Channel modelling… Introduction

• To design, implement and operate efficient optical

communication systems, it is imperative that the
characteristics of the channel are well understood.
• Characterization of a communication channel is
performed by its channel impulse response, which is
then used to analyse and combat the effects of channel
distortions.
• A considerable amount of work has been published on
the channel characterization, covering both
experimental measurement and computer modelling of
indoor and outdoor systems.
• The power penalties directly associated with the
channel may be separated into two factors, these being
optical path loss and multipath dispersion.
Outdoor Channel
Atmospheric Channel
• The atmospheric channel is a very complex and
dynamic environment that can affect the
characteristics of the propagating optical beam
(laser in most cases), thus resulting in optical losses
and turbulence-induced amplitude and phase
fluctuation.
• The properties of the atmospheric channel is
random in nature; therefore its effects can be
characterized by statistical means.
• There are a number of models. Depending on the
type of the model and its accuracy, one can estimate
the received optical irradiance at the receiver.
• The concentric layers around the surface of the
Earth are broadly classified into two regions:
homosphere and heterosphere.
• The homosphere covers the lower layers ranging
from 0–90 km. Heterosphere lies above
homosphere above 90 km.
The homospheric region of atmosphere is composed
of various gases, water vapors, pollutants, and other
chemicals. Maximum concentrations of these particles
are near the Earth surface in the troposphere that
extends up to 20 km.

The density of particles decreases with the altitude up

through the ionosphere (region of upper atmosphere
that extends from about 90 to 600 km and contains
ionized electrons due to solar radiations).
These ionized electrons form a radiation belt around
the surface of the Earth
These atmospheric particles interact with all signals
that propagate through the radiation belt and lead to
deterioration of the received signal due to
absorption and scattering.
• Absorption is the phenomenon where the signal
energy is absorbed by the particles present in the
atmosphere resulting in the loss of signal energy and
gain of internal energy of the absorbing particle.
• In scattering, there is no loss of signal energy like in
absorption, but the signal energy is redistributed (or
scattered) in arbitrary directions.
• Both absorption and scattering are strongly dependent
upon operating wavelength and will lead to decrease in
received power level.
• These effects become more pronounced when the
operating wavelength of the transmitted signal is
comparable with the cross sectional dimensions of the
atmospheric particles.
Figure shows transmittance (or attenuation) effects as a
function of wavelength using MODTRAN software
package.
The output of MODTRAN in clear weather conditions is
plotted for wavelength up to 3 µm.
• It is clear from this figure that peaks in attenuation at
specific wavelengths is due to absorption by
atmospheric particles and therefore the choice of
wavelength has to be done very wisely in the high
transmissive band for FSO communication links.
• The atmospheric condition in FSO channel can be
broadly classified into three categories, namely, clear
weather, clouds, and rain.
• Clear weather conditions are characterized by long
visibility and relatively low attenuation.
• Cloudy weather conditions range from mist or fog to
heavy clouds and are characterized by low visibility,
high humidity, and large attenuation.
• Rain is characterized by the presence of rain droplets of
variable sizes, and it can produce severe effects
depending upon rainfall rate.
• Various atmospheric conditions can be represented by
size of the particle (i.e., cross-sectional dimension
relative to operating wavelength) and the particle
density (i.e., volumetric concentration of the particles).
• Cloudy weather conditions range from mist or fog to
heavy clouds and are characterized by low visibility, high
humidity, and large attenuation.
• It is seen that conditions may vary from high density and
small particle size like in the case of mist and fog to low
density and large particle size during heavy rain.
Atmospheric Channel Losses
• The loss mechanisms in an FSO are virtually identical to
those in a line-of-sight RF (microwave and millimetre)
channel but the fading level is higher than the RF Signals.
• Optical signal propagating through a free space channel is
very sensitive to the atmospheric conditions such as fog,
rain and so forth.
• For optical radiation travelling through the atmospheric
channel, the interaction between the photons and the
molecular constituent of the atmosphere cause some of
the photons to be extinguished while particulate
constituents scatter the photons.
• These events ultimately result in a power transmission
loss mechanism otherwise described by the Beer–
Lambert law (BLL).
• BLL law describes the transmittance of an optical
field through the atmosphere as function of the
propagation distance.
• In addition, optical radiation traversing the
atmosphere spreads out (beam divergence) due to
the diffraction, causing the size of the received beam
to be greater than the receiver aperture.
• These factors, combined with others are responsible
for the difference between the transmitted and the
received optical powers.
• Beam divergence can be minimized by employing a
very narrow coherent laser Source.
Atmospheric absorption losses can be reduced by
adopting wavelengths that lie in the low-loss
‘transmission windows’ in the visible or infrared
bands.
The concentrations of matter in the atmosphere,
which result in the signal attenuation, vary
spatially and temporally, and will depend on the
current local weather conditions.
For a terrestrial FSO link transmitting an optical
signal through the atmosphere, the optical power
reaching the receiver Pr is related to the
transmitted power Pt by
• where τod is optical depth. The fraction of the power
transmitted in the optical link is defined as the
transmittance and is given by

