Lecture 5: Large-Scale Path Loss

Chapter 4 – Mobile Radio Propagation:

Large-Scale Path Loss
I. Problems Unique to Wireless (not wired) systems:

 Paths can vary from simple line-of-sight to ones

that are severely obstructed by buildings,
mountains, and foliage.
 Radio channels are extremely random and
difficult to analyze.
 Interference from other service providers
 out-of-band non-linear Tx emissions

2
 Interference from other users (same network)
 CCI due to frequency reuse
 ACI due to Tx/Rx design limitations & large #
users sharing finite BW
 Shadowing
 Obstructions to line-of-sight paths cause areas of
weak received signal strength

3
 Fading
 When no clear line-of-sight path exists, signals are
received that are reflections off obstructions and
diffractions around obstructions
 Multipath signals can be received that interfere with
each other
 Fixed Wireless Channel → random & unpredictable
 must be characterized in a statistical fashion
 field measurements often needed to characterize radio
channel performance

4
 Wireless Communication Channels

From “Wireless Communications” Edfors, Molisch, Tufvesson

 Communications over wireless channels suffer from multi-path propagation

 Multi-path channels are usually frequency selective
 OFDM supports high data rate communications over frequency selective
channels
 ** The Mobile Radio Channel (MRC) has
unique problems that limit performance **
 A mobile Rx in motion influences rates of
fading
 the faster a mobile moves, the more quickly
characteristics change

6
II. Radio Signal Propagation

7
 The smoothed line is the average signal
strength. The actual is the more jagged line.
 Actual received signal strength can vary by
more than 20 dB over a few centimeters.
 The average signal strength decays with
distance from the transmitter, and depends on
terrain and obstructions.

8
 Two basic goals of propagation modeling:
1) Predict magnitude and rate (speed) of received
signal strength fluctuations over short
distances/time durations
 “short” → typically a few wavelengths (λ) or
seconds
 at 1 Ghz, λ = c/f = 3x108 / 1x109 = 0.3 meters
 received signal strength can vary drastically by 30
to 40 dB

9
 small-scale fluctuations → called _____ (Chapter 5)
 caused by received signal coming from a sum of
many signals coming together at a receiver
 multiple signals come from reflections and
scattering
 these signals can destructively add together by being
out-of-phase

10
2) Predict average received signal strength for
given Tx/Rx separation
 characterize received signal strength over distances
from 20 m to 20 km
 Large-scale radio wave propagation model models
 needed to estimate coverage area of base station
 in general, large scale path loss decays gradually
with distance from the transmitter
 will also be affected by geographical features like
hills and buildings

11
 Free-Space Signal Propagation
 clear, unobstructed line-of-sight path → satellite and
fixed microwave
 Friis transmission formula → Rx power (Pr) vs. T-R
separation (d)

12
 where
 Pt = Tx power (W)
 G = Tx or Rx antenna gain (unitless)
 relative to isotropic source (ideal antenna which
radiates power uniformly in all directions)
 in the __________ of an antenna (beyond a few meters)
 Effective Isotropic Radiated Power (EIRP)
EIRP = PtGt
 Represents the max. radiated power available
from a Tx in the direction of max. antenna gain,
as compare to an isotropic radiator

13
 λ = wavelength = c / f (m). A term is related to
antenna gain.
 So, as frequency increases, what happens to the
propagation characteristics?

 L = system losses (antennas, transmission lines

between equipment and antennas, atmosphere, etc.)
 unitless
 L = 1 for zero loss
 L > 1 in general

14
 d = T-R separation distance (m)
 Signal fades in proportion to d2
 We can view signal strength as related to the
density of the signal across a large sphere.
 This is the surface area of a sphere with radius d.
 So, a term in the denominator is related to distance
and density of surface area across a sphere.

15
 Path Loss (PL) in dB:

16
 d2 → power law relationship
 Pr decreases at rate of proportional to d2
 Pr decreases at rate of 20 dB/decade (for line-of-
sight, even worse for other cases)
 For example, path loses 20 dB from 100 m to 1 km
 Comes from the d2 relationship for surface area.

