Radio Propagation and Network Planning

Radio Propagation and Network Planning
Multipath Propagation
Cairo September 2008

Sergio Baldanza
Sergio.baldanza@nsn.com
Spectral Bandwidth of the digital signal
Spectrum of a rectangular binary Signal
Higher Bandwidth efficiency achieved by : Nyquist filtering

: Multiple phase modulation procedure
Page 2
Nyquist filtering – Inter-symbol interference
Nyquist Bandwidth
BNY = 0.5 x fbit
Page 3
Cosine – roll-off filtering
•An ideal rectangular channel cannot

be achieved
•The signal can be reconstructed
without ambiguities if the spectral
function has half of the amplitude at
the Nyquist bandwidth position.
•Cosine roll-off filters
B = (1 + r) * BN
r = roll-off factor
BN = Nyquist bandwidth
Page 4
Reduction of the bandwidth - Modulation
Bandwidth is directly
proportional to the Symbol Rate
Complexity of the modulation

increases the criticity of the
system
Special equalization measures

are required
Page 5
Eye diagram
Eye diagram for :
a) 4 QAM modulation
b) 16 QAM modulation
c) 128 QAM modulation
Roll off factor = 0.4
Page 6
Received Signal Level - General
Received signal level, affected by small-scale and large-scale fluctuations
Page 7
Atmosphere refraction – Snell’s law
Snell’s law : n1 x sin Θ1 = n2 x sin Θ2
An EM ray entering a medium with different refraction index “n”

is reflected and refracted according to the Snell’s law
Page 8
Received Signal Level - General
Besides the dominant component,

the Rx antenna receives a large
number of reflected and scattered
waves. The Rician K-factor is
defined as the ratio of signal
power in dominant component
over the local-mean power of any
other component. For K
decreasing to zero, the Rician
model reduces to the Rayleigh
model
Page 9
Multipath propagation - origin
Page 10
Rx level vs Time and Frequency
Page 11
Impairements due to multipath propagation
Page 12
Page 13
Multipath fading events
Page 14
Multipath – Statistical distribution
From the Rx field recordings the

cumulative statistical distributions
are obtained.
The median value of Rx field is

assumed as reference
Normally monthly distributions are

considered
Page 15
Multipath – Statistical distribution
Page 16
Statistical distribution of fadings
For attenuation > 15/20 dB

the variations follow the :
Log (p) = - P0 x A
To which corresponds a
Rayleigh distribution of the
Rx field
Page 17
Concept of P0 – Multipath occurrence factor
P0 A0
Page 18
Multipath occurrence factor – P0
Page 19
Multipath occurrence factor - examples
Page 20
P0 calculation methods
General formula : P0 = K x Q x Fa x Lb
Formula from Bell Labs ( Rep. 338 ) Traditional

This formula is the result of field measurements in th US,
reported by A. Vigants and W.T. Barnett in several papers
Formula from ITU-R P.530-8

This formula is based on the concept of PL (probability
Widely used
that the gradient is lower than -100 Nunit/km) and on
various terrain coefficients
Formula from ITU-R P.530-9

This formula is based on the concept of ΔN not exceeded Official ITU, but …..
for 1% of the year and area roughness coming from a
geographic database
Page 21
P0 formula and Outage – Rec. ITU-R P.530-8
Geoclimatic and terrain factor
K = 5.0 10-7 x 10-0.1(Co-CLAT-CLON) x PL1.5 Where :
PL = Percentage of time that the refractivity gradient in the lowest

100 m of the atmosphere is more negative than –100 N units/km
Path inclination
 p  hr – he /d
Outage
–1.4
p w  Κ d 3.6 f 0.89  1   p   10 – A / 10 %
 
Page 22
P0 formula and Outage – Rec. ITU-R P.530-9
Geoclimatic and terrain factor
3.9  0.003dN1
K  10 sa  0.42 Where :
dN1 = The point refractivity gradient in the lowest 65 m of the

atmosphere not exceeded for 1% of an average year
Sa = The standard deviation of terrain heights (m) within a

110 km x 110 km area with a 30 s resolution
Outage
pw  Kd 3.2 (1  | p | ) 0.97 10 0.032 f  0.00085hL  A / 10 %
Page 23
Effect of multipath distortion
Page 24
Page 25
Effects of multipath on Digital Radio Systems
The flat attenuation reduces :
• The S/N ratio at the Receiver

• The C/I ratio in case of un-correlated Interferences
The linear distortion causes :
• Increasing of the inter-symbol interference

• Carrier phase shift, with consequent interference
between orthogonal signals I and Q
The increase of BER is the first effect of these

phenomena
To analytically estimate the probability to have a value of BER
higher than a fixed value, It is necessary to define a channel model
Page 26
Multipath Channel Model
The application of a suitable Channel Model permits to predict the

reduced quality percentage of time of the radio links due to
the presence of multipath.
It is necessary to define :
• A mathematical function H ( jω ) well representing the Channel Model
•A joint probability distribution of the H ( jω ) parameters, conditioned

to the multipath occurrence
•A suitable scale factor representing the occurrence probability in the

reference time period ( normally one month)
Page 27
The most used Model is a symplied 2 rays model
The Channel is characterized by two rays with

amplitude 1 and a1 respectively.
The second is delayed in respect
to the first of τ1
If τ1 is small (2πB τ1 <<< 1), with B as the channel

bandwidth, the transfer function can be
approximated as :
H ( jω ) = a ( 1- b e- j (ω – ω0 ) τ ) with τ =~ τ1
The Transfer function is a Cycloid with minima in :

