Winter 2023 ECSE-610 1

2. Path Loss and

Shadowing

Prof. Benoit Champagne

Dept. of Electrical & Computer Engineering

McGill University, Montreal, Canada

January 10, 2023

Winter 2023 ECSE-610 2

Introduction
▪ The wireless radio channel poses a severe challenge as a
medium for reliable high-speed communications.
▪ In this chapter we characterize the variations in RX signal
power over distance due to path loss and shadowing:
▪ Path loss is caused by dissipation of the power radiated by the TX
as well as by effects of the propagation channel, and occurs over
long distances (100-1000m)
▪ Shadowing is caused by obstacles between the TX and RX that
further attenuate signal power and occurs over shorter distances
(10–100 m for outdoor, less for indoor).
▪ These variations are referred to as large-scale effects.
▪ Chapter 3 deals with RX power variations due to constructive
and destructive addition of multipath signal components:
▪ These variations occur over very short distances, on the order of
the wavelength, and are referred to as small-scale effects.
Winter 2023 ECSE-610 3

Received power versus distance

Winter 2023 ECSE-610 4

Overview
2.1 Radio Wave Propagation
2.2 Transmit and Receive Signal Models
2.3 Free-space Path Loss
2.4 Ray tracing
2.5 Empirical Path-loss Models
2.6 Simplified Path-loss Model
2.7 Shadow Fading
2.8 Combined Path Loss and Shadowing
2.9 Outage Probability
Winter 2023 ECSE-610 5

2.1 Radio Wave Propagation

Winter 2023 ECSE-610 6

Goal and approaches

▪ In the context of wireless communications, the main goal of
radio propagation studies is to make prediction about the
received signal at a given location in a given environment.
▪ To this end, information is required about:
▪ Properties of TX signal (bandwidth, power).
▪ Locations of TX and RX.
▪ Propagating environment (geometry, EM properties, etc.).
▪ Predictions about the RX signals can be of a different nature:
▪ Total power or power spectral density (if wideband).
▪ Signal amplitude resulting from combination of multipaths.
▪ Two different approaches exist to obtain this information:
▪ Solution of Maxwell’s equations.
▪ Simplified ray tracing.
Winter 2023 ECSE-610 7

Maxwell’s equations
▪ These equations can be used to describe how electric
and magnetic fields are generated in a given environment
by charges or current distributions (antennas):
Winter 2023 ECSE-610 8

Maxwell’s equations (cont’d)

▪ In theory the solution of these equations along with proper
boundary conditions can provide a precise description of
the EM field at any point in space.
▪ In order to obtain accurate predictions, very precise
information is required about:
▪ The TX source signal (antenna, signal excitation, etc.)
▪ The transmission environment (geometry, dielectric/magnetic
properties, etc.)
▪ In radio communications applications:
▪ It is difficult to obtain this information with sufficient precision.
▪ The solution of the equations can only be approximated
using computer intensive numerical approaches.
Winter 2023 ECSE-610 9

Ray-tracing techniques
▪ These techniques approximate the propagation of EM waves
by representing the wavefronts as simple particles.
▪ The model determines the reflection and refraction effects on
the wavefront while more complex scattering and diffraction
phenomena are accounted for by simplified models.
▪ Ray tracing can be used to calculate both:
▪ Signal amplitude;
▪ Power profile.
▪ Its application requires:
▪ Sufficiently precise description of the propagation environments in
terms of geometry and dielectric properties of reflective surfaces.
▪ Extensive computer-based calculation for accurate prediction in
practical (i.e., complex) environments.
Winter 2023 ECSE-610 10

Main propagation mechanisms

▪ Ray tracing can take into account the following
transmission phenomena:
▪ Free space transmission: Straight line propagation of signal
between the TX and RX in the absence of obstructions.
▪ Reflection: Change of direction of a wavefront bouncing at an
interface between two media with different EM properties.
▪ Refraction: Change of direction of a wavefront as traversing
the interface between two media with different EM properties.
▪ Diffraction: Bending of a wavefront around the corner of an
obstacle into a region of geometrical shadow.
▪ Scattering: Diffuse reflection of a wavefront at it passes
through a clutter of non-uniform obstacles.
Winter 2023 ECSE-610 11

