This action might not be possible to undo. Are you sure you want to continue?
RADIOWAVE PROPAGATION
This chapter is a presentation of the basic principles and algorithms related to radiowave
propagation used in radiorelay transmission. Both loss and attenuation algorithms as well as
fade prediction models for different fading mechanisms are presented and discussed. The
chapter also includes a presentation of the basic concepts of main propagation mechanisms,
Fresnel zone, equivalent and true Earth radii and the decibel scale. Diversity, hardware failure
and passive repeaters are also presented.
TABLE OF CONTENTS
1 The decibel 1
1.1 A relative comparison 1
1.2 Some motivations for using decibels 1
1.3 Absolute comparisons 1
1.4 The comparison of field quantities 2
1.5 Power and field quantity ratios 3
2 The main propagation mechanisms 3
2.1 Propagation along the earth’s surface 4
3 The Fresnel zone and clearance 4
3.1 Definition 4
3.2 The Fresnel ellipsoid 4
3.3 Clearance 5
4 Equivalent and true earth radii 6
4.1 Earthradius factor 6
4.2 Comparing the equivalent and true Earth surface 7
5 Prediction models 7
6 The prediction cycle 8
7 The loss/attenuation block 8
7.1 Freespace loss 8
7.1.1 Definition 8
7.1.2 Freespace loss between two isotropic antennas 9
7.2 Atmospheric gases 10
7.2.1 Definition 10
7.2.2 The troposphere 11
7.2.3 Chemical composition 11
7.2.4 Absorption peaks 11
7.2.5 Calculating total gas attenuation 11
7.2.5.1 Oxygen (dry air) 12
7.2.5.2 Water vapor 13
7.2.5.3 Total gas attenuation 15
7.3 Reflection 16
7.3.1 Ground reflection interference 16
7.3.2 The problems of handling reflection 17
7.3.3 Reflection coefficient 18
7.3.4 The Fresnel reflection coefficient 18
7.3.5 Divergence factor 18
7.3.6 Correction factor 19
7.3.7 Rough estimation of the total reflection coefficient 19
7.3.8 Calculation of the position of the reflection point 20
7.3.9 Optimum height difference 22
II
7.4 Precipitation 23
7.4.1 Types of precipitation 23
7.4.2 Precipitation: now 24
7.4.3 Precipitation: hail 24
7.4.4 Precipitation: fog and haze 25
7.4.5 Precipitation: rain 25
7.4.6 Cumulative distribution of rain 25
7.4.7 Obtaining Rain Intensity (the former ITUR model) 25
7.4.8 Rain zones  diagram 26
7.4.9 Obtaining rain intensity (current ITUR model) 26
7.4.10 The calculation of the specific rain attenuation 29
7.4.11 Calculating total rain attenuation 32
7.4.12 Calculating total rain attenuation for 0.01% 33
7.5 Obstruction  diffraction 33
7.5.1 Definition 33
7.5.2 Knifeedge obstructions 34
7.5.3 Knifeedge loss curve 35
7.5.4 Typical knifeedge losses 36
7.5.5 Singlepeak method 36
7.5.6 Triplepeak method 37
7.5.7 Smoothly spherical earth 40
7.5.8 Typical losses resulting from smoothly spherical earth 41
7.5.9 Clearance and path geometry 41
7.5.9.1 The Earth bulge 41
7.5.9.2 Path geometry 42
7.5.9.3 The height of the lineofsight 43
7.5.9.4 Clearance of the LOS 43
7.5.9.5 Antenna height 43
7.5.9.6 Obstacle penetration 44
7.5.10 Vegetation 44
7.6 The Link Budget 44
7.6.1 Path loss 45
7.6.2 Fade margin 45
7.6.3 Power diagram 46
7.6.4 Effective fade margin 46
8 The fading block 47
8.1 Definition 47
8.2 General cause 47
8.3 General classification 47
8.4 Classification based on source 48
8.5 The concept of outage 48
8.6 Rain fading (current ITUR model) 48
8.6.1 Calculation of the fade margin based on a yearly basis 48
8.6.2 Outage due to rain fading  annual basis 49
8.6.3 Transformation between yearly and worst month basis 50
8.6.3.1 From yearly basis to worst month 50
8.6.3.2 From worst month to yearly 51
8.6.4 Presentation of the rain fading models in diagram form 51
8.7 Multipath fading 52
8.7.1 Flat and frequency selective fading 52
8.7.2 The effects of multipath propagation 53
8.7.3 Measures taken against multipath fading 54
8.8 Flat fading (former ITUR model): small percentages of time 54
8.8.1 Introduction 54
8.8.2 Fade occurrence factor 55
8.8.3 Flat fading and quality (error performance) 55
8.8.4 Estimation of the geoclimatic factor 55
8.8.5 Inland Links 55
8.8.5.1 Antenna altitude coefficient 56
8.8.5.2 Latitude coefficient 57
III
8.8.5.3 Longitude coefficient 57
8.8.5.4 Climatic factor p
L
57
8.8.6 Coastal Links 58
8.8.6.1 Coastal links over/near large bodies of water 58
8.8.6.2 Coastal links over/near mediumsized bodies of water 59
8.8.6.3 Links at other regions 59
8.8.7 Link and terrain parameters – overview 60
8.8.7.1 Estimation of the path inclination 60
8.8.8 Outage due to flat fading 61
8.8.9 Range of values for the climatic factor p
L
61
8.9 Flat fading (current ITUR model): small percentages of time 62
8.9.1 Method for detailed link design 62
8.9.1.1 Geoclimatic factor 62
8.9.1.2 Parameters 62
8.9.1.3 Outage due to flat fading 64
8.9.2 Method for quick link design 64
8.9.2.1 Geoclimatic factor 64
8.9.2.2 Outage due to flat fading 65
8.9.3 Method for small percentage of time  conclusion 65
8.9.4 Method for all percentages of time 66
8.9.5 Range of validity for the flat fading method 68
8.10 Reduction of crosspolar discrimination 69
8.10.1 XPD outage due to multipath propagation 69
8.10.2 XPD outage due to precipitation 71
8.11 Outage due to frequency selective fading 72
8.11.1 General aspects 72
8.11.2 The prediction model 73
8.12 Refraction fading 75
9 Diversity 76
9.1 Basic concepts 76
9.2 The definition of the improvement factor 77
9.3 Improvement factor for space diversity 78
9.4 Improvement factor for frequency diversity 78
9.5 The calculation of outage when employing diversity 79
9.6 Prediction of outage using diversity 79
9.6.1 Space diversity 79
9.6.2 Frequency diversity 81
9.6.3 Space and frequency diversity with two receivers 81
9.6.4 Space and frequency diversity with four receivers 81
10 Prediction of total outage 82
11 Hardware failure 83
11.1 The calculation of the radiolink system’s MTBF 83
11.2 Nonredundant systems 84
11.3 Redundant systems 85
11.4 Hardware failure per path 86
12 Passive repeaters 87
12.1 The basic concepts 87
12.2 Path calculation when using passive repeaters 88
IV
RADIOWAVE PROPAGATION
Ericsson AB 1
4/038 02LZU 102 152, Rev B, June 2003
1 The decibel
1.1 A relative comparison
It is usual in radio technique that two different values or entities are
compared with one another. The comparison between two levels
(expressed in power) by taking the ratio is a classical example. The
decibel is a measure of the relationship between two power levels.
Decibel is abbreviated as dB and is defined as follows
[ ]
2
1
10
log . 10 dB
P
P
A =
(1)
where P
1
and P
2
are the power levels being compared.
Note that the decibel is a measure of a relationship and has no actual
physical significance. The decibel is therefore not a measure of a
physical entity.
One decibel corresponds approximately to the smallest variation in
sound volume that can be discerned by the human ear.
1.2 Some motivations for using decibels
Some of the motivations behind the widespread use of the decibel are:
• The decibel is convenient to use since the direct relationship
between radiorelated power levels covers a wide range of
numerical values. The logarithmic nature of the relationship
between two powerlevels results in values that is easy to handle.
• Addition or subtraction operations can be easily performed on
logarithmic values, simplifying the handling of amplification and
attenuation.
• The manner in which human sensory organs perceive differences in
the sensory impressions of varying intensity that they receive is in
fact logarithmic.
1.3 Absolute comparisons
The decibel concept defined above is related to the quotient of two
values, and provides no information as to the absolute value of these
entities. An absolute comparison between two power levels can
however be performed if a reference value is employed, for example the
W (Watt) or mW (milliWatt), referred to respectively as dBW and
dBm.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
2 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
[ ]
Watt 1
log . 10 dBW
10
P
A =
(2)
where P is the power in Watt.
[ ]
milliWatt 1
log . 10 dBm
10
P
A =
(3)
where P is the power in milliWatt.
Since,
10
dBW
10
W 1
=
P
(4)
And
10
dBm
10
mW 1
=
P
(5)
the result obtained following division is

.

\

=
10
dBm  dBW
10
W 1
mW 1
(6)
Or

.

\

−
=
10
dBm  dBW
3
10
W 1
W 1
(7)
Giving
10
dBm  dBW
3 = −
(8)
Or
30 dBW dBm + =
(9)
1.4 The comparison of field quantities
The decibel concept can be generalized to also include the comparison
between field magnitudes. The term field quantity refers to a quantity
whose square is proportional to power. Examples of field quantities are
electrical voltages, currents and field strengths.
RADIOWAVE PROPAGATION
Ericsson AB 3
4/038 02LZU 102 152, Rev B, June 2003
The application of the decibel concept results in
[ ]
( )
( )
⋅ =
2
1
10
quantity Field
quantity Field
log 20 dB A
(10)
1.5 Power and field quantity ratios
Power and field quantities, lying between 10
3
and 10
3
are expressed in
their equivalent decibel values in Table 1.
Power
ratios
dB Field quantity
ratios
dB
1 000=10
3
30
1 000=10
3
60
100=10
2
20
100=10
2
40
10=10
1
10
10=10
1
20
9
9.5
9
19
8
9
8
18
7
8.5
7
17
6
8
6
16
5
7
5
14
4
6
4
12
3
5
3
9.5
2
3
2
6
1
0
1 0
1/2
3
1/2
6
1/4
6
1/4
12
1/8
9
1/8
18
1/10=10
1
10
1/10=10
1
20
1/100=10
2
20
1/100=10
2
40
1/1000=10
3
30
1/1000=10
3
60
Table 1: Power and field ratios.
2 The main propagation mechanisms
Most of the propagation mechanisms are affected by climactic
conditions. When calculating the transmission quality and availability
of radio networks, the significance of the various mechanisms varies as
a function of the radio spectrum. The following propagation
mechanisms may however be considered as the most notable:
• Freespace
• Diffraction
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
4 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
• Refraction
• Absorption
• Scattering
• Reflection
2.1 Propagation along the earth’s surface
An electromagnetic wave traveling close to and along the surface of the
earth is affected by the following factors:
• Electrical properties of the earth’s surface
• Earth’s curvature
• Atmosphere
• Earth’s topography
• Vegetation
3 The Fresnel zone and clearance
Expressions for the calculation of the earth bulge, the height of the line
ofsight, the clearance of the line of sight, the antenna heights and
obstacle penetration are given in section 7.5.9.
3.1 Definition
Fresnel zones are specified employing an ordinal number that
corresponds to the number of halfwavelength multiples that represents
the difference in radio wave propagation path from the direct path. The
first Fresnel zone is therefore an ellipsoid whose surface corresponds to
one halfwavelength path difference and represents the smallest volume
of all the other Fresnel zones.
The first Fresnel zone contains almost all the energy that is transmitted
between the antennas and is therefore of great significance in the
calculation of the attenuation caused by obstructing bodies.
3.2 The Fresnel ellipsoid
The Fresnel zone is an ellipsoid having its focal points at the antenna
points A and B as illustrated in Figure 1. The radius of the first Fresnel
zone, R, is a function of the distance between A and B, the distance
between any point M on the ellipsoid and the frequency. The radius of
the first Fresnel zone is indirectly proportional to frequency and the
higher the frequency the narrower the Fresnel zones.
RADIOWAVE PROPAGATION
Ericsson AB 5
4/038 02LZU 102 152, Rev B, June 2003
! d
A
= Distance from antenna A to point M, km
! d
B
= Distance from antenna B to point M, km
! R = Radius of the Fresnel zone at point M, m
! d
A
+ d
B
= d = Distance between antennas A and B, km
M
A
B
R
d
B
d
A
! f = Frequency, MHz
( )
d f
d d d
R
A A
⋅
− ⋅
⋅ = 547
( )
d f
d d d
R
A A
⋅
− ⋅
⋅ = 32 17 .
GHz
Figure 1: Fresnel zone between two stations located on an equivalent
earth surface (the ray beam is straight).
3.3 Clearance
The refractive properties of the atmosphere are not constant. The
variations of the refraction index in the atmosphere (expressed by the
earthradius factor k) may force terrain irregularities to totally or
partially intercept the Fresnel zone. Clearance can be described as any
criterion to insure that the antenna heights are sufficient so that in the
worst case of refraction (for which k is minimum!) the receiver antenna
is not placed in the diffraction region.
h
c
h
c
= LOSclearance
Figure 2: The clearance of the line of sight.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
6 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
The direct path between the transmitter and the receiver needs a
clearance above the ground or any obstruction of at least 60% of the
radius of the Fresnel zone in order to insure freespace propagation.
Clearance values have to fit the local climate conditions.
Clearance can be considered by applying “clearance criteria” that are
climate dependent or by properly handling diffractiondiffraction fading
(ktype fading).
Advices:
1) The higher the frequency the smaller the Fresnel zone and
consequently more vulnerable to nonLOS effects (object attenuation).
2) Low kvalues lower the LOS (demand higher antenna heights) but
offer better protection against interference from other stations. Higher
kvalues give higher LOS (demand lower antenna heights) but expose
the link to interference from other stations.
3) The most common discrepancy arises when the radius of the first
Fresnel zone is not compensated for its vertical projection. The more
inclined the path is the more correction is required.
4 Equivalent and true earth radii
4.1 Earthradius factor
In simple terms, one can describe the ray beam between two antennas
by employing an imagined propagation path that directly links the two
antennas. In freespace this path would describe a straight line, a so
called optical lineofsight.
If instead, the antennas are placed on a spherical body surrounded by an
atmosphere (as in the case for the earth), wave propagation will be
affected by variations in atmospheric refractive index as the wave
travels through the various atmospheric layers. The ray beam will now
not follow the optical lineofsight, but will describe a curved line
between the two antennas. The form of the curve will vary as a function
of variations in the refractive index of the atmosphere traversed by the
wave.
To simplify the description of this curved ray beam, the concept of
equivalent earth surface having an equivalent earth radius, R
e
, has been
introduced. Defined as follows:
R k R
e
⋅ =
(11)
RADIOWAVE PROPAGATION
Ericsson AB 7
4/038 02LZU 102 152, Rev B, June 2003
where
k = Earthradius factor
R = True earth radius (6.37·10
6
m)
The earthradius factor is a function of the refractive index gradient. For
normal atmosphere (i.e., atmosphere in which the refractive index
gradient decreases linearly with altitude), the kvalue is 4/3 if the
refractive index gradient is 39 Nunits/km.
NOTE: kvalues are determined by employing the diagram in
Chapter 152 after selecting the appropriate value of the refractive
gradient in Chapter 153.
4.2 Comparing the equivalent and true Earth surface
The equivalent earth surface is that earth surface that would be required
for the ray beam between the antennas to lie along a straight line, see
Figure 3. A beam that travels outside of the optical lineofsight must
bend downwards in order to become a straight line, which is equivalent
to enlarging the earth’s radius, i.e. reducing the curvature of the earth.
The earthradius factor, k, describes exactly the degree to which the
earth’s radius would have to be changed in order that the ray beam
describe a straight line.
True earth surface
Optical lineofsight
True ray beam
R
Equivalent ray beam
Equivalent earth surface
Optical lineofsight
R
e
= k · R
Figure 3: The equivalent and the true earth surface.
5 Prediction models
Prediction models for the purpose of performing fading prognoses are
almost always empirical (comes from the Greek word empeiria
meaning experience), i.e., they are not founded on theoretical
considerations but are only built upon observation and experience.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
8 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
Empirical models are arrived as the result of the application of
mathematical regression techniques on measurement data and therefore
result in a relationship that describes a variable’s dependency under
certain given conditions.
Empirical prediction models often provide a fair description of the
fading process for distances and frequencies that lie within the data
ranges for which measurements have actually been collected. Their
application to other distances and frequency ranges may, on the other
hand, result in significant error.
6 The prediction cycle
Figure 4 of Chapter 2 (RadioRelay Transmission Overview) illustrates
the four blocks of the prediction cycle: loss/attenuation, fading,
frequency planning and quality and availability. In this chapter, two
blocks will be studied, namely the loss/attenuation and the fading
blocks.
7 The loss/attenuation block
The loss/attenuation block is composed of three main contributions:
branching, propagation, and “others”.
The branching contribution comes from the hardware required to
delivery the transmitter/receiver output to the antenna, for instance,
waveguides as well as splitters and attenuators.
The propagation contribution comes from the losses due to the Earth
atmosphere and the terrain, for instance, freespace as well as gas,
precipitation (mainly rain), ground reflection, and obstacle.
The “others” contribution has a somewhat unpredictable and sporadic
character, for instance, sandstorm as well as fog, clouds, smoke, and
moving objects crossing the path. In addition, poor equipment
installation and unsuccessful antenna alignment may give rise to
unpredictable losses. The “others” contribution is normally not
calculated but it can be accounted in the planning process as an
additional loss and then being part of the fade margin.
7.1 Freespace loss
7.1.1 Definition
Freespace wave propagation implies that the effects caused by
disturbing objects and other obstacles that are located at sufficiently
long distances are assumed to be negligible.
RADIOWAVE PROPAGATION
Ericsson AB 9
4/038 02LZU 102 152, Rev B, June 2003
7.1.2 Freespace loss between two isotropic antennas
Electromagnetic waves are attenuated when propagating between two
points geometrically separated from each other. The attenuation is
inversely proportional to the square of distance and gives the freespace
loss that stands for most of the total attenuation caused by wave
propagation effects. Freespace loss is always present and it is
dependent on distance and frequency. The freespace loss between two
isotropic antennas is currently derived from the relationship between
the total output power from a transmitter and the received power at the
receiver. Its value is illustrated in. The resulting expression is
λ
π d
A
bf
⋅ ⋅
⋅ =
4
log 20
(12)
where
A
bf
= Freespace loss, dB
d = Distance from the transmitting antenna, km
λ = Wavelength, m
Following the transformation of wavelength into frequency
(c=2.99792500·10
8
m/s) and entering of the actual units, the following
expression is attained
f d A
bf
log 20 log 20 5 . 92 ⋅ + ⋅ + =
(13)
where
A
bf
= Freespace loss, dB
d = Distance from the transmitting antenna, km
f = Frequency, GHz
The freespace loss (dB) as a function of distance (km) is illustrated in
Figure 4 in the frequency range 1 to 50 GHz.
1GHz
15
5
20
30
40
10
50
0 10 20 30 40 50
Distance, km
170
160
150
140
130
120
110
100
90
80
F
r
e
e

