Large–Scale Propagation Loss

1. Wave propagation models

1. Large–Scale Path Loss Model (includes shadowing) :
It is the decay in signal strength due to distance between Tx and Rx. Remember the formula
Pr(d)=Pr(d o)×(d o/d)n. this decay is gradual with distance and slow in nature and can be noticed
for relatively long distances and long time intervals.
Shadowing is the random variation in signal strength which is also slow in nature and gradual
with distance.
2. Small–Scale fading:
It is the fast and random variations of signal strength due to multipath propagation and
movement of mobile station or the surrounding environment. It occurs over small distances
(fractions of a wavelength) and is fast in nature.

Path loss+shadowing
Path loss+shadowing
+fading

Path loss

2. Free Space Propagation Model

By free space we mean that the path between Tx and Rx is clear line–of–sight (no obstructions),
and that there are no objects in the surrounding of this path.
When this is the case, the received power is given by Friis formula (assuming no loss in antennas or
other equipment):

Pt G t G r 2
Pr (d) 
4d 2
Pt and Pr are transmitted and received powers (Watt), Gt and Gr are transmitter and receiver antenna
gains (dimensionless), d is distance between transmitter and receiver (m), and λ is the wavelength
(m).

Friis formula is valid only for far–field (or Fraunhofer region), that is when d is very large
compared to λ or to largest dimension (D) of transmitter antenna–by largest dimension (D) we mean
the largest distance between any two points on the surface of antenna, for rectangular shaped
antenna it’s the diagonal while for elliptical shaped it’s the major axis….etc.

Far–field is given by:

2D 2
d


Gain of antenna is given by:

4 A e
G
2

Ae is the effective area of antenna (m2)–front area of a dish receiver for example. Ae is then

G2
Ae 
4

The quantity P tGt is called the Effective Isotropic Radiated Power (EIRP). Directional transmitter
antenna concentrates power in the direction of propagation and reduces power transmitted in other
directions. When compared to isotropic antenna, the amount of power radiated in the direction of
propagation is therefore more than that radiated by isotropic antenna (of course the power radiated
in other directions is less than that radiated by isotropic antenna, directional antenna only
concentrates power and does NOT add power to wave)

EIRP  Pt G t Watt 
The power density (in Watt/m2) at the receiver antenna (just before it is collected by antenna) P d is
given by

EIRP Pt G t
Pd (d )  
4d 2 4 d 2

The received power Pr(d) (in Watt) is the product of the power density Pd (Watt/m2) by receiver
antenna effective area Ae (m2)

Pt G t G r 2 Pt G t G r 2
Pr (d)  Pd (d)A e  
4d 2 4 4d 2
which is the Friis formula again.

Power density Pd(d) and electric field intensity E(d) at distance (d) from transmitter antenna are
related by:

[E (d)]2 [ E (d)] 2
Pd (d)  
R fs 120

The path loss, PL(d), is the ratio of transmitted to received power and is greater than one.
 Pt (4d) 2
PL(d )  
Pr (d) G t G r 2

In dB the formula becomes

 ( 4d ) 2   G t G r 2 
PL (d )dB  10 log  2
 10 log  2 
G t G r    4d  

If we know the received power P(d o) at distance do (d o in far–field), then we can compute the
received power Pr(d) at any distance d>do.
2
d 
Pr (d)  Pr (d o ) o 
d 
In dB
Pr (d )dB  Pr (d o )dB  20 log(d o )  20 log(d )

Usually do is taken to be 1m for indoor propagation and 100m or 1km for outdoor propagation.

dB, dBW, and dBm

–For power calculations:
dB (or dBW) is a unit of power compared to one Watt and is given by

PdB  PdBW   10 log(PWatt )

While dBm is a unit of power compared to one milli–Watt

PdBm  10 log(PmW)  10 log(103  PWatt )  PdBW   30

–For power ratios (like gain or loss):

Only dB is used (NOT dBW NOR dBm)

