1

EE6501
Wireless Communications
Instructor: Dr. Jehan Zeb, [email protected]
2 Announcements

Agenda for Today

➢ Introduction to Radio Propagation models
➢ 2-ray propagation model
➢ Simplified Pathloss model
➢ Calculation of path loss exponent (MMSE fit to empirical readings)
➢ Shadowing and pathloss
 Questions!!!

4
5

Tutorial on Radio
Propagation - Large-
Scale Path Loss
 Propagation Models
6
 We are interested in propagation characteristics and
models for waves with frequencyy in range: few MHz to a
few GHz
 Modeling radio channel is important for:
 Link Budget
 Link budget design determines fundamental quantities such as
transmit power requirements, coverage areas, and battery life
 It is determined by the amount of received power that may be
expected at a particular distance or location from a transmitter
 Finding modulation and coding schemes to improve the channel
quality
 Determine the maximum channel capacity

 Time dispersion
 It arises because of multi-path propagation where replicas of
transmitted signal reach the Rx with different propagation delays due
to the propagation mechanisms that are described earlier.
 Time dispersion nature of the channel determines the maximum data
rate that may be transmitted without using equalization.
 7 Radio Propagation Mechanisms

 The physical mechanisms

that govern radio
propagation are complex
and diverse but are
generally attributed to the
three main factors
 Reflection
 Diffraction
 Scattering
8 Radio Propagation Models

 As the mobile moves away from the transmitter over larger

distances, the local average received signal will gradually
decrease. This is called large-scale fading
 Typically the local average receiver power is computed by averaging
signal measurements over a measurement track of 5-40 lambda
(displacement of 1-10m track)
 The models that predict the mean signal strength of an arbitrary
receiver transmitter (T-R) separation distance are called large-scale
propagation models
 Useful for estimating the coverage area of the transmitters
9 Radio Propagation Models

 As the mobile moves over small distances, the instantaneous

received signal fluctuates rapidly giving rise to small-scale fading
 The reason is that the signal is the sum of many contributors coming
from different directions and since the phases of these signals are
random the sum behave like a noise (distribution of which is known as
Raleigh fading)
 In small scale fading, the received signal power may change as much
as 3-4 orders of magnitude (30-40dB) when the receiver is only moved a
fraction of the waveleght
10 Radio Propagation Mechanisms
 The received signal strength can be factored into
11
12 Propagation Models

 Radio propagation models can be derived

 Empirical methods – collect measurement, fit curves
 Analytical methods – model the propagation mechanisms
mathematically and derive equations for path loss
 Long distance path loss model
 Empirical and analytical models show that received signal
power decreases logarithmically with distance for both indoor
and outdoor channels
13 Transmit and Receive Signal Model

 Let u(t) be the baseband modulated transmitted signal

 The received signal will have a similar form:

{
r (t ) = Re v(t )e j 2 πf c t }
where complex baseband signal v(t) will depend on
channel through which u(t) propagates.

In particular, if s(t) is transmitted through a time-invariant

channel then v(t) = u(t) ∗ c(t), where c(t) is the equivalent
lowpass channel impulse response for the channel
14
Pathloss
 Suppose u(t) of pwr Pt is transmitted through a given
channel, with corresponding recd signal r(t) of pwr Pr,
averaged over random variations due to shadowing.
 We define the linear path loss of the channel as the ratio
of transmit pwr to receive pwr
Pt
PL =
Pr
 We define PL of the channel as the dB value of the linear
path loss or, equivalently, the difference in dB between
the transmitted and received signal power
 Pt 
PL ( dB ) = 10 log10  
 Pr 
15 Free-Space Propagation Model
 Used to predict the received signal strength when
transmitter and receiver have clear, unobstructed LOS
path between them.
 The received power decays as a function of T-R
separation distance raised to some power.
 Path Loss: Signal attenuation as a positive quantity
measured in dB and defined as the difference (in dB)
between the effective transmitter power and received
power.
16 Free-Space Propagation Model
 Free space power received by a receiver antenna
separated from a radiating transmitter antenna by a
distance d is given by Friis free space equation:

