78 log 18.33log 40.94MHz MHzf f
C
0|69.55 26.16log 13.82log
44.9 6.55log
MHz b m
b
A f h a h
B h
hb and hm
in meter
2014-02-14 Fredrik Tufvesson - ETIN10 10
The COST 231-Walfish-Ikegami model
The Okumura-Hata model is not suitable for micro cells or small
macro cells, due to its restrictions on distance (d > 1 km).
The COST 231-Walfish-Ikegami model covers much smaller
distances, and it is better suitable for calculations on small cells and covers
the 1800 MHz band as well.
Frequency
Distance
Mobile station height
Base station height
800 – 2000 MHz
0.02 – 5 km
1 – 3 m
4 – 50 m
2014-02-14 Fredrik Tufvesson - ETIN10 11
The COST 231-Walfish-Ikegami model
How to calculate prop. loss
0 msd rtsL L L L
Free space Roof-top to street Building multiscreen
Assumptions: - a Manhatten grid, constant building height, and a flat tarrain.
- the effect of waveguiding through street canyons is not included
L0 is a function of d (KM),f0(MHz)
Lmsd is a function of Δhb, d, f0, b
Lrts is a function of w, f0, Δhm, the orientation of the street
Details about calculations can be found in the appendix.
2014-02-14 Fredrik Tufvesson - ETIN10 12
Motley-Keenan indoor model
PL PL0 10n logd/d0FwallFfloor
For indoor environments, the attenuation is heavily affected by the
building structure. Walls and floors play an important rule.
distance dependent
path loss sum of attenuations
from walls, 1-20
dB/wall
sum of attenuation from the
floors (often larger than wall
attenuation)
• Site specific => requires the location of BS, MS, and the building plan.
• Neglects propagation paths that go around the walls (through corridors)
2014-02-14 Fredrik Tufvesson - ETIN10 13
Wideband models
(Tapped Delay Line Models)
• Tapped delay line model often used
• Often Rayleigh-distributed taps, but might include LOS
and different distributions of the tap values
• Mean tap power determined by the power delay profile
• Popular cases: N=2, no LOS => the two-path channel, the simplest delay-dispersion model
LOS + one fading tap => widely used for satellite channels
1
, expN
i i i
i
h t t j t
2014-02-14 Fredrik Tufvesson - ETIN10 14
Models for the Power Delay Profile
• Often described by a single exponential decay
• though often there is more than one “cluster”
( / ) 0( )
0sc
exp SP
otherwise
0,
,
( ) 0( )
0
cck
sc kck k
PP
SP
otherwise
log( ( ))scP
log( ( ))scP
delay spread
where k is the cluster’s index
2014-02-14 Fredrik Tufvesson - ETIN10 15
arrival time
• If the bandwidth is high, the time resolution is large so we
might resolve the different multipath components
• Need to model arrival time
• The Saleh-Valenzuela model:
• The D-K-model:
hl0
L
k0
K
k,lTlk,l
cluster arrival time (Poisson)
ray arrival time (Poisson)
S1 S2
arrival rate: 0 ( )t
0 ( )K t
2014-02-14 Fredrik Tufvesson - ETIN10 16
Standardized Channel Models
COST 207 model for GSM
The COST 207 model specifies:
FOUR power-delay profiles for different
environments.
FOUR Doppler spectra used for different
delays.
IT DOES NOT SPECIFY PROAGATION LOSSES FOR THE
DIFFERENT ENVIRONMENTS!
2014-02-14 Fredrik Tufvesson - ETIN10 17
Wideband models
COST 207 model for GSM
[ ]s
[ ]P dB
0
10
20
300 1 [ ]s
[ ]P dB
0
10
20
300 1 2 3 4 5 6 7
[ ]s
[ ]P dB
0
10
20
300 5 10 [ ]s
[ ]P dB
0
10
20
30100 20
Four specified power-delay profiles
RURAL AREA TYPICAL URBAN
BAD URBAN HILLY TERRAIN
2014-02-14 Fredrik Tufvesson - ETIN10 18
Wideband models
COST 207 model for GSM
Four specified Doppler spectra
max max0
,s iP
max max0
max max0 max max0
CLASS GAUS1
GAUS2 RICE
0.5 si 0.5 s 2 si
2 si Shortest
path in
rural areas
,s iP
,s iP ,s iP
2014-02-14 Fredrik Tufvesson - ETIN10 19
GAUS2 GAUS1 CLASS
Wideband models
COST 207 model for GSM
[ ]s
[ ]P dB
0
10
20
300 1 [ ]s
[ ]P dB
0
10
20
300 1 2 3 4 5 6 7
[ ]s
[ ]P dB
0
10
20
300 5 10 [ ]s
[ ]P dB
0
10
20
30100 20
RURAL AREA TYPICAL URBAN
BAD URBAN HILLY TERRAIN
Doppler spectra:
First tap
RICE
here
2014-02-14 Fredrik Tufvesson - ETIN10 20
GAUS2 GAUS1 CLASS
Wideband models
COST 207 model for GSM
[ ]s
[ ]P dB
0
10
20
300 1 [ ]s
[ ]P dB
0
10
20
300 1 2 3 4 5 6 7
RURAL AREA TYPICAL URBAN
Doppler spectra:
First tap
RICE
here
2014-02-14 Fredrik Tufvesson - ETIN10 21
Wideband models
ITU-R model for 3G
The ITU-R model specifies:
SIX different tapped delay-line channels for
three different scenarios (indoor, pedestrian, vehicular).