 P Watt  
Power ratiodB  10 log 1   10 logP1 Watt   10 logP2 Watt   P1 dBW   P2 dBW 
 P2 Watt  

The same power ration results if we divide mW over mW or Watt over Watt

 P mW 
Power ratiodB  10 log 1   10 logP1 mW  10 logP2 mW  P1 dBm  P2 dBm
P
 2 mW  
In other words

Power ratiodB  P1 dBW   P2 dBW   P1 dBm  P2 dBm

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example 4.1
f=900MHz, largest dimension of transmitter antenna D=1m, find far–field distance?
Solution
d > (2D2/λ)  d > (2(1)2/(3×108/900×106)  d > 6m
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
 ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example 4.2
Pt=50 Watts, Gt=Gr=1, f=900 MHz find a) Pt in dBm b) P t in dBW c) Pr(100m) and Pr(1km).
Assume free space propagation
Solution
Pt[dBm]=10log(50×1000)=10log(50)+10log(1000)=17+30= 47dBm
Pt[dBW]=10log(50)=10log(50)=17dBW
or
Pt[dBW]= Pt[dBm]–30 = 47–30= 17dBW
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

3. Propagation Mechanisms
Reflection, diffraction, and scattering.

–Reflection: when wave falls on a smooth surface which is also large compared to wavelength,
ex: earth surface, buildings, walls,…etc.

–Diffraction: bending of propagating wave when hitting sharp edges, ex: mountain peaks, edges
of walls…etc.

–Scattering: when wave falls on rough surface or when dimension of surface is small compared to
wavelength, ex: street signs, tree leafs, walking people…etc.

4. Reflection

When a wave travels from one medium to another, part of it is reflected back into the first medium
and the other part is refracted (or transmitted) through the second medium. The amount of
reflected or transmitted parts depend on polarization (parallel or perpendicular) of wave,
permittivity (ε) and permeability (μ) of both media, and angle of incident θi. In conductors, it
depends on conductivity (σ) and frequency (f) as well.

We will consider transverse EM plane waves. In such waves E and H are orthogonal to each other
and direction of propagation is orthogonal to plane containing E an H. We define plane of
incidence as the plane containing the incident, the reflected, and the transmitted waves. In the
figure shown it is the plane of the paper.
A wave is said to be parallel polarized if the electric field E is contained in the plane of incidence,
figure (a).A wave is said to be perpendicular polarized if the electric field E is perpendicular to the
plane of incidence, figure (b).
We are interested in two cases, when first medium is air or vacuum (ε1=εo, μ1=μo, σ1=0) and
second medium is perfect dielectric (ε2=εrεo and μ2=μo σ2=0). The other case is when first medium
is air or vacuum and the other medium is a perfect conductor (σ2= ∞)

In both cases,

i   r

And in the case of reflection from perfect dielectric

11 sin(90  i )   2  2 sin(90   t )

Or, since we assumed μ1=μ2=μo, ε1=εo, and ε2=εrεo

sin(90  i )   r sin(90   t )

Or

cos(i )   r cos( t )

Here θi, θr, and θt are measured with respect to surface of boundary (NOT the normal of
boundary).

What about the amount of the reflected wave?

At the surface of boundary we have the following relation

E r   Ei

Γ is called Fresnel reflection coefficient, and its value depends on polarization of incident wave.

For parallel polarized wave (again we are assuming μ1=μ2=μo, ε1=εo, and ε2=εrεo)

 r sin(i )   r  cos 2 (i )

|| 
 r sin( i )   r  cos 2 (i )

While for perpendicular polarized wave we have

sin( i )   r  cos 2 (i )

 
sin(i )   r  cos 2 (i )

Note that εr ≥1 for any medium

What if E field is neither contained in, nor perpendicular to surface of incidence?