s
p
a
c
e
l
o
s
s
,
d
B
Figure 4: The freespace loss as a function of distance for eight different
frequencies.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
10 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
Advices:
1) If the distance is doubled while maintaining constant frequency, the
freespace loss is increased by 20·log 2= 6 dB. The same applies to a
doubling of the frequency while maintaining a constant distance. In
other words, an additional attenuation of 6 dB will be caused for every
doubling of either the distance or the frequency.
2) Comparing to other kind of loss, freespace loss gives the major
contribution. Expressed in the GHz range, the freespace loss has a
minimum of approximately 92 dB. If it is expressed in the MHz range
the minimum is 92 dB – 60 dB = 32 dB (1 GHz = 1000 MHz → 20·log
1000 = 60 dB).
3) This relatively small increase of freespace attenuation by only 6 dB
with increased distance might give the impression that long paths can
easily be obtained by simply increasing the transmitter output power, or
the receiver sensitivity or the antenna gain. This is not so easy to
accomplish because the total path attenuation is also determined by
other negative contributions, for example gas attenuation.
4) Cellplanners commonly refer to halfwave dipole antenna gains.
Comparing to the above presentation for which the gain of an “ideal”
isotropic antenna is 1 (0 dB), the gain of a halfwave dipole antenna is
1.64 (2.15 dB). Considering both stations of a radio link, the difference
between freespace loss comparison using isotropic and halfwave
dipole antennas is about 4.30 dB.
7.2 Atmospheric gases
7.2.1 Definition
The atmosphere, up to an altitude of 3040 km, consists of two layers:
• Troposphere
• Stratosphere
An often sharply demarcated transition layer referred to as the
tropopause separates the troposphere and stratosphere.
It is within this troposphere in which all weatherrelated processes
(precipitation, cloud formation, electrical storms, etc.) arise.
The troposphere lies at an altitude of 10 km over the earth’s medium
latitudes and somewhat less over its poles. At the equator, the
troposphere lies at an altitude varying between 16 and 18 km above the
earth’s surface.
RADIOWAVE PROPAGATION
Ericsson AB 11
4/038 02LZU 102 152, Rev B, June 2003
7.2.2 The troposphere
The troposphere consists of approximately 9/10 of the earth’s
atmospheric mass, and aside from variations in moisture content,
density and temperature, its constitution is more or less constant
throughout its volume. This layer contains just a few notable elements
and their compounds, which are of significance in the propagation of
radio waves.
7.2.3 Chemical composition
Nitrogen and oxygen molecules account for approximately 99% of the
total volume. From the propagation point of view, it is suitable to
consider the atmosphere as being a mixture of two gases, dry air and
water vapor.
The chemical composition of the earth’s atmosphere is illustrated in
Table 2.
Chemical Composition of the Earth’s Atmosphere, %
N
2
O
2
Ar CO
2
Ne He Kr Xe H
2
78.09 20.93 0.93 0.03 0.00018 5.2⋅10
4
1.0⋅10
4
8.0⋅0
6
<5⋅10
5
Table 2: The chemical composition of the earth’s atmosphere.
7.2.4 Absorption peaks
Water and dry air (oxygen) result in the following absorption peaks:
• Water (H
2
O) displays absorption peaks at the following radio
frequencies: 22,235 GHz, 183,310 GHz and at 323.8 GHz. In
addition, even greater absorption occurs at higher frequencies,
where the propagation of IR and visible light transmission are
primarily affected.
• Oxygen molecules (O
2
) displays absorption peaks at the following
radio frequencies: 5070 GHz (a complex system of absorption
peaks lie in this frequency band), 118.75 GHz and at 367 GHz.
7.2.5 Calculating total gas attenuation
In what follows, the algorithms for the calculation of the specific
attenuation due to oxygen (dry air) and water vapor will be described
stepbystep.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
12 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
7.2.5.1 Oxygen (dry air)
Two atmospheric parameters are involved in the calculation of the
specific attenuation of oxygen: the atmospheric pressure and the
temperature.
The atmospheric pressure is normalized to the value at see level (1013
hPa) by
1013
p
r
p
=
(14)
where r
p
is the normalization factor and p (hPa) the pressure of the
atmosphere at a certain altitude. A “normal atmosphere” is the
atmosphere where the pressure at the see level is 760 mmHg, which
corresponds to 1 atm or 1013.25 hPa. The nonSI unity is bar (100
kPa).
The temperature is normalized to a mean value of 15 °C by
( ) t
r
t
+
=
273
288
(15)
where r
t
is the normalization factor and t is the temperature (°C).
The following parameters are determined:
( ) [ ] 1 1 5663 . 1 5106 . 0 5050 . 0
1
e 7665 . 6
− − ⋅ −
⋅ ⋅ ⋅ =
t
r
t p
r r η
(16)
( ) [ ] 1 1 5496 . 0 8491 . 0 4908 . 0
2
e 8843 . 27
− − ⋅ − −
⋅ ⋅ ⋅ =
t
r
t p
r r η
(17)
( ) 5 . 3 ln
ln
1
2


.

\

=
η
η
a
(18)
1
4
η
a
b =
(19)
( )
( ) [ ]
t
r
t p
O r r
− ⋅ − −
⋅ ⋅ ⋅ =
1 5280 . 2 6032 . 1 4954 . 1 '
e 128 . 2 54 γ
(20)
Finally, the specific attenuation due to oxygen for frequencies equal or
lower than 54 GHz is given by
RADIOWAVE PROPAGATION
Ericsson AB 13
4/038 02LZU 102 152, Rev B, June 2003
( )
( )
3 2
2 2 2
3 2
10
54
54 ' 3429 . 0
36 . 0
34 . 7
−
⋅ ⋅
+ −
⋅ ⋅
+
⋅ ⋅ +
⋅ ⋅
= f
b f
b
r r f
r r
a
O
t p
t p
O
γ
γ
(21)
where f is the frequency and the other parameters are defined earlier.
7.2.5.2 Water vapor
In the calculation of the specific attenuation due to water vapor, one
more atmospheric parameter is required: water vapor content (g/m
3
).
NOTE: Water vapor content can be selected from the charts
included in Chapter 156.
However, in combination with a given temperature, the watervapor
content selected from the charts might not be physically consistent with
the appropriate value correspondent to the vapor saturation pressure. In
other words, the watervapor pressure cannot exceed the vapor
saturation pressure at the temperature considered. To avoid this
common mistake, one more atmospheric parameter has been introduced
in the stepbystep calculation: relative humidity (%).
The vapor saturation pressure, p
s
, is solely dependent on the
temperature and is given by

.

\

+
⋅
⋅ =
97 . 240
502 . 17
e 1121 . 6
t
t
s
p
(22)
The relative humidity of the atmosphere, RH, is given as the ratio
between the water vapor pressure in the atmosphere, p
H2O
, and the
vapor saturation pressure, p
s
.
100
2
⋅ =
s
O H
p
p
RH
(23)
Solving the above expression for the vapor pressure it is obtained
s O H
p
RH
p ⋅ =
100
2
(24)
The water vapor content (watervapor density) can be derived from the
general gas equation. It is given by
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
14 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
15 . 273
7 . 216
2
+
=
t
p
O H
ρ
(25)
The following parameters are determined
ρ ξ ⋅ + ⋅ ⋅ = 0061 . 0 9544 . 0
69 . 0
1 t p w
r r
(26)
ρ ξ ⋅ + ⋅ ⋅ = 0067 . 0 95 . 0
64 . 0
2 t p w
r r
(27)
ρ ξ ⋅ + ⋅ ⋅ = 0059 . 0 9561 . 0
67 . 0
3 t p w
r r
(28)
ρ ξ ⋅ + ⋅ ⋅ = 0061 . 0 9543 . 0
68 . 0
4 t p w
r r
(29)
ρ ξ ⋅ + ⋅ ⋅ = 006 . 0 955 . 0
68 . 0
5 t p w
r r
(30)
( )
( )
2
2
22
235 . 22
235 . 22
1
+
−
+ =
f
f
g
(31)
( )
( )
2
2
557
557
557
1
+
−
+ =
f
f
g
(32)
( )
( )
2
2
557
557
557
1
+
−
+ =
f
f
g
(33)
( )
( )
2
2
752
752
752
1
+
−
+ =
f
f
g
(34)
Finally, the specific attenuation of water vapor for frequencies equal or
lower than 50 GHz is given by
[ ] { }
4 2 5 . 2 5 . 8 3 2 2
10 10 76 . 1 10 13 . 3
− − −
⋅ ⋅ ⋅ + + + + ⋅ + ⋅ ⋅ ⋅ + ⋅ ⋅ ⋅ = ρ ρ γ f E D C B A r r r r
t t t p w
(35)
where
( ) ( )
( )
2
1
2
1 23 . 2
22 1
42 . 9 235 . 22
e 84 . 3
w
r
w
f
g
A
t
ξ
ξ
⋅ + −
⋅ ⋅ ⋅
=
−
(36)
RADIOWAVE PROPAGATION
Ericsson AB 15
4/038 02LZU 102 152, Rev B, June 2003
( ) ( )
( )
( ) ( )
( )
2
3
2
1 4385 . 6
3
2
2
2
1 7 . 0
2
29 . 6 226 . 321
e 078 . 0
48 . 9 31 . 183
e 48 . 10
w
r
w
w
r
w
f f
B
t t
ξ
ξ
ξ
ξ
⋅ + −
⋅ ⋅
+
⋅ + −
⋅ ⋅
=
− −
(37)
( ) ( )
( )
( ) ( )
( )
2
1 09 . 1
5
2
4
2
1 6 . 1
4
380
e 36 . 26
22 . 9 153 . 325
e 76 . 3
−
⋅ ⋅
+
⋅ + −
⋅ ⋅
=
− −
f f
C
t t
r
w
w
r
w
ξ
ξ
ξ
(38)
( ) ( )
( )
( ) ( )
( )
2
1 17 . 0
557 5
2
1 46 . 1
5
557
e 7 . 883
448
e 87 . 17
−
⋅ ⋅ ⋅
+
−
⋅ ⋅
=
− −
f
g
f
D
t t
r
w
r
w
ξ ξ
(39)
( ) ( )
( )
2
1 41 . 0
752 5
752
e 6 . 302
−
⋅ ⋅ ⋅
=
−
f
g
E
t
r
w
ξ
(40)
7.2.5.3 Total gas attenuation
Specific attenuation (dB/km) for water vapor and oxygen (dry air) are
separately calculated and then added together to give the total specific
attenuation.
( ) d A
w O G
⋅ + = γ γ
(41)
where
A
G
= Total gas attenuation, dB
γ
w
= Specific absorption due to the effects of water vapor, dB/km
γ
o
= Specific absorption due to the effects of oxygen (dry air),
dB/km
d = Path length, km
The specific attenuation is strongly dependent of frequency,
temperature and absolute or relative humidity of the atmosphere as is
illustrated in Figure 5.
Figure 5: The total specific atmospheric attenuation.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
16 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
Advices:
1) If gas absorption is calculated as a function of relative humidity and
temperature, be aware both parameters are reciprocally consistent.
2) If local values of temperature are available, select the average
summer temperature.
3) For tropic climate nearby large bodies of water, annual temperature
charts can be employed for the selection of temperature.
4) Frequency bands located close to the strong features of water vapor
(23 GHz) and oxygen (5060 GHz) are strongly affected by attenuation.
Depending on the values of temperature and humidity, the specific
attenuation can be as much as 1 or 2 dB/km. This gives a large negative
contribution to the fade margin, then making the quality and availability
objectives harder to be accomplished. On the other hand, such high
values of specific attenuation are also of benefit because it provides
valuable shielding to cochannel interference.
5) Considering the “atmospheric isolation” described above as a
positive contribution, the use of high frequency systems will improve
the efficiency of spectrum utilization through the enhanced opportunity
for multiple frequency reuse for shortdistance communication systems
operating within the same part of the frequency band.
7.3 Reflection
Reflection on the earth surface may give rise to multipath propagation.
Depending on the path geometry, the direct ray at the receiver may be
interfered with the groundreflected ray and the “reflection loss” can be
significant. Since the refraction properties of the atmosphere are
constantly changing (kvalue changes), the reflection loss varies
(fading).
Due to its fading characteristics, reflection loss is normally not included
in the linkbudget because it may lead to heavy over or under
dimensioning. Some rough estimations of reflection loss as a function
of the total reflection coefficient is described below.
7.3.1 Ground reflection interference
The respective field strength components of the direct and reflected
waves interfere with one another at the receiver. Receiver interference
due to ground reflection is the result of the reception of the resultant
field strength, i.e., the vector addition of the field components.
RADIOWAVE PROPAGATION
Ericsson AB 17
4/038 02LZU 102 152, Rev B, June 2003
Signal strength is dependent on the total reflection coefficient (resulting
from dielectric constant, conductivity and polarization) and the total
phase shift (resulting from antenna height, path length, earthradius
factor, frequency and the phase angle of the reflection coefficient).
Generally, Figure 6 illustrates two extreme cases:
1) How the highest value of signal strength, A
MAX
, varies with the total
reflection coefficient. This case illustrates amplification, i.e., the field
strength components have the exact same direction, a phase angle of 0°.
2) How the lowest value of signal strength, A
MIN
, varies with the total
reflection coefficient. This case illustrates a loss, i.e., the field strength
components are directed opposite to one another, a phase angle of 180°.
Figure 6: The signal strength as a function of the total reflection
coefficient. The highest value of signal strength is obtained for a phase
angle of 0° and the lowest value for a phase angle of 180°.
7.3.2 The problems of handling reflection
The handling of reflection is difficult and complicated, particularly due
to the uncertainties and measurements of the following parameters:
• High frequencies mean short wavelengths (at 23 GHz, the
wavelength ≈ 1.3 cm)
• Terrain data accuracy can affect the total reflection coefficient
which in effect, consists of three factors, of which one is directly
coupled to the degree of irregularity of the terrain
• Antenna height cannot be determined with sufficient accuracy since
the height database has its limitations
• Earthradius factor
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
18 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
7.3.3 Reflection coefficient
The total reflection coefficient for a smooth spherical surface consists
of three elements: Fresnel reflection coefficient, divergence factor and
correction factor.
7.3.4 The Fresnel reflection coefficient
The Fresnel reflection coefficient for a smooth flat surface is
dependent on frequency, grazing angle, polarization and ground
characteristics (from the dielectric and conductivity constant). Figure 7
shows the Fresnel reflection coefficient’s absolute value for sea water
as a function of grazing angle, two different frequencies and both
horizontal and vertical polarization.
Figure 7: The Fresnel reflection coefficient as a function of the grazing
angle for seawater.
7.3.5 Divergence factor
The divergence factor is applied to the Fresnel reflection coefficient
when approximating the earth’s surface as being spherical. Its value is
a function of antenna height, earth radius factor and the path length.
The divergence factor increases as both the difference in antenna
heights, transmitterreceiver, and the value of the earth radius factor
increase  it decreases with hop length (longer distances along the
earth’s surface must be considered as being an arc).
RADIOWAVE PROPAGATION
Ericsson AB 19
4/038 02LZU 102 152, Rev B, June 2003
7.3.6 Correction factor
The correction factor accounts for the surface irregularities (roughness)
in different types of ground formations. Table 3 illustrates the
approximate values of the correction factor for different ground
surfaces at two different frequencies, 1 and 10 GHz.
Groundsurface types
ρ
s
1 GHz
ρ
s
10 GHz
Sea, lake, mirrorface ice field 0.951 0.901
Snow & ice field, frozen soil, naked damp
ground
0.850.95 0.800.90
Damp field, flat and large scale agricultural
and cattle breeding land
0.750.85 0.650.80
Flat grass land, flat field with thin bush,
desert
0.550.75 0.450.65
Gently rolling terrain, savanna, partitioned
plowed fields and pasture
0.350.55 0.250.45
Rolling terrain, forest, thick forest against
sandy wind, wind break, medium or small
city area, area where a bank or a high way
transverses the radio path near the reflection
point
0.180.35 0.090.25
Terrain with outstanding undulation,
undulated forest, medium or small city with
high rise buildings, area with large factories,
stadiums located to transverses the radio
path near the reflection point
0.080.18 0.040.09
Mountainous area, area with a deep ridge to
shield the reflected area
0.040.18 <0.04
Table 3: Approximate values of the correction factor for different
ground surface types.
7.3.7 Rough estimation of the total reflection coefficient
The Fresnel reflection coefficient is very close to 1 for small grazing
angles, regardless of frequency and polarization. Ordinarily, grazing
angles, in connection with radio links, lie between 1 and 10 mrad which
is equivalent to 1/1000 and 1/100, respectively, of the relationship
between the antenna height and the hop length (both are to be specified
in the same units). The Fresnel reflection coefficient for a surface
having good reflective characteristics may lie in the vicinity of 0.90.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
20 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
The value of the divergence factor may also lie around 0.90. For
example the divergence factor is 0.91, for a 30 kilometer hop and a
height difference of 30 m between the antennas and k=1.33. If the
height difference is increased to 330 m, the divergence factor increases
to 0.97 for the same k value. If the hop length is decreased to 15 km, the
divergence factor increases to 0.99 for a height difference of 30 m and a
k value of 1.33.
The value of the correction factor varies with frequency and ground
surface type in accordance with the Table 3. For very smooth surfaces,
e.g., the surface of a body of water, the correction factor is
approximately 0.90.
The total reflection coefficient for a spherical and very smooth surface
can be approximated to 0.90⋅0.90⋅0.90 ≅ 0.73. From the diagram in
Figure 6, the reflection loss is approximately 12 dB.
Estimations can be easily performed if one assumes that the values of
both the Fresnel reflection coefficient and divergence factor lie close to
0.90 and then apply the correction factor value given in the Table 3 for
the different ground surface types.
7.3.8 Calculation of the position of the reflection point
The calculation of the position of the reflection point is primarily a
geometric problem and the result is therefore presented in connection
with the presentation of the path profile. The groundreflected beam
path and the reflection point’s position are clarified.
There are two different methods available for the calculation of the
reflection point’s position. The simplest algorithm avoids the numerical
solution of thirddegree equation and is therefore employed in here. The
following intermediate parameters are calculated initially:
Intermediate parameter c
B A
B A
h h
h h
c
' '
' '
+
−
=
(42)
where
c = Intermediate parameter m
h′
A
= Antenna height at station A, m
h′
B
= Antenna height at station B, m
Intermediate parameter m
RADIOWAVE PROPAGATION
Ericsson AB 21
4/038 02LZU 102 152, Rev B, June 2003
( )
3
2
10 ' ' 4
−
⋅ + ⋅ ⋅
=
B A e
h h R
d
m
(43)
where
m = Intermediate parameter
d = Distance between stations A and B, km
R
e
= Equivalent earth radius, km
h′
A
= Antenna height at station A, m
h′
B
= Antenna height at station B, m
Intermediate parameter b
( )