Notch frequencies = ω0 + 2kπ / τ
Page 28
Multipath channel model - Cycloid
Page 29
“Flat” contribution
The probability that the signal is lower than a “Threshold”

depends on the Flat Fade Margin of the system :
 FM10th
Pf  P0  10
“Selective contribution”
The selective contribute is calculated by solving the following

triple integral :
Ps     p(b, f 0 , ) db df 0 d
S
Page 30
Statistical distribution of the parameters

Page 31
Calculation methods - multipath
Different methods exist, some recognized by International

Bodies (ITU – ETSI), some others used by single Operators or
Administrators.
Two main families :
a) Methods based on the channel model
b) Methods based on parametric models

(They are based on the measurement of appropriate parameters
of fading , whose statistic is correlated with the statistic of
the link behavior.)
These methods are generally more precise, but their application
with difficulty applies to general scenarios.
Page 32
Frequency selective fadings
Page 33
Effects of frequency selective fadings
Page 34
Signature
Page 35
Signature measurement
Page 36
Signature measurement
Page 37
Vectorial addition of two signals
Page 38
Vectorial addition of two signals
Page 39
Minimum phase and non minimum phase
•The distintion is based on the

group delay behaviour
In the two rays models depends on the

reference :
If the reference is made to the ray delay :

Minimum phase rotation
Τ>0
0 < b < 1 - minimum phase
1 < b < inf. – non minimum phase
If the ref. is made to the ray amplitude :
B<1
Τ > 0 - minimum phase
Τ < 0 – non minimum phase Non-Minimum phase rotation
Page 40
Signature
Page 41
Depolarization due to multipath
Page 42
Cross Polarization - definitions
XPI = 20 log E11 / E21 - XPD = 20 log E11 / E12
Page 43
Depolarization due to multipath
Antenna diagrams XPD degr. mechanisms

Page 44
XPD degradation – experimental results
The joint equi-probability

distribution of XPD and the
flat copolar attenuation CPA
is shown in the figure
The median curve of the

distribution is assumed
as the model of the
phenomenon
Page 45
Multipath XPD degradation model
For deep fading,

50
XPD = C – CPA C
40
Following Rec. 530,

30
C=XPDg + 5 + Q.
XPD (dB)
20
Where XPDg is the minimum XPD 10

of the antennas ca 7 dB
FM
0
 k  
Q  10 log xp 
 P0  -10
0 10 20 30 40 50
CPA (dB)
Q is a coefficient defined
In the Recommendation
Page 46

Radio Propagation and Network Planning

Uploaded by

Copyright:

Available Formats

Radio Propagation and Network Planning

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Radio Propagation and Network Planning

Uploaded by

Copyright:

Available Formats

Radio Propagation and Network Planning

Cairo September 2008

Spectrum of a rectangular binary Signal

Higher Bandwidth efficiency achieved by : Nyquist filtering

•An ideal rectangular channel cannot

•Cosine roll-off filters

Complexity of the modulation

Special equalization measures

Eye diagram for :

Roll off factor = 0.4

Received signal level, affected by small-scale and large-scale fluctuations

Snell’s law : n1 x sin Θ1 = n2 x sin Θ2

An EM ray entering a medium with different refraction index “n”

Besides the dominant component,

From the Rx field recordings the

The median value of Rx field is

Normally monthly distributions are

For attenuation > 15/20 dB

Formula from Bell Labs ( Rep. 338 ) Traditional

Formula from ITU-R P.530-8

Formula from ITU-R P.530-9

Geoclimatic and terrain factor

K = 5.0 10-7 x 10-0.1(Co-CLAT-CLON) x PL1.5 Where :

PL = Percentage of time that the refractivity gradient in the lowest

Geoclimatic and terrain factor

dN1 = The point refractivity gradient in the lowest 65 m of the

Sa = The standard deviation of terrain heights (m) within a

pw  Kd 3.2 (1  | p | ) 0.97 10 0.032 f  0.00085hL  A / 10 %

The flat attenuation reduces :

• The S/N ratio at the Receiver

The linear distortion causes :

• Increasing of the inter-symbol interference

The increase of BER is the first effect of these

The application of a suitable Channel Model permits to predict the

• A mathematical function H ( jω ) well representing the Channel Model

•A joint probability distribution of the H ( jω ) parameters, conditioned

•A suitable scale factor representing the occurrence probability in the

The most used Model is a symplied 2 rays model

The Channel is characterized by two rays with

If τ1 is small (2πB τ1 <<< 1), with B as the channel

The Transfer function is a Cycloid with minima in :

The probability that the signal is lower than a “Threshold”

The selective contribute is calculated by solving the following

Statistical distribution of the parameters

Different methods exist, some recognized by International

Two main families :

a) Methods based on the channel model

b) Methods based on parametric models

•The distintion is based on the

In the two rays models depends on the

If the reference is made to the ray delay :

If the ref. is made to the ray amplitude :

XPI = 20 log E11 / E21 - XPD = 20 log E11 / E12

Antenna diagrams XPD degr. mechanisms

The joint equi-probability

The median curve of the

For deep fading,

Following Rec. 530,

Where XPDg is the minimum XPD 10