Appendix A: Representation of
Bandpass Signals and Channels

Material developed in additional notes

Winter 2023 ECSE-610 12

2.2 Transmit and Receive

Signal Models
Winter 2023 ECSE-610 13

Need for bandpass model

▪ Our models are developed mainly for signals in the UHF
and SHF bands:
▪ UHF: Ultra-high frequency band, from 0.3 to 3 GHz
▪ SHF: Super-high frequency band from 3 to 30 GHz.
▪ This range of frequencies is desirable for wireless system
operation because of its propagation characteristics and
relatively small required antenna size (size ∝ 𝜆).
▪ Real modulated and demodulated signals are often
represented as the real part of a complex signal in order
to facilitate analysis.
▪ This model gives rise to the equivalent lowpass
representation of bandpass signals, which we use for our
transmitted and received signals (Appendix A).
Winter 2023 ECSE-610 14

Transmitted signal
▪ Equivalent complex low-pass (analytic) representation:
𝑠 𝑡 = Re 𝑢 𝑡 𝑒 𝑗2𝜋𝑓𝑐𝑡
= 𝑠𝐼 𝑡 cos 2𝜋𝑓𝑐 𝑡 − 𝑠𝑄 𝑡 sin(2𝜋𝑓𝑐 𝑡)
= 𝜌 𝑡 cos 2𝜋𝑓𝑐 𝑡 + 𝜃 𝑡
▪ Carrier signal: 𝑒 𝑗2𝜋𝑓𝑐 𝑡
▪ Complex envelope:
𝑢 𝑡 = 𝑠𝐼 𝑡 + 𝑗𝑠𝑄 𝑡 = 𝜌 𝑡 𝑒 𝑗𝜃(𝑡)
𝜌 𝑡 = 𝑢(𝑡) ≥ 0 (amplitude)
𝜃 𝑡 = ∡𝑢 𝑡 ∈ 0,2𝜋 (phase)
▪ Power relation (𝑓𝑐 𝑇 ≫ 1):
1 𝑇/2 1 𝑇/2
𝑃𝑡 = න 𝑠 𝑡 2 𝑑𝑡 =
න 𝑢 𝑡 2 𝑑𝑡
= 12𝑃𝑢
𝑇 −𝑇/2 2𝑇 −𝑇/2
Winter 2023 ECSE-610 15

Signal propagation
▪ According to Maxwell’s equation, the signal 𝑢 𝑡 𝑒 𝑗2𝜋𝑓𝑐 𝑡
emitted by the TX antenna generates wavefronts that
propagate through the air at the speed 𝑐 ≅ 3 × 108 m/s.
▪ For plane wave propagation (no polarization), we have
𝐸 𝒙, 𝑡 = 𝐸0 𝑒 𝑗(𝜔𝑐 𝑡−𝒌⋅𝒙)
where
▪ 𝒙: position vector in space
2𝜋
▪𝒌= 𝒆: wave vector 𝒇𝒄 𝝀
𝜆
100MHz 3m
▪ 𝒆: unit direction vector
𝑐 1GHz 30cm
▪𝜆 = : wavelength
𝑓𝑐 10GHz 3cm
▪ 𝜔𝑐 = 2𝜋𝑓𝑐 : angular frequency
Winter 2023 ECSE-610 16

Received signal
▪ Practical channels ℋ are linear while the noise is additive:

▪ If the RX and TX are fixed and other properties of the

transmission environments remain constant (stationary),
the channel behaves as an LTI system and therefore:
𝑟 𝑡 = ℎ 𝑡 ∗ 𝑠(𝑡) + 𝑛(𝑡)
where
▪ ℎ(𝑡): channel impulse response
▪ 𝑛(𝑡): additive noise (thermal noise, interfering signals, etc.)
Winter 2023 ECSE-610 17

Received signal (cont’d)

▪ For (real) bandpass input signal 𝑠 𝑡 with complex
envelope 𝑢(𝑡), we can write:
𝑟 𝑡 = 𝑅𝑒 𝑣(𝑡)𝑒 𝑗2𝜋𝑓𝑐𝑡 + 𝑛(𝑡)
𝑣 𝑡 = ℎ𝑙 𝑡 ∗ 𝑢 𝑡
where
▪ ℎ𝑙 (𝑡): equivalent low-pass channel impulse response.
▪ In practice, due to changes in the propagation
environment, the radio channel is time-varying.
▪ Motions of TX, RX or other objects induce small changes in
the carrier frequency, also known as Doppler effect.
▪ This requires the use of a TV impulse response:
∞