• The optical depth and optical atmospheric

transmittance are related to the atmospheric
attenuation coefficient and the transmission path
length L by the Beer– Lambert’s law as
where γt(λ) and T(λ,L) represent the total
attenuation/extinction coefficient (m^−1) and the
transmittance of the atmosphere at wavelength λ,
respectively.
γ is an optical source (i.e laser) wavelength specific
and depends on the path integrated distribution of
atmospheric constituents along the LOS and is
defined as γ = −10 log10T = 4.43 τod. For τod of 0.7, the
corresponding loss is 3 dB.
The attenuation of the optical signal in the
atmosphere is due to scattering and absorption
introduced by the molecular constituents (gases) and
aerosols.
• The aerosol is made up of tiny particles of various
shapes ranging from spherical to irregular shapes
suspended in the atmosphere.
• Generally speaking, the atmospheric attenuation
coefficient can be expressed as

where αml and αal are respectively the absorption

coefficient for the molecular and aerosol, and βml
and βal are the scattering coefficients for the
molecular and aerosol, respectively.
• Absorption and scattering of light by particles
present in the atmosphere is a complex
process involving Mie scattering and
nonselective scattering by large size particles
(such as fog, haze and rain) and Rayleigh
scattering by smaller particles.
Empirical Approach
• The attenuation due to the fog depends on factors such as the
location, particle size distribution, liquid water content and
average particle diameter.
• As the particle concentration and size distribution vary from
one location to another, this makes the task of predicting fog-
induced attenuation a challenging one.
• Generally, the fog particle size distribution is not available and
not reported in the standard meteorological data.
• Therefore, the fog-induced attenuation of the optical signal is
only predicted using empirical models derived from the
experimental observations.
• The empirical model uses the visibility V data in order to
characterize the density of fog.
The link visibility (i e, the meteorological visual range, MVR) is
used to measure the attenuation due to the fog. The very
fundamental law to measure the fog density indirectly on the
basis of the visibility is the Koschmieder law.
It defines the visibility as the distance to an object at which the
visual contras/ transmittance of an object drops to a certain
value of the visual/ transmittance threshold Tth level of the
original visual contrast (100%) along the propagation path.
The meteorological visibility V (km) can therefore be expressed in
terms of the atmospheric attenuation coefficient βλ and Tth as

Tth varies from 0.0077 to 0.06, where the smaller value defines a
larger MVR for a certain atmosphere environment. Mostly used
values of Tth are 2% and 5%
• The scattering coefficient of fog as earlier mentioned is a
parameter that is dependent on the wavelength of the
propagating optical beam and V.
• The following is a model developed by Kruse to link all these
parameters together.