Pr(d) = (PtGtGrλ2) / (4πd)2 1

 Pt is transmitted power
 Pr(d) is the received power
 Gt is the transmitter antenna gain (dimensionless quantity)
 Gr is the receiver antenna gain (dimensionless quantity)
 d is T-R separation distance in meters
 λ is wavelength in meters.
17 Free-Space Propagation Model
 Gain of an antenna G is related to its effective aperture Ae by:

G = 4πAe / λ2 2
Effective aperture Ae is related to physical size of
antenna,
where λ is given by

λ = c/f = 2πc / ωc 3
 f is carrier frequency in Hertz

 ωc is carrier frequency in radians per second.

 c is speed of light in meters/sec
18 Free-Space Propagation Model
 Path loss, which represents signal attenuation as positive
quantity measured in dB, is defined as the difference (in
dB) b/w the effective transmitted power and received
power
4
PL(dB) = 10 log (Pt/Pr) = -10log[(GtGrλ2)/(4π)2d2]
(You can drive this from equation 1)

5
 If antennas have unity gains (Gt = Gr = 1)

PL(dB) = 10 log (Pt/Pr) = -10log[λ2/(4π)2d2] 6

 Summarizing Free-space propagation model
19
 Consider a signal transmitted through {
r (t ) = Re c(t ) * u (t )e j 2 πf c t }
free space to a Rx located at = Re{αδ(t − τ) * u (t )e j 2 πf c t }
distance d from Tx. Assume there are {
= Re αu (t − τ)e j 2 πf c (t − τ ) }
no obstructions b/w Tx & Rx
 λ Gl 
j 2 πf c ( t − τ ) 
 The channel model associated with = Re u (t − τ)e 
 4 πd 
this transmission is called a line-of-
sight (LOS) channel  λ G −
j 2 πd  
j 2 πf c t 
= Re  u (t − τ)e λ
e
l

 4 πd  

Pwr transmitted in signal u(t) is Pt, ratio of received to transmitted

power is
2
 λ Gl 
Pr (d ) = Pt  
 4πd 
 
20
21 Validity of Free-Space Propagation
Model
 For Friis equation to hold, distance d should be in the
far-field of the transmitting antenna.
 The far-field, or Fraunhofer region, of a transmitting
antenna is defined as the region beyond the far-field
distance df
7
 df = 2D2/λ
D is the largest physical dimension of the
antenna.
Additionally, df >> D and df >> λ
22 Free-Space Propagation Model – Reference
Distance d0
 It is clear the Eq 1 does not hold for d = 0.
 For this reason, models use a close-in distance d0 as the
received power reference point.
d0 should be >= df
d0 should be smaller than any practical distance a
mobile system uses
 Received power Pr(d), at a distance d > d0 from a
transmitter, is related to Pr at d0, which is expressed as
Pr(d0).
 The power received in free space at a distance greater
than d0 is given by:

Pr(d) = Pr(d0)(d0/d)2 d >= d0 >= df 8

23 Free-Space Propagation Model
 Expressing the received power in dBm and dBW

Pr(d) (dBm) = 10 log [Pr(d0)/0.001W] + 20log(d0/d) 9

where d >= d0 >= df and Pr(d0) is in units of watts.

Pr(d) (dBW) = 10 log [Pr(d0)/1W] + 20log(d0/d) 10

where d >= d0 >= df and Pr(d0) is in units of watts.

 Reference distance d0 for practical systems:

 For frequncies in the range 1 - 2.5 GHz
 1 m in indoor environments
 100m-1km in outdoor environments
24
TWO RAY MODEL
25

 The free space propagation model does not take into

account signal reflections.
 The signal transmitted by a land-based antenna is
received by another land-based antenna in the
presence of a strong ground reflection.
 The model is quite accurate for predicting received
signal strength over several km, using tall antennas
 Derivation
26
27  Let the channel 2-ray channel impulse response be

c(t ) = α1δ(t − τ1 ) + Rα 2 δ(t − τ 2 )

where, α1 & α 2 are the channel coefficient for the direct path and the reflected path respectively
τ1 & τ 2 are the channel delays for the direct path and the reflected path respectively

R is the ground reflection coefficient

 The received signal is defined as:

{
r (t ) = Re c(t ) * u (t )e j 2 πf c t }
= Re{α1δ(t − τ1 ) * u (t )e j 2 πf c t + Rα 2 δ(t − τ 2 ) * u (t )e j 2 πf c t }
{
= Re α1u (t − τ1 )e j 2 πf c (t − τ1 ) + Rα 2 u (t − τ 2 )e j 2 πf c (t − τ 2 ) }
 λ Gl j 2 πf c ( t − τ1 ) λ Gl 
j 2 πf c ( t − τ 2 ) 
= Re u (t − τ1 )e +R u (t − τ 2 )e 
 4 πl π
4 ( x + x '
) 
 λ G −
j 2 πl
λ G −
j 2 π ( x + x ')  
π
= Re  u (t − τ1 )e λ + R − τ λ

l l j 2 f t
u (t ) e e c

π π +
2
 
'
4 l 4 ( x x ) 
28  The received signal is defined as:
 λ G −
j 2 πl
λ G −
j 2 π ( x + x ')  
j 2 πf c t
r (t ) = Re u (t − τ1 )e λ + R − τ λ

l l
u ( t ) e e 
π π +
2
 
'
4 l 4 ( x x ) 
 λ Gl − j 2 πl  u (t − τ ) u (t − τ ) −
j 2 π ( x + x ') j 2 πl
+  π

r (t ) = Re e λ  1
+R 2
e λ λ
e c 
j 2 f t

 4π x+x '
 l  
If transmitted signal is Narrowband
signals relative to delay spread, u (t − τ1 ) ≈ u (t − τ1 ) ≈ u (t )

 λ Gl j 2 π( f c t − l )  1 R −
j 2 π( x + x '−l ) 

= Re  e λ
u (t )  + e λ

 4π  l x + x
'
  ∆φ = phase difference
between the two paths

 λ Gl j 2 π( f c t − l )  2 π( x + x '−l )
1 R − j∆φ   ∆φ =
= Re  e λ
u (t )  + e 
λ
 4 π  l x + x '
 
  With this appr, recd pwr of the two-ray model for
29 narrowband transmission is
2 2
 λ  1 R
Pr = Pt Gl   + e − j∆φ
A
 
4 π l x + x '

for asymptotically large Tx-Rx separation distance d, x + x’ ≈ l ≈ d, and R ≈ −1.

2
Pr  λ  − j∆φ
2
= Gl   1− e
Pt  4πd 
2
 λ 
 1 − {cos(−∆φ) + j sin(−∆φ)}
Pr 2
= Gl 
Pt  4πd 
As, ∆φ ≈ 0, cos(∆φ) ≈ 1 & sin(∆φ) ≈ ∆φ
2
 λ 
 1 − {1 − j∆φ}
Pr 2
= Gl 
Pt  4πd 
2
 λ 
= Gl   (∆φ) B
2

 4πd 
30

hr x’ hr
d

l
Δφ can be represented in terms of transmitter and receiver antenna heights
using the geometry of 2-ray model. Using geometry we can show that

x + x ' − l = (ht + hr ) 2 + d 2 − (ht − hr ) 2 + d 2

When d is very large compared to ht + hr (Taylor series approximation)
2 π( x + x '−l ) 4 πht hr
∆φ = ≈
λ λd
31 Substituting these approximations into B yields that, in this asymptotic limit,
the received signal power is approximately
2
 λ Gl 
Pr = Pt   (∆φ) 2
 4πd 
 
2 2
 λ Gl   4πh h  2  Gl ht hr 
Pt = Pt    t r
 = Pt   C
 4πd   λd   d2 
   
or, in dB, we have
Pr ( dBm ) = Pt ( dBm ) + 10 log10 (Gl ) + 20 log10 (ht hr ) − 40 log10 (d )

Thus, in the limit of asymptotically large d, recd pwr falls off inversely with fourth
power of d and is independent of the wavelength λ.
32
33 Intuition
 Previous plot uses f = 900 MHz, ht = 50 m, hr = 2 m, Gl = Gr =1
 For small distances, d < ht, the two rays add constructively and
the recd pwr slowly increases
 For d > ht and up to a certain critical distance dc, the wave
experiences constructive and destructive interference of the 2
rays resulting in maxima and minima
 These maxima & minima are referred to as small-scale fading
 At dc final maxima is reached from where onwards power falls
off proportionally with d-4
 Rapid falloff for d > dc is due to the fact that the signal only
combine destructively & are out of phase by approximately π
 dc can be obtained by setting ∆φ = π, this results in dc = 4 ht hr /λ
34 Path Loss Exponent for Different
Environments
Environment Path Loss Exponent, n
Free space 2