.

\

+
⋅
⋅
⋅
⋅ + ⋅
⋅
+
⋅ =
3
1
3
2
3
acos
3
1
3
cos
3
1
2
m
m c
m
m
b
π
(44)
The position of the reflection point is calculated from
( ) b
d
d
A
+ ⋅ = 1
2
(45)
and
A B
d d d − =
(46)
where
d
A
= The distance between station A and the reflection point, km
d
B
= The distance between station B and the reflection point, km
d = The distance between stations A and B, km
b = The intermediate parameter as above
Advices:
1) The grazing angle of radiorelay paths is normally very small,
currently lower than 1 degree.
2) It is strongly recommended to avoid ground reflection. This can be
achieved by “shielding” the path against the indirect ray.
3) Vertical polarization gives less loss. For large grazing angles the
difference between vertical and horizontal polarization is substantial.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
22 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
4) Changing the antenna heights can move the location of the reflection
point. This approach, usually known as the “HiLo technique”, force the
reflection point to move closer to the lowest antenna by affecting the
height of the higher antenna. The grazing angle increases and the path
becomes less sensitive to kvalue variations.
5) Space diversity normally provides good protection against reflection.
It is currently applied for paths over open water surfaces.
6) The contribution due to “reflection loss” is NOT automatically
included in the link budget, but in the case reflection cannot be avoided
the fade margin may be adjusted by including this contribution as
“additional loss” in the link budget.
7.3.9 Optimum height difference
To calculate the optimum height difference between the diversity
antennas, one first calculates the height difference between two adjacent
points along the mast, at which signal strength is a minimum (or a
maximum). This calculation is naturally performed for both stations, A
and B. For example, assume that an antenna is mounted on a mast at a
given position, i.e., at a given height. As the antenna is moved from this
starting position, the resultant signal strength (the sum of the signal
strengths of the direct and the phaseshifted reflected waves) will either
increase to a maximum or decrease to a minimum depending on the
direction of movement. The distance between the points in which
minimum (or maximum) signal strength is measured is the distance
referred to above.
3
2
'
'
10
74 . 12
1
2
3 . 0
⋅
⋅
−
⋅
⋅
=
k
d
h
f
d
h
B
B
A
δ
(47)
3
2
'
'
10
74 . 12
1
2
3 . 0
⋅
⋅
−
⋅
⋅
=
k
d
h
f
d
h
A
A
B
δ
(48)
where
δh´
A
= Height difference between the two maximum/minimums at
station A, m
RADIOWAVE PROPAGATION
Ericsson AB 23
4/038 02LZU 102 152, Rev B, June 2003
δh´
B
= Height difference between the two maximums/ minimums at
station B, m
h´
A
= Antenna height above the point of reflection at station A, m
h´
B
= Antenna height above the point of reflection at station B, m
d
A
= Distance between station A and the point of reflection, km
d
B
= Distance between station B and the point of reflection, km
d = Distance between station A and B, km
f = Frequency, GHz
k = Earthradius factor
The distance between the stations and the point of reflection is
calculated as described in accordance to section 7.3.8.
The distance required between the diversity antennas is then calculated
as follows:
2
'
A
A
h
h
δ
δ =
(49)
2
'
B
B
h
h
δ
δ =
(50)
where
δh
A
= Height difference between the antennas at station A, m
δh
B
= Height difference between the antennas at station B, m
7.4 Precipitation
7.4.1 Types of precipitation
Precipitation can take the form of:
• Rain
• Snow
• Hail
• Fog and haze
In common for all of the above forms of precipitation is the fact that
they all consist of water particles (haze can also consist of small solid
particles). Their distinctions lie in the distribution of the size and form
of their water drops. Rain attenuation is, however, the main contributor
in the frequency range used by commercial radio links.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
24 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
Sharp demarcations between these forms of precipitation is however not
always apparent. ”Intermediate” states can very well occur.
7.4.2 Precipitation: now
Attenuation is only caused by wet snow.
The attenuation caused by dry snow can be considered as negligible for
frequencies below 50 GHz.
Snow cover on antennas and radomes, socalled ice coatings, can result
in two problems:
• Increased attenuation
• Deformation of the antenna’s field radiation diagram
Both cases result in the reduction of the input signal strength at the
receiving station.
Antenna ice coating can of course be alleviated by electrically heating
the antennas, however the disadvantages are unfortunately greater than
the advantages. Some of the disadvantages are:
• Antennas must be held warm constantly, there is otherwise the risk
that melted snow forms to ice
• Electrical heating may be interrupted in the event of an electrical
power loss
• There is no knowledge as to the impact, or its degree, on an
antenna’s field radiation diagram due to electrical heating
7.4.3 Precipitation: hail
The effects of hail on radio connections are first apparent when hail
particle sizes approach the size of radio waves, for example, 150 mm (2
GHz), 9.6 mm (31 GHz) and 6 mm (50 GHz). Hail particle sizes greater
than 10 mm are however quite rare.
Measurements made in Sweden show that the deepest fading lasted for
just under 5 minutes and was less than 10 dB.
Hailstorms can therefore not be considered as availability limiting
factor, since they occur quite infrequently together with other forms of
precipitation.
RADIOWAVE PROPAGATION
Ericsson AB 25
4/038 02LZU 102 152, Rev B, June 2003
7.4.4 Precipitation: fog and haze
Measurements performed in Sweden show that the deepest fading that
can be related to heavy fog and haze amounted to between 4 and 7 dB.
Fog and haze can therefore not be considered as availability limiting
factor, since both fog and haze occur quite infrequently together with
other forms of precipitation.
7.4.5 Precipitation: rain
Attenuation due to rain is the generally responsible for two principal
attenuation mechanisms: absorption and scattering caused by the
raindrops.
The extent of the attenuation due to rain is primarily a function of
• Form of the rain drops
• Size distribution of the rain drops
The most common form of falling raindrops under the influence of air
resistance is the oblate form (not exactly ellipsoidal). This causes
horizontally polarized waves to attenuate more than vertically polarized
waves.
7.4.6 Cumulative distribution of rain
Due to the rapid timevariation of rain, the measured cumulative
distribution of rain intensity is heavily dependent of the integration time
selected for the measuring process. The rain intensity statistical
distributions used in ITUR reports are assumed to be the results of
measurements or transformations corresponding to an integration time
of 1 minute. The instantaneous rain intensity (which is extremely
difficult to measure) is however more suitable from a networkplanning
standpoint.
For the purpose of availability calculations, one is however interested in
the cumulative distribution of rain intensity (rainfall rate), i.e., that
percentage of time during which a given level of rain intensity is
attained or exceeded.
Normally, the reference level applied to rain intensity is the rain
intensity that is exceeded during 0.01% of the time, which is often
designated as R
0.01
.
7.4.7 Obtaining Rain Intensity (the former ITUR model)
NOTE: The former ITUR procedure for obtaining rain intensity
values employed worldwide charts of rain zones in which the world
is subdivided into 15 different rain zones, see Chapter 15 10.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
26 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
NOTE: Once the proper rain zone is geographically found, then its
correspondent rain intensity expressed in mm/h for different
fractions of time (%) is obtained from the table in Chapter 1411.
For instance, three rain zones cover Sweden, C, E and G and three rain
zones, K, N and P, cover Brazil.
7.4.8 Rain zones  diagram
The cumulative distributions listed in the previous table are illustrated
in diagram form in Figure 8. The curves represent ITUR’s 15 different
rain zones covering the entire earth. The distribution of rain intensity
(mm/h) represents a percentage of time that is equivalent to the
attainment or exceeding of a given rain intensity. The Yaxis to the
right shows the time percentage expressed in minutes per year.
0 50 100 150 200 250
Rain intensity, mm/h
0.001
0.010
0.100
1.000
2
3
4
5
6
7
8
9
2
3
4
5
6
7
8
9
2
3
4
5
6
7
8
9
P
e
r
c
e
n
t
a
g
e
o
f
t
i
m
e
r
a
i
n
i
n
t
e
n
s
i
t
y
i
s
e
x
c
e
e
d
e
d
,
%
5.26
52.56
525.60
5256.00
M
i
n
u
t
e
s
/
y
r
P
N
Q
L
M
K H
F
G
E
J
A B
C
D
P
e
r
c
e
n
t
a
g
e
o
f
t
i
m
e
r
a
i
n
i
n
t
e
n
s
i
t
y
i
s
e
x
c
e
e
d
e
d
,
%
Figure 8: The rain zones represented as cumulative distributions.
7.4.9 Obtaining rain intensity (current ITUR model)
The new ITUR rain intensity procedure, also known as Baptista
Salonen model, is conditioned to the following aspects:
1. High quality, long integrationtime (few hours) and high spatial
resolution (about one grid point per 100 km)
2. Models for transforming long integrationtime rain data to short
integrationtime rain data
RADIOWAVE PROPAGATION
Ericsson AB 27
4/038 02LZU 102 152, Rev B, June 2003
The new procedure does not demand any rain zone chart and rain
intensity (rainfall rates) is directly calculated as a function of the
geographical location of the site. Rain intensity values are not any
longer representative for a certain major rain region but represent local
values.
The basic of the new ITUrainfall model is the rain intensity data that is
now available from two different raindata programs: 1) Global
Precipitation Climate Project (GPCPdata) and 2) European Center for
MediumRange Weather Forecast (ECMWFdata).
The new model is derived in two steps. First, suitable functions
describing properly the rainfall rate distributions in tropical and mid
latitude climates are derived. This function is expressed as follows:
( )
( ) R c
R b
R a
e P p
⋅ +
⋅ +
⋅ ⋅ −
⋅ =
1
1
0
(51)
where p is the annual probability that the rain intensity R (mm/h) is
exceeded, P
0
is the rain probability obtained from statistical data and a,
b and c are parameters.
The next step is to optimize the above parameters by employing
empirical functions. The difference between predicted and measured
rainfall rates is minimized. The rain intensity data used in the
optimization is from ITUR databases covering a large amount of sites
all over the world at different climates.
The probability of rain P
0
is approximated by the following expression:


.

\

− ⋅ =
⋅
6
0.0117 
6 0
e 1
r
S
P
M
r
P P
(52)
where M
s
(mm) is the annual rainfall amount of stratiformtype rains
and P
r6
(%) is the probability of rainy 6 hours periods.
The annual probability that the rain intensity R (mm/h) is obtained from
the previous expression
A
C A B B
R
⋅
⋅ ⋅ − + −
=
2
4
2
(53)
where
b a A ⋅ =
(54)
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
28 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003


.

\

⋅ + =
0
ln
P
p
c a B
(55)


.