TWO channels per scenario (one short and one
long delay spread).
TWO different Doppler spectra (uniform & classical),
depending on scenario.
THREE different models for propagation loss (one
for each scenario).
The standard deviation of the log-normal shadow
fading is specified for each scenario.
The autocorrelation of the log-normal shadow
fading is specified for the vehicular scenario.
2014-02-14 Fredrik Tufvesson - ETIN10 22
Wideband models
ITU-R model for 3G ns
2014-02-14 Fredrik Tufvesson - ETIN10 23
Directional channel models
• The spatial domain can be used to increase the spectral
efficiency of the system
– Smart antennas
– MIMO systems
• Need to know directional properties
– How many significant reflection points?
– Which directions?
– Model incoming angle (direction of arrival) and outgoing angle
(direction of departure) to scatterers
• Model independent from specific antenna pattern
2014-02-14 Fredrik Tufvesson - ETIN10 24
Double directional impulse response
ht, r TX, r RX,,, 1
Nr
ht, r TX, r RX,,,
TX position RX position
delay direction-of-departure
direction-of-arrival
ht, r TX, r RX,,, |a|e j
number of multipath components
for these positions
2014-02-14 Fredrik Tufvesson - ETIN10 25
Physical interpretation
lW
BS Y MS
2014-02-14 Fredrik Tufvesson - ETIN10 26
Angular spread
l
double directional delay power spectrum
angular delay power spectrum
angular power spectrum
power
Es, ,,s, ,, Ps, ,,
DDDPS, ,Ps ,,,d
ADPS,DDDPS ,,GMSd
APSAPDS,d
P APSd
2014-02-14 Fredrik Tufvesson - ETIN10 27
Geometry-based stochastic channel models
• Assign positions for scatterers
according to given distributions
• Derive impulse response given
the scatterers and distributions
for the signal properties.
• Used in the COST 259
model, COST 273,
COST 2100, WINNER
3GPP/3GPP2
28
Geometry-Based Stochastic Channel Model
(GSCM)
BS
MS 1
Cluster
Local cluster
Local cluster
Cluster
MS 2
• Create an ”imaginary” map for radio wave scatterers
(clusters)
2014-02-14 Fredrik Tufvesson - ETIN10
Courtesey: K. Haneda, Aalto Uni.
2014-02-14 Fredrik Tufvesson - ETIN10 29
MIMO channel models
• Channel matrix
• Signal model
ℎ 𝜏 =
ℎ11(𝜏) ℎ12(𝜏)ℎ21(𝜏) ℎ22(𝜏)
⋯ ℎ1𝑀𝑇𝑥(𝜏)
⋯ ℎ2𝑀𝑇𝑥(𝜏)
⋮ ⋮ℎ𝑀𝑅𝑥1(𝜏) ℎ𝑀𝑅𝑥2(𝜏)
⋱ ⋮⋯ ℎ𝑀𝑅𝑥𝑀𝑇𝑥(𝜏)
𝑦 𝑡 = ℎ 𝜏 𝑥(𝑡 − 𝜏)
𝜏
• Kronecker model 𝐻 =1
𝐸 𝑡𝑟 𝐻𝐻∗𝑅𝑅𝑥1/2𝐺𝐺𝑅𝑇𝑥1/2
2014-02-14 Fredrik Tufvesson - ETIN10 30
Deterministic modeling methods
• Solve Maxwell’s equations with boundary conditions
• Problems:
– Data base for environment
– Computation time
• “Exact” solutions
– Method of moments
– Finite element method
– Finite-difference time domain (FDTD)
• High frequency approximation
– All waves modeled as rays that behave as in geometrical optics
– Refinements include approximation to diffraction, diffuse scattering,
etc.
2014-02-14 Fredrik Tufvesson - ETIN10 31
Ray launching
• TX antenna sends out rays in different directions
• We follow each ray as it propagates, until it either
– Reaches the receiver, or
– Becomes too weak to be relevant
• Propagation processes
– Free-space attenuation
– Reflection
– Diffraction and diffuse scattering:
each interacting object is source
of multiple new rays
• Predicts channel in a whole
area (for one TX location)
2014-02-14 Fredrik Tufvesson - ETIN10 32
Ray tracing
• Determines rays that can go
from one TX position to one
RX position
– Uses imagining principle
– Similar to techniques known
from computer science
• Then determine attenuation
of all those possible paths
Example: Ray tracing
2014-02-14 Fredrik Tufvesson - ETIN10 33
Required base
station power to
connect to a
WCDMA cell phone.
Example from
Stuttgart. Courtesey:
Awe-communications
Example: Ray tracing
2014-02-14 Fredrik Tufvesson - ETIN10 34
Coverage for a
WCDMA cell phone.
Example from Stuttgart.
Courtesey: Awe-
communications
Propagation Models