Here, λo is the maximum spectrum of the solar band, δ is the

particle size-related coefficient and its values are given as
Figure a and b shows the atmospheric attenuation coefficient βλ
against the visibility for a range of wavelengths using the Kruse
model for Tth of 2% and 5%, Respectively.
In Figure a for V of 500 m the attenuation difference at 0.69
and 1.55 μm is 10 dB increasing to 28 dB for the dense fog at V
of 100 m.
However, in Figure b with Tth = 5%, the attenuation difference
is 4 dB at V of 500 m increasing to 28 dB at V of 100 m.

a) Tth = 2% b) Tth = 5%
The Kruse model was originally developed for dense
haze and is not very suitable to study the effect of fog
for V = 1 km.
Further study based on the empirical data in region V <
1 km has indicated that βλ is wavelength independent
for V < 0.5 km. This led to the Kim model which gives
the values of δ as
The visibility range values under different weather conditions are
presented in Table, while Figure shows the attenuation
coefficient values based on the Kim model against the visibility for
a range of wavelengths.
According to this model, in moderate-to-dense fog
conditions, using higher wavelength for the FSO links
offers no advantage over shorter ones since the
attenuation coefficient is wavelength independent in
such conditions.
Further work on theoretical as well as practical
investigation of this model for the low visibility is
required to confirm that the attenuation is really
independent of the wavelength for moderate-to-dense
fog regimes as the Kim model was derived from
calculation of exact Mie theory and by considering
dense haze and fog.
• Optical attenuation in foggy environments poses a great
challenge for a reliable operation of FSO links.
• In order to guarantee a reliable and 100% availability therefore,
adequate link margin must be provided to account for the fog
attenuation.
• Naboulsi proposed relations for the attenuation caused by the
radiation and advection fog at a wavelength of 0.69 -1.55 μm
and a visibility range of 0.05 - 1 km. These are given by

This shows that the attenuation due to fog is higher at the 1.55 μm than 0.785
μm which is in contradiction to the Kim’s model for V > 0.5 km.
Figure shows the comparison of the attenuation due to fog
using Kim and Naboulsi models for Tth of 2% and 5% at
wavelengths of 830, 1550 and 2000 nm.
The attenuation difference between the advection and the
convection fog is very close to each other. The Naboulsi
advection and convection fog models presents higher
attenuation due to fog compared to the Kim model.
For Tth of 2% the behaviour of the Kim model is very similar to
the Naboulsi models (see Figure b).
The Naboulsi models deviate from the Kim model for
wavelengths higher than 900 nm (Figure c and d).
The reason for this deviation is due to the experimental data
collected from different locations.
• Figure a–d shows the fog attenuation (dB/km) for the most
important fog models for 0.5 μm < λ < 1.55 μm and V = 0.1,
0.25, 0.5 and 1 km.
• It shows the correlation for a range of 0.5 μm < λ < 0.9 μm
and for V = 1 km. These models are also verified
experimentally showing a good relation with the
experimental data.
• Kim’s model underestimates the optical attenuation
compared to the Naboulsi model, but the difference is small
for 0.5 μm < λ < 0.9 μm.
• Kim and Naboulsi convection models are best to estimate
the attenuation of the optical signal in fog for the
wavelength range of 0.5 - 0.9 μm and for V = 1 km.
 The loss for rain and snow are defined as

The power loss due to the rain and snow are so low compared to
that due to the Mie scattering, they still have to be accounted for in
the link margin during the link budget analysis.

A typical rainfall of 2.5 cm/h could result in an attenuation of ~6

dB/km while a typical value for attenuation due to light snow to
blizzard is 3–30 dB/km.
 Beam Divergence
• One of the main advantages of FSO systems is the
ability to transmit a very narrow optical beam, thus
offering enhanced security.
• But due to diffraction, the beam spreads out. This
results in a situation in which the receiver aperture is
only able to collect a fraction of the beam. The
remaining uncollected beam then results in beam
divergence loss.
• For relatively low data rate systems, a typical value
for the beam divergence is 68 mrad and at higher data
rates, 2 mrad beam divergence is commonly used.
However, what really determines the size of the beam
divergence are the optical channel and the pointing
jitter.
Considering the arrangement of an FSO communication link of
Figure, and by invoking the thin lens approximation to the
diffuse optical source whose irradiance is represented by Is, the
amount of optical power focused on the detector is derived as