Urban area cellular radio 2.7 to 3.5

Shadowed urban cellular radio 3 to 5

In building line-of-sight 1.6 to 1.8

Obstructed in building 4 to 6

Obstructed in factories 2 to 3
35 Long distance path loss model
 The average large-scale d n
path loss for an arbitrary T-R PL(d ) ∝ ( )
d0
separation is expressed as a
d
function of distance by using PL(dB)= PL(d 0 ) + 10n log( )
a path loss exponent n: d0

PL(d ) denotes the average large - scale path loss

 The value of n depends on at a distance d (denoted in dB)
the propagation
environment: for free space
it is 2; when obstructions are
present it has a larger value.
36 Shadowing
 The PL equation does not consider the fact the
surrounding environment may be vastly different at two
locations having the same T-R separation
 This leads to measurements that are different than the
predicted values obtained using the previous Path Loss
equation.
 Measurements show that for any value d, the path loss
PL(d) in dBm at a particular location is a random variable
(RV)
 Distribution needs to be known?
37
38 Log-normal Shadowing- Path Loss
Then adding this random factor:

PL(d )[dB] = PL(d ) + X σ

d
PL(d )[dB] = PL(d 0 ) + 10n log( ) + X σ
d0
PL(d ) denotes the average large-scale path loss (in dB) at a distance d.

Xσ is a zero-mean Gaussian (normal) distributed random variable (in dB)

with standard deviation s (also in dB).

PL(d 0 ) is usually computed assuming free space propagation model

between transmitter and d0 (or by measurement).
Above equation takes into account the shadowing affects due to
cluttering on the propagation path. It is used as the propagation
model for log-normal shadowing environments.
 Why Log-normal Shadowing?
39
 When shadowing is dominated by attenuation from blocking objects, the
attenuation of a signal as it travels through an object of depth d is appr
equal to
s ( d ) = e − αd

α is attenuation constant that depends on object’s materials &

dielectric properties. Assuming equal α for all blocking objects, & ith
blocking object w/ random depth di, then attenuation of a signal as it
propagates is
s (d t ) = e ∑i
−α d
= e − αd t
i

where d t = ∑id i . If there are many objects b/w Tx and Rx, then by CLT,
dt can be approximated by a Gaussian random variable.
Thus, log s(dt) = αdt will have a Gaussian distribution w/ mean μ &
standard deviation σ. The value of σ will depend on environment.
40 Log-normal Shadowing- Received Power

 The received power in log-normal shadowing environment is given by the

following formula

Pr (d )[dBm] = Pt [dBm] − PL(d )[dB]

 d 
Pr (d )[dBm] = Pt [dBm] −  PL(d 0 )[dB] + 10n log( ) + X σ [dB]
 d0 
 The antenna gains are included in PL(d).
41 Log-normal Shadowing, n and σ

 The log-normal shadowing model indicates the recd

power at a distance d is normally distributed with a
distance dependent mean and with a standard
deviation of σ

 In practice the values of n and σ are computed from

measured data using linear regression so that the
difference between the measured data and estimated
path losses are minimized in a mean square error sense.
42 Outage Probability under Path Loss & Shadowing

 The combined effects of path loss and shadowing have

important implications for wireless system design
 Typically a target minimum received power level Pmin below
which performance becomes unacceptable (e.g. the voice
quality in a cellular system is too poor to understand)
 However, with shadowing the received power at any given
distance from the transmitter is log-normally distributed with
some probability of falling below Pmin
 Outage probability pout(Pmin, d) with path loss & shadowing is
the probability that the received power at a given distance
d, Pr(d), falls below Pmin: pout(Pmin, d) = p(Pr(d) < Pmin)

 Pmin − Pr (d )
P(Pr (d ) ≤ Pmin ) = 1 − Q  
 σ 
43 Outage Probability under Path Loss & Shadowing