𝑣 𝑡 = න ℎ𝑙 𝜏; 𝑡 𝑢 𝑡 − 𝜏 𝑑𝜏
0
Winter 2023 ECSE-610 18

Doppler effect
▪ For simplicity, we make the following assumptions:
▪ The TX transmits pure carrier signal 𝑒 𝑗𝜔𝑐 𝑡 , 𝜔𝑐 = 2𝜋𝑓𝑐 .
▪ The TX is located in the far-field of the RX, so that the
plane-wave narrowband model is valid:
𝐸 𝒙, 𝑡 = 𝐸0 𝑒 𝑗(𝜔𝑐 𝑡−𝒌⋅𝒙) (1)
where
▪ 𝒙 is the position vector of the RX.
▪ The RX is moving at constant speed along a straight line,
𝒙 ≡ 𝒙 𝑡 = 𝒙0 + 𝒗𝑡 (2)
where
▪ 𝒙0 is the initial position of TX.
▪ 𝒗 is the velocity vector of TX.
Winter 2023 ECSE-610 19

Doppler effect (cont’d)

▪ Upon substitution of (2) in (1), the electric field at the RX
can be expressed as
′ 𝑗𝜔𝑐′ 𝑡
𝑟 𝑡 = 𝐸 𝒙 𝑡 , 𝑡 = 𝐸0 𝑒
where we find:
▪ 𝐸0′ = 𝐸0 𝑒 −𝑗𝒌⋅𝒙0
▪ 𝜔𝑐′ = 2𝜋𝑓𝐷′ = 2𝜋 𝑓𝑐 + 𝑓𝐷
▪ The frequency shift, also called Doppler shift, is given by
1
𝑓𝐷 = −2𝜋 𝒌⋅𝒗
▪ Denoting by 𝜃 the angle between vectors 𝒌 and 𝒗, we can
equivalently write:
𝑓𝐷 = −𝜆𝑣 cos 𝜃 = −𝑣𝑐𝑓𝑐 cos 𝜃
▪ If source moves towards TX, 𝜋/2 < 𝜃 ≤ 𝜋 and 𝑓𝐷 > 0.
▪ If source moves away from TX, 0 < 𝜃 < 𝜋\2 and 𝑓𝐷 < 0.
Winter 2023 ECSE-610 20

Path loss
▪ Assume no noise and LTI channel, and define:
𝑇/2
▪ TX power: 𝑃𝑡 =
1
𝑇
‫׬‬−𝑇/2 𝑠 𝑡 2 𝑑𝑡 = 12𝑃𝑢
𝑇/2
▪ RX power: 𝑃𝑟 =
1
𝑇
‫׬‬−𝑇/2 𝑟 𝑡 2 𝑑𝑡 = 12𝑃𝑣
▪ We define the (linear) path loss as the ratio:
𝑃𝑡 𝑃𝑢
𝑃𝐿 = =
𝑃𝑟 𝑃𝑣
▪ In decibel scale, we have:
𝑃𝑡
𝑃𝐿 (𝑑𝐵) = 10 log10 = 𝑃𝑡 (𝑑𝐵) − 𝑃𝑟 (𝑑𝐵)
𝑃𝑟
▪ When the reference power for the decibel scale is one milliwatt, we use
the notation dBm instead of dB.
▪ The path gain is defined as the inverse of 𝑃𝐿 :
1
𝑃𝐺 = , 𝑃𝐺 (𝑑𝐵) = −𝑃𝐿 (𝑑𝐵)
𝑃𝐿
Winter 2023 ECSE-610 21

2.3 Free-space path loss

Winter 2023 ECSE-610 22

Free space propagation

▪ Consider the equivalent low-pass representation of the TX
and (noise-free) RX signals as follows:
𝑠 𝑡 = Re 𝑢 𝑡 𝑒 𝑗2𝜋𝑓𝑐𝑡
𝑟 𝑡 = Re 𝑣 𝑡 𝑒 𝑗2𝜋𝑓𝑐𝑡
▪ In free space propagation, also called line-of-sight (LOS)
propagation, we have
𝜆
𝑣 𝑡 = 𝐺𝑙 𝑒 −𝑗2𝜋𝑑Τ𝜆 𝑢(𝑡 − 𝜏)
4𝜋𝑑
(1) (2) (3)
where
▪ 𝑑: distance between the TX and RX.
▪ 𝐺𝑙 = 𝐺𝑡 𝐺𝑟 : product of TX and RX antenna gains.
▪ 𝜏 = 𝑑/𝑐: propagation delay.
Winter 2023 ECSE-610 23

