EE 463 - Wireless

EE 4260
Communications
Dr. Ahmed Iyanda Sulyman
Associate Professor
Electrical Engineering Department
King Saud University

Chapter 3: Mobile radio

propagation: Large-scale path loss
1. Introduction
2. Deterministic models for Mobile radio
propagation
3. Statistical models - shadowing
4. Determination of % coverage area
5. Empirical Outdoor and Indoor propagation models

1
 1 – Introductions
 Unlike wired channels that are stationary and
predictable, mobile radio channels are extremely
random and not easily analyzed or predicted.

Reflected
Reflected
Path #1
Path # 2

Direct path or line of sight (LOS)

Base Mobile user

Reflected
station Path # 3

 How do we predict these random propagation

channel effects on the transmitted signals in mobile
communication systems?. [needed for planning & designs]

1 – Introductions

• Propagation models have been developed for

predicting the average received signal strength
(RSS) at a given distance from the transmitter, as
well as the variations in the RSS.
 Large-scale propagation models: Deterministic,
Shadowing, and Empirical models for predicting the
RSS as function of distance from BS.

 Small-scale fading models: Statistical models for

predicting random variations in the average RSS.

2
 1 – Introductions
 Large scale propagation models:
 Predict the mean signal strength over large
transmitter-receiver distances (thousands of meters) .
 Useful in estimating radio coverage area of
transmitter (cell) as well as interference among cells.
1

0.8

Coverage U(r) 0.6

 Factors:
0.4
1. Free space
propagation loss 0.2
(path loss),
0
2. Ground reflections, 200 400 600 800 1000 1200
Distance(m) (Linear scale)
3. Diffractions and scattering,
4. Shadowing (caused by mobile environment).
5

1 – Introductions
• Small scale propagation models:
 Characterize rapid fluctuations of the received signal strength
over very short distances (a few wavelengths) or over short
time durations (seconds).
 Factors:
1) multipath propagation, 2). Speed of the mobile, 3). Speed
of the surrounding objects, 4). Bandwidth of transmitted
signals
• Small-scale fading
are sometimes called
multipath fading Prec
because the
contribution of Pt
multipath
phenomenon to this
fading type is huge.
log(d ) 6

3
 1 – Introductions
• Three basic mechanisms impacting both Large
scale & Small scale propagations:
 Reflection: occurs when a propagating electromagnetic
wave falls on objects much larger than its wavelength,
causing reflections {e.g. Earth surface, buildings, walls}.
 Diffractions: occurs when radio path between
transmitter and receiver is obstructed. The emergent
wave over the obstacle edge is bent at an angle different
from the incident angle on the obstacle, impacting on the
energy & distance traverse by the wave after diffraction.
 Scattering: occurs when propagating electromagnetic
wave hits objects along its path, whose dimensions are
small compared to its wavelength, and when the number
of obstacles per unit space is too large {e.g. downtown
area}, causing scattering of the wave.
7

2 – Deterministic models
 Free space propagation model:
 Used to predict received signal strength when the
transmitter and receiver have a clear, unobstructed
line-of-sight (LOS) path between them.
 Satellite communication systems and micro-wave
line-of-sight radio links typically undergo free space
propagations.
 The power received by a receiving antenna separated
from the transmitting antenna by a distance d, is
given by  Pt = transmitted power
 Prec(d) = received power at
distance d
Pt Gt Gr  2
Prec d  
 Gt = transmitting antenna gain
watts  Gr = receiving antenna gain
(4 ) 2 d 2 L  L = system loss factor, L ≥1
  =c/f = wavelength of the
transmitted carrier signal.
8

4
 2 – Deterministic models
 Free space propagation model:
 Example 3.1:
 If a transmitter applies 50W of power, to a unity gain
antenna with a 900 MHz carrier frequency, (a) find
the received power (in dB) at a free space distance of
100m from the antenna, (b) what is Prec(10km)?.
Assume unity gain for the receiver antenna.
 Ans: 50(1)(1)(3 / 9) 2
Prec ( d ) 
 ( 4 ) 2 d 2 (1)
 Prec(100m)=-54.5dB, Prec(d=10km)=-94.5dB

 The effective isotropic radiated power (EIRP) of

an antenna is given by EIRP=Pt*Gt.
 An isotropic (ideal) antenna has unity gain (G =1).
9