Beam divergence
where At and Ar are the transmitter and receiver
effective aperture areas, respectively. while As is the
area of the optical source and Lp is the path length.
This clearly shows that a source with high radiance
Is/As and wide apertures are required in order to
increase the received optical power.
For a non-diffuse, small source such as the laser, the
size of the image formed at the receiver plane is no
longer given by the thin lens approximation; it is
determined by diffraction at the transmitter aperture.
The diffraction pattern produced by a uniformly
illuminated circular aperture of diameter dt is known to
consist of a set of concentric rings.
 The image size is said to be diffraction limited when the radius
of the first intensity minimum or dark ring of the diffraction
pattern becomes comparable in size with the diameter dim of
the normally focused image. That is

Therefore, the source diameter is

where fl is the focal length, u is the distance from the source to

the centre of the transmitter lens.
If an image is made through a small aperture, there is a point at which the
resolution of the image is limited by the aperture diffraction.
The equation above shows that for diffraction to be the sole cause
of beam divergence (diffraction limited), ds < 1.22 λfl/dt < 1.22.
A laser being inherently collimated and coherent normally
produces a diffraction-limited image.
The diffraction limited beam divergence angle in radians is thus
given by b  λ/dt. If the transmitter and receiver effective
antenna gains are, respectively, given by

And the free-space path loss is given by

The received optical power then becomes

where the radiation solid angle b   (b )^2 /4. The diffraction
limited beam spreading/ geometric loss in dB is thus given by

(B)

The same result given by Equation above can be obtained by

substituting Aim = θbLp for the image size in Pr = Pt Ar /Aim.
In order to reduce the diffraction-limited beam
divergence, a beam expander of the type shown in
Figure, in which the diffracting aperture has been
increased, can then be used. The beam expander reduces
the beam divergence loss and increases the received
signal power in the process.

Typical beam expander

For most practical sources, the beam divergence angle is
usually greater than that dictated by diffraction and for a
source with an angle of divergence θs, the beam size at a
distance Lp away is (dt + θsLp).
The fraction of the received power to the transmitted power is
therefore given as

where dr is the receiving lens diameter.

The geometric loss in decibels is thus

(C)
The beam spreading loss for the diffraction limited
source given by Equation B is, expectedly, lower than
that of the non-diffraction limited case represented by
Equation C, this is because the image size is smaller by dt
in the diffraction limited case.

From the foregoing, it is clear that a source with a very

narrow angle of divergence is preferable in terrestrial
FSO systems.

It should however be mentioned that, wide divergence

angle sources are desirable in short-range FSO links to
ease the alignment requirement, compensate for
building sway and eliminate the need for active tracking
systems at the expense of increased geometric loss
apparently.
 A typical FSO transceiver has optical beam divergence in the
range of 2–10 mrad and 0.05–1.0 mrad (equivalent to a beam
spread of 2–10, and 5 cm to 1 m, respectively, at 1 km link
range) for systems without and with tracking, respectively.

Optical And Window Loss

This class of power loss includes losses due to imperfect lenses
and other optical elements used in the design of both the
transmitter and receiver.
It accounts for the reflection, absorption and scattering due to
the lenses in the system. The value of the optical loss Lop, can be
obtained from the component manufacturer.
It apparently depends on the characteristics of the equipments
and quality of the lenses used.
• For FSO transceivers installed behind windows
within a building, there exists an additional optical
power loss due to the window glass attenuation.
• Although (glass) windows allow optical signals to
pass through them, they contribute to the overall
power loss of the signal.
• Uncoated glass windows usually attenuate 4% per
surface, because of reflection.
• Coated windows display much higher losses and its
magnitude is wavelength dependent.
 Pointing Loss
• Additional power penalty is usually incurred due to lack
of perfect alignment of the transmitter and receiver.
• The resulting power loss is catered for by including
pointing/ misalignment loss LP in the link budget
analysis.
• The transmitter is directly pointed at the receiver, the
pointing loss is 0 dB.
• For short FSO links (<1 km), this might not be an issue
but for longer link ranges, this can certainly not be
neglected.
• Misalignments could result from building sway or
strong wind effect on the FOS link head stand.
Including all the losses the received optical power then becomes