Interpretation
▪ (1) Effect on amplitude:
▪ In free space, signal attenuation decays in 1/𝑑

▪ (2) Effect on carrier:

▪ This term represents the carrier phase shift due to the
propagation delay:
𝑒 −𝑗2𝜋𝑓𝑐 𝜏 = 𝑒 −𝑗2𝜋𝑑Τ𝜆
▪ In practice, 𝑑 ≫ 𝜆, and this term cannot be neglected
▪ (3) Effect on envelope:
▪ This term represents the delay of the complex envelope
▪ If 𝜏 ≪ 1/𝐵, i.e. narrowband case, we have 𝑢(𝑡 − 𝜏) ≅ 𝑢(𝑡)
▪ In this case, the delay has little effect on the envelope and
can be neglected
Winter 2023 ECSE-610 24

Path loss
▪ From the previous equation, we have:
2
1 𝑇/2 𝜆
𝑃𝑣 = න 𝐺𝑙 𝑒 −𝑗2𝜋𝑑Τ𝜆 𝑢(𝑡 − 𝜏) 𝑑𝑡
𝑇 −𝑇/2 4𝜋𝑑
2
𝜆 1 𝑇/2 2 𝑑𝑡 =
𝜆2
𝐺𝑙
= 𝐺𝑙 න 𝑢 𝑡 2
𝑃𝑢
4𝜋𝑑 𝑇 −𝑇/2 4𝜋𝑑
▪ Hence, the path loss is
𝑃𝑢 4𝜋𝑑 2
𝑃𝐿 = = 2
𝑃𝑣 𝜆 𝐺𝑙
▪ The path loss is proportional to the distance squared
(doubling distance incurs a 6dB loss in RX power).
▪ The dependence of the RX power on 1/𝜆2 can be related to
𝐺𝑟 𝜆2
the effective area of the RX antenna: 𝐴𝑒 = .
4𝜋
Winter 2023 ECSE-610 25

2.4 Ray Tracing

Winter 2023 ECSE-610 26

Overview
▪ A radio signal transmitted from a fixed TX will encounter multiple
objects that produce reflected, diffracted, or scattered copies:
▪ These copies, known as multipath signal components, can be
attenuated in power, delayed in time, and shifted in phase and/or
frequency with respect to the LOS path.
▪ The multipath signals are summed (coherently) at the RX, which
produces distortion and power variations in the RX signal.
▪ In ray tracing, we assume a finite number of objects with known
locations, geometry and dielectric properties.
▪ Ray-tracing techniques approximate the propagation of EM waves
by representing the wavefronts as simple particles.
▪ The effects of reflection, diffraction, and scattering on the wavefront
are approximated using simple geometric equations instead of
Maxwell’s more complex wave equations.
Winter 2023 ECSE-610 27

Overview (cont’d)
▪ There exists many computer programs based on ray tracing to
help engineers and researchers with network planning and
simulation model development for in indoor and outdoor, e.g.:
▪ NYUSIM
▪ REMCON
▪ WinProp
▪ DeepMIMO
▪ In these programs, computer graphics are combined with aerial
photographs (outdoor) or architectural drawings (indoor) to
obtain a 3D geometric picture of the environment.
▪ The following sections describe several ray-tracing models of
increasing complexity that can be used to develop a better
understanding of channel propagation:
▪ The two-ray model
▪ The ten-ray model
▪ General model with scattering and diffraction
Winter 2023 ECSE-610 28

2.4.1 Two-ray Model

Winter 2023 ECSE-610 29

Model description
▪ The two-ray model is used when a single ground
reflection dominates the multipath effect.
▪ The received signal consists of two components:
▪ LOS component: TX signal propagating through free space.
▪ Reflected component : TX signal reflected off the ground.
 Winter 2023 ECSE-610 30