2 – Deterministic models
 Free space propagation model (referenced to d0):
 Equation above is not valid when d=0, therefore large-
scale propagation models use a close-in reference
distance, d0, as a known received power reference point.
 The received power at any distance d ≥ d0 is then given by
2
 d 
Prec d   Prec d 0   
 d0 
Received power decays as sq. of d, or -20dB/dec
Prec(d0) is obtained by measurements.
 The path loss: Is the difference (in dB) between the
effective transmitted power, and the received power:
PL(dB) = Pt(dB) - Prec(dB).
 Gt G r 2 
 The path loss for free space model is PL ( dB )   10 log  

 4  d L 
2 2

10

5
 2 – Deterministic models
 Received Power and E-magnetic Field:
 The power received at distance d, is related to the
electric field energy as.
2
Pt G t G r  2 E G r 2
Prec d   
4 2 d 2 480  2 watts

 |E| is the magnitude of the radiating portion of the

electric field (in V/m).

11

2 – Deterministic models
 Free space propagation model:

 Example 3.2:
 Assume a receiver is located 10km from a 50W
transmitter. The carrier frequency is 900MHz, free
space propagation is assumed, Gt=1, and Gr=2. Find
(a) the power at the receiver, (b) the magnitude of the
E-field at the receiver antenna, (c) the rms voltage
applied to the receiver input assuming that the receiver
antenna has a purely real impedance of 50Ω and is
perfectly matched to the receiver (i.e. no loss).
 Ans: (a) Prec(d)=-91.5dB, (b) |E|=0.0039V/m, (c)
Prec(d)= V2/R: thus V=0.187mV.

12

6
 2 – Deterministic models
 Ground reflection (two-ray) model:

 In a mobile radio channel, a single direct path

between the base station and a mobile is not always
achieved, and hence the free space propagation
model above is in most cases inaccurate when used
alone.
 The 2-ray ground reflected model considers both the
direct path and a ground reflected propagation path
between transmitter and receiver, and this model is
found to be reasonably accurate for predicting the
large-scale signal strength over distances of several
kilometers for mobile radio system with tall towers
(over 50m).

13

2 – Deterministic models
 Ground reflection (two-ray) model:

 Illustration:

Transmitter

ht Receiver

hr

14

7
 2 – Deterministic models
 Ground reflection (two-ray) model:

 The received power at distance d from the transmitter

for this model is
Pt G t G r h t2 h r2
Prec ( d ) 
d4
 The path loss for the 2-ray model is

PLdB (d )  40 log d  10 log Gt  10 log Gr  20 log ht  20 log hr 

15

2 – Deterministic models
 Ground reflection (two-ray) model:

 Example 3.3
 A mobile is located 5km from a 10W base station
transmitter uses an antenna with a gain of 2.55dB to
receive cellular radio signals. Given that the base
station antenna is an isotropic radiator and is 50m tall
while the mobile station antenna is 1.5m above the
ground, estimate the received power using two-ray
model.

 Ans: Prec(d) = 10log10[PtGt Gr h2t h2r / d4] = -98 dB

16

8
 2 – Deterministic models
 The path loss model:

 Some classical propagation models have emerged

overtime, and these are used in practice to predict
large-scale coverage for mobile communications
system designs.

 This approach uses the path loss models to estimate

the received signal level as a function of distance, and
thus predict signal levels at different locations.

17

2 – Deterministic models
 The path loss model:

 Log-distance path loss models.

 Both theoretical and measurement-based propagation
models indicate that average received signal power
decrease logarithmically with distance.
 Measurements have shown that the average received
power, for any mobile communication environment,
can be expressed as n
 d 
Prec d   Prec d 0   
 d0 
 where n is the path loss exponent which indicates
the rate at which the signal power decays with
distance for a given environment. Note that n = 2 for
the free space model.
18

9
 2 – Deterministic models
 The path loss model:
 The average large-scale path loss (signal attenuation)
can thus be expressed as
 d 
P L dB d   P L dB d 0   10 n log  
 d0 
 Path loss increases with distance at a rate proportional
to n, or 10ndB/dec (similarly for Prec(d) ).