\

=
0
ln
P
p
C
(56)
For p>P
0
, R(p)=0
The parameters a, b and c are empirically optimized and finally given
by the expressions:
1 . 1 = a
(57)
( )
0
22932 P
M M
b
s c
⋅
+
=
(58)
where M
c
(mm) is the annual rainfall amount of convectivetype rains.
b c ⋅ = 5 . 31 (59)
The users of the new ITU rainfall rate model are, however, not forced to
calculate the parameters M
s
, M
c
and P
rg6
since they are calculated and
stored in data files at the ITUR. The files are as follows:
ESARAINPR6.TEXT ! contains the numerical values of the
parameter P
r6
.
ESARAIN_MC.TXT ! contains the numerical values of the parameter
M
c
.
ESARAIN_MS.TXT ! contains the numerical values of the parameter
M
s
.
The values of the parameters M
s
, M
c
and P
r6
are stored as 121rows and
241columns matrix (121x241) corresponding to each point in a grid
system.
It is assumed that the Earth surface is divided in a grid having a 1.5
degree resolution, that is, every square has a side of 1.5 degrees and for
interpolation purposes each square can be considered as a plane (flat)
square. The longitude and latitude of the Earth determine every point
forming the grid. The values of the parameters M
s
, M
c
and P
r6
are given
from the above databases for every grid point.
RADIOWAVE PROPAGATION
Ericsson AB 29
4/038 02LZU 102 152, Rev B, June 2003
The values of the longitude and latitude for all grid points are also
stored as 121rows and 241columns matrix (121x241) and can be
obtained from data files at the ITUR. The files are as follows:
ESARAINLON.TXT ! contains the longitude values for each grid
point.
ESARAINLAT.TXT ! contains the latitude values for each grid point.
For each specific grid point (LON
i
, LAT
j
) there will be M
sij
, M
cij
and P
r6ij
corresponding values.
Parameter values for other geographical locations than the grid points
given in the above matrices are obtained by twodimensional
interpolation technique.
NOTE: Rain intensity values all over the world are displayed in the
charts presented in Chapter 1512.
Figure 9 gives an overview of the geographical distribution of rain
intensity for 0.01% of time, R
0.01
.
Figure 9: Distribution of rain intensity R
0.01
all over the world. High
rain intensity regions are encountered in the dark regions.
7.4.10 The calculation of the specific rain attenuation
The calculation of specific rain attenuation is performed in two steps:
• First, a calculation is performed of the values of the coefficients
corresponding to certain assumptions concerning the distribution of
raindrop size, form, temperature and type of polarization
(horizontal/vertical)
• Then, a calculation is performed of the specific rain attenuation for
a given instantaneous rain intensity
Calculate the coefficients as follows
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
30 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
( ) ( )
2
2 cos cos
2
τ θ ⋅ ⋅ ⋅ − + +
=
V H V H
f
k k k k
k
(60)
( ) ( )
f
V V H H V V H H
f
k
k k k k
⋅
⋅ ⋅ ⋅ ⋅ − ⋅ + ⋅ + ⋅
=
2
2 cos cos
2
τ θ α α α α
α
(61)
where
k
H
, α
H
, k
V
, α
V
= Frequency dependent coefficients.
θ = The path’s elevation angle
τ = The polarization tilt angle relative to the horizontal plane
NOTE: Frequency dependent coefficients are provided in Chapter
159.
The calculation of specific rain attenuation (dB/km) is performed as
follows
f
R k
f R
α
γ ⋅ =
(62)
where
k
f
, α
f
= Frequency dependent coefficients
R = Rain intensity, mm/h
The specific rain attenuation that is exceeded during 0.01% of the time,
can be calculated by relating the rain intensity to the reference level
0.01%, i.e.,
f
R k
f R
α
γ
01 . 0
01 . 0
⋅ =
(63)
Figure 10 illustrates specific rain attenuation (dB/km) that are exceeded
during 0.01% of the time as a function of frequency (GHz) for three
different values of rain intensity, R
0.01
, for both horizontal and vertical
polarization.
RADIOWAVE PROPAGATION
Ericsson AB 31
4/038 02LZU 102 152, Rev B, June 2003
Figure 10: Specific rain attenuation exceeded during 0.01% of the time
as a function of frequency.
Figure 11 illustrates the specific rain attenuation (dB/km) that are
exceeded during 0.01% of the time as a function of rain intensity for
four different values of frequency (GHz), for both horizontal and
vertical polarization.
Figure 11: Specific rain attenuation exceeded during 0.01% of the time
as a function of rain intensity.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
32 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
Figure 12 illustrates the specific rain attenuation (dB/km) that is
exceeded during 0.01% of the time as a function of rain intensity for
horizontal (H) and vertical (V) polarization at 23 GHz.
At 23 GHz and horizontal polarization, the specific rain attenuation at
R
0.01
=30 mm/h is almost twice the value at R
0.01
=12 mm/h.
Figure 12: Specific rain attenuation exceeded during 0.01% of the time
as a function of rain intensity for a frequency of 23 GHz.
7.4.11 Calculating total rain attenuation
The total rain attenuation for a radio link path can be calculated as
follows, if the statistical distribution of the rain cells along the path is
known
eff R R
d A ⋅ = γ
(64)
where
A
R
= Total rain attenuation, dB
γ
R
= Specific rain attenuation, dB/km
d
eff
= Effective path length, km
The effective path length is calculated as follows
r d d
eff
⋅ =
(65)
where
RADIOWAVE PROPAGATION
Ericsson AB 33
4/038 02LZU 102 152, Rev B, June 2003
d = Actual path length, km
r = Reduction factor
The reduction factor is arrived at as follows
0
1
1
d
d
r
+
=
(66)
The factor 1/d
0
is coupled to rain intensity for the 0.01% reference
level. d
0
is then
01 . 0
015 . 0
0
e 35
R
d
⋅ −
⋅ =
(67)
The reduction factor accounts for the extensions of rain cells and
transforms actual path lengths to equivalent path lengths along which
the rain can be regarded as having a uniform distribution.
7.4.12 Calculating total rain attenuation for 0.01%
The total rain attenuation that is exceeded 0.01% of the time can be
calculated if the rain intensity is related to the 0.01% reference level, as
follows
eff R R
d A ⋅ =
01 . 0 01 . 0
γ
(68)
where
A
R 0.01
= Total rain attenuation that is exceeded during 0.01% of the
time, dB
γ
R0.01
= Specific rain attenuation that is exceeded during 0.01% of
the time, dB/km
d
eff
= Effective path length, km
The total rain attenuation that is exceeded during 0.01% of the time is
used later in the calculation of unavailability caused by rain.
7.5 Obstruction  diffraction
7.5.1 Definition
Diffraction is the responsible mechanism for obstacle loss/attenuation.
In fact, obstacle loss is also known in the literature as “diffraction loss”
or “diffraction attenuation”.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
34 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
Depending on the shape, size and electrical properties of the obstacle,
diffraction calculations can be cumbersome and timeconsuming. Since
radiorelay paths normally require LOS, relatively simple methods for
calculating the obstacle loss are currently employed. One powerful,
although simple method for calculation of obstacle loss, is the single
peak method that is based on the knifeedge approximation. This
method can easily be extended to comprise the three most significant
peaks inside the Fresnel zones.
Obstruction loss is calculated based on the path’s geometry and on the
actual frequency used.
The geometry is a function of:
• Topography
• The antenna’s height above ground level
• The earthradius factor, k
Different kvalues result in different obstruction loss values. Small k
values result in the greatest obstruction loss due to the fact that the
beam tends to bend more towards the ground surface, or expresses in
another manner, the obstruction penetrates deeper into the Fresnel zone.
7.5.2 Knifeedge obstructions
A knifeedge obstruction is one that consists of an individual
obstruction having negligible length in the direction of the radio wave’s
propagation path as illustrated in Figure 13. The loss contributed by
such an obstruction is derived from the knifeedge loss curve, which is
a physically derived function.
A
B
Equivalent earth surface
r
1F
h
LOS
v < 0
v > 0
Figure 13: Knifeedge obstruction showing the obstruction’s height
relative the free lineofsight.
RADIOWAVE PROPAGATION
Ericsson AB 35
4/038 02LZU 102 152, Rev B, June 2003
In the case of knifeedge obstructions, the obstruction loss value, A
H
is
only dependent on the parameter ν, which is defined as the
obstruction’s relative penetration of the Fresnel zone:
F
LOS
r
h
1
= ν
(69)
where
h
LOS
= The obstruction’s height above the free lineofsight
r
1F
= The Fresnel zone’s radius at the point of the obstruction
The parameter ν, as defined above, differs by a factor of 2 ≅ 1.41
from the definition in Rep. 7153, vol. 5, which means a difference of
approximately 13 dB in obstruction loss for the particular value of ν.
The height of the obstruction over the free lineofsight may be defined
as
h
LOS
= (ground elevation + height of the tree line or building
height)  the height of the free lineofsight
7.5.3 Knifeedge loss curve
The loss caused by an obstruction is arrived at from the knifeedge loss
curve, which is a physically derived function. Knifeedge loss A
H
as a
function of the relative penetration ν, is shown in Figure 14.
Figure 14: Knifeedge loss as a function of the relative penetration
parameter.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
36 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
When performing path calculations, realistic degrees of Fresnel zone
penetration are often considered as lying in the interval from 0.5 to 2,
which means calculation of obstruction losses based on the diagram
insertion above.
For ν≥10, obstruction losses are calculated as follows:
( ) 10 log 20 16 ≥ ⋅ + = ν ν
H
A
(70)
where
A
H
= Obstruction loss, dB
ν = The obstruction’s relative penetration of the Fresnel zone
7.5.4 Typical knifeedge losses
Figure 15 illustrates a few typical examples of loss values (dB) for the
knifeedge function.
0
0
6
12
16
20
Figure 15: Typical loss values (dB) resulting from the knifeedge
function.
7.5.5 Singlepeak method
The singlepeak method calculates the value of the obstruction loss as
the greatest knifeedge obstruction loss attained as a result of an
individual obstruction lying along the path as illustrated in Figure 16.
The algorithm defines those peaks in the path profile between station A
and station B that penetrate the Fresnel zone. The penetration, ν, of
every peak is calculated relative to the Fresnel zone along the free line
ofsight, AB. The corresponding knifeedge loss, A
H
, is calculated as if
only one peak existed along the path. The greatest loss value that is
found along the path is returned as the sought obstruction loss value.
RADIOWAVE PROPAGATION
Ericsson AB 37
4/038 02LZU 102 152, Rev B, June 2003
A
B
Figure 16: In the singlepeak method the obstruction loss is taken as the
greatest knifeedge obstruction loss lying along the path.
The singlepeak method is, as is obvious, a pure application of the
knifeedge model. It works best for paths that have one dominant peak.
The results of the model are less reliable for more realistic paths having
a number of significant peaks.
7.5.6 Triplepeak method
Simply stated, the triplepeak method may be described as a calculation
of the obstruction loss value along the propagation path, based on the
sum of the three largest knifeedge losses.
The algorithm involves an initial calculation of the obstruction loss
based on the singlepeak method, as described earlier. This first
calculation of the single knifeedge loss represents the first contribution,
A
1
, to the total obstruction loss.
The path profile is then split at that the point, M, which resulted in the
largest knifeedge loss, see Figure 17. The peak of point M is regarded
as being a common antenna or termination point along the partial paths
AM and MB. If the peak consists of trees, then the mast height of the
fictitious antenna is set to the height of the trees, otherwise the mast
height is set to zero. In the event that the fictitious antenna attains a
height beneath the original free lineofsight, AB, then the mast height is
instead set so that the antenna exactly reaches the free lineofsight.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
38 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
A
B
M
Figure 17: The path profile after the first split.
The partial paths, AM and MB to the left and right of the located peak,
M, are each searched for two new paths in the same manner as was the
original path. Note that the partial paths, as illustrated in the figure
above, generally have other free linesofsight and Fresnel zones than
does the original path. Each partial path results in a separate knifeedge
loss value. The higher of the two values will represent the second
contribution, A
2
, to the total obstruction loss.
The particular partial path is then subdivided at the peak, N, that
resulted in the highest knifeedge loss, see Figure 18. The resultant
partial paths are then each searched in the same manner as was the
original partial paths. The third and final contribution, A
3
, to the total
obstruction loss is the largest knifeedge loss resulting from one of the
partial paths AN, NM, and MB.
The total obstruction loss, A
H
, is obtained by summing the three
contributions described above, A
1
, A
2
and A
3
.
3 2 1
A A A A
Obst
+ + =
(71)
RADIOWAVE PROPAGATION
Ericsson AB 39
4/038 02LZU 102 152, Rev B, June 2003
A
B
N
M
Figure 18: The path profile after the second split.
The triplepeak method is entirely empirical, but it has proven to work
well in actual applications. It works better than the singlepeak method
in many path profile scenarios, since it accounts for more than only the
highest peak along the path.
The difference between the triplepeak method and a hypothetical
repetition (three times) of the singlepeak method, lies in the fact that
secondary peaks in the triplepeak method will contribute less than the
primary peak considering the peaks’ penetration of the original Fresnel
zone. This is a function of two factors:
• The partial paths are always shorter than the full path
• The partial paths’ free linesof sight always lies higher than (or at
the same level as) the original full path
A shorter path results in a smaller Fresnel zone radius. Higher free line
ofsight results in a relatively lower peak free lineofsight. Together,
these factors result in a smaller relative penetration. The result is that
the secondary peaks cause lower obstruction losses.
The triplepeak method, as it is applied here, is a further development of
the original multiplepeak method introduced by Deygout, ”Multiple
KnifeEdge Diffraction of Microwaves”, IEEE Trans. Ant. Prop. vol.
AP14, 1966.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
40 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
7.5.7 Smoothly spherical earth
In the case of smoothly spherical earth (flatearth), the obstruction is
represented by an smooth surface, such as a sea or lake, penetrating the
Fresnel zone. Losses are calculated using a simple function that may be
derived from empirical considerations. The geometry of the smoothly
spherical earth is illustrated in Figure 19.
d
r
d
d
B
d
A
h
B
h
A
B
A
Figure 19: The geometry of the smoothly spherical earth.
The loss calculation is performed in accordance with the Cheriex
method. First the distances to the radio horizon from both antennas are
calculated as follows
A A
h R k d ⋅ ⋅ ⋅ ⋅ ≅
−3
10 2
(72)
B B
h R k d ⋅ ⋅ ⋅ ⋅ ≅
−3
10 2
(73)
where
d
A
= Distance from station A to the radio horizon, km
d
B
= Distance from station B to the radio horizon, km
h
A
= Antenna height at station A, m
h
B
= Antenna height at station B, m
k = Earthradius factor
R = True earth radius (≅6370 km)
The distance between both radio horizons may be easily calculated as
( )
B A r
d d d d + − ≅
(74)
where
d = Distance between station A and B, km
RADIOWAVE PROPAGATION
Ericsson AB 41
4/038 02LZU 102 152, Rev B, June 2003
The obstruction loss for evenly curved earth is calculated as
r Obst
d k f A ⋅ ⋅ ⋅ + ≅
−
3
2
3
112 . 0 20
(75)
where
A
Obst
= Obstruction loss, dB
f = Frequency, MHz
7.5.8 Typical losses resulting from smoothly spherical earth
Figure 20 illustrates typical loss values (dB) for smoothly spherical
earth for a path of 50 kilometers and a frequency of 2.2 GHz.
40
10
20
Figure 20: Typical loss values (dB) resulting from a smoothly spherical
earth.
For grazing linesofsight, i.e., the antennas have the same horizon (d
A
+
d
B
= d), the loss is 20 dB, which applies regardless of frequency and
path length.
7.5.9 Clearance and path geometry
7.5.9.1 The Earth bulge
The local height of the Earth bulge (h) is dependent on the kvalue. The
parameter h is very important for clearance purposes. The shadow
region in Figure 21 covers its local value.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
42 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
x
h
max
d
2
d
1
y
y=d/2
y=d/2
k· R
k· R
k· Rh
max
h
O
A
B
M
N
Figure 21: The local height of Earth bulge.
The local height of the Earth bulge is given by
k
d d
h
⋅
⋅
=
74 . 12
2 1
(76)
where the distances d
1
and d
2
are normally expressed in km and h in
meters.
The local height of the Earth bulge is inversely proportional to the
earthradius factor. For high kvalues, the Earth surface is close to a
plane surface while for low kvalues the Earth surface becomes more
curved and may penetrate the radio path.
7.5.9.2 Path geometry
In what follows, clearance, obstacle penetration and antenna height will
be discussed. Figure 22 displays the path geometry for which the path
parameter clearance c is depicted. The height of the lineofsight is x,
the bulge of the Earth is h and the height of the obstacle above the earth
surface is h
3
. The other parameters have their current designation. The
antenna heights are represented as “total” heights, that is, both the
terrain and the actual antenna heights are included.
RADIOWAVE PROPAGATION
Ericsson AB 43
4/038 02LZU 102 152, Rev B, June 2003
h
1
d
2
x
d
1
h
2
h
2
x
xh
1
θ
θ
d
h
3
h
c
Figure 22: Path geometry (the path is drawn on a equivalent earth).
7.5.9.3 The height of the lineofsight
The height of the lineofsight with respect to the antenna heights is
given by the following expression
1 1
1 2
h d
d
h h
x + ⋅
−
=
(77)
where h
1
and h
2
are given in m and d and d
1
in km.
7.5.9.4 Clearance of the LOS
The clearance of the LOS (the height above the obstacle) is given as
follows
1
2 1
1
1 2
1
74 . 12
h
k
d d
d
d
h h
h c −
⋅
⋅
− ⋅
−
+ =
(78)
where d, d
1
and d
2
are expressed in km and c, h
1
, h
2
and h
3
are
expressed in m.
7.5.9.5 Antenna height
The antenna height as a function of the obstacle height and the Earth
radius factor for a given clearance is easily obtained from the above
expression by setting h
2
= H,
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
44 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003

.

