Scribd
Upload
57 views

Dr. Yahya Ali

Dr. Yahya Ali's lecture discusses long-term shadow fading and small-scale multipath fading. Long-term shadow fading is caused by terrain obstructions like hills or man-made structures and results in signal power fluctuations over distances of tens to hundreds of meters. Small-scale multipath fading causes amplitude and power variations over distances of a wavelength due to constructive and destructive interference from multiple signal paths. The lecture also introduces the Rayleigh and Rician fading models and how they relate to the distribution of the received signal envelope.

Uploaded by

Sakena Abbas
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
57 views

Dr. Yahya Ali

Dr. Yahya Ali's lecture discusses long-term shadow fading and small-scale multipath fading. Long-term shadow fading is caused by terrain obstructions like hills or man-made structures and results in signal power fluctuations over distances of tens to hundreds of meters. Small-scale multipath fading causes amplitude and power variations over distances of a wavelength due to constructive and destructive interference from multiple signal paths. The lecture also introduces the Rayleigh and Rician fading models and how they relate to the distribution of the received signal envelope.

Uploaded by

Sakena Abbas
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

Lecture 3

Dr. Yahya Ali


1. Long-term shadow fading:
Note1
 Long-term shadow fading is, in general, caused by variations
in radio signal power due to signal encounters with terrain
obstructions, such as hills, or man-made obstructions, such as
buildings. Measured signal power may thus differ substantially
at different locations, even though at the same radial
distance from a transmitter. Long-term fading results in
fluctuations of the field strength of the radio signal around the
average power, the area-mean power, which is given by the
expression PT
 g(d)GTGR, as noted earlier. We can say that shadowing
represents medium-scale fluctuations of the radio field
strength occurring over distances from tens to hundreds of
meters.
Note 2

 Shadowing is mainly caused by terrain and


buildings. As a result, since each geographical
area has its own unique terrain, buildings, building
shapes and density, as well as different building
materials, the standard deviation σ in the
Gaussian model depends on the environment.
Due to this phenomenon, it is highly important to
perform field measurements and carefully
characterize an environment prior to deployment
of cellular systems. In this book we assume a
known value of σ, in the range of 6–10 dB, for
simplicity in modeling.
Note 3

 The two-ray propagation model is the simplest one that


one could adopt in demonstrating the effect on the
average received power of multiple rays due to
reflection, diffraction, and scattering, impinging on a
receiver. It treats the case of a single reflected ray.
Despite the relative simplicity of this model, it has been
found to provide reasonably accurate results in macro
cellular systems with relatively high BS antennas and/or
line of sight conditions between BS and receiver. The
two-ray model assumes that the transmitted em wave
reaches the (non-moving) receiver directly through a
line-of-sight path, and indirectly by perfect reflection
from a flat ground surface.
2. :Small-scale multipath fading
Note 1
 Amplitude and power variations occur over
distances of wavelengths, hence the reference to
small-scale fading. Specifically, with Rayleigh
statistics, the probability density function fα(α), α ≥
0, of the random variable α is given by

 with σ2r an adjustable Rayleigh parameter. It is


readily shown that the second moment of the
Rayleigh distribution is E(α2) = 2σ2.
Note2

 The average received power is just PT


g(d)GTGR. This implies taking the expectation
of the instantaneous received signal power
PR with respect to the two random variables x
and α. Doing so, we find that E(α2) = 1. Hence
we must set σ2r= 1/2, as indicated. An
alternative approach is to take expectations
with respect to x and α in the dB received
power expression . It is clear then that E(x) = 0
and E(α2) = 1. We shall also show that the
instantaneous local-mean received power PR
is, in this case, an exponentially distributed
random variable.
Note3
 If the transmitted signal is again taken to be an unmodulated sinewave at
frequency fc Hz, we can write the superposition of the L rays in a complex
phasor form extending