 The combined effects of path loss and shadowing have

important implications for wireless system design
 Typically a target minimum received power level Pmin below
which performance becomes unacceptable (e.g. the voice
quality in a cellular system is too poor to understand)
 However, with shadowing the received power at any given
distance from the transmitter is log-normally distributed with
some probability of falling below Pmin
 Outage probability pout(Pmin, d) with path loss & shadowing is
the probability that the received power at a given distance
d, Pr(d), falls below Pmin: pout(Pmin, d) = p(Pr(d) < Pmin)

 Pmin − Pr (d ) 
P(Pr (d ) ≤ Pmin ) = 1 − Q  
 σ 
44

 The probability that the received signal level will exceed a certain value γ
can be calculated from the cumulative density function as:

 γ − Pr (d ) 
Pr[Pr (d ) > γ ] = Q 
 σ 
 
 The probability that the received signal level will fall below a certain value γ
can be calculated as:

 γ − Pr (d )   Pr (d ) − γ 
Pr[Pr (d ) < γ ] = 1 − Pr[Pr (d ) > γ ] = 1 − Q  = Q 
 σ   σ 
   
45 Outage Probability under Path Loss & Shadowing

 The combined effects of path loss and shadowing have

important implications for wireless system design
 Typically a target minimum received power level Pmin below
which performance becomes unacceptable (e.g. the voice
quality in a cellular system is too poor to understand)
 However, with shadowing the received power at any given
distance from the transmitter is log-normally distributed with
some probability of falling below Pmin
 Outage probability pout(Pmin, d) with path loss & shadowing is
the probability that the received power at a given distance
d, Pr(d), falls below Pmin: pout(Pmin, d) = p(Pr(d) < Pmin)

 Pmin − Pr (d )
P(Pr (d ) ≤ Pmin ) = 1 − Q  
 σ 
46
47
48
 Empirical Path Loss Models
49
 Most wireless channels cannot be accurately modeled by
free-space path loss or ray tracing.
 A No of path loss models, developed over years, predict
path loss in typical wireless environments – large urban
macrocells, urban microcells & more recently, inside
buildings
 These models are mainly based on empirical
measurements over a given distance in a given frequency
range and a particular geographical area or building.
 However, applications of these models are not always
restricted to environments in which empirical
measurements were made, which makes the accuracy of
such empirically-based models applied to more general
environments somewhat questionable.
 Nevertheless, many wireless systems use these models as a
basis for performance analysis
 Empirical Path Loss Models
50
 Most wireless channels cannot be accurately modeled by
free-space path loss or ray tracing.
 A No of path loss models, developed over years, predict
path loss in typical wireless environments – large urban
macrocells, urban microcells & more recently, inside
buildings
 These “Empirical
models are Path-Loss
mainly basedModels”
on empirical
Please go through the book
measurements over a given distance in a given frequency
range and a particular geographical area or building.
 However, applications of these models are not always
restricted to environments in which empirical
measurements were made, which makes the accuracy of
such empirically-based models applied to more general
environments somewhat questionable.
 Nevertheless, many wireless systems use these models as a
basis for performance analysis
 Reference d0 = 100 m
51
Distance from Received
Tx Power
Example 100 m (reference) 0 dBm
200 m -20 dBm
1000 m -35 dBm
a) Find MMSE estimate for n 3000 m -70 dBm
b) Find σ about the mean value
c) Use the above model to estimate the received power at d =
2000 m
d) Predict the likelihood that the received signal level at 2 km
will be greater than -60 dBm
e) Calculate the distance d (mtrs) beyond which the outage
probability is less than 1% (Assume Pmin = -60 dBm)
52 a
.
53
54

The actual received power at d=2 km is a

Guassian random variable with mean -57.2 dBm
and
a standard deviation σ = 6.2 dB
55
  Pmin − Pr (d ) 
56 e. P(Pr (d ) ≤ Pmin ) = 1 − Q  
 σ 
 Pmin − Pr (d )
1 − Q  ≤ 0.1
 σ 
 Pmin − Pr (d )
Q  > 0.9
 σ 
 Pmin − Pr (d ) −1
  > Q (0.9)
 σ 

Solve the equation to find the mean received power for the
outage probability to be less than 1% - (Answer = - 46.2 dBm)

Use this, to measure the distance d (mtrs), beyond which

outage is less than 1% (Answer = 1122 mtrs)