\

−
⋅
⋅
+ + + =
1
2 1
3
1
1
74 . 12
h
k
d d
h c
d
d
h H
(79)
7.5.9.6 Obstacle penetration
Now, assume the height of the lineofsight is lower than the height of
the obstacle and gives an obstacle penetration expressed by b meters
above the lineofsight. Thus, the height of the lineofsight x is
properly expressed as
b h h x − + =
3
(80)
Employing the above expressions, the obstacle penetration is obtained
according to
d
d
h h
h
k
d d
h b ⋅
−
− −
⋅
⋅
+ =
1 2
1
2 1
3
74 . 12
(81)
with the parameters expressed as before.
7.5.10 Vegetation
For unexpected obstacle intercepting the Fresnel zone, for instance
growing vegetation, the additional loss can be calculated using the
method recommended by the ITUR.
Advices:
1) Highresolution path profiles and careful site (and path) surveys are
important tasks in the planning process to avoid unexpected obstacle
attenuation.
2) Vegetation is continuously growing. What seems to be LOS today
might not be LOS “tomorrow”!
7.6 The Link Budget
The Link budget is the process of adding and subtracting gain and
losses of a radiorelay path, see Figure 23. The main output of link
budget calculations is the signal level at the receiver (dBm), the path
loss and the fade margin. In most applications, the same duplex radio
setup is applied to both stations forming the radiorelay path. Thus, the
calculation of the received signal level is independent of direction.
RADIOWAVE PROPAGATION
Ericsson AB 45
4/038 02LZU 102 152, Rev B, June 2003
7.6.1 Path loss
The path loss is the sum of all losses and gains between the
transmitter’s and the receiver’s antenna contacts and is calculated as
follows:
ARx ATx F L Obst G bf S
G G A A A A A A − − + + + + =
∑
(82)
where
A
S
= Path loss, dB
A
bf
= Freespace loss, dB
A
G
= Gas attenuation, dB
A
Obst
= Obstruction loss, dB
A
L
= Additional loss, dB
A
F
= Antenna feeder loss, dB
G
Atx
= Transmitter antenna gain, dBi
G
Arx
= Receiver antenna gain, dBi
7.6.2 Fade margin
Under interferencefree conditions, the fade margin is defined as the
difference between the received signal level under ”normal” wave
propagation conditions (fadefree time) and the receiver’s threshold
level at a given biterror level, i.e.,
Tr R
P P M − =
(83)
where
M = Fade margin, dB
P
R
= Receiver signal level, dBm
P
Tr
= Receiver threshold level, dBm
Receiver signal level is calculated as the difference between the
transmitter’s output power and the path loss, i.e.,
S Tr R
A P P − =
(84)
where
P
R
= Receiver signal level, dB
P
Tr
= Transmitter output power, dBm
A
S
= Path loss, dB
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
46 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
7.6.3 Power diagram
A power diagram is a schematic approach to the illustration of the
effects on a transmitter’s radiated power as it propagates towards a
receiving station as shown in Figure 23. Concepts such as fade margin
and receiver threshold value are also included in the definition.
Antenna Splitter
Transmitter /
Receiver
Transmitter /
Receiver
Wave guide
Wave guide
Waveguide
Antenna Splitter
Transmitter /
Receiver
Transmitter /
Receiver
Wave guide
Wave guide
Waveguide
Output
Power
B
ra
n
c
h
in
g
L
o
sse
s
A
n
t
e
n
n
a
G
a
i
n
P
r
o
p
a
g
a
t
i
o
n
L
o
s
s
e
s
A
n
t
e
n
n
a
G
a
i
n
Branching Losses
Receiver Thresh.
Value
Received Power
Unfaded
Fade
Margin
T
r
a
n
s
m
i
t
t
e
d
P
o
w
e
r
Figure 23: The power diagram showing possible losses from the
transmitter to the receiver. The fade margin is indicated as the
difference between the received power and the receiver’s threshold
value.
7.6.4 Effective fade margin
The receiver’s threshold value as defined earlier only applies under
negligible or interferencefree conditions. In reality, this is however not
the case. A certain interference contribution is almost always present
when performing path calculations, which usually affects availability
results.
The interference contribution can be interpreted as degradation in the
receiver’s threshold value, i.e., threshold degradation. The effective
fade margin is therefore defined as the difference between the fade
margin and the threshold degradation. The effective fade margin is used
later in availability calculations.
Interference calculations provide the value of the threshold degradation.
RADIOWAVE PROPAGATION
Ericsson AB 47
4/038 02LZU 102 152, Rev B, June 2003
Advices:
1) The main purpose of the link budget is to calculate the fade margin
and “delivery” its value to the fading block.
2) The fade margin is calculated with respect to the receiver threshold
level for a given biterror ratio (BER). The threshold level for BER=10

6
for MINI LINK equipment is about 4 dB higher than the threshold
level for BER= 10
3
. Consequently, the fade margin is 4 dB larger for
BER=10
6
than for BER=10
3
. For other equipment than MINI LINK, it
is generally common to have one dB of threshold level for each decade
of BER.
3) The fade margin is NOT an input parameter for “tuning” path
proprieties in the design of microwave links. Appropriate planning of
microwave networks relies on the values of quality and availability
objectives.
8 The fading block
8.1 Definition
Fading is often defined as a variation in signal strength over time, phase
or polarization. Fading is normally the result of changes in the physical
properties of the atmosphere or due to ground or water reflections.
8.2 General cause
Fading can be caused by the occurrence of an isolated phenomenon, one
that is solely responsible for its appearance. It is however more
common that fading appears in one and the same hop as the result of a
combination of various phenomenon that interact with one another,
leading to the degradation of signal quality and availability. Climate,
topography and surroundings can vary to such great degrees that fading
often depends on the aggregate effects of numerous phenomenon.
8.3 General classification
Fading can be classed as follows:
• Source
• Propagation attributes
• Time variation
• Statistical distribution
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
48 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
8.4 Classification based on source
The phenomenon of fading is often classified based on the source of the
phenomenon. Source can be divided into four primary groups:
• Atmospheric fading: absorption, refraction and turbulence.
• Groundbased fading: geology, the roughness of the surrounding
terrain, propagation path differences due to tides or variations in
snow depth, obstructions due to variations in vegetation
• ”Manmade” fading: obstruction or reflection caused by boats,
aircraft and temporary constructions sites, antenna vibration.
• Mixed fading: due to the occurrence of atmospheric inversion layers
and the reflection they cause.
8.5 The concept of outage
Outage is generally defined as the probability that a predefined bit
error ratio is exceeded during a certain measured period. The concept of
outage comprises quality (error performance) and availability events
that are referred to biterror ratio.
8.6 Rain fading (current ITUR model)
8.6.1 Calculation of the fade margin based on a yearly basis
The fade margin that is exceeded during different periods of time based
on a yearly basis is calculated as follows
( ) P
R P
P A M
log 043 . 0 546 . 0
01 . 0
12 . 0
⋅ + −
⋅ ⋅ =
(85)
where
A
R0.01
= Total rain attenuation that is exceeded 0.01% of the time,
dB
M
P
= Fade margin that is exceeded p% of the time, dB
P = Percentage of the time during which 0.001 < P < 1%
The total attenuation for 0.01% of time, A
0.01
, is calculated as a function
of the rain intensity (rainfall rate) for 0.01% of time, R
0.01
, and the
effective path length by equation (68).
The attenuation exceeded for a certain percentage of time can be
referred to as the fade depth. If we adapt the fade margin, M, to be as
much as the fade depth, then A
p
can be replaced by M in both
expressions above.
RADIOWAVE PROPAGATION
Ericsson AB 49
4/038 02LZU 102 152, Rev B, June 2003
In the previous ITUmodel, the above expression was valid for all
values of latitude and longitude. In the new revision of the ITUR
recommendation, however, the above expression is modified to fit
different values of the latitude. Thus, for radio links located at latitudes
equal or greater than 30° (North and South) the above expression is still
applied. On the other side, for latitudes lower than 30°N and 30°S (60°
belt along the equator), the current model is expressed as
( ) p p
p
A
A
log 139 . 0 855 . 0
01 . 0
07 . 0
⋅ + −
⋅ =
(86)
with the parameters defined as previously.
Compared to the previous model, the new model presented does not
provide any remarkable improvement. In addition, it seems to be
statistically inconsistent since it gives higher p values than the model
used for latitudes equal to or greater than 30°N and 30°S.
When discussing both models for calculating the probability
(percentage of time) that the fade margin will be exceeded, transmission
network planners are encouraged to stress the inconsistency of the new
model to be used in the 60° belt along the equator. Particularly, the fact
the model does not provide any remarkable improvement.
8.6.2 Outage due to rain fading  annual basis
The prediction model for the rain fading across a particular area is a
cumulative distribution over fade margin. It calculates the probability
that a given fade margin will be exceeded.
The probability that a given fade margin M is exceeded, on an annual
basis, can be attained from the previous mathematical expression by
solving the equation for the fraction of time, P. The empirical
prediction model for rain fading becomes


.

\

⋅ ⋅ + + −
=
M
A
R
p
01 . 0
12 . 0 log 172 . 0 29812 . 0 546 . 0 628 . 11
10
(87)
where
p = Percentage of time that a given fade depth M (fade margin) is
exceeded in the average year, %
A
R0.01
= Total rain attenuation that is exceeded 0.01% of the time,
dB
M = Fade margin, dB
The outage is finally obtained by transforming the annual value given
by the above expression from percentage to ratio as follows:
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
50 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
100
p
P =
rain
(88)
Where P
rain
is the probability (expressed in ratio) of exceeding the fade
margin M in the average year and p (expressed in %) as given by
expression (87).
8.6.3 Transformation between yearly and worst month basis
8.6.3.1 From yearly basis to worst month
The transformation from an annual probability to one based on a worst
month is obtained as follows
rain
P Q p ⋅ =
w
(89)
where
p
w
= Probability (expressed in ratio) of exceeding the fade margin
M during the average worst month
P
rain
= Probability (expressed in ratio) of exceeding the fade margin
M in the average year
Q = Conversion factor (climatic constant), 12> Q >1
The probability p
w
and P
rain
are referred to the same threshold level.
The conversion factor Q is expressed as a function of P
rain
and the
climatic parameters Q
1
and β. In the range of interest for microwave
planning, Q is given by following expression
% 3
12
for
1
1
1
< < 
.

\

⋅ = P
Q
P Q Q

rain
β
β
(90)
Substituting (90) in (89), the transformation from yearly basis to worst
month basis is given by
β −
⋅ =
1
1 rain
P Q p
w
(91)
where p
w
and P
rain
are defined as above.
NOTE: The values of the climatic parameters Q
1
and β can be
found in Chapter 1513.
For ”global planning” purposes the parameters Q
1
and β are specified
by ITUR as follows:
RADIOWAVE PROPAGATION
Ericsson AB 51
4/038 02LZU 102 152, Rev B, June 2003
Q
1
= 2.85
β = 0.13
8.6.3.2 From worst month to yearly
The transformation from a yearly probability to worstmonth
probability is obtained from expression (91)
β β − −
−
⋅ = 1
1
1
1
1 w rain
p Q P
(92)
The selection of the climatic parameters when transforming annual
worstmonth time percentage to average annual time percentages may
in some applications of microwave design have an important effect.
The range of validity of the conversion model is strongly dependent on
the climate and should be known by microwave designers.
8.6.4 Presentation of the rain fading models in diagram form
Figure 24 illustrates the rain fading models (worst month and on a
yearly basis) for different values of the quotient between total rain
attenuation exceeding 0.01% of the time (A
R0.01
) and the fade margin,
M. When the quotient is equal to 0.155, outage is set to 8·10
7
.
0 1 10
2 3 4 5 6 7 8 9 2 3 4 5 6 7 8
A
R0.01
/M
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
P
e
r
c
e
n
t
a
g
e
o
f
t
i
m
e
t
h
e
f
a
d
e
m
a
r
g
i
n
i
s
e
x
c
e
e
d
e
d
,
%
Rain fading model
annual basis
Rain fading model
worst month
1
Figure 24: The rain fading models for worst month and on a yearly
basis.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
52 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
8.7 Multipath fading
Figure 25 illustrates a multipath scenario.
Atmospheric layer
Figure 25: Multipath propagation illustrated by three radio beams:
Beams that are reflected by the atmosphere or the ground travel a longer
distance than do direct beams. Dependent on the size of the time delays
and the employed channel bandwidth, fading can either be
• Flat or
• Frequency selective
In general:
• Fading due to rain, for frequencies below 10 GHz, may be
considered as negligible in comparison with fading due to multipath
propagation, which is often dominant below 10 GHz.
• Fading due to multipath propagation, for frequencies above 10 GHz,
may be considered as negligible in comparison with fading due to
rain, which is often dominant above 10 GHz.
• A good rule of thumb is however, that there exists a crossover
region between the frequencies of 10 and 18 GHz, and a point at
which fading due to rain and multipath propagation are of about the
same order of magnitude.
8.7.1 Flat and frequency selective fading
Flat fading implies that there does not exist any noticeable local
variation within the transmitted frequency band, see Figure 26, i.e.,
fading has the same degree throughout the band.
Frequency selective fading implies that there does exist a noticeable
variation within the transmitted frequency band (see the figure below).
RADIOWAVE PROPAGATION
Ericsson AB 53
4/038 02LZU 102 152, Rev B, June 2003
f
B
A
f B
A
Flat Frequency selective
Figure 26: Flat and frequency selective fading.
The extent of the influence of multipath propagation on a radio link
system depends on whether the system is analog or digital and whether
the fading is flat or frequency selective.
8.7.2 The effects of multipath propagation
The effect of flat fading for digital and analog connections is similar.
Signal level decreases and quality degrades. Continued quality
degradation will eventually lead to the breakdown of the connection.
Digital systems usually exhibit a somewhat higher tolerance to flat
fading than do analog systems.
In the case of base band, analog link connections utilize frequency
multiplexing in which each channel of N channels contains a small
(B/N) frequency band.
For frequency selective fading, signal levels vary locally within the
frequency band, both in amplitude and phase. The result, in the case of
analog connections, is that a number of channels may attain signal
levels that are so low that connection within these channels is virtually
impossible. The connection can, however, be maintained at a lower
capacity.
In the case of base band, digital link connections utilize the entire
frequency band, B, of all channels in a timemultiplexed manner. This
means that every channel has a time slot and synchronism is therefore
required for system management purposes.
Inband variations in the case of a timeduplexed digital connection
represents a loss of information, the connection loses synchronization,
resulting in the fact that the connection can no longer be maintained.
The disturbance affects all channels and is abated, only after
synchronization is once again established.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
54 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
8.7.3 Measures taken against multipath fading
Measures that are aimed at suppressing fading due to multipath
propagation can be divided into three categories:
• Diversity
Diversity implies that a signal reaches the receiver via a number (at
least two) of different alternatives, the purpose being that the received
signals are to be uncorrelated. Examples of diversity are frequency,
space, path, polarization and angle.
• Adaptive equalizer
The purpose of adaptive equalizer (in both the time and frequency
domains) is the equalization of signal amplitude and phase.
• System
By modifying system parameters or other system attributes, one can
attain an improvement in system tolerance to multipath fading. For
example, improvements can, in some cases, be achieved through the
modification of path geometry or by simply changing the antenna.
8.8 Flat fading (former ITUR model): small percentages of
time
In what follows, the probability of exceeding a given fade margin
(outage) due to flat fading will be studied. Small percentage of time
means “large” fade margins.
8.8.1 Introduction
Flat fading, also known as singlefrequency, frequency independent or
narrowband fading, can generally be predicted for any part of the
world. The method relies on the prediction of the distribution at large
fade depths in the average worst month. Unlike the former prediction
method, the present method normally employed for large fade depths
does not take into account the path profile and, therefore, is suitable for
initial planning, licensing or design purposes.
In addition, there is a method applicable for all fade depths, in which
the method for large fade depths and an interpolation procedure for
small fade depths are employed
RADIOWAVE PROPAGATION
Ericsson AB 55
4/038 02LZU 102 152, Rev B, June 2003
8.8.2 Fade occurrence factor
Worldwide measurements and statistical compilations of fading events
indicate that the probability the received level fades F dB below the
freespace follows the following expression
10
10
F
flat
p
−
∝
(93)
where the fade depth F is normally interpreted as the fade margin (M).
The above expression can be rewritten as
10
0
10
F
flat
p p
−
⋅ =
(94)
where the factor p
0
is usually given as a function of climatic and path
parameters and is obtained as
8.8.3 Flat fading and quality (error performance)
Generally, multipath fading is considered as negligible for frequencies
above 10 GHz for which rain is the dominating fading mechanism. The
significance of multipath fading in rain abundant regions may even be
negligible for frequencies lower than 10 GHz.
Multipath fading (flat fading and frequency selective) is normally the
main contributor to Severely Errored Seconds (SES). Comparing flat
and frequency selective fading, the former is usually more frequent in
narrow bandwidth systems.
8.8.4 Estimation of the geoclimatic factor
The geoclimatic factor is strongly dependent on the geographical path
location, antenna altitude and size of bodies of water in the vicinity of
the path (refraction anomalies). The geoclimatic factor is calculated for
known terrain (inland and coastal links) or for unknown terrain.
8.8.5 Inland Links
Inland links are links for which
• The entire path profile is above 100 m altitude (with respect to mean
sea level) or beyond 50 km from the nearest coastline or
• Part or all the entire path profile is below 100 m altitude (with
respect to mean sea level) and entirely within 50 km of the
coastline, but having an intervening height of land higher than 100
m between the link and the coastline.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
56 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
If the above conditions are not met, the link is considered as coastal link
and the parameter “coastal fraction” (r
c
) is employed, see 8.8.6.1.
Links passing over a river or a small lake should normally be classed as
passing over land.
Normally, measured values of the geoclimatic factor K
i
are not
available, but it can be estimated according to
( ) 5 . 1 1 . 0 7
0
10 10 0 . 5
L
C C C
i
p K
Lon Lat
⋅ ⋅ ⋅ =
− − ⋅ − −
(95)
Where
C
0
= Antenna altitude coefficient, dB
C
Lat
= Latitude coefficient, dB
C
Lon
= Longitude coefficient, dB
p
L
= Percentage of time the refractivity gradient in the lowest 100
m of the atmosphere is lower than –100 N units/km in the estimated
average worst month, %
8.8.5.1 Antenna altitude coefficient
The values of the antenna altitude coefficients are classified according
to:
• Low altitude antenna: lowerantenna altitude less than 400 m above
mean sea level
• Medium altitude antenna: lowerantenna altitude in the range 400
700 m above mean sea level
• High altitude antenna: lowerantenna altitude higher than 700 m
above mean sea level
The terrain type is classified according to:
• Plains
• Hills
• Mountains
The C
0
values are displayed in Table 4 for links located on known
terrain and in Table 5 for links located on unknown terrain.
RADIOWAVE PROPAGATION
Ericsson AB 57
4/038 02LZU 102 152, Rev B, June 2003
Low altitude
antenna (0400 m)
Medium altitude
antenna (400700 m)
High altitude antenna
(above 700 m)
Plains Hills Plains Hills Plains Hills Mountains
0 3.5 2.5 6 5.5 8 10.5
Table 4: Antenna altitude coefficient values for links located on known
terrain.
When the type of the terrain is not known, the following table gives the
C
0
values for planning purpose.
Low altitude
antenna (0400 m)
Medium altitude
antenna (400700 m)
High altitude antenna
(above 700 m)
1.7 4.2 8.0
Table 5: Antenna altitude coefficient values for links located on
unknown terrain.
8.8.5.2 Latitude coefficient
The latitude coefficient is given for three latitude regions according to
C
Lat
= 0 for 53 °S ≥ ξ ≤ 53 °N
C
Lat
= 53 + ξ for 53 °N or °S < ξ < 53 °N or °S
C
Lat
= 7 for ξ < 60 °N or °S
8.8.5.3 Longitude coefficient
C
Lon
= 3 for longitudes of Europe and Africa
C
Lon
= 3 for longitudes of North and South America
C
Lon
= 0 for all other longitudes
8.8.5.4 Climatic factor p
L
The specific value of the refractivity gradient, p
D
= 157 N units/km,
represents the boundary between superrefraction and ducting, thus
becoming the probability for the occurrence of a radio duct. Unlike p
D
,
p
L
values are readily available in the literature. In addition, it has been
found that p
D
and p
L
1.5
are highly correlated. Therefore p
L
values are
currently employed in the estimation of the geoclimatic factor.
NOTE: The p
L
values for the entire world are obtained from the
charts presented in Chapter 155.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
58 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
The highest value, expressed in %, obtained from the four maps should
be used for planning purposes. An exception is when planning for
latitudes greater than 60 °N or 60 °S when the maps of May and August
should be used.
8.8.6 Coastal Links
Coastal links are links having a fraction r
c
of the path profile
• Less than 100 m above a body of water
• Within 50 km of its coastline
• No height of land above the 100 m altitude (relative to the mean
altitude of the body of water in question) between the fraction of the
path profile and the coastline.
8.8.6.1 Coastal links over/near large bodies of water
The size of large bodies of water is considered with respect to several
known examples:
• English Channel
• North Sea
• Large reaches of the Baltic and Mediterranean Seas
• Hudson Strait
• Other bodies of water of similar size
Normally, measured values of the geoclimatic factor K are not
available, but it can be estimated according to
( )
i cl
K r K r
l
K K K
cl c i c
≥ =
⋅ + ⋅ −
n whe 10
log log 1
(96)
K
i
when K
cl
< K
i
ξ ⋅ − ⋅ − −
⋅ ⋅ =
011 . 0 1 . 0 4
0
10 10 3 . 2
C
cl
K
(97)
where K
i
is given by expression (95) and C
0
is obtained from Table 4.
The condition K
cl
< K
i
occurs in a few regions at low and mid latitudes.
The parameter “coastal fraction” (r
c
) is defined as the fraction of the
path profile below 100 m altitude above the mean sea level of the body
of water in question and within 50 km of the coastline, without
intervening height above 100 m altitude. If r
c
= 100%, the link is
completely unhidden, while for r
c
= 0%, the path is completely hidden.
RADIOWAVE PROPAGATION
Ericsson AB 59
4/038 02LZU 102 152, Rev B, June 2003
8.8.6.2 Coastal links over/near mediumsized bodies of water
The size of mediumsized bodies of water is considered with respect to
several known examples
• Bay of Fundy (East Coast of Canada)
• Strait of Georgia (West Coast of Canada)
• Gulf of Finland and other bodies of water of similar size
Normally, measured values of the geoclimatic factor K are not
available, but it can be estimated according to
( )
i cm
K r K r
m
K K K
cm c i c
≥ =
⋅ + ⋅ −
n whe 10
log log 1
(98)
K
i
when
K
cm
< K
i
( )
cl i
K K
cm
K
log log 5 . 0
10
+ ⋅
=
(99)
where K
i
is given by expression (95) and C
0
is given by Table 4. The
condition K
cm
< K
i
occurs in a few regions at low and mid latitudes. The
parameter r
c
is as above.
When the size of the body of water classification (medium or large) in
question is not easy applicable, then the geoclimatic factor should be
estimated according to
( ) ( )
cl cm c i c
K K r K r
K
log log 5 . 0 log 1
10
+ ⋅ ⋅ + ⋅ −
=
(100)
where the parameters are defined above.
8.8.6.3 Links at other regions
There are regions consisting of extensive area of lakes or “water
systems”. Typical example is the region of lakes in southern Finland.
Links not located in coastal areas but near vast area of lakes are
considered as coastal areas and the geoclimatic factor should be
estimated according to
( ) [ ]
cm c i c
K r K r
K
log log 2 5 . 0
10
⋅ + ⋅ − ⋅
=
(101)
where the parameters are defined above.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
60 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
8.8.7 Link and terrain parameters – overview
When topography databases are not available, the alternative “unknown
terrain” and the corresponding three antenna altitude coefficient
alternatives should be employed. When topography databases are
available, the parameter input for the flat fading prediction can
generally be structured according to Figure 27.
Unknown
terrain
Known
terrain
1) Low altitude antenna (0400m)
2) Medium altitude antenna (400700m)
3) High altitude antenna (above 700m)
Coastal Links
Inland Links
Links at other
regions
1) Over/near large
bodies of water
2) Over/near medium
sized bodies of water
1) Low altitude antenna (0400m)
a) Hills
b) Plains
2) Medium altitude antenna (400700m)
a) Hills
b) Plains
3) High altitude antenna (above 700m)
a) Hills
b) Plains
c) Mountains
Figure 27: The structure of the parameter input in the flat fading
prediction function.
8.8.7.1 Estimation of the path inclination
Path inclination is calculated as follows
d
h h
B A
−
= ε
(102)
where
ε = Path inclination, mrad
h
A
= Antenna height + ground elevation at the transmitter, m
RADIOWAVE PROPAGATION
Ericsson AB 61
4/038 02LZU 102 152, Rev B, June 2003
h
B
= Antenna height + ground elevation at the receiver, m
d = Path length, km
A general rule of thumb is that rays will penetrate a duct without being
significantly reflected when the path slope is approximately greater than
0.4° (7 mrad). This would correspond approximately to a path length of
7 km and an antenna height difference of 50 m or a path length of about
3 km and an antenna height difference of 20 m.
8.8.8 Outage due to flat fading
The percentage of time p
w
(%) that the fade margin M is exceeded in the
average worst month is calculated as follows,
( )