 Note: Negative “loss” = “gain”

17
 Example:
 Path loss can be computed in terms of a link budget
calculation.
 Compute path loss as a sum of dB terms for the
following:
 Unity gain transmission antenna.
 Unity gain receiving antenna.
 No system losses
 Carrier frequency of 3 GHz
 Distance = 2000 meters


18
 Close in reference point (do) is used in large-scale models

 do : known received power reference point - typically 100 m or

1 km for outdoor systems and 1 m for indoor systems
 df : far-field distance of antenna, we will always work problems
in the far-field

2D2
df  d f  D d f  


 D: the largest physical linear dimension of antenna

19
 Reference Point Example:
 Given the following system characteristics for large-
scale propagation, find the reference distance do.
 Received power at do = 20 W
 Received power at 5 km = 13 dBm

 Using Watts:

 Using dBm:

20
III. Reflections

 There are three basic propagation mechanisms

in addition to line-of-sight paths
 Reflection - Waves bouncing off of objects of large
dimensions
 Diffraction - Waves bending around sharp edges of
objects
 Scattering - Waves traveling through a medium
with small objects in it (foliage, street signs, lamp
posts, etc.) or reflecting off rough surfaces

21
 Reflection occurs when RF energy is incident upon
a boundary between two materials (e.g air/ground)
with different electrical characteristics
 Permittivity µ
 Permeability ε
 Conductance σ
 Reflecting surface must be large relative to λ of RF
energy
 Reflecting surface must be smooth relative to λ of
RF energy
 “specular” reflection

22
Fresnel reflection coefficient Γ
 The amount of energy reflected to the amount of
energy incidented is represented by Fresnel
reflection coefficient Γ, which depends upon the
wave polarization, angle of incidence and
frequency of the wave.
 For example, as the EM waves can not pass
through conductors, all the energy is reflected
back with angle of incidence equal to the angle
of reflection and reflection coefficient Γ = −1.

23
 What are important reflecting surfaces for
mobile radio?

 Fresnel reflection coefficient → Γ

 describes the magnitude of reflected RF energy
 depends upon material properties, polarization, &
angle of incidence

24
IV. Ground Reflection (2-Ray) Model
 Good for systems that use tall towers (over 50 m
tall)
 Good for line-of-sight microcell systems in urban
environments

25
 ETOT is the electric field that results from a combination of a
direct line-of-sight path and a ground reflected path

 is the amplitude of the electric field at distance d

 ωc = 2πfc where fc is the carrier frequency of the signal
 Notice at different distances d the wave is at a different phase
because of the form similar to

26
 For the direct path let d = d’ ; for the reflected path
d = d” then

 for large T−R separation : θi goes to 0 (angle of incidence

to the ground of the reflected wave) and
Γ = −1
 Phase difference can occur depending on the phase
difference between direct and reflected E fields
 The phase difference is θ∆ due to Path difference , ∆
= d”− d’, between

27
 From two triangles with sides d and (ht + hr) or (ht – hr)

28
 ∆ can be expanded using a Taylor series
expansion

29
 which works well for d >> (ht + hr), which means
and are small

30
 the phase difference between the two arriving
signals is

E0 d 0   
ETOT (t )  2 sin  
d  2
  2 hr ht
  0.3 rad
2 d
E0 d0 2 hr ht k
ETOT (t )  2  2 V/m
d d d

31
 For d0=100meter, E0=1, fc=1 GHz, ht=50 meters, hr=1.5 meters, at t=0

32
 note that the magnitude is with respect to a
reference of E0=1 at d0=100 meters, so near 100
meters the signal can be stronger than E0=1
 the second ray adds in energy that would have been
lost otherwise
 for large distances it can be shown
that

33
34
V. Diffraction

 RF energy can propagate:

 around the curved surface of the Earth
 beyond the line-of-sight horizon
 Behind obstructions
 Although EM field strength decays rapidly as
Rx moves deeper into “shadowed” or
obstructed (OBS) region
 The diffraction field often has sufficient
strength to produce a useful signal

35
 Huygen’s principle says points on a wavefront
can be considered sources for additional
wavelets.

36
 The wavefront on top of an obstruction generates
secondary (weaker) waves.

37
 The difference between the direct path and
diffracted path, call excess path length

 Fresnel-Kirchoff diffraction parameter

 The corresponding phase difference

38
39
 The excess total path length traversed by a ray
passing through each circle is nλ/2