 The resultant phasor appearing at the receiver is then given by

 The term t0 = d/c represents the average delay encountered previously. Each
of the L rays has a random variation about this delay, denoted by the random
variable τk. We assume the random phase ωcτk is uniformly distributed between
0 and 2π, as is the random additional phase θk. The sum of the two random
phases, θk − ωcτk ≡ φk, is then uniformly distributed between 0 and 2π as well.
The amplitude term ak of the kth ray is taken to be a real number and random
as well. (A mobile’s movement will introduce additional Doppler shift phase
terms, to be discussed shortly.)
 As real part

 The instantaneous local-mean received power PR is proportional to


the time average of 𝑆𝑅2 𝑡

 with c a proportionality constant. This is precisely the power given by


(𝑃𝑅 ), which includes the three effects of propagation distance,
shadow fading, and multipath fading, the last the effect on which
we are focusing. Expanding (𝑆𝑅 𝑡 ) by trigonometry, we get
Note 4
 It is important to note now that for large L the random variables x and y,
each defined as the sum of L random variables, become Gaussian
 distributed, by the Central Limit Theorem of probability. In particular, since ak
and φk are independent random variables, and the φks were assumed to
be uniformly distributed, with zero average value, it is left to the reader to
show that the expectations of the various rvs are given by

 The Gaussian variables x and y are thus zero-mean and have the same
variance σ2R. In addition, it is readily shown from the orthogonality of cos φk
and sin φk that E(xy) = 0. Here we have taken expectations with respect to
the short-distance (order of wavelengths) random variations among the
various path components only. We are essentially conditioning on the
 much longer-distance variations due to shadow fading.
Note5
 We can write this expression in the form PR = α2p, with p the local-
mean power term due to shadow fading defined earlier in its dB
form.

 Taking the expectation with respect to the multipath, shorter-


distance, random variation only (i.e., conditioning on the
longer-distance shadow fading variations), we have, from

and the definition of σ2R above

𝑐
c again a proportionality constant. Since 𝛼 = a and a is Rayleigh
2𝑝

distributed, it is clear that α must be Rayleigh distributed as well


 We have used the expectation symbol E( ) here to indicate, implicitly,
averaging with respect to the multipath term only, to obtain the local-
area mean power p. Further averaging with respect to this power would
then give us the average received power PR. This verifies the statement
made earlier that E(α2)=1, the term pdB is sometimes referred to as the
local-mean power, to distinguish it from the area-mean power, and to
indicate that it differs from the instantaneous received power PR which
 incorporates the effect of the more rapidly varying multipath fading. The
probability density function for pdB is thus written as follows

 as well as the shadow-fading probability density function expression


(𝑓 𝑃𝑑𝐵 ) in which we have the average of pdB given by the dB average
power.
Note6
 We continue our analysis. Since gaussian uncorrelated
variables are independent, x and y, as defined by (x,y),
are independent as well. Again using simple
trigonometry, we rewrite (SR(t)) in the form

 where a2 = x2 + y2, and θ = tan−1 y/x. Hence E(a2) =


E(x2) +E(y2) = 2σ2R. Note that a is the (random)
envelope or amplitude of the received signal SR (t).
Given x and y zero-mean gaussian, it can be shown that
the signal envelope a is Rayleigh distributed (Schwartz,
1990), as given by
Note7
 Note that the random amplitude a now differs from its
earlier definition in the Rayleigh case above. It is now
given by a2 = (A+x)2 + y2. An analysis similar to that
carried out in the Rayleigh case shows that, with x and
y independent gaussian random variables, the
probability distribution of the random amplitude a is
given by the following Ricean distribution (Schwartz,
1990):

 I0(z) is the modified Bessel function of the first kind and
zero order, defined as

 The Rayleigh distribution is thus a special case of the
Ricean distribution. It may readily be shown that, for
A2/2σ2R >> 1, the Ricean distribution approaches the
gaussian distribution, centered about the value
A.
 connecting PR and the signal envelope or
amplitude a. It is left for the reader to showthat this
distribution,in the Ricean model case, is given by

The parameter is called the Ricean K-factor. Use has been


made, in deriving

The Ricean K factor is related very closely to the ratio of the


average received line-of-sight or direct-ray signal power to the
average received signal
power through scattered rays. Measurements made in microcellular
environments show
K falling in the range of 6 dB–30 dB.

You might also like