In this case we decompose the wave into two components, parallel and perpendicular and deal
with each component separately.
 If we plot the magnitudes of Γ|| and Γ┴ as function of θi for fixed values of εr we get the following
graphs

Note that there is a value of θi that makes Γ|| equal to zero, this value of θi is called the Brewster
angle θB. (for Γ┴ there is no such value of θi that makes it zero)

Solving for Γ|| = 0, we get

 r 1 1
B  
 2r  1 r 1

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example 4.4
Find Γ|| and Γ┴ when θi approaches zero and medium 1 is free space and medium 2 is perfect
dielectric.
Solution:
 r sin( 0)   r  cos 2 (0)
||   1
 r sin(0)   r  cos 2 (0)

sin(0)   r  cos 2 (0)

   1
2
sin( 0)   r  cos (0)

This is an important result since for wireless communications usually the distance between Tx
and Rx is very large compared to antenna heights and the angle of incidence θ i is very small and can
be approximated as zero.
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Reflection from perfect conductors

When medium 2 is a perfect conductor (σ = ∞), then Γ|| =1 and Γ┴ = –1. Both are constants and are
independent of θi.
5. Ground Reflection (2–ray) Model

In the previous chapter we used n=4 for path loss exponent, in this section we’ll see why. We will
derive the 2–ray formula from free space formula.
First we need to know how the electric filed decays with distance (d) in free space. So
2
EIRP E (d o )
Pd (d o )  
4d o2 120

2
EIRP E (d )
Pd (d )  
4d 2 120
Dividing the two equations we get
do
E (d )  E (d o )
d

Remember that E(d) and E(do) are the amplitudes of a sinusoidal electric component of an EM wave
and are always positive.
Now consider the situation in which Rx antenna receives a line–of–sight (LOS) and a ground
reflected versions of the same transmitted signal as shown.

Each one of the two paths alone is represented by free–space model. Heights of Tx and Rx antennas
are ht and h r. The ground distance between Tx and Rx towers is (d), while the LOS distance
between Tx and Rx antennas is (d’) which can be more than (d) because h t and hr are not necessarily
equal. The distance traveled by the reflected path is (d”) which is clearly larger than (d’). We define
Δ=d”–d’.

  d   d  d 2  ( h t  h r ) 2  d 2  (h t  h r ) 2

Δ can be approximated by (this is the first approximation)

2h t h r

d
If the magnitude of a free–space model electric field is given at some reference point (d o) then the
magnitudes of LOS and reflected paths are

d o E odo
E LOS  E (d o ) 
d d

do E d
E g   E (d o )  o o
d  d
What about phase–shift? Phase–shift between LOS path and reflected path is given by


   2


θΔ can be calculated using the exact or the approximate values of Δ.

So if the electric field received by LOS is considered to be at zero phase as E o d o cos t  then the
c
d
electric field received by reflected path is  E o d o cos t   
c 
d 

We will use phasor diagram, but before that, if (d) is very large compared to ht and hr then (d), (d’)
and (d”) are close to each other and hence

Eo d o E od o Eo d o
 
d d d

In other words the amplitudes of the two received signals are almost equal. Still though, there is a
significant difference in the phases.

The total electric field at the receiver is the phasor sum of the LOS and reflected waves. We will
assume that (d) is large and so θi ≈ 0 and Γ= –1.

E od o E d
E TOT (d)  E LOS  E g  0   o o  
d d
Eodo  E od o
 0   
d d
E d
 o o 10  1 
d
E d
 o o 10  1(   )
d
E d
 o o 12  12  2(1)(1) cos(   )
d
E d
 o o 2  2 cos(  )
d
E d  
 2 o o sin   
d  2 
 When θΔ/2<0.3 rad, we can use the approximation sin(θ Δ/2)≈ θΔ/2 (this is 2nd approx.). Note that
when

  2h t h r 2h t h r 20h t h r

 0 .3   0.3   0.3  d  d
2  d 0.3 

In this case

Eodo   E d  E d  E d 2h t h r 4h t h r E o d o 1

E TOT (d)  2 sin     2 o o   2 o o 2 o o 
d  2  d 2 d  d d  d2

What’s important here is that ETOT now decays as (1/d 2) not as (1/d) as was the case with free–space
model. Moreover, ETOT now depends on (λ), which was not the case for free–space.
Since we used free–space model to find E(do), the power density at (do) can be written as