 Given that we know PL(d) at any location (e.g. by

measurements or by PL equation), and we are given
the transmitted power, we can estimate the received
power as:
Prec , dB d   Pt ,dB  PL dB d 
 Exercise: show that this equation for Prec(d) results in
same expression for Prec(d) shown in the previous slide.
19

2 – Deterministic models
 The path loss model:
 Table 3.1: Path loss for different mobile environments

Environment Path loss exponent “n”

Free space 2
Urban area (cellular radio) 2.7 to 3.5

Shadowed urban cells 3 to 5

In building, line-of-sight 1.6 to 1.8
In building, obstructed path 4 to 6
In factory, obstructed path 2 to 3

20

10
 3 – Statistical path loss
models - shadowing
 The deterministic model expresses the path loss as a
function of the distance, d, only. It does not take into
account the changes of the surrounding environment.
 Statistical models suggest that the path loss at a
particular distance, d, is a random variable (Gaussian in
dB, with mean PLdB(d) ):
 d 
PL dB ( d )  P L dB ( d )  X   P L dB ( d 0 )  10 n log    X 
 d0 

0 mean Gaussian variable

Gaussian variable with P LdB (d ) mean

 Note: Prec(d) in previous slide becomes the average value,
and  is estimated in practice by measurements.
This model is called log-normal shadowing. 21

3 – Statistical path loss

models - shadowing
 Lognormal shadowing (cont’d):
 Xσ is a zero-mean Gaussian distributed random
variable in dB (log-normal) with standard deviation σ,
also in dB.
 In practice, the values of the path loss exponent n and
the path loss variance σ2 (or standard deviation σ) are
computed from measured data, such that the mean
squared error (MSE) between the measured and
estimated path loss is minimized.
 Note: PL(d) is a random variable with Gaussian (normal)
distribution in dB.
 Note: Prec(d) is also a r.v. with Gaussian (normal)
distribution in dB.

22

11
 3 – Statistical path loss
models - shadowing
 Lognormal shadowing (cont’d):
 Let X be a Gaussian r. v., then the statistical distribution
or probability distribution function (pdf) of X is given by

f X x  
1 
exp  
x  
2


2  2 2 
 

f X x 
 x   
2
1
μ = average value of X exp( )
2  2 2

σ = standard deviation

σ2 = variance

x
 23

3 – Statistical path loss

models - shadowing
• Lognormal shadowing (cont’d):
 Thus,
 x  P L ( d ) 2
f PL d   x  1 
 exp   
2   2 2 
 

  y  Prec (d )2 
f Precd   y 
1
exp 



2   2 2 

 Using statistical information, we can determine the

probability that the received signal power Prec (d), or the
path loss PL(d), will exceed (or fall below) a given level

24

12
 3 – Statistical path loss
models - shadowing
• Example:

• The probability that the received signal power (in dB) will
exceed a certain value  1 is given by

Pr Prec ( d )   1   f Prec ( d ) x  dx
1

1   x   2 
 
1 2 
exp  
 2 2 
 dx

• No easy closed-form solution.

• We use Q-function to calculate the answer.

25

3 – Statistical path loss

models - shadowing
• The Q-function

1  x2 
Q( x)   exp    dx
 2  2 
• Properties:
• (1) Q  1   1  Q   1 

• (2) Q 0   0 . 5

• (3) Q     1 , Q   0
  x  1  x 
• (4) Q x   1
2 1  erf     2 erfc  
  2    2

exp  y  dy =Error function,
 exp y  dy =Compl. Error function
x

 
erf ( x )  2 2
erfc( x)  2 2

0 x

26

13
 3 – Statistical path loss
models - shadowing
x Q(x) x Q(x)
• Table of 0.0 0.50000 2.0 0.02275

Q-function: 0.1 0.46017 2.1 0.01786

0.2 0.42074 2.2 0.01390

0.3 0.38209 2.3 0.01072


1  x2  0.4 0.34458 2.4 0.00820
Q( x)   exp   dx
2
0.5 0.30854 2.5 0.00621
  2 0.6 0.27425 2.6 0.00466

0.7 0.24196 2.7 0.00347

• Values of
0.8 0.21186 2.8 0.00256

0.9 0.18406 2.9 0.00187

Q(x), for 1.0 0.15866 3.0 0.00135

0x 7, are 1.1 0.13567 3.1 0.00097

shown in 1.2 0.11507 3.2 0.00069

the table. 1.3 0.09680 3.3 0.00048

1.4 0.08076 3.4 0.00034

1.5 0.06681 3.5 0.00023

1.6 0.05480 3.6 0.00016

1.7 0.04457 3.7 0.00011

1.8 0.03593 3.8 0.00007

1.9 0.02872 3.9 0.00005

27

3 – Statistical path loss

models - shadowing

• The probability that the received signal power (in dB) will
exceed a certain value  1 can be computed interms of the
Q-function by substituting y=(x-µ)/ in slide 25 as