40
 The diffraction gain due to the presence of a knife
edge, as compared the the free space E-field

41
42
43
44
45
46
47
48
VI. Scattering
 Received signal strength is often stronger than that
predicted by reflection/diffraction models alone
 The EM wave incident upon a rough or complex
surface is scattered in many directions and provides
more energy at a receiver
 energy that would have been absorbed is instead reflected to
the Rx.
 Scattering is caused by trees, lamp posts, towers, etc.
 flat surface → EM reflection (one direction)
 rough surface → EM scattering (many directions)

49
50
VII. Path Loss Models

 We wish to predict large scale coverage using

analytical and empirical (field data) methods
 It has been repeatedly measured and found that
Pr @ Rx decreases logarithmically with
distance

PL (d) = (d / do )n where n : path loss exponent or

PL (dB) = PL (do ) + 10 n log (d / do )

51
 “bar” means the average of many PL values at a
given value of d (T-R sep.)
 n depends on the propagation environment
 “typical” values based on measured data

52
 At any specific d the measured values vary
drastically because of variations in the
surrounding environment (obstructed vs. line-
of-sight, scattering, reflections, etc.)

 Some models can be used to describe a

situation generally, but specific circumstances
may need to be considered with detailed
analysis and measurements.

53
 Log-Normal Shadowing
PL (d) = PL (do ) + 10 n log (d / do ) + Xσ
 describes how the path loss at any specific location may vary
from the average value

 has a the large-scale path loss component we have already

seen plus a random amount Xσ.

54
 Xσ : zero mean Gaussian random variable, a “bell curve”
 σ is the standard deviation that provides the second
parameter for the distribution
 takes into account received signal strength variations
due to shadowing
 measurements verify this distribution
 n & σ are computed from measured data for different
area types
 any other path loss models are given in your book.
 That correlate field measurements with models for different
types of environments.

55
56
Log-normal Shadowing, n and σ
 The log-normal shadowing model indicates the
received 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.

57
Example of determining n and σ
 Assume Pr(d0) = 0dBm
and d0 is 100m
 Assume the receiver
power Pr is measured at
distances 100m, 500m,
1000m, and 3000m,
 The table gives the
measured values of
received power

58
 We know the measured values.
 Lets compute the estimates for received power at
different distances using long distance path loss
model.
 Pr(d0) is given as 0dBm and measured value is
also the same.
 mean_Pr(d) = Pr(d0) – mean_PL(from_d0_to_d)
 Then mean_Pr(d) = 0 – 10logn(d/d0)
 Use this equation to computer power levels at
500m, 1000m, and 3000m.
59
 Average_Pr(500m) = 0 – 10logn(500/100) = -6.99n
 Average_Pr(1000m) = 0 – 10logn(1000/100) = -10n
 Average_Pr(3000m) = 0 – 10logn(3000/100) = -14.77n
 Now we know the estimates and also measured actual
values of the received power at different distances
 In order approximate n, we have to choose a value for
n such that the mean square error over the collected
statistics is minimized

60
Path Loss Models
Path-Loss Models
 The most general case of signal reception might consist of a direct path,
reflected paths, diffracted paths, and scattered paths (which makes
mathematical analysis cumbersome)
 Path-Loss models are empirical models that are based on fitting curves or
analytical expressions that recreate a set of measured data
 Note:
 A given empirical model might only be valid within the environment where the
measurements used to estimate such model have been taken

62
Log-Distance Path-Loss Model
Theoretical and Measurement-based Propagation suggest that the
average received signal power decreases logarithmically with
distance

PL (d): Average path-loss for an arbitrary separation

n : Path-loss exponent

63
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

64
 Log-normal Shadowing
 Distance between two nodes alone cannot fully explain the signal
strength level at the receiver
 Shadowing has been introduced as a means to model the variation
of signal propagation behavior between two different signal paths
assuming the same propagation distance

PL (d): Path-loss model for an arbitrary separation d

Xσ : Shadowing parameter (zero mean Gaussian distributed random
variable in dB with standard deviation σ also in dB)