Pt G t E o2
Pd (d o )  
4d o2 120

This means that we can write

2
4d o2 E 2o E o d o 
Pt G t  
120 30

What about the TOTAL power density resulting from ETOT which is composed of both ELOS and Eg?
2
E2  4h t h r E o d o 1  1
Pd (d)  TOT   
120   d 2  120

Finally the received power is

2 2
E 2TOT G r 2  4h t h r E o d o 1  G r 2 E o d o  G r h 2t h 2r Pt G t G r h 2t h 2r
Pr (d)  Pd (d)A e     
120 4   d 2  480 2 30 d4 d4

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example:4.6
d=5km, Gr=2.55 dB, do=1km, Eo=10–3V/m, f=900MHz, h t=50m, h r=1.5m. Find Pr(5km)?
Solution:
E d 2 G r h 2t h 2r
Pr (d)  o o
30 d4

P (5km) 
110 3
2
 1000 10 0.255 50 2 (1.5) 2
 0.54  10 12 Watt
r 4
30 5000
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
6. Diffraction
Any wave (including EM waves) that is
obstructed by an obstacle can propagate
around this obstacle and reach into the
shadow region, this phenomenon is
called diffraction.

Diffraction can be explained by

Huygen’s principle which states that: shadow region
we can think of the propagating EM obstacle
covered by
diffracted waves
wave as small point sources located at small sources
the wave–front, these small point
sources radiate small EM waves. If we
sum the vectors of these small EM
waves, we get the original EM wave.
Huygen’s principle is a simplified
model to deal with EM waves.

Wave front Wave front

Knife–edge diffraction model

To study diffraction mathematically, we will use a simplified method in which we represent

mountains, buildings, …etc by a knife–edge. The knife edge is supposed to be very thin and extends
infinitely in and out of the page plane.

The question is: what is the value of the electric field with the presence of the knife–edge obstacle?

The answer is in terms of the free–space electric field Eo which is the electric field in absence of the
knife–edge obstacle or any other obstacle including earth surface. So we will use the free space
model to find Eo.

Pt G t E 2o
Pd (d )  
4d 2 120

30Pt G t
Eo 
d

Now we find Ed, the electric field with the presence of the knife–edge.

E d  F( v)E o

F(v) is called the complex Fresnel integral.

How to find F(v)?

First we define (v), the Fresnel–Kirchoff diffraction parameter. In the figure above we assume hr<ht
(it can be also ht<h r). We simplify the geometry by subtracting the minimum of ht and hr from all
heights given in the figure. What’s important is the line–of–sight distance between transmitter and
knife–edge which is approximately equal to ground distance (d1), also the line–of–sight distance
between receiver and knife edge which is also very near to (d 2), (here we are assuming that d1 and
d2 are much larger than ht and hr). Important too is the height of the knife–edge above the line–of–
sight (h). Note that (h) can be zero if knife–edge is on line–of–sight, or it can be negative if knife–
edge is under line–of–sight.

Now we define (v) as

2d1  d 2 
vh
 d1 d 2

At the end we calculate F(v), the Fresnel complex integral, as

small sources
(Huygen’s)

Diffracted waves (vector summation

of small waves from small sources)
hobs
ht–h r
ht
hr
d1 d2

h=h obs–hr
≈d 1
ht–h r ≈d2

d1 d2

≈d 1 h=0
ht–h r ≈d 2

d1 d2

≈d 1

≈d 2
ht–h r (h) negative

d1 d2
 Fv  
1  j  exp(  jt 2 )dt
2  2
v

Don’t worry! We will not evaluate this difficult integral. Instead, we will use a graph that represent
the dB value of F(v) versus (v). The dB value of F(v) is called the diffraction gain Gd(v).