1   y2 
Pr Prec d    1    exp   dy
 
1
2  2 


   
 Q 1 
  

• Similarly, the probability that the received signal power is

below  1 is

1   y2     1     
PrPrec d    1    exp  dy  Q    1  Q 1 
 
 2  2       
1

28

14
 3 – Statistical path loss
models - shadowing
• Outage Probability under shadowing: due to shadowing, the
received power at any given distance d from the BS may fall
below the min usable power required by the MS, Pmin.
• We define the outage probability Pout(Pmin , d) as
 P  Prec d 
Pout ( Pmin , d )  Pr Prec d   Pmin   1  Q  min 
  
 P  ( Pt  10 log10 K 0  10n log10 ( d / d 0 )) 
 1  Q min 
  
where K0 (dB)=20log(/4d0), and Pt(dB)+K(dB)=Prec(d0) dB.
• Example 3.4: Find the outage probability at 150m from a BS
for a channel based on the combined path loss and shadowing
models, assuming transmitted power of Pt=10mW, minimum
power requirement of Pmin=-110.5dBm, K0(dB)=-31.54dB,
d0=1m, σ =3.65, and n=3.71. Ans: Pout = 0.0121
• Note: Pout = 1% is a typical target in wireless system designs.
29

3 – Statistical path loss

models - shadowing
• Example 3.5
• A mobile receiver in a cellular system detects 1 m W
signal at a distance d0=1m from a serving base station.
Assuming hexagonal cell geometry, and cell radius
R=470m in a 7-cell reuse system, compute:
• (1) The average received signal power at the edge of the
cell (given that the path loss exponent for the
environment is n=3).
• (2) The probability that a call at the edge of the cell is not
lost (assuming log-normal shadowing with variance   16), 2

given that the minimal acceptable signal level to ensure a

call is not lost is    108 dB .
• Ans: (1) Prec (d)=10log{ Prec(d0)[R/d0]-n }= -110.17 dB
(2) Pr[Prec(R)>=-108dB]=Q[(-Prec(R))/]=Q(0.5425)
Hence Pr[call not lost] = Pr[Prec(R)>=-108dB]= 0.3

30

15
 4 – Determination of %
coverage area
 Useful Cell coverage area:
 It is clear that due to random
effects of shadowing, some
locations within a coverage area
will be below a particular desired
received signal threshold. BS

 Figure here illustrates the

challenges that shadowing
poses in cellular system design.

 Effective (useful) coverage area

now looks like amoeba-like
shape, and worse still this shape
is constantly changing due to the
random nature of shadowing!.
31

4 – Determination of %
coverage area
 Useful coverage area:
 For a coverage area of radius R from a BS, let  be the
desired useable received signal threshold.
 It can be shown that the % of useful service area (i. e., the
% of area with the ave. received signal   ), given a known
likelihood of coverage at the cell boundary is

 1  2 ab    1  ab   
 1  erf a   exp 
1
U ( )   2  1  erf   
2   b   b  

 Proof: see handout [pages 141-142 in Ref. 1 only shows the

final result].

a     Pt  P L ( d 0 )  10 n log R / d 0  /  2 b  10n log e  /  2

    Prec ( R )  /  2
32

16
 4 – Determination of %
coverage area
 By choosing the signal level such that Prec R    ,
i.e. a =0, we have 1500

  1  
1000

1  1 
U ( )  1  exp  2   1  erf     500

2  b   b  

Distance (m)
0

-500

(entire cell is usefully covered ?) -1000

-1500
-1500 -1000 -500 0 500 1000 1500
Distance (m)

 In terms of the Q-function, we have respectively for

these two cases [Ref. 2, chapter2]:
 2  2 aˆ bˆ   2  aˆ bˆ 
U (  )  Q aˆ   exp  Q 
 

 bˆ 2   bˆ 
1  2   2 
U (  )    exp  2  Q    aˆ     Prec ( R)  /  , bˆ  10 n log e  / 
 2  bˆ   bˆ  
33