where ηt is the transmit optics efficiency, ηA is the aperture

illumination efficiency of the transmit antenna, λt is the transmit
wavelength.
Latm and Lpol are the fractional loss due to absorption of the
transmitting medium and mismatch of the transmit and receive
antenna polarization patterns, respectively, the term Ar /4 
(Lp)^2 is the fraction of power collected by Ar if the transmitter
is an ideal isotropic radiator.
ηr is the receiving optics collecting efficiency, Ltp and Lrp is the
transmitter and receiver pointing losses, respectively.
Table shows a typical link budget for 2.5 Gbps for a 2 km link and 1550 nm
wavelength.
• Unfortunately, due to the extreme complexity
involved in mathematically modelling of atmospheric
turbulence, a universal model valid for all the
turbulence regimes does not currently exist. models
describe the pdf statistics of the irradiance
fluctuation.
• Three most reported models for irradiance
fluctuation in a turbulent channel are the log-normal,
gamma–gamma and negative exponential models.
• Their respective ranges of validity, as reported in the
literature, are in the weak, weak-to-strong and
saturate regimes.
• Atmospheric turbulence results in random fluctuation
of the atmospheric refractive index n along the path of
the optical field/radiation traversing the atmosphere.
• This refractive index fluctuation is the direct end
product of random variations in atmospheric
temperature from point to point.
• These random temperature changes are a function of
the atmospheric pressure, altitude and wind speed.
The smallest and the largest of the turbulence eddies
are termed as the inner scale l0, and the outer scale L0,
of the turbulence, respectively. l0 and L0 are typically
of the order of a few millimetres and several metres,
respectively.
• These weak lens-like eddies shown graphically in
Figure result in randomized interference effect
between different regions of the propagating beam
causing the wavefront to be distorted in the
process.
• Known effects of atmospheric turbulence include:
beam spreading, beam wander and beam
scintillation.
• The beam wander effect caused by a largescale
turbulence, and diffraction effects are often
assumed to be insignificant, in particular, when the
receiver aperture diameter D is greater than the
size of Fresnel zone (L/k)^0.5, where k = 2π/λ is the
optical wave number.
• Since the change of beam wander-induced
fluctuation is very slow, it can be readily taken care of
using the tracking schemes.
• In an attempt to model the turbulent atmospheric
channel, the widely accepted ‘Taylor hypothesis’,
which says that the turbulent eddies are fixed or
frozen and can only move in their frozen form with
the transverse component of the mean local wind will
be followed.
• This hypothesis means that the temporal variations
in the beam pattern or its statistical properties are
caused by the component of the local wind that is
perpendicular to the beam direction of propagation.
• In addition, the temporal coherence time τo of
atmospheric turbulence is reported to be of the
order of milliseconds.
• This value is very large compared to the duration of
a typical data symbol, thus the turbulent
atmospheric channel can be described as a ‘slow
fading channel’ since it is static over the duration of
a data symbol.
• The relationship between the temperature of the
atmosphere and its refractive index is given by
Equation (D) while for most engineering applications,
the rate of change of the refractive index with
respect to channel temperature is represented by
Equation (E).
 (D)

(E)

where Pas is the atmospheric pressure in millibars, Te is the

effective temperature in Kelvin and λ the wavelength in microns
Near the sea level,
The contribution of humidity to the refractive index fluctuation is
not accounted for in Equation ( D) because this is negligible at
optical wavelengths.
The position and time-dependent index of refraction denoted by
nas(r,t) can be expressed as the sum of its free-space (no
turbulence) value nas0, and a turbulence induced random
fluctuation component nas1 (r, t).
Thus