G d v   20 log F( v)

Here we use 20log[] and not 10log[] since it is not a power ratio.

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example:4.7
λ=1/3 m, d 1=1km, d2=1km, find diffraction loss for a) h=25m, b) h=0, c) h=–25m.

Solution:
a) 2d1  d 2  21000  1000
vh  25  2.74
d1d 2 (1 / 3)(1000)(1000)

Using the graph we find diffraction gain Gd to be approximately –22 dB (or diffraction loss is 22
dB)

b) h=0 so v=0, using graph the diffraction loss is 6dB

c) 2d1  d 2  21000  1000

vh  25  2.74
d1d 2 (1 / 3)(1000)(1000)

Using the graph we find diffraction loss to be approximately 0 dB

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example:4.8
f=900 MHz 100m
a) find diffraction loss
b) find obstacle height required to h
induce 6dB diffraction loss. 50m 75m
Solution:
a) h=75m–25m×2km/12km=70.8m 25m

210000  2000
v  70.8  4.25
(1 / 3)(10000)(2000)
10km 2km
Using graph we find that diffraction
gain Gd≈–26dB or diffraction loss is
26dB.

b) From graph we find that to get diffraction loss of 6dB (Gd≈–6dB) we need h=0, which means that
the obstacle edge is exactly on the LOS between Tx and Rx. This means that
hobs=25+25m×2km/12km=29.17m.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Fresnel Zone
The path length difference between
line–of–sight and diffracted signal (Δ)
 d 12  h 2
is given by  d 22  h 2
ht h
 d 2
1 
 h 2  d 22  h 2  d1  d 2  ≈d 1
LOS hr
2 ≈d 2
h  d1  d 2 
  
2  d1 d 2 

The phase difference between line–of–

d1 d2
sight signal and diffracted signal is
given by

2 h 2  d1  d 2 
   
   d1d 2 

We are looking for values of (h) that makes the phase difference between LOS and diffracted path
equal to multiples of π, that is   n , which correspond to constructive and destructive
interference cases. These values of (h) are called Fresnel distances of order (n), or simply rn.
Solving for   n we get

nd1d 2
rn 
d1  d 2 
 If we find rn for all possible d1, d2 (remember d1+d2=d), we end up with an ellipsoid for each (n).
These ellipsoids have the transmitter and receiver antennas at their foci as shown.

n=3, or φ=3π n=2, or φ=2π

n=1, or φ=π

Recall that
2d1  d 2 
vh
 d1 d 2
then
 2
 v
2
Now, what are the values of v (or h) after which F(v) keep decreasing below 0dB? (in other words,
the diffracted signal strength keeps falling below LOS strength?)
From the graph we find that after v = –0.8, F(v) keeps decreasing below 0dB. This value of (v)
corresponds to

d1d 2  0 .8 d1d 2
h  0.8   0.57r1
2d1  d 2  2 d1  d 2 
In other words, in order to guarantee diffracted signal to be equal in strength to LOS signal, we
have to keep obstacles outside 57% of the first Fresnel zone, or approximately 55% (sometimes it’s
taken as 60%).

7. Scattering

Scattering is the reflection of EM wave from a “rough” surface, in this case the EM wave does not
reflect in one direction like in smooth reflection, but scatters in different directions. Question is,
when is a surface considered to be rough or smooth?

Rayleigh criterion answers this question. When the maximum separation between top and bottom
parts of the surface, called (h) is greater than the critical height (h c), the surface is rough, otherwise
it is smooth. Critical height (h c) is given by


hc 
8 sin  i
 θi h

In this case, the reflected electric field intensity is less than that in the case of smooth (specular)
reflection. So the reflection coefficient must be multiplied by scattering loss factor (ρs)

rough   s  ρ s<1

When the wave is scattered from a rough surface, the received power is computed in two steps. First
from transmitter to the rough surface for which we find the power density as

Pt G t
Pd (d1 ) 
4d12

RCS(m2)