4 – Determination of %
coverage area
 U() can be evaluated for a large number of values of
σ and n as shown below
Pr[ Prec r    ]
1
SIGNAL ABOVE THRESHOLD, U()
FRACTION OF TOTAL AREA WITH

0.9
0.85
0.9
0.8
0.75
0.8 0.7
0.65
0.6
0.7
0.55
0.5
0.6

0.5
0 1 2 3 4 5 6 7 8

 n

34

17
 4 – Determination of %
coverage area
 Example 3.6
 If n = 4 and σ = 8dB, and if 75%
boundary coverage is desired (i.e.,
75% of time the signal is to exceed
BS
the threshold, , at the boundary),
then the area coverage is 90% U()
(U()=0.9).
 If n=2 and σ = 8dB (σ/n=4), a 75%
boundary coverage provides 86%
area coverage BS
 If n=3 and σ = 9dB (σ/n=3), then U()
50% boundary coverage provides
71% area coverage
Cell coverage overlap
 See previous slide for chart. at boundary to avoid gaps

35

4 – Determination of %
coverage area
 Example 3.7
 Four received power measurements were taken at distances of
100m, 200m, 1km, and 3km from a transmitter. These
measured values are shown in the table below. It is assumed
that the path loss for these measurements follows the log-
normal shadowing model, where d0 =100m:
 (a) find the minimum mean square error (MMSE) estimate for
the path loss exponent n.
 (b) calculate the standard deviation about the mean value.
 (c) estimate the received power at d=2km using the resulting
model.
 (d) predict the likelihood that the received signal level at 2km
will be greater than -60dBm.
 (e) predict the % of area within a 2km radius cell that receives
signals greater than –60dBm, given the result in part (d).
36

18
 4 – Determination of %
coverage area
 Example 3.7

Distance from Received Power

Transmitter
100 m 0 dBm
200 m -20 dBm
1000 m -35 dBm
3000 m -70 dBm

 Ans: see handout or lecture notes

37

4 – Determination of %
coverage area
 Example 3.8:
 (a)Find the coverage area for a cell with the combined path loss
and shadowing model of Example 3.4, a cell radius of 600m, a BS
transmit power of Pt=100mW=20dBm, and a minimum received
power requirement of Pmin=-110 dBm.
 (b) Repeat part (a) for the case Pmin=-120 dBm.

 Ans:
 (a) Prec ( R )  Prec (d 0 )  10n log(600)  114.6dBm , a   110  114 . 6
3 . 65  1 . 26 , b  37.1( 0.434 )
3.65  4.41 ,
and
 2  2 (1 . 26  4 . 41 )   2  (1 . 26 )( 4 . 41 ) 
U (  )  Q 1 . 26   exp   Q  
 ( 4 . 41 ) 2   4 . 41 
 Thus, U ( )  0 . 59

 (b) U ( )  0 . 988
38

19
 5 – Empirical Outdoor and
Indoor Propagation Models
 The Log-normal Shadowing model gives more accurate path
loss estimation than the Log-distance model as it accounts for
the variations in the received power over different times for the
same Transmitter-receiver (T-R) separation.
 However, such variation will be different for different terrain
types, like Deserts/Forests, etc. This effect is not captured by
the Log-normal shadowing model.
 Several outdoor propagation models have been developed that
account for specific terrain types.
 These are known as Empirical Models as they are empirical
formulas derived from measurements.
 We cover in this chapter the most popular ones such as: the
Okumura model, Hata model, COST-Hata model, and more
recently the SUI model (Stanford University Interim model) for
IEEE 802.16 (used in 4G,Wimax & LTE, cellular planning) [3-8].
 Also, Modified SUI model for Millimeter Wave bands (5G) [9].
39

5 – Empirical Outdoor and

Indoor Propagation Models
 Okumura Model (developed by Okumura in Tokyo, [Ref. 3])
 Using extensive measurements, Okumura developed a set of
curves giving the median (50%) attenuation relative to free
space, in an urban area over a quasi-smooth terrain.
 Useful for BS antenna heights of 30-1000m and MS antenna
height of  3m (>> antenna heights in use today).
 This model can be used to predict signal attenuations for:
 Frequencies in the range 150 MHz to 1920 MHz, but it is
typically extrapolated up to 3000 MHz (covers 2G/3G bands).
 Distances of 1-100 km from BS (typical cell radius is 1.3km).
 This model is widely used for signal prediction in urban areas
(simple and accurate, but no analytical explanation).