.

\

−
−
⋅ + ⋅ ⋅ ⋅ =
10
4 . 1 89 . 0 6 . 3
10 1
M
f d K p ε
ω
(103)
where
K = Geoclimatic factor
d = Path length, km
f = Frequency, GHz
ε = Path slope, mrad
M = Fade margin, dB
The multipath occurrence factor (expressed in ratio) corresponding to
the percentage of time of exceeding M = 0 dB in the average worst
month is given by
100
0
w
p
P =
(104)
8.8.9 Range of values for the climatic factor p
L
The range of the climatic factor and its impact on the fading results is
examined. The probability to exceed fade margin as a function of path
length is displayed for different p
L
settings, ranging from 1% (areas of
high latitudes) to approximately 40% (specific areas in the vicinity of
the equator). Fade margin close to 30 dB and frequency 7 GHz are
normally frequent values in many link applications. The path is
considered horizontal through all calculations, thus giving somewhat
more pessimistic probabilities to exceed fade margin.
Considering the actual parameter settings, the corresponding probability
range to exceed fade margin is extremely large, approximately three
orders of magnitude, see Figure 28.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
62 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
0 10 20 30 40
10
6
10
5
10
4
10
3
10
2
10
1
P
r
o
b
a
b
i
l
i
t
y
t
o
e
x
c
e
e
d
f
a
d
e
m
a
r
g
i
n
,
%
"
#
M = 30 dB
C
Lat
= 0 dB
C
Lon
= 3 dB
f = 7 GHz
ε = 0 deg.
p
L
(%)
" 1
% 5
& 10
' 20
( 30
) 40
* 50
+ 60
# 70
%
&
'
(
C
0
= 0 dB
Path length, km
Figure 28: The probability range to exceed fade margin for climatic
factor in the range 1% and 40%.
8.9 Flat fading (current ITUR model): small percentages of
time
The new ITUR model still employs the concept of geoclimatic
parameter but compared to former models, its calculation demands two
new parameters: point refractivity and surface roughness. The model
gives two alternatives for link design: detailed and quick link design.
8.9.1 Method for detailed link design
8.9.1.1 Geoclimatic factor
The geoclimatic factor is estimated according to the following
expression
42 . 0 d 003 . 0 9 . 3
1
10
− ⋅ − −
⋅ =
a
N
s K
(105)
where dN
1
is the point refractivity gradient and the surface roughness s
a
.
8.9.1.2 Parameters
8.9.1.2.1 Point refractivity
The point refractivity gradient dN
1
is the refractive gradient in the
lowest 65 m of the atmosphere not exceeded for 1% of an average year.
RADIOWAVE PROPAGATION
Ericsson AB 63
4/038 02LZU 102 152, Rev B, June 2003
Values of the point refractivity gradient are stored in the data
DNDZ_01.TXT at the ITUR.
The values of dN
1
are stored as 121rows and 241columns matrix
(121x241) corresponding to each point in a grid system.
It is assumed that the Earth surface is divided in a grid having a 1.5
degree resolution, that is, every square has a side of 1.5 degrees and for
interpolation purposes each square can be considered as a plane (flat)
square. The longitude and latitude of the Earth determine every point
forming the grid. The values of the parameters dN
1
are given from the
above database for every grid point.
The values of the longitude and latitude for all grid points are also
stored as 121rows and 241columns matrix (121x241) and can be
obtained from data files at the ITUR. The files are as follows:
DNDZ_LON.TXT ! contains the longitude values for each dN
1
grid
point.
DNDZ_LAT.TXT ! contains the latitude values for each dN
1
grid
point.
For each specific grid point (LON
i
, LAT
j
) there will be dN
1ij
corresponding values.
Values of dN
1
for other geographical locations than the grid points
given in the above matrices are obtained by twodimensional
interpolation technique.
NOTE: Values of dN
1
all over the world are displayed in the charts
presented in Chapter 154.
Figure 29 shows the variations of the point refractivity gradient all over
the world where darker regions symbolize strong refractivity activities.
Figure 29: Point refractivity gradients. Stronger refractivity activities
are encountered in darker regions.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
64 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
8.9.1.2.2 Surface roughness
The surface roughness s
a
is defined as the standard deviation of terrain
heights (m) within a 110 km x 110 km area with a 30 arc seconds
resolution.
The elevation data is found at the Global Topographic 30 arc seconds
(GTOPO30) database. It is a global Digital Elevation Model (DEM)
with a horizontal grid spacing of 30 arc seconds (approximately 1
kilometer). GTOPO30 was derived from several raster and vector
sources of topographic information. For easier distribution, GTOPO30
has been divided into tiles, which can be selected from the map shown
in http://edcdaac.usgs.gov/gtopo30/gtopo30.html.
8.9.1.3 Outage due to flat fading
The percentage of time p
w
(%) that the fade margin is exceeded in the
average worst month is given as follows
( )
10
00085 . 0 032 . 0 97 . 0
2 . 3
10 1
M
h f
p w
L
d K p
− ⋅ − ⋅ −
⋅ + ⋅ ⋅ = ε
(106)
where
K = Geoclimatic factor (obtained in section 8.9.1.1)
d = Path length, km
f = Frequency, GHz
ε = Path slope, mrad
h
L
= Altitude of the lower antenna, m
M = Fade margin, dB
The accuracy of the above model gives an overall standard deviation of
error in fading predictions that is 5.7 dB.
8.9.2 Method for quick link design
8.9.2.1 Geoclimatic factor
The geoclimatic factor is estimated according to the following
expression
1
0029 . 0 2 . 4
10
dN
K
⋅ − −
=
(107)
RADIOWAVE PROPAGATION
Ericsson AB 65
4/038 02LZU 102 152, Rev B, June 2003
where dN
1
is the point refractivity gradient. The parameter dN
1
is
described in section 8.9.1.2. Note that the surface roughness s
a
is not
employed in the method for quick link design.
8.9.2.2 Outage due to flat fading
The percentage of time p
w
(%) that the fade margin is exceeded in the
average worst month is given as follows
( )
10
001 . 0 033 . 0 2 . 1
0 . 3
10 1
M
h f
p w
L
d K p
− ⋅ − ⋅ −
⋅ + ⋅ ⋅ = ε
(108)
where the parameters are defined in the previous section.
The accuracy of the above model gives an overall standard deviation of
error in fading predictions that is 5.9 dB.
The outage due to flat fading is finally obtained in accordance with
equation (104),
100
w
ns
p
P =
(109)
where P
ns
is expressed in ratio and p
w
(%) is determined by (106) or
(108), as appropriate.
8.9.3 Method for small percentage of time  conclusion
The wide range of the parameter values has considerable effects on the
flat fading results. In fact, the influence of some parameters on the
model is rather noticeable.
Figure 30 displays two extreme planning alternatives: “easy” and
“difficult”. Typical for the “easy” alternative are lower frequencies and
low p
L
values, high altitude antenna situated on mountains in the
vicinity of the equator at longitude corresponding to North and South
America. Typical for the “difficult” alternative are higher frequencies
and high p
L
values, low altitude antenna situated on plains far from the
equator at longitudes corresponding to Europe and Africa. The range
displayed in Figure 30 is approximately three orders of magnitude and
it might be wider for regions with stronger refraction properties (high p
L
values).
Since flat fading is one of the major contributors to severely errored
seconds (SES), the wide range of parameter values plays a decisive part
when dimensioning the path length to fulfill the actual error
performance objectives. The transmission network planners should
therefore select the parameter values with caution.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
66 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
0 10 20 30 40
Path l ength, km
10
6
10
5
10
4
10
3
10
2
10
1
P
r
o
b
a
b
i
l
i
t
y
t
o
e
x
c
e
e
d
f
a
d
e
m
a
r
g
i
n
,
%
C
0
= 10.5 dB
p
L
= 1%
C
Lat
= 0 dB
ε = 0 deg
M = 30 dB
C
Lon
= 3 dB
p
L
=40%
f = 7 GHz
C
0
= 0 dB M = 30 dB
C
Lat
= 7 dB
ε = 0 deg C
Lon
= 3 dB
f= 2 GHz
Figure 30: Two extreme planning alternatives: “easy” (the lowest
curve) and “difficult”.
8.9.4 Method for all percentages of time
The prediction method described below combines an empirical
interpolation procedure between the deep fading region of the
distribution and 0 dB with the prediction method for small percentages
of time described in the previous section.
The interpolation procedure is performed in the following steps:
1) The outage due to flat fading given by equation (106) and (108) is a
function of the fade margin M. In this first step the intercept of the
deepfading distribution is obtained by setting M = 0. The multipath
occurrence factors for detailed and quick design are p
0
, respectively
( )
L
h f
p
d K p
⋅ − ⋅
−
⋅ + ⋅ ⋅ =
00085 . 0 032 . 0
97 . 0
2 . 3
0
10 1 ε
(110)
( )
L
h f
p
d K p
⋅ − ⋅
−
⋅ + ⋅ ⋅ =
001 . 0 033 . 0
2 . 1
0 . 3
0
10 1 ε
(111)
where p
0
, is expressed in percentage (%) and the other parameters are
specified as above.
2) There will be a transition between the deepfading distribution given
by equation (106) and (108) and the shallow fadingdistribution. The
fade depth for which the transition occurs is given by
RADIOWAVE PROPAGATION
Ericsson AB 67
4/038 02LZU 102 152, Rev B, June 2003
0
log 2 . 1 25 p A
t
⋅ + =
(112)
Where p
0
is given by equation (110) and (111), as appropriate.
3A) If M ≥ A
t
, calculate the percentage of time p
w
(%) that M is
exceeded in the average worst month by
10
0
10
M
w
p p
−
⋅ =
(113)
The above equation is similar to equation (106) and (108).
3B) If M < A
t
, calculate the percentage of time p
t
(%) that M is
exceeded in the average worst month by
10
0
10
t
A
t
p p
−
⋅ =
(114)
The above equation is similar to equation (106) and (108) in which M =
A
t
.
4) For the transition given by the coordinate (A
t
, p
t
), the parameter q
a
’ is
calculated as follows
( )
)
`
¹
¹
´
¦
−
− ⋅ − =
100
100
ln log
20
' t
t
a
p
A
q
(115)
5) A new parameter q
t
is obtained from q
a
’ and the transition fade A
t
.
( )


.

\

+ ⋅ −
⋅


.

\

⋅ +
−
=
−
⋅ −
−
800
10 3 . 4
10 10 3 . 0 1
2
20
016 . 0
20
'
t
A
A
A
a
t
A q
q
t
t
t
(116)
6) The parameter q
a
corresponding to the required fade margin M is
determined as follows
[ ]


.