Expression of received signal

▪ By liner superposition, we find 𝑟 𝑡 = Re 𝑣 𝑡 𝑒 𝑗2𝜋𝑓𝑐 𝑡 where
𝜆 𝐺0 Τ𝜆 𝐺1
𝑣 𝑡 = { 𝑢(𝑡 − 𝜏0 ) 𝑒 −𝑗2𝜋 𝑙0 +𝑅 𝑢(𝑡 − 𝜏1 ) 𝑒 −𝑗2𝜋𝑙1 Τ𝜆 }
4𝜋 𝑙0 𝑙1
▪ The 1st term (subscript 0) corresponds to the LOS path:
𝑙0 = 𝑙 Length of LOS path
𝐺0 = 𝐺𝑎 𝐺𝑏 Product of antenna gains
𝜏0 = 𝑙 Τ𝑐 Propagation delay

▪ The 2nd term (subscript 1) corresponds to reflected path:

𝒍𝟏 = 𝒙 + 𝒙′ Length of reflected path
𝐺1 = 𝐺𝑐 𝐺𝑑 Product of antenna gains
𝜏1 = (𝑥 + 𝑥 ′ )Τ𝑐 Propagation delay
𝑅 Ground reflection coefficient
Winter 2023 ECSE-610 31

Narrowband assumption
▪ Let Δ𝑡 ≜ 𝜏1 − 𝜏0 denote the difference in travel time
between the LOS and reflected paths.
▪ In effect Δ𝑡 represents the delay spread of the 2-ray channel.

▪ Under the narrowband assumption for multipath transmis-

sion, i.e. Δ𝑡 ≪ 𝐵−1 , we find:
▪ 𝑢 𝑡 − 𝜏1 = 𝑢 𝑡 − 𝜏0 − Δ𝑡 ≈ 𝑢(𝑡 − 𝜏0 )
▪ In the book, they further assume 𝑢 𝑡 − 𝜏0 ≈ 𝑢(𝑡).

▪ Hence, the received signal simplifies to

𝜆 −𝑗2𝜋𝑙Τ𝜆 𝐺0 𝐺1 −𝑗Δ𝜙
𝑣 𝑡 = 𝑒 +𝑅 ′
𝑒 𝑢(𝑡)
4𝜋𝑑 𝑙 𝑥+𝑥
where we define Δ𝜙 = 2𝜋 𝑥 + 𝑥 ′ − 𝑙 /𝜆 , i.e., the phase
difference between the LOS and reflected paths.
Winter 2023 ECSE-610 32

Path loss calculation

▪ Based on above expression relating complex envelopes of TX
and RX signals, we obtain the following general expression for
the path loss:
2 2
𝑃𝑢 𝜆 𝐺0 𝑅 𝐺1 −𝑗Δ𝜙
𝑃𝐿 = = + 𝑒
𝑃𝑣 4𝜋 𝑙 𝑥 + 𝑥′
2𝜋 𝑥+𝑥 ′ −𝑙
where Δ𝜙 = .
𝜆
▪ From Maxwell’s equations with boundary condition (assuming
plane wave with oblique incidence), it can be shown that
sin 𝜃 − 𝑍
𝑅=
sin 𝜃 + 𝑍
where 𝑍 = 𝑓(𝜖𝑟 , 𝜃) and 𝜖𝑟 is the dielectric constant of the ground
(typical value 𝜖𝑟 ≈ 15).
Winter 2023 ECSE-610 33

Asymptotic limit
▪ As the distance 𝑑 between the TX and RX increases,
interesting simplifications occur in above expression.
▪ Assumption 1: The condition 𝑑 ≫ max ℎ𝑡 , ℎ𝑟 implies
▪ 𝜃≈0
▪ 𝑅 ≈ −1
▪ 𝐺0 ≈ 𝐺1 ≡ 𝐺
▪ Δ𝜙 ≈ 4𝜋ℎ𝑡 ℎ𝑟 Τ𝜆𝑑 (follows from Taylor expansion of (2.13))
▪ Under Assumption 1, we find:
2
𝜆 −𝑗Δ𝜙 2
𝑃𝐿 = 𝐺 1− 𝑒
4𝜋𝑑
Winter 2023 ECSE-610 34

Asymptotic limit (cont’d)

▪ For the two-ray model, we define the critical distance
4ℎ𝑡 ℎ𝑟
𝑑𝑐 =
𝜆
▪ Assumption 2: The condition 𝑑 ≫ 𝑑𝑐 implies that
▪ 𝑒 −𝑗Δ𝜙 ≈ 1 − 𝑗𝜋 𝑑𝑐 Τ𝑑 (obtained using Taylor expansion)
▪ In practice, min ℎ𝑡 , ℎ𝑟 > 𝜆, so that Assumption 2 implies
Assumption 1
▪ When both assumptions are satisfied, we find
𝐺ℎ𝑡2 ℎ𝑟2
𝑃𝐿 =
𝑑4
▪ That is, the RX power decays in the 4th power of 𝑑.
Winter 2023 ECSE-610 35