65
Received Power in Path-Loss Models

d d
PT - PL  d 
4 3

d
d

Positio
n
Index
1 2 1 2 3 4

66
Received Power in Path-Loss Models

PR  d 

d d X1
X 4
PT - PL  d  X3
4 3 X 2
d
d

Positio
n
Index
1 2 1 2 3 4

67
Reception Quality

PR  d 

d d X1
X 4
PT - PL  d  X3
4 3 X 2

d
d

Positio
n
Index
1 2 1 2 3 4
γ: Desired received power threshold

68
Pr  PR  d   γ   Pr  X σ  PT  PL  d   γ 
   
 Probability of Bad Reception Quality
Pr  PR  d      Pr  X   PT  PL  d    
   
 x2 
  2 
1  2σ
Pr  X σ  xth   e  
dx Xσ follows a normal distribution
2
2πσ xth with zero mean and standard
x deviation σ
Let z =   fX  x 
σ 
 z2 
  
1  2
Pr  X σ  xth    e  
dz
2π  xth  σ2
 
 σ 

x  1  xth 
Pr  X σ  xth   Q  th 
 2 erfc  
 σ   2σ 

Pr  PR  d   γ   Q 

 PT  PL  d   γ 

 xth
x

 σ 
 
 z2 
 -  
Note: Q(x)= 1 2
 2 
e π x
  -u 2
dz erfc(x)= e du
69 2π x
Percentage of Coverage Area
 Due to the random effects
of shadowing some
locations within the
coverage area will be h
below a particular desired R’
received signal level R

 So, its better to compute

how the boundary coverage PR  d    0  d  R'
area relates to the percent
of area covered within the PR  d    0  d  R'
boundary
R: Radius of Coverage Area
required for Transmitter

70
Calculation of Percentage of Coverage Area

Assume h (height of antenna) is Negligible, then,

U(γ) depicting the percentage of area with
received signal strength equal to or exceeding γ
dA
may be calculated as follows
r
1 r
U γ    P r  PR  r   γ  d A dθ
πR 2
2π R R
1
U γ     P r  P  r  
R γ  r d r d θ
πR 2 0 0


   PT  PL  r  
Pr  PR  r      Q  

   R: Radius of Coverage Area
  required for Transmitter

1
Pr  PR  r      erfc 

   PT  PL  r   
2  2 
 

1
Pr  PR  r      erfc   T 
    P  PL  d   10nlog  r d   
0 0
 
2  2 
71  
 Calculation of Percentage of Coverage Area

1
Pr  PR  r      erfc   T  0 0 
    P  PL  d   10nlog  r d   

2  2 
 

1
Pr  PR  r      erfc   T  0 0 
    P  PL  d   10nlog  R d   10nlog  r R   

2  2 
 
2π R
1
U γ     Pr  P  r   γ  rdrdθ
R
πR 2 0 0
R 2π R
1 2
U γ   0  R
Pr  P  r   γ 
 r d r 0 dθ   Pr  P  r   γ  rdr
R
πR 2 R2 0
R
2 1  r 
U  γ   2  r  erfc  a  b ln  d r
R 0 2  R

a  
γ   PT  PL  d 0   10 n log  R d 0  
  b
10 n log e 
2σ 2σ
72
Calculation of Percentage of Coverage Area

It can be shown that

1   1  2ab    1  ab  
U γ    1  e rf  a   ex p  2  1  erf   
2   b   b  

By choosing the signal level such that

P R R   γ  i .e ., a  0
Therefore for the case when
Boundary Coverage = 50 %

1   1   1  
U γ    1  e x p  2  1  e r f    
2   b   b  
73
Calculation of Percentage of Coverage Area

“Wireless Communications:
Principles and Practice 2nd
Edition”, T. S. Rappaport,
Prentice Hall, 2001

74
Outdoor Propagation Models
 Longley-Rice Model (Read)

 Durkin’s Model (Read)

 Okumura’s Model

 Hata Model

 PCS extension to Hata Model

 Walfisch and Bertoni

75 (Read)
Okumura’s Model
 Okumura’s model is one of the most widely used models
for signal predictions in urban and sub-urban mobile
communication areas

 This model is applicable for frequencies ranging from 150

MHz to 1920 MHz

 It can cover distances from 1 km to 100 km and it can be

used for base station heights starting from 30m to 1000m

 The model is based on empirical data collected in detailed

propagation tests over various situations of an irregular
terrain and environmental clutter

76
 Okumura’s Model

L50  dB   L F  A mu  f , d   G  h te   G  h re   G AREA

 L50 is the median value or 50th percentile value of the propagation path loss
 LF is the free space propagation path loss
 Amu is the median attenuation relative to free space
 GAREA is the gain due to the type of environment
 G(hte) is the base station antenna height gain factor
 G(hre) is the mobile antenna height gain factor