\

+ ⋅ + ⋅ ⋅
⋅ + + =
−
⋅ −
−
800
10 3 . 4 10 10 3 . 0 1 2
20
016 . 0
20
M
q q
M
t
M
M
a
(117)
7) Finally, the percentage of time p
w
(%) for which the fade margin M is
exceeded in the average worst month is calculated as follows
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
68 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
− ⋅ =
⋅
−
−
20
10
e 1 100
M
a
q
p
ω
(118)
If p
0
given by equation (110) or (111) is p
0
<2000, the equation (118)
will be a monotonic function p
w
(M) and can easily be employed for
obtaining M for a given value of p
w
.
The prediction methods for small percentages of time and various
percentages of time are compared in Figure 31 for a path length of 20
km and with the previous parameter setting. For fade margin values
normally employed in most link applications, the output of both
methods are comparable, then making the method for small percentages
of time more suitable since it is relatively straightforward.
0 10 20 30 40 50
Fade margi n, dB
10
5
10
4
10
3
10
2
10
1
10
0
10
1
10
2
P
r
o
b
a
b
i
l
i
t
y
t
o
e
x
c
e
e
d
f
a
d
e
m
a
r
g
i
n
,
%
C
0
= 0 dB
p
L
= 5%
f = 7 GHz
d = 20 km
ε = 0 deg
C
Lat
= 0 dB
Small percentages
of time
Various percentages
of time
C
Lon
= 0 dB
Figure 31: Comparison between the prediction methods for small
percentages of time and various percentages of time
8.9.5 Range of validity for the flat fading method
The current ITUR flat fading prediction model described above has
been derived from multiple regressions on available fading data for 251
links in various geoclimatic regions of the world. The parameter ranges
are as follows:
185 ≥ d ≥ 7.5 km
37 ≥ f ≥ 0.45 GHz
RADIOWAVE PROPAGATION
Ericsson AB 69
4/038 02LZU 102 152, Rev B, June 2003
37 ≥ ε
2 300 ≥ h
L
≥ 17 m
150 ≥ dN
1
≥ 860 Nunit/km
850 ≥ s
a
≥ 6 m (for s
a
< 1 use a lower limit of 1 m)
The models are expected to be valid for frequencies to at least 45 GHz
while the lower frequency limit is estimated to be inversely proportional
to the path length and can be roughly estimated by
d
f
15
min
=
(119)
8.10 Reduction of crosspolar discrimination
Crosspolar discrimination (XPD) maybe reduced when a fraction of
the energy transmitted in one polarization state is transferred to the
orthogonal polarization state. The consequence of this effect is to cause
interference between two channels. In addition, crosspolarization may
arise from characteristics of the antenna systems in each terminal.
Reduction of crosspolar discrimination (XPD) maybe caused by rain
and other hydrometers as well as under periods of multipath
propagation. In what follows, the effects affecting crosspolar
discrimination due to multipath propagation and rain is described.
8.10.1 XPD outage due to multipath propagation
Multipath propagation combined with crosspolarization patterns of the
antennas contributes to lower the crosspolarization discrimination
(XPD) for small percentages of time. Reductions in the XPD can be
calculated with the following steps:
1) Determine the following parameter
35 for 40
35 for 5
0
¦
¹
¦
´
¦
>
≤ +
=
g
g g
XPD
XPD XPD
XPD
(120)
where XPD
g
is the minimum XPD at boresight for both transmitting and
receiving antennas guaranteed by the manufacturer.
2) Evaluate the multipath activity parameter as follows
( )
75 0
0
2 0
e  1
.
P .  ⋅
= η
(121)
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
70 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
where P
0
is the multipath occurrence factor given by equation (110) or
(111),as appropriate.
3) Determine the parameter Q as follows:


.

\
 ⋅
⋅ − =
0
log 10
P
k
Q
XP
η
(122)
where η and P
0
are given as described above and k
XP
is given as follows
antenna transmit for two e 0.3  1
antenna transmit one for 75 . 0
2
6 
10 4 
¦
¹
¦
´
¦
⋅
=


.

\

⋅ ⋅
λ
t
s
XP
k
(123)
where S
t
(m) is the vertical antenna separation and λ is the carrier
wavelength given by
f
2997925 . 0
= λ
(124)
Where λ is expressed in m and f in GHz.
4) Estimate the parameter C
Q XPD C + =
0
(125)
where XPD
0
is given by equation (120) and Q by equation (122).
5) Determine the probability of outage due to multipath cross
polarization
10
0
10
XPD
M
XP
P P
−
⋅ =
(126)
where P
0
is the multipath occurrence factor given by equation (110) or
(111),as appropriate, and M
XPD
is as follows
XPIC th wi
XPIC without
0
0
¦
¦
¹
¦
¦
´
¦
+ −
−
=
XPIF
I
C
C
I
C
C
M
XPD
(127)
RADIOWAVE PROPAGATION
Ericsson AB 71
4/038 02LZU 102 152, Rev B, June 2003
where C
0
/I is the carriertointerference ratio for a given reference BER.
XPIF is a crosspolarization improvement factor measured in
laboratory. This factor expresses the difference between cross
polarization (XPI) at sufficiently large carriertonoise ratio (typically
35 dB) and at a specific BER for systems with and without crosspolar
interference canceller (XPIC).
8.10.2 XPD outage due to precipitation
Depolarization phenomena induced by rain depend on rain intensity,
raindrop size distribution and the tilting angle.
1) If CPA is considered as a rain induced level, the unconditional XPD
cumulative distribution is obtained as follows
( ) ( ) CPA f V U XPD log ⋅ − =
(128)
where XPD is expressed in dB and the coefficients U and V(f) are
normally dependent on a number of variables and empirical parameters.
Assuming LOSpaths having small elevation angles and horizontal or
vertical polarization, the coefficients U and V(f) can be obtained as
( ) f U U log 30
0
⋅ − =
(129)
GHz 20 8 for 8 . 12
19 . 0
≤ ≤ ⋅ = f f V
(130)
GHz 35 20 for 6 . 22 ≤ < = f V
(131)
where U
0
is normally around 15 dB.
2) Determine the path attenuation A
0.01
as given by equation (68).
3) Evaluate the equivalent path attenuation A
P
as follows

.

\

+
−
=
V
XPIF
I
C U
P
A
0
10
(132)
where A
P
is expressed in dB, U and V(f) are given by (129), (130) and
(131), respectively. C
0
/I is the carriertointerference ratio for a given
reference BER and XPIF (dB) is the crosspolarized improvement
factor for the reference BER. If an XPIC equipment is not employed,
XPIF=0.
4) Evaluate the parameters m and n as follows
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
72 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
otherwise 40
40 if
12 . 0
log 3.26 2
01 . 0
¦
¹
¦
´
¦
≤


.

\

⋅
⋅
=
m
A
A
m
P
(133)
where A
P
is given by equation (132) and A
0.01
is given by equation (68)
2
4 23 . 161 7 . 12 m
n
⋅ − + −
=
(134)
The values of n must be in the range –3 and 0.
5) Finally, the outage probability P
XPR
(expressed in ratio) due to
precipitation effects is obtained as follows
( ) 2
10
−
=
n
XPR
P
(135)
8.11 Outage due to frequency selective fading
8.11.1 General aspects
In the case of frequency selective fading, waves from different paths
interfere with one another at the receiver. The different propagation
waves can often be the result of ground reflections, reflections in the
ducting layer or propagation in layers having highly positive refraction
gradients. Layers having horizontal structures can also result in
frequency selective fading.
Occasionally, the various constituents unite with one another so that a
fieldstrength minimum arises in the case of certain frequencies.
Atmospheric layer movement (changes in path geometry) causes these
minimums to shift across the frequency band. The speed of such
shifting can vary from tens to hundreds of MHz/sec.
Frequency selective fading is often characterized with by specifying the
difference in propagation time between the direct and indirect waves.
The propagation time differences are in turn a function of certain path
parameters (e.g., path length and path inclination) as well as
meteorological parameters.
The first cause of frequency selective fading is inband distortion,
which can be described with the aid of the transfer function’s slope
within the frequency band.
RADIOWAVE PROPAGATION
Ericsson AB 73
4/038 02LZU 102 152, Rev B, June 2003
Measurements performed in Sweden and simulations employing the 3
path model have shown that a slope of a minimum of 0.22 dB/MHz is
attained within bandwidth B if the relative delay, τ, fulfills the
following relation
B
50
≥ τ
(136)
where
τ = Relative delay, ns
B = Bandwidth, MHz
The choice of the 0.22 dB/MHz threshold for the transfer function’s
slope is directly related to the fact that inband distortion has proven to
cause system outage at values as low as 0.2 dB/MHz.
For a bandwidth B=50 MHz, the relative delay, τ, becomes ≥ 1 ns and
multipath fading is classified as being frequency selective. For systems
having smaller bandwidths, the relative delay is longer for a given path
length which means that the system becomes less sensitive to frequency
selective fading, since longer relative delays are less probable than
shorter relative delays.
A rule of thumb is that multipath fading, for radio links having
bandwidths less than 40 MHz and path lengths less than approximately
30 km, is described as being flat instead of frequency selective.
The prediction of frequency selective fading is a very difficult task.
There exist many different prediction models, the results of which
unfortunately deviate considerably from one another. With the
exception of the fact that contributions from flat and frequency selective
fading are not weighted but are additive, the prediction models
described here is the model specified in the ITUR F.1093
recommendation.
8.11.2 The prediction model
The probability of the occurrence of frequency selective fading is
obtained by
mp s s
P p
/
⋅ =η
(137)
where
Ps/mp = Probability of the occurrence of fading caused by
intersymbol interference during multipath fading
η = Probability of the occurrence of multipath fading
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
74 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
The propagation parameter η is empirically obtained by the following
expression


.

\

⋅ −
− =
4
3
0
2 . 0
e 1
P
η
(138)
where P
0
(expressed in ratio) is the multipath occurrence factor and is
expressed by expression (104).
Different types of distributions can characterize the echo delay τ. This
method employs an empirical relation, which assumes exponentially
distributed delays. Thus, it is expressed as follows
n
m m
d

.

\

⋅ =
50
0
τ τ
(139)
where
τ
m
= Mean value of the echo delay, ns
τ
m0
= Mean relative delay for a standard path of 50 km, ns
D = Path length, km
n = normalization exponent with values in the range of 1.3 and 1.5.
The mean relative delay for a standard path, τ
m0
, is usually about 0.7
seconds for exponentially distributed delays.
The probability of the occurrence of fading due to intersymbol
interference during multipath fading can be written as follows
( )
r
m
B
b
mp s
W p C
P
τ
τ
2
20
/
2 10 1 ⋅ ⋅ ⋅ ⋅ ⋅
=
−
(140)
where
P
s/mp
= Probability of the occurrence of fading due to intersymbol
interference during multipath fading
C = Constant factor
p
b
(1) = The value of p
b
when b=1
W = Signature width, GHz
B = Signature depth, dB
τ
r
= Reference delay for λ
a
(average of linear signature), ns
RADIOWAVE PROPAGATION
Ericsson AB 75
4/038 02LZU 102 152, Rev B, June 2003
The value of the product C⋅p
b
(1) is usually 2.16 and the value of the
reference delay τ
r
is 6.3 ns.
Inserting (138) and (140) in (137) the selective outage probability P
s
(expressed in ratio) is obtained by
( )
r
m
B
b
s
W p C
P
τ
τ η
2
20
2 10 1 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
=
−
(141)
where the parameters are as previously defined.
8.12 Refraction fading
Refraction fading, also known as ktype fading, is characterized by the
fact that a lower earthradius factor, k, causes the effective earth radius
to be less (the curvature of the effective earth surface becomes larger).
This, in turn, may cause earth surface irregularities (buildings,
vegetation, mountains, etc.) to penetrate the first Fresnel zone and cause
obstruction attenuation. The lower the values of the earthradius factor
the smaller the effective earth radius and the greater the obstruction
attenuation.
The probability of refraction fading is therefore coupled to obstruction
attenuation for a given value of earthradius factor. Since the earth
radius factor is not constant, the probability of refraction fading is
calculated based on the cumulative distribution of the earthradius
factor.
The probability of refraction fading is calculated in four steps:
1. A table is first constructed containing probabilities that the various k
values will not be exceeded. A kvalue distribution table for any
specific p
L
factor is employed, in order to interpolate the probabilities
for given kvalues.
2. Obstruction attenuation values for the given kvalues in the above
table are calculated based on an algorithm selected by the user.
3. The calculated obstruction attenuation values then replace the k
values in the table. The earlier table is transformed into a new table
containing the probabilities in which the different obstruction
attenuation values will be exceeded.
4. Finally, fade margin is coupled to obstruction attenuation by
applying the link budget and the probabilities are calculated for which
fade margin values are exceeded.
P
dr
represents the outage due to refractiondiffraction fading.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
76 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
9 Diversity
9.1 Basic concepts
Systems that include doubled sets of equipment or systems that transmit
the same signal in parallel over two or more radio channels having
different frequencies are referred to as being redundant systems. In this
respect, redundancy and diversity have the same meaning.
Diversity receiving is an effective means of reducing the effects of,
above all, multipath fading, where the received signal is the vector
addition of multipath components that primarily vary in time, phase and
angle of arrival.
Random signal variations often occur during very short periods of time
and may very well be described with the aid of the Rayleigh
distribution. One utilizes the fact that deep fade in radio channels that
transmit the same information but are sufficiently separated in, for
example, frequency and/or space, have low correlation. The lower the
correlation, the higher is the improvement gained by the use of
diversity. In practice, good improvement can already be noticed at a
correlation of 0.6. Diversity is therefore a method that provides
statistically independent multipath components at the receiver.
Two fading depth statistical distributions are compared when measuring
the diversity improvement on a radio connection: one for the path with
diversity and one for the path without diversity. The measured
improvement can be expressed in two different ways: diversity gain and
diversity improvement.
Figure 32 illustrates two fadedepth statistical distributions, for one and
the same path. Points A (without diversity) and C (with diversity)
correspond to two different fading levels having the same probability.
The measured improvement is referred to as diversity gain and is
expressed in dB. Points A (without diversity) and B (with diversity)
correspond to two different probabilities for the same level of fading
and is referred to as the improvement and is expressed as a factor.
Diversity improvement can therefore be expressed as the ratio of two
probabilities.
RADIOWAVE PROPAGATION
Ericsson AB 77
4/038 02LZU 102 152, Rev B, June 2003
Probability of exceeding the fading depth, %
F
a
d
i
n
g
d
e
p
t
h
,
d
B
Without diversity
With diversity
improvement
g
a
i
n
A
B
C
10
2
10
3
10
4
10
5
10
6
10
7
10
20
30
40
50
0
Figure 32: Two fadedepth statistical distributions for one and the same
path, without and with diversity.
9.2 The definition of the improvement factor
Diversity is primarily utilized to reduce the effects of multipath fading.
The improvement factor can therefore be associated with the statistical
cumulative distribution of fading depth during the year’s worst month
in accordance with the prediction model for flat multipath fading.
In the calculation of the improvement factor for digital links, the
expression for analogue or narrowband systems is adjusted in the case
of frequency diversity. For space diversity, however, it is used the same
expression in the calculation of the improvement factor for both analog
and digital or narrowband systems.
As shown in Figure 32, the improvement factor can be defined as the
ratio of two probabilities: with the use of diversity and without the use
of diversity. Note however, that the improvement implies that the
probability of a given event will be even less, i.e., outage will increase.
( )
( ) M P
M P
I
with
without
=
(142)
where
I = Improvement factor
P
without
(M) = Probability that the fading depth will be greater than
or equal to M dB during the worst month for a path without
diversity, (%)
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
78 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
P
with
(M) = Probability that the fading depth will be greater than or
equal to M dB during the worst month for a path with diversity,
(%)
9.3 Improvement factor for space diversity
The improvement factor due to the use of space diversity is calculated
here using the same algorithm for both analog and digital links
( )
[ ]
10
04 . 0
10 e 1
04 . 1
0
48 . 0 12 . 0 87 . 0
G M
p d f S
I
∆ −
⋅ ⋅ ⋅ ⋅ −
⋅ − =
− −
(143)
where
I = Improvement factor for analog and digital links
s = Vertical separation between the antennas, m
f = Frequency, GHz
d = Path length, km
M = Fade margin, dB
∆G= Difference in antenna gain between the two antennas, dB
The multipath occurrence factor p
0
(%) is calculated by the expression
(110) or (111), as appropriate.
The prediction model is considered as giving valid results within the
following interval:
3 ≤ S ≤ 23 m
2 ≤ f ≤ 11 GHz
43 ≤ d ≤ 240 km
The model’s validity for values outside of these boundaries is not
certainly known, but there is some reason to believe that it may be valid
to path lengths as short as 25 km.
9.4 Improvement factor for frequency diversity
The frequencydiversity improvement factor for a 1+1 analogue system
or path without strong surface reflections can be calculated as follows:
10
10
80
M
ns
f
f
d f
I ⋅


.