RX power versus distance

▪ Fig. 2.5 shows a plot of 𝑃𝐿 vs 𝑑 for the following parameters:
𝐺0 = 𝐺1 = 1, ℎ𝑡 = 50m, ℎ𝑟 = 2m, 𝑓𝑐 = 900MHz (𝜆 = 33.3cm)
Winter 2023 ECSE-610 36

2.4.2 Ten-ray Model

Winter 2023 ECSE-610 37

Model description
▪ This model assumes a linear street with buildings on both sides
along with TX and RX antennas close to street level:
▪ The building-lined street acts as a dielectric canyon.
▪ In theory, an infinite number of reflected rays can reach the RX.
▪ In practice, signal energy is dissipated with each reflection, and
signal paths corresponding to more than 3 reflections can
generally be ignored.
▪ Experiments show that a model of 10 reflected rays closely
approximates signal propagation through the dielectric canyon:
Winter 2023 ECSE-610 38

Path loss calculation

▪ Received signal:
9
𝜆 𝐺𝑙 −𝑗2𝜋𝑙Τ𝜆
𝑅𝑖 𝐺𝑖
𝑣 𝑡 = 𝑢 𝑡 𝑒 +෍ 𝑢 𝑡 − 𝜏𝑖 𝑒 −𝑗2𝜋𝑥𝑖 Τ𝜆
4𝜋 𝑙 𝑥𝑖
𝑖=1
▪ 𝑙, 𝐺𝑙 : length and antenna gain for LOS ray
▪ Without loss in generality, we set 𝜏0 = 0 for LOS ray.
▪ 𝑥𝑖 , 𝐺𝑖 : length and antenna gain for 𝑖th ray
▪ 𝜏𝑖 = (𝑥𝑖 −𝑙)/𝑐: relative delay for 𝑖th ray
▪ 𝑅𝑖 : product of relection coefficients for 𝑖th ray
▪ Received power in narrowband:
2
2 9
𝜆 𝐺𝑙 𝑅𝑖 𝐺𝑖 −𝑗Δ𝜙
𝑃𝑟 = 𝑃𝑡 +෍ 𝑒 𝑖
4𝜋 𝑙 𝑥𝑖
𝑖=1
▪ Δ𝜙𝑖 = 2𝜋(𝑥𝑖 − 𝑙)/𝜆: relative phase difference for 𝑖th ray
Winter 2023 ECSE-610 39

2.4.3 General Ray Tracing (GRT)

Winter 2023 ECSE-610 40

Introduction
▪ GRT can be used to predict field strength and delay spread for
any building configuration and antenna placement.
▪ The building database (height, location, dielectric properties)
and the TX and RX locations must be specified exactly.
▪ The GRT method uses geometrical optics to trace the
propagation of the various signal components:
▪ LOS and reflected (as before)
▪ Diffracted components
▪ Scattered components
▪ No limit to the number of multipath components:
▪ The strength of each component is derived explicitly based on
geometry and dielectric properties.
▪ In general, the LOS and reflected paths provide the dominant
components, since diffraction and scattering losses are high.
Winter 2023 ECSE-610 41

Diffraction
▪ Diffraction occurs when the TX signal “bends around” an object
in its path to the RX.
▪ Diffraction phenomena can be observed under various
conditions of propagation:
▪ Hilly or irregular terrain.
▪ Presence of building edges.
▪ Obstructions blocking LOS.
▪ Diffraction can be characterized using the geometrical theory of
diffraction (GTD), but this approach is too complex.
▪ Wedge diffraction simplifies the GTD by assuming the
diffracting object is a wedge:
▪ Can be used to characterize diffraction around street corners.
▪ Still requires a numerical solution and is not commonly used.
▪ Diffraction is most commonly modeled by the Fresnel knife-
edge diffraction model because of its simplicity.
Winter 2023 ECSE-610 42

Fresnel knife-edge diffraction model

▪ The geometry of this model is shown below:

▪ The diffracting object is assumed to be asymptotically thin.