40

20
 5 – Empirical Outdoor and
Indoor Propagation Models
 Okumura Model
 The Okumura curves were
developed from measurements
using vertical omni-directional
antennas at both the BS
(htx=200m) and MS (hrx=3m),
and are plotted vs frequency, in
the range 100 MHz to 1920
MHz., and as a function of the
distance from the BS in the
range 1km to 100km.
 The model can be expressed as:

PL50,dB (d )  PLF ,dB (d )  Amu ( f , d )  G(htx )  G(hrx )  GAREA

Okumura, et al, “Field Strength and Its Variability in
PL50=50% (median) value of propagation path loss
VHF and UHF Land Mobile Service,” Riview Electrical
PLF = Free space propagation loss
Communication Laboratory, Sept. 1968 © IEEE.
Amu= median attenuation relative to free space
Insights: upto 700-900MHz – modest attenuations,
G(x) or GX = gain factor due to x
Cellular service above 1GHz – severe attenuations
41

5 – Empirical Outdoor and

Indoor Propagation Models
 To determine path loss using
Okumura’s model, first calculate
the free space path loss for the
Transmitter-Receiver (T-R)
separation of interest, then
estimate the extra attenuation
for signal propagation in urban area
from the Okumura curves, and add
it to the free space loss along with
correction factors to account for
terrain types and antenna heights.
 Okumura found also that G(htx) and
G(hrx) vary as:
htx
G ( htx )  20 log( ) , 1000m  htx  30m Correction factor GAREA for different terrain types
200
hrx h
G (hrx )  10 log( ) , hrx  3m, and G ( hrx )  20 log( rx ) , 3m  hrx  10m
3 3
42

21
 5 – Empirical Outdoor and
Indoor Propagation Models
 Okumura Model
 Example 3.9 (see page 153 in the text).
 Find the median path loss using the Okumura’s model for
d=50km, htx=100m, hrx=10m in a suburban environment. If
the base station transmitter radiates an EIRP of 1kW at a
carrier frequency of 900MHz, find the power at the receiver
(assume a unity gain receiving antenna).
 Ans: PLF=-10log10(2/(4)2 d2)=125.5dB, with =c/f=(1/300).
 From the Okumura curves/ Equations:
Amu(900MHz, (50km))=43dB, GAREA = 9dB, G(htx)=-6dB,
G(hrx)=10.46dB, and thus PL50 (dB)=155.04dB.
 Therefore, the median received power is:
Prec (d)= EIRP(dBm) – PL50(dB) + Grec(dB)
=60dBm -155.04dB +0dB = -95.04dBm
43

5 – Empirical Outdoor and

Indoor Propagation Models
 Hata Model [Ref. 4]:
 The Hata model is an empirical formulation of the graphical
path loss data provided by Okumura (to avoid reading error).
 Hata presented the urban area propagation loss as a standard
formula and supplied correction Equations for application to
other situations.
 The standard formula for median path loss in urban areas is
given by
PL50,dB (urban )  69.55  26.16 log( f c )  13.82 log(htx )  a ( hrx )  ( 44.9  6.55 log htx ) log d

fc = frequency (in MHz) ˜ 150MHz to 1500MHz

htx = BS antenna height (in meters) ˜ 30m to 200m
hrx = MS antenna height (in meters) ˜ 1m to 10m
d = T-R separation (typically 1-20 km)
a(hrx) = correction factor for MS antenna height, which is a
function of the coverage area

44

22
 5 – Empirical Outdoor and
Indoor Propagation Models
 Hata Model:
 For small & medium sized city,
a ( hrx )  [1 .1 log( f c )  0 .7 ]hrx  [1 .56 log( f c )  0 .8] dB

 For a large city,

a ( hrx )  8 .29 [log( 1 .54 hrx )]2  1.1 dB f c  300 MHz
a ( hrx )  3.2[log( 11 .75 hrx )] 2  4 .97 dB f c  300 MHz
 To obtain the path loss in suburban area & rural areas,
the standard Hata formula above is modified respectively as
PL 50 ,dB ( suburban )  PL 50 ( urban )  2[log( f c / 28 )] 2  5 .4
PL50 ,dB ( rural )  PL50 ( urban )  4.78[log( f c )]2  18.33 log( f c )  40.94

 accurate for d >1km (good for large cells in 1G, 2G, 3G,
but not for smaller cell sizes & higher freq used in 4G).
45