\
 ∆
⋅
⋅
=
(144)
where
I
a
= Improvement factor
RADIOWAVE PROPAGATION
Ericsson AB 79
4/038 02LZU 102 152, Rev B, June 2003
f = Carrier frequency, GHz
∆f = Frequency separation, GHz. If ∆f > 0.5 GHz use ∆f = 0.5
d = Path length, km
M = Fade margin, dB
The prediction model is considered as giving valid results within the
following interval:
30 ≤ d ≤ 70 km
2 ≤ f ≤ 11 GHz
∆f /f ≤ 5 %
The validity of the model for values outside of these boundaries is
unknown.
9.5 The calculation of outage when employing diversity
When calculating the improvement brought about by the use of
diversity, it is important to remember that the probability that is referred
to above is the outage due to flat multipath fading for the worst month.
Following the calculation of the improvement factor, the outage in
conjunction with the use of diversity can be attained from equation
(142).
( )
( )
I
M P
M P
without
with
=
(145)
9.6 Prediction of outage using diversity
In what follows, methods for the calculation of outage due to multipath
fading using space and frequency diversity are presented. The basic for
both methods is the principle illustrated in Figure 32 and given by
equation (142).
9.6.1 Space diversity
The method presented below is employed for outage predictions using
maximumpower combiners. Other combiners may give better
performance. The method can be summarized in the following steps:
1) Determine the multipath activity factor as given by equation (121).
2) Determine the square of the nonselective correlation coefficient
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
80 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
η
ns ns
ns
P I
k
⋅
− =1
(146)
where I
ns
(improvement factor) and P
ns
(expressed in ratio) are given by
equation (143) and (109), respectively. The multipath activity factor as
given by equation (121).
3) Determine the square of the selective correlation coefficient as
follows
( )
( )
( )
9628 . 0 for 1 3957 . 0  1
9628 . 0 0.5 for 1 195 . 0  1
5 . 0 for .8238 0
5136 . 0
1 log 13 . 0 109 . 0 2
¦
¹
¦
´
¦
> ⋅
≤ < ⋅
≤
=
− ⋅ −
w w
w
r
w
w
s
r r
r r
r
k
w
(147)
where the correlation coefficient r
w
of the relative amplitudes is given
by
( )
( )
26 . 0 for 1 .6921 0  1
26 . 0 for 1 9746 . 0  1
2
034 . 1
2
2
170 . 2
2
w
¦
¹
¦
´
¦
≥ ⋅
≤ ⋅
=
ns ns
ns ns
k k
k k
r
(148)
4) Applying equation (145), determine the outage for flat fading (non
selective fading) using space diversity
ns
ns
dns
I
P
P =
(149)
where P
dns
is the outage (expressed in ratio) for flat fading when using
diversity, P
ns
(expressed in ratio) is the outage for flat fading without
space diversity given by equation (109) and I
ns
the improvement factor
given by equation (143).
5) Determine the outage for frequency selective fading using space
diversity
( )
2
2
1
s
s
ds
k
P
P
− ⋅
=
η
(150)
RADIOWAVE PROPAGATION
Ericsson AB 81
4/038 02LZU 102 152, Rev B, June 2003
where P
ds
is the outage (expressed in ratio) for frequency selective
fading when using diversity, P
s
(expressed in ratio) is the outage for
frequency selective fading without space diversity given by equation
(141), η is the multipath activity factor as given by equation (121) and
k
s
2
is given by (147).
6) Finally, calculate the total outage probability for multipath fading
when employing space diversity as follows
( )
3
4
75 . 0 75 . 0
dns ds d
P P P + =
(151)
where P
d
is expressed in ratio.
9.6.2 Frequency diversity
The method for the calculation of outage due to multipath fading using
frequency diversity is similar to the method employed for space
diversity except in step 2 where the improvement factor I
ns
in equation
(146) is calculated in accordance with equation (144).
All the other steps follow the method described in the previous section
for space diversity.
9.6.3 Space and frequency diversity with two receivers
The method for the calculation of outage due to multipath fading using
a combination of space and frequency diversity with two receivers is
similar to the method employed for space diversity, except for the
calculation of the nonselective correlation coefficient k
ns
which is
given in accordance to the following expression
f ns s ns ns
k k k
, ,
⋅ =
(152)
where k
ns,s
and k
ns,f
are the nonselective correlation coefficients for
space and frequency diversity, respectively. The correlation coefficient
k
ns,s
is obtained by using equation (146) using I
ns
as given by equation
(143) and k
ns,f
is also obtained by using equation (146) but employing
I
ns
as given by equation (144).
All the other steps follow the method described section 9.6.1 for space
diversity.
9.6.4 Space and frequency diversity with four receivers
The method for the calculation of outage due to multipath fading using
a combination of space and frequency diversity with four receivers is
somewhat similar to previous methods.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
82 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
1) Evaluate the multipath activity factor η by using equation (121).
2) Calculate the diversity parameter m
ns
as follows
( ) ( )
2
,
2
,
3
1 1
f ns s ns ns
k k m − ⋅ − ⋅ = η
(153)
Where the correlation coefficients k
ns,s
and k
ns,f
are obtained as
described in the previous section.
3) Evaluate the nonselective (flat) outage probability P
dns
as follows
ns
ns
dns
m
P
P
4
=
(154)
where P
dns
is expressed as ratio, P
ns
(expressed in ratio) is obtained
from equation (109) and the parameter m
ns
is given by (153).
4) Evaluate the square of the nonselective (flat) correlation coefficient
k
ns
as follows
( ) ( )
2
,
2
,
2
1 1 1
f ns s ns ns
k k k − ⋅ − ⋅ − = η
(155)
Where η is given by equation (121) and the correlation coefficients k
ns,s
and k
ns,f
as before.
5) Determine the selective correlation coefficient k
s
as given by
equation (147) and (148).
6) The selective outage probability P
d
(expressed in ratio) is obtained
by
( )
2
2
2
1
− ⋅
=
s
s
dns
k
P
P
η
(156)
where P
s
(expressed in ratio) is the outage for frequency selective
fading without diversity given by equation (141).
7) Finally, the total outage probability P
d
(expressed in ratio) for
multipath fading when employing diversity with four receives is given
by equation (151).
10 Prediction of total outage
Now the outages due to multipath propagation and rain, with and
without diversity (protection) is calculated as follows
RADIOWAVE PROPAGATION
Ericsson AB 83
4/038 02LZU 102 152, Rev B, June 2003
employed is diversity if
¹
´
¦
+
+ + +
=
P P
P P P P
P
XP d
dr XP s ns
t
(157)
Where the first row is for unprotected systems (diversity is not used)
and the second row is for protected systems (diversity is used). All the
contributions are calculated in the previous sections.
11 Hardware failure
Hardware failure is calculated for systems with and without
redundancy. Passive redundancy applies to redundant systems
configurations including monitored hot standby.
The word standby, as used here, implies that a ”reserve” component is
connected when replacing one that has failed, i.e., passive redundancy.
Hot refers to the fact that the ”reserve” component functions optimally
from the point at which it is introduced into the system, no ”warm
up/switchover” is therefore required. Monitored implies electronic
control/supervision.
11.1 The calculation of the radiolink system’s MTBF
The MTBF (Mean Time Between Failure) for a particular equipment
can be arrived at both theoretically and practically. Theoretical MTBF
values are attained in accordance with certain reliability models that are
applied to the equipment. This is performed on a component level and
is a highly complex operation since many different parameters may be
involved. Practical MTBF is, on the other hand, somewhat simpler to
estimate via lifecycle testing aided by the collection of reliability data
from the equipment that is included in numerous systems. However, to
arrive at reliable MTBF values, it is very important that the number of
investigated units (the population), from which reliability data is
collected, is sufficiently large.
Modern radiolink equipment exhibits very high availability. MTBF
values between 10 and 15 years are no longer unusual. Radiolink
manufacturers often specify the total MTBF of their radiolinks, where
the MTBFs of the individual components (Mux, BB, MF and RF units,
power supplies, etc.) are included. It is however important that one
always check which elements are included in the equipment’s MTBF
before starting unavailability calculations.
The system’s total mean time between failures, MTBF
S
, can be
expressed as a function of the component’s mean time between failure,
MTBF
i
, as follows
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
84 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
∑
=
=
n
i i
S
MTBF
MTBF
1
1
1
(158)
where
MTBF
s
= System’s total mean time between failure, years
MTBF
i
= Component’s individual mean time between failure, years
n = Number of components in the system
11.2 Nonredundant systems
The probability of hardware failure for nonredundant systems is
calculated as follows
100
8760
8760
⋅
+
=
MTTR
MTBF
MTTR
p
S
s
(159)
where
p
s
= Probability of hardware failure for a nonredundant system, %
MTBF
s
= System’s total mean time between failure, years
MTTR = The mean time to restore, hours
The factor (1/8760) transforms MTTR from hours into years and
consequently has the same units as MTBF
S
. Mean time to repair, MTTR
(Mean Time To Restore), is defined as the duration of the interruption.
As a rule, this time consists of the travel time required between a
manned supervising station and the station containing the failed
equipment plus the actual repair time. It is important to note that the
waiting time that always arises in connection with the ordering and
delivery of spare parts is often not included in MTTR. The mean time to
restore concept assumes that spare parts are always available when
failure occur.
MTTR may be considered as a measure of a system’s maintainability
and is always, for practical purposes, specified in hours.
Figure 33 illustrates, for four different values of MTTR, the probability
of hardware failure of a nonredundant system as a function of MTBF.
RADIOWAVE PROPAGATION
Ericsson AB 85
4/038 02LZU 102 152, Rev B, June 2003
0 3 6 9 12 15
MTBF, years
0.00
0.02
0.04
0.06
0.08
0.10
P
r
o
b
a
b
i
l
i
t
y
o
f
h
a
r
d
w
a
r
e
f
a
i
l
u
r
e
,
%
MTTR= 6 hours
MTTR= 12 hours
MTTR= 24 hours
MTTR= 48 hours
0.10
Figure 33: Hardware failure of a nonredundant system as a function of
the MTBF.
11.3 Redundant systems
The probability of hardware failure for redundant systems is calculated
as follows
100
1
8760
8760 8760
⋅
+
⋅
+
⋅
⋅
⋅
=
u
u
u
S S
s
MTTR
MTTR
MTBF
MTTR
MTBF
MTTR
MTBF
MTTR
p
(160)
where
p
r
= Probability of hardware failure of a redundant system, %
MTBF
s
= System’s total mean time between failure of one the
duplicated equipment, years
MTBF
u
= Mean time between failure of the nondoubled (non
redundant) equipment (baseband distributor + switch), years
MTTR = Mean time to restore of one of the doubled (redundant)
units of a redundant system, hours
MTTR
u
= Mean time to restore of the nondoubled (non redundant)
equipment (baseband distributor + switch), hours
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
86 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
The function of the switchunit is to automatically switch traffic from
failed equipment to equipment that is in proper operating condition. The
comments above therefore only apply under the premise that a switch
unit fault does not cause total system failure due to the fact that the
switch is required for system recovery. This means that traffic continues
via the remaining operational equipment even if the switchunit and one
of the doubled components are not operating.
The probability of hardware failure of a redundant system (monitored
hot standbyconfiguration) as a function of MTBF has been calculated
for four different values of MTTR and is illustrated in the figure below.
In the calculation, the values of MTBF
u
and MTTR
u
, for the non
doubled equipment (the base band distributor + switch), have been set
to 10 years and 12 hours, respectively.
0 3 6 9 12 15
MTBF, years
10
6
10
4
10
4
P
r
o
b
a
b
i
l
i
t
y
o
f
h
a
r
d
w
a
r
e
f
a
i
l
u
r
e
,
%
MTTR= 6 hours
5
MTTR= 12 hours
MTTR= 24 hours
MTTR= 48 hours
Figure 34: The probability of hardware failure of a redundant system as
a function of the MTBF.
11.4 Hardware failure per path
The probability of hardware failure is calculated per path. The
assumption is, however, that the input parameters apply to the entire
radiolink system in question. This means, therefore, that the MTBF of
the radiolink system includes all individual components, including
their respective MTBF values, for both the transmitter and the receiver
in a duplexsetup configuration.
RADIOWAVE PROPAGATION
Ericsson AB 87
4/038 02LZU 102 152, Rev B, June 2003
The calculation of the probability of hardware failure is performed
separately for each station. The total probability of hardware failure for
each direction of the path (go and return) is obtained by adding the
hardware failure contributions of both stations.
The calculated probability of hardware failure is given in percent per
year.
The following parameters are required:
• Configuration: redundant, nonredundant or no calculation of
unavailability due to hardware failure.
• The path’s mean time between failure for the doubled equipment,
years
• The path’s mean time between failure for the nondoubled
equipment, years
• The mean time to restore for the doubled units of a redundant
system, hours
• The mean time to restore for the nondoubled units, hours
12 Passive repeaters
12.1 The basic concepts
Passive repeaters generally consist of larger mirrored surfaces than
those of reflectors and are often used either on mountain peaks as plane
reflectors or in conjunction with certain applications when they are
referred to as backtoback reflectors.
Backtoback reflectors are discussed here. These reflectors consist of
two parabolic antennas, often mounted backtoback on a low mast and
are connected to one another by a short wave guide/cable. There exists
therefore no direct connection from either a transmitter or receiver to
these parabolic antennas. They are primarily used in cities where line
ofsight transmission is not possible due to buildings and other
obstructions.
Path calculation in connection with the use of backtoback reflectors
requires the calculation of
• Received and radiated power of the repeaters
• Attenuation
• Fade margin
The intention behind the use of backtoback reflectors is consequently
to influence the final quality of the path.
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
88 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
12.2 Path calculation when using passive repeaters
An added attenuation arises when using passive repeaters between two
stations due to the fact that the path between A and B is calculated as
two independent paths: from station A to the repeater R, see Figure 35,
and from the repeater to station B.
The added attenuation consists of obstruction, gas attenuation and free
space loss and affects the total path attenuation between A and B.
A
B
R
d
A
d
B
Figure 35: A passive repeater consisting of two parabolic antennas
mounted backtoback.
Before calculating the added attenuation, one must first calculate the
received and the radiated power of the repeater.
A repeater can be considered as being a directional antenna, both in the
directions of the transmitter and receiver antennas. The input signal at
R, from station A, is calculated as being
AR bf
R A A R
L
G G P P
,
1
⋅ ⋅ ⋅ =
(161)
where
P
R
= Power received by the receiving antenna at R (the repeater)
P
A
= Radiated power of the transmitting antenna at A
G
A
= Antenna gain of the transmitting antenna at A
G
R
= Antenna gain of the repeater R
L
bf,AR
= Freespace loss between the transmitting antenna and the
repeater
RADIOWAVE PROPAGATION
Ericsson AB 89
4/038 02LZU 102 152, Rev B, June 2003
Freespace loss L
bf
,
SR
between A and R can be written as
2
,
4

.

\
 ⋅ ⋅
=
λ
π
A
AR bf
d
L
(162)
where
λ = Wavelength, m
d
A
= Distance between the transmitter antenna and the repeater, m
If the repeater reflects the received power, P
R
, in the direction of the
receiver, B, the received power at B is
RB bf
B R R R
L
G G P P
,
'
1
⋅ ⋅ ⋅ =
(163)
where
P
R’
= Power at the receiver antenna B
P
R
= Radiated power of the repeater
G
R
= Antenna gain of the repeater R
G
B
= Antenna gain of the receiver B
L
bf,RM
= Freespace loss between the repeater and the receiver
antennas
Freespace loss L
bf
,
RM
between R and B can be written as
2
,
4

.

\
 ⋅ ⋅
=
λ
π
B
RB bf
d
L
(164)
where
d
B
= Distance between the repeater and the receiver antennas, m
λ = Wavelength, m
Note that freespace loss, L
bf
,
AR
and L
bf
,
RB
are transformed to dB
following logarithmic conversion, see equation (12).
The total path loss is calculated as follows
RB AR B A KB KA
RB G AR G RB H AR H RB bf AR bf S
G G G G A A
A A A A A A A
− − − − − +
+ + + + + + =
− − − − , ,
(165)
where
A
S
= Total path attenuation, dB
RADIO TRANSMISSION NETWORK AND FREQUENCY PLANNING
90 Ericsson AB
4/038 02LZU 102 152, Rev B, June 2003
A
bf,AR
= Freespace loss for the partial path AR, dB
A
bf,BR
= Freespace loss for the partial path BR, dB
A
HAR
= Obstruction loss for the partial path AR, dB
A
HBR
= Obstruction loss for the partial path BR, dB
A
GAR
= Gas attenuation for the partial path AR, dB
A
GBR
= Gas attenuation for the partial path BR, dB
A
KA
= Feeder loss at station A, dB
A
KB
= Feeder loss at station B, dB
G
A
= Antenna gain at station A, dBi
G
B
= Antenna gain at station B, dBi
G
AR
= Antenna gain for the antenna at R facing station A, dBi
G
BR
= Antenna gain for the antenna at R facing station B, dBi
The fade margin for a path using backtoback antennas is calculated as
follows
S th Tr
A P P M − − =
(166)
where
M = Fade margin, dB
P
Tr
= Transmitter’s output power, dBm
P
th
= Receiver’s threshold for a given biterror ratio, dBm
A
S
= Total path loss, dB
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.