≈d2
≈d1

d1 d2

The power density at the reflector is multiplied by its effective area, called the Radar Cross Section
(RCS) in m2, which takes into account the reflection coefficient and the loss factor. The total power
reflected by the surface is then equal to
PG
Prefl  Pd (d1 )  RCS  t 2t  RCS
4d1

In the second step, this power is re–radiated towards the receiver, and the received power is given
by
P P G  RCS  G r 2 Pt G t G r 2  RCS
Pr (d 2 )  refl2 A Rx  t t 2 
4d 2 ( 4d1 )(4d 22 )(4) 4 3 d12 d 22
8. Practical Path Loss Models

Log–distance Model

We know that the path loss exponent (n) is 2 for free–space and 4 for two–ray model. In general (n)
can have other values depending on the environment surrounding the wave path. We define power
“loss” as

PL(d )[dB]  Pt [dB]  Pr (d)[dB]

 Pt [dB]  Pr (d o )[dB]  Pr (d o )[dB]  Pr (d )[dB]
 d 
 PL(d o )[dB]  10n log 
 do 

PL(do) is determined by site measurement or by using free–space model (although (n) is not
necessarily 2, we will use the free–space model as an approximation for PL(d o) only).

We want to apply the above formula in real life. In practice, the received power in a given
environment is not exactly equal to that calculated by this formula which was derived for ideal
cases. Two locations within the same environment (same path loss exponent n) and having same
distance (d) between Tx and Rx may have different actual path losses. In practice, random clutter
and variations between any two locations within the same environment cause these differences.
These random variations are called shadowing. Since shadowing is random, we will use statistical
models to describe it.

The above formula only gives a statistical “average value”, or “mean value”, of large scale path
loss. The exact path loss depends on shadowing which is random in nature. Shadowing is described
by probability density function. We will use over–bars to indicate that these formulas only give
average values and not exact. Thus, the average power loss is

 d 
PL(d )[dB]  PL(d o )[dB]  10n log 
 do 

And the average received power is

Pr (d)[dB]  Pt [dB]  PL(d)[dB]

Log–normal Shadowing

The random variations due to shadowing are Gaussian (or Normally) distributed around the mean
value. Any Gaussian distribution needs two parameters to be identified, mean–value and standard
deviation. The mean value (at distance (d) from transmitter) is given by Pr (d)[dB] , and the standard
deviation is given in dB as (σ[dB]). Note that both the mean and standard deviation are in dB. The
Probability Density Function (pdf) of the received signal at distance (d) is thus

1   x  P (d)[dB]2 
f X (x)  exp r
2

2 2   
 
The probability that the received signal, P r(d)[dB], is greater than a specific value, γ[dB] (in dB
too), is given by

Pr[Pr (d)   ]   f X ( x )dx


Again, we do not need to evaluate this integral; instead, we will use Q–function tables.

   Pr (d ) 
Pr[Pr (d)   ]  Q 
  

Of course, the probability that Pr(d) is less than γ is

   Pr (d)   P (d )   
Pr[Pr (d )   ]  1  Pr[Pr (d)   ]  1  Q   Q r 
     
Always remember: Pr(d), γ, and σ all are in dB.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Example:4.9(c,d) modified!
The received power at some reference d o=100m was found to be 0dBm, n=4.4, σ =6.875dB. c) find
the average received power at d=2kmd) find Pr[Pr(2km)>–60dBm)

Solution:
 2km 
Pr (2km)[dB]  Pr (100m)[dB]  10(4.4) log   57.25 dB
 100m 
  60dBm  (57.25dBm) 
Pr[Pr (d )  60dBm]  Q   Q(0.4)  1  Q(0.4)  1  0.34458  0.65542
 6.875dB 
Note that the term (–60dBm–(–57.25dBm)) = –2.75dB (not dBm!) which is the same if we convert
the –60dBm to –90dB and the –57.25dBm to –87.25dB then subtract them!
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––