5 – Empirical Outdoor and

Indoor Propagation Models
 COST-Hata Model [Ref. 5]:
 Hata model is not good for micro-cell with radius < 1km, or for
cellular systems using frequencies > 1.5GHz (like 3G, 4G, and
some 2G systems).
 The European committee named COST 231 (cooperative for
scientific & technical research 231), developed an extension to
the Hata model for higher frequencies up to 2 GHz as
PL50,dB (urban)  46.3  33.9 log( fc )  13.82log(htx )  a(hrx )  (44.9  6.55loghtx ) logd  CM
where C M  3dB for large urban centres (0 dB for small city / suburban areas)
 fc is now in the range 1500MHz to 2000MHz. Other
parameters remain the same as in the original Hata model.
 This model is known as COST-Hata model and is widely used.
 However, the problem of inaccurate indoor and micro-cell (d<
1km) propagation models remains.
46

23
 5 – Empirical Outdoor and
Indoor Propagation Models
 4G: SUI Model for IEEE802.16 (WiMAX) system [6]
 The expression for calculating the path loss according to
the Stanford University Interim (SUI) model is given by
d
PL  A  10n log10 ( )  X f  X h  s, for d  d 0
d0
 where d is the distance (in meters) between BS and
receiving antenna of the MS, d0 is set to 100m.
 Xf is the correction parameter for the frequencies above
2GHz: X f  6.0 log ( f )
2000
 Xh is the correction parameter for the receiver antenna
height:
hr h
X h  10 .8 log ( ), for terrai n A, B, X h  20 .0 log ( r ) , for terrai n C
2 2
 S is a lognormal distributed factor that is used to account
for the shadow fading owing to trees and other clutters
and has a value between 8.2dB and 10.6dB.
47

5 – Empirical Outdoor and

Indoor Propagation Models
 SUI Model for IEEE 802.16: Model Parameters

 A is given by
Terrain Path loss Area Vegetation
a b c

4 d 0
A  20 log 10 ( )
 Type A Highest Hilly Very dense 4.6 0.0075 12.6

λ = wavelength in meters
Type B Moderate Flat Very dense 4 0.0065 17.1

 n is given by
c
n  a  bh BS  Type B Moderate Hilly Rare 4 0.0065 17.1
h BS

 hBS is the BS ant. height (m) Type C Lowest Flat Rare 3.6 0.005 20

 a b and c depend on the Note: since basic Wimax assumed Backhaul

terrain types (see the table) service, vegetations & hills are emphasized.

48

24
 5 – Empirical Outdoor and
Indoor Propagation Models
 SUI Model for IEEE 802.16 [Ref. 6]:
 SUI model can be used for base station antenna height
from 10m to 80m, the receiving antenna height (hr)
between 2m and 10m and the cell radius between
0.1km and 8km.
 SUI model can be used for path loss predictions in rural,
suburban, and urban areas via the three different terrain
types: Hilly, flat , dense vegetations, etc., in the table.
 SUI model can be used to estimate the path loss for
transmissions in the frequency range 1900MHz to
11GHz.
 Since 3G & 4G typically operate in 2/2.5, 3/3.5, upto
6GHz range, using the SUI model for link budget in these
networks give accurate result [Ref. 7].
 Other 4G: 3GPP Models for LTE system [see Ref. 8]
 Modified SUI for mmWave in 5G: [Ref.9, given in class]
49

5 – Empirical Outdoor and

Indoor Propagation Models
 Indoor Propagation Models:
 In general, indoor channels may be classified as either line-of-
sight (LOS) or obstructed (OBS), with varying degrees of clutters.
Several models have been proposed, but popular ones are
 (1) Log-distance Path Loss Model:
 Indoor path loss has been shown by many researchers to obey
the distance power law (same as slide#20):
d 
PLdB ( d )  PLdB ( d 0 )  10 n log    X 
 d0 
 where n depends on the surroundings/building type, and Xσ is a
normal random variable (in dB) with standard deviation σ.
 See next table for values of n and Xσ measured in different building types.