In accordance with the Taylor’s ‘frozen-flow’ hypothesis,

which implies that the temporal variations of the index of
refraction of the channel are mainly due to the transverse
component of the atmospheric wind, the randomly
fluctuating part of Equation above can then be written as

where vw(r) is the local wind velocity perpendicular to the field

direction of travel.
In order to understand the structure of atmospheric
turbulence, it is convenient to adopt the energy cascade
theory of turbulence. As per this theory, when the wind
velocity is increased, the Reynolds number exceeds the
critical value.
This results in the local unstable air masses called
turbulent eddies with their characteristic dimensions
slightly smaller than, and independent of, the parent
flow.
Under the influence of the inertial forces, the larger size
eddies break up into smaller eddies until the inner scale
size l0 is reached. The family of eddies bounded above by
the outer scale L0 and below by the inner scale l0 forms
the inertial subrange.
The outer scale denotes the scale below which the turbulent
properties are independent of the parent flow.
Typically, the outer scale is in the order of about 10 to
100m and is usually assumed to grow linearly with the
height above the ground.
The inner scale is of the order of 1 to 10 mm near the
ground, but it is of the order of centimeters or more in the
troposphere and stratosphere. Scale size smaller than the
inner scale l0 belongs to the viscous dissipation range.
In this range, the turbulent eddies disappear, and the
remaining energy is dissipated in the form of heat. This
phenomenon is known as Kolmogorov theory of turbulence
and is depicted in Fig.
• The mathematical model of the atmospheric
turbulence and its effects on the optical beam
propagation assumes that the fluctuations in the
atmospheric parameter are stationary random
processes having statistically homogeneous and
isotropic nature.
• Within this mathematical framework, the structure
function in the inertial range satisfies the universal
two-thirds power law, i.e., it follows r^(2/3)
dependence, where r refers to the spatial scale
defined as
where r1 and r2 refer to position vectors at two points
separated by distance r in space. The structure function for a
random variable x(r)is then given as

The random variable x (r) is assumed to have a mean and a

superimposed fluctuating component that is represented as

(2)

In the above equation, the first term in the angle bracket is the
slowly varying mean component and second term the random
fluctuations.
Using Eq. (2), the structure function in Eq. (1) can be rewritten
as

The first term in above Eq. becomes zero for a stationary

random process. This makes structure function a useful
parameter for describing the random fluctuations.
With the atmosphere as the propagation medium, these
random fluctuations can be associated with any of these
parameters, i.e., velocity, temperature, and refractive index.
The structure function for wind velocity Dv (rv) is given as
• where v1 and v2 are the velocity components at
two points separated by distance r and (Cv)^2 the
velocity structure constant (in m^4/3 /s^2) that
measures the amount of energy in the
turbulence.
• Similarly, the structure function for the
temperature is given as Dt(rt ) = (Ct)^2 r^2/3,
where (Ct)^2 is the temperature structure
constant (in deg^2/m^2/3).
• The turbulence in the atmosphere also results
from the random fluctuations of the atmospheric
refractive index n due to variations in
temperature and pressure along the propagation
path in the atmosphere.
In general, the refractive index of the atmosphere at any point
r in space can be expressed as sum of the average and the
fluctuating terms, i.e.,

where n0 = n (r)  1 is the mean value of index of refraction

and n’(r) represents the random deviation of n (r ) from its
mean value. Therefore, the above equation can be rewritten
as

The index of refraction of the atmosphere is related to

temperature and pressure of the atmosphere and is given as
where  is the wavelength in m, P’ the atmospheric pressure in
mbar, and T’ the temperature of the atmosphere in Kelvin.
Changes in the optical signal due to absorption or scattering by
the molecules or aerosols are not considered here. The structure
function for refractive index, Dn (r), can be expressed as
Dn (rn ) = (Cn )^2r^(2/3 ),

where (Cn )^2 is called refractive index structure constant and is a

measure of the strength of fluctuations in the refractive index.
(Cn )^2 is related to temperature structure constant, (Ct )^2as
The (Ct )^2 is determined by taking the measurements
of mean square temperature between two points
separated by a certain distance along the propagation
path (in deg^2/m^2/3).