77
Okumura’s Model: Amu Curves

“Wireless Communications:
Principles and Practice 2nd
Edition”, T. S. Rappaport,
Prentice Hall, 2001

78
Okumura’s Model: GArea Curves

“Wireless Communications:
Principles and Practice 2nd
Edition”, T. S. Rappaport,
Prentice Hall, 2001

79
Okumura’s Model: G(hte), G(hre)
 The empirical model of Okumura assumed hte =
200m, hre = 3m

 h te 
G  h te   2 0 lo g   3 0 m  h te  1 0 0 0 m
 200 
 h re 
G  h re   1 0 lo g   h re  3 m
 3 
 h re 
G  h re   2 0 lo g   3 m  h re  1 0 m
 3 

80
Hata Model

L50  urban  dB   69.55  26.26log  fc   13.82log  h te   a  h re 

  44.9  6.55log  h te   log  d 

 L50 is the median value or 50th percentile value of the propagation

path loss
 fc (in MHz) is the frequency (15MHz to 1500MHz)
 hte is the effective transmitter height in meters (30m to 200 m)
 hre is the effective transmitter height in meters (1m to 10 m)
 d is the T-R separation in Km
 a(hre) is the correction factor for effective mobile (i.e., receiver)
antenna height which is a function of the size of the coverage area
81
 Hata Model: a(hre)
 For a Medium sized city, correction factor is given by:

a  h re   1.1log  f c   0.7  h re  1.56log  f c   0.8 dB

 For a Large city, correction factor is given by:

a  h re   8.29  log 1.54h re    1.1

2
dB for f c  300MHz

a  h re   3.2  log 11.75h re    4.97

2
dB for f c  300MHz

82
Hata Model
 Path loss in suburban area, the equation is modified as

L 50  dB   L50  urban   2 log  f c / 28    5.4

2

 For path loss in open rural areas, the formula is modified as

L 50  dB   L50  urban   4.78  log  f c    18.33log  f c   40.94

2

 Hata Model is well-suited for Large cell mobile systems

83
 PCS Extension to Hata Model
 An extended version of the Hata model developed by COST-231
working committee for 2 GHz range

L50  urban  dB   46.3  33.9log  f c   13.82log  h te   a  h re 

  44.9  6.55log  h re   log  d   CM

 fc is the frequency (1500MHz to 2000 MHz)

 hte is the effective transmitter height in meters (30m to 200 m)
 hre is the effective transmitter height in meters (1m to 10 m)
 d is the T-R separation in Km (1 Km to 20 Km)
 CM=0 dB for medium sized city and suburban areas, CM=3 dB for
metropolitan centers
84
Indoor Propagation Models
 The indoor radio channel differs from the traditional mobile
radio channel in the following aspects
 Much smaller distances
 Much greater variability of the environment for a much smaller
range of T-R separation distances
 Difficult to ensure far-field radiation

 Propagation within buildings is strongly influenced by

specific features such as
 Building layout
 Construction materials
 Building type
 Open/Closed doors
 Locations of antennas

85
Partition Losses (Same Floor)

“Wireless Communications:
Principles and Practice 2nd
Edition”, T. S. Rappaport,
Prentice Hall, 2001

86
Partition Losses between Floors

“Wireless Communications:
Principles and Practice 2nd
Edition”, T. S. Rappaport,
Prentice Hall, 2001

87
Log-Distance Pathloss Model
 The lognormal shadowing model has been shown to be
applicable in indoor environments

88
 Ericsson Multiple Breakpoint Model

Upper bound
Lower bound on the path-
on the path- loss
loss

“Wireless Communications:
Principles and Practice 2nd
Edition”, T. S. Rappaport,
Prentice Hall, 2001

89
Attenuation Factor Model
 This was described by Seidel S.Y. It is an in-building propagation
model that includes
 Effect of building type
 Variations caused by obstacles

 nSF represents the path-loss exponent for the same floor

measurements
 FAF represents the floor attenuation factor
 PAF represents the partition attenuation factor for a specific
obstruction encountered by a ray drawn between the transmitter and
receiver

90
Attenuation Factor Model
 FAF may be replaced by an exponent that accounts for the
effects of multiple floor separation

 nMF represents the path-loss exponent based on measurements

through multiple floors

91