50

25
 5 – Empirical Outdoor and
Indoor Propagation Models
 Illustrations: Building Frequency (MHz) n  (dB)
Path loss Retail Stores 914 2.2 8.7
exponent, and Grocery Store 914 1.8 5.2
standard
Office, hard partition 1500 3.0 7.0
deviation for
Office, soft partition 900 2.4 9.6
indoor
propagations Office, soft partition 1900 2.6 14.1

(Table 4.6 in Factory LOS

Ref 1). Textile / Chemical 1300 2.0 3.0
Textile / Chemical 4000 2.1 7.0
Paper / Cereals 1300 1.8 6.0
 n ranges Metalworking 1300 1.6 5.8
from 1.6-3.3 Suburban Home
Indoor Street 900 3.0 7.0
  ranges Factory Obstructed
from 3-14dB Textile / Chemical 4000 2.1 9.7
Metalworking 1300 3.3 6.8

51

5 – Empirical Outdoor and

Indoor Propagation Models
 Indoor Attenuation Factor Model:
 In-building, site-specific, propagation effects can be obtained by
adding experimental data for floor and partition losses to an
analytical or empirical path-loss model, PL(d), above, to accurately
estimate the received signal strength for indoor users [Ref. 1,2].
Nf Np
PL dB , indoor ( d )  PL dB , outdoor ( d )   FAF i   PAF i
i 1 i 1

 where FAF and PAF denotes respectively the floor attenuation

factor for the ith floor, and the partition attenuation factor for the ith
partition traversed by the signal.
 Nf and Np are the numbers of floors and partitions traversed by
the signals.

52

26
 5 – Empirical Outdoor and
Indoor Propagation Models
 Illustration 1: Typical Building FAF(dB)  (dB) # Loc.
FAF measurements Office Building 1:
(Table 4.5 in Ref 1). Via One Floor 12.9 7.0 52
Via Two Floors 18.7 2.8 9
 Generally speaking,
Via Three Floors 24.4 1.7 9
attenuation per floor is
Via Four Floors 27.0 1.5 9
greater at the first floor,
than the incremental Office Building 2:
attenuation caused by Via One Floor 16.2 2.9 21
each additional floor. Via Two Floors 27.5 5.4 21
 After five or six floor, Via Three Floors 31.6 7.2 21
only little extra FAF will
be experienced.
Partition type Partition Losses
 Illustration 2: Typical Cloth partition 1.4
Double plasterboard wall 3.4
partition losses for Foil insulation 3.9
different materials Concrete wall 13
(Table 2.1 in Ref 2). Aluminum siding 20.4
All metal 26
53

References
1. T.S. Rappaport, “Wireless Communications: principles and practice,” Second Ed.,
Prentice HallPTR, Upper Saddle River, NJ 07458, USA.
2. A. Goldsmith, “Wireless Communications,” Cambridge University Press, New York, NY
10013-2473, USA.
3. Okumura, et. al., “Field Strength and Its Variability in VHF and UHF Land Mobile
Service,” Review Electrical Communication Laboratory, Sept. 1968.
4. Masaharu Hata, “Empirical Formula for Propagation Loss in Land Mobile Radio
Services,” IEEE Trans. Veh. Technol, vol. VT-29, No. 3, pp.317-325, Aug. 1980.
5. COST Action 231, “Digital mobile radio towards future generation systems, final Report,”
Tech. Rep., European Communities, EUR 18957, 1999.
6. V. Erceg et. Al., “Channel models for fixed wireless application,” Tech. Rep., IEEE 802.16
Broadband Wireless Access working group, Jan 2001.
7. V. S. Abhyawardhana, et. Al., “Comparison of empirical propagation path loss models for
fixed wireless access systems, Proc., IEEE Veh. Technol. Conf., pp. 73-77, 2005.
8. http://www.3gpp.org/LTE
9. A. I Sulyman, A. Nassar, M. K. Samimi, G. R. MacCartney, T. S. Rappaport, A. Alsanie,
“Radio Propagation Path Loss Models for 5G Cellular Networks in the 28GHz and 38GHz
Millimeter-Wave Bands, IEEE Commun. Mag., Sept. 2014

54

27
 Micro-Projects

Topic proposal due: just after Midterm#1

Final Report / Presentations: Week # 14
S/N Student Name ID Micro project Topics
1
2
3
4
5
6
7
8

55

56

28