It is obvious from Eq. above that refractive index

structure parameter can be obtained by measuring
temperature, pressure, and temperature spatial
fluctuations along the propagation path.
• All the expressions for the structure function
are defined for the inertial subrange, i.e., for l0
<< r << L0.
• Out of all the structure constants referred
above, the refractive index structure constant,
i.e., (Cn )^2 , is considered the most critical
parameter along the propagation path in
characterizing the effects of atmospheric
turbulence.
• Depending on the size of turbulent eddy and
transmitter beam size, three types of
atmospheric turbulence effects are observed:
The Scintillation Effect
The irradiance fluctuations within the cross section of
the received beam after propagating through
turbulent atmosphere are commonly referred as
“scintillation” and are measured in terms of
scintillation index (or normalized variance of
irradiance) (I )^2 .
It causes loss of signal-to-noise ratio and induces
deep signal fades. The (I )^2 is defined as
where I is the irradiance (intensity) at some point in the detector
plane and the angle bracket   denotes an ensemble average.
For X amplitude of the transmitted optical beam, the received
irradiance, I, at the receiver takes the form as

where I0 is the intensity without turbulence. From above two

Eqs. , (I )^2 in terms of log-amplitude variance, (X )^2 is given
by

Further, the variance of log irradiance (also called Rytov variance)

(R)^2 is related to (I )^2 as below
In weak turbulence, scintillation index is expressed as

and

where k is wave number (2/). It is clear from above two Eqs. that for a
weak turbulence conditions, longer wavelength will experience lesser
irradiance fluctuations for a given link range.
Scintillation index for strong turbulence is given by

Clearly, this Eq. shows that for strong turbulence, smaller wavelength will
experience lesser irradiance fluctuations.
Various studies have been performed to develop the
mathematical model for the probability density function (PDF)
of the randomly fading received signal irradiance.
These studies have led to various statistical models that can
describe the turbulence-induced scintillation over a wide
range of atmospheric conditions.
For weak turbulence, (I )^2 < 1, and the irradiance statistics is
given by lognormal model. This model is widely used due to
its simplicity in terms of mathematical computations. The PDF
of the received irradiance, I, is given as
where  is the mean of ln (I). Since (I )^2 = 4 (x)^2 ; the above
lognormal pdf can be rewritten as

When the strength of turbulence increases, lognormal pdf

shows large deviation as compared to experimental data.
Therefore, lognormal statistics is not appropriate model in
case of strong fluctuation regimes.
For strong turbulence, (I )^2 >1, and the field amplitude is
Rayleigh distributed which in turn leads to negative
exponential statistics for received irradiance.
Its pdf is given by

where I0 is the mean irradiance. In this case, (I )^2 1, and this
happens only far into saturation regime.
The gamma-gamma distribution is used to successfully describe
the scintillation statistics for weak to strong turbulence. In this
model, the normalized irradiance, I, is defined as the product of
two independent random variables, i.e., I = IX + IY , where IX and
IY represent large-scale and small-scale turbulent eddies and
each of them following a gamma distribution.
This leads to gamma-gamma distribution whose pdf
is given as

where Ka (.) is the modified Bessel function of second kind of

order a. The parameters α and β are the effective number of
small-scale and large-scale eddies of the scattering environment
and are related to the atmospheric conditions through the
following expressions
where

The parameter k = 2π/λ is the optical wave number, DR the

diameter of the receiver collecting lens aperture, and R the
link range in meters, (Cn)^2 refractive index structure
parameter whose value varies from 10^-13 m^-2/3 for strong
turbulence to 10^-17 m^-2/3 for weak turbulence.
Since the mean value of this turbulence model is E [I] = 1 and
the second moment is given by E [I^2] = (1+1/ α) (1+1/β) ,
therefore, scintillation index (SI), that gives the strength of
atmospheric fading, is defined as