▪ Model does not consider conditions such as polarization,
conductivity, and surface roughness
▪ The path loss associated with this model is a function of the so-
called Fresnel-Kirchhoff diffraction parameter:
2(𝑑 + 𝑑 ′ )
𝑣=ℎ
𝜆𝑑𝑑′
▪ However, calculation of EM field at RX location is very complex.
▪ Result cannot be found in closed form in general.
Winter 2023 ECSE-610 43

Fresnel knife-edge model (cont’d)

▪ A common approximation for knife-edge diffraction path loss (in
dB) relative to LOS is given by

▪ Received signal for knife-edge diffraction:

𝜆 −𝑗2𝜋 𝑑+𝑑 ′ Τ𝜆 𝑢(𝑡
𝑣 𝑡 = 𝐿(𝑣) 𝐺𝑑 𝑒 − 𝜏𝑑 )
4𝜋 𝑑 + 𝑑 ′
▪ 𝐺𝑑 : antenna gain factor
▪ 𝜏𝑑 = Δ𝑑/𝑐: delay relative to (blocked) direct path
𝜆𝑣 2
▪ Δ𝑑 = 𝑑 + 𝑑′ −𝑙 ≅ where 𝑙 = length of the direct path.
4
Winter 2023 ECSE-610 44

Scattering
▪ Scattering is a physical process whereby radio waves are
forced to deviate from a straight trajectory due to localized non-
uniformities in the medium through which they pass.

▪ For a given TX and RX location and scatterer properties (size,

shape, composition, dielectric, etc.), precise calculation of the
received signal amplitude or power at the RX is very difficult.
▪ A common and useful approximation is given by the bistatic
radar equation.
Winter 2023 ECSE-610 45

Bistatic radar model

▪ The model assumes that the signal propagates from the
TX to the scatterer based on free-space propagation.
▪ The signal is then re-radiated by the scatterer with
transmit power equal to 4𝜋𝜎/𝜆2 times the received power.
▪ Parameter 𝜎 is the so-called radar cross-section (in m2).
▪ Class development…
Winter 2023 ECSE-610 46

General ray tracing model

▪ The GRT method uses geometrical optics to trace the
propagation of the various signal components:
▪ LOS and reflected (as before)
▪ Diffracted components
▪ Scattered components
▪ The general expression for the complex envelope of the RX
signal takes the following form:
𝑣 𝑡 = 𝑣𝐿𝑂𝑆 𝑡 + ෍ 𝑣𝑅𝐸𝐹,𝑖 𝑡 + ෍ 𝑣𝐷𝐼𝐹,𝑗 𝑡 + ෍ 𝑣𝑆𝐶𝐴,𝑘 (𝑡)
𝑖 𝑗 𝑘
where the expressions for the various terms have already
been derived.
Winter 2023 ECSE-610 47

2.5 Empirical Path-loss Models

Winter 2023 ECSE-610 48

Okumura model
▪ One of the most common models for signal prediction in large
urban macrocells is the Okumura model.
▪ This model is applicable over distances of 1–100 km and
frequency range of 150–1500 MHz.
▪ Okumura used extensive measurements of base station-to-
mobile signal attenuation throughout Tokyo to develop curves
giving median attenuation relative to free space.
▪ The empirical path-loss formula of Okumura at distance 𝑑 and
carrier frequency 𝑓𝑐 is given by
𝑃𝐿 = 𝐿 𝑓𝑐 , 𝑑 + 𝐴 𝑓𝑐 , 𝑑 − 𝐺 ℎ𝑡 − 𝐺 ℎ𝑟 − 𝐺𝑃𝑆
▪ 𝐿(𝑓𝑐 , 𝑑): free-space path loss
▪ 𝐴 𝑓𝑐 , 𝑑 : median attenuation from paper
▪ 𝐺 ℎ𝑡 , 𝐺(ℎ𝑟 ), 𝐺𝑃𝑆 : various correction factors
Winter 2023 ECSE-610 49

Hata model
▪ The Hata model is an empirical formulation of the graphical
path-loss data provided by Okumura and is valid over the same
range of frequencies, 150–1500 MHz.
▪ This empirical model simplifies calculation of path loss because
it is a closed-form formula and is not based on empirical curves
for the different parameters.
▪ The standard formula for empirical path loss in urban areas
under the Hata model is
𝑃𝐿 = 𝐴1 + 𝐴2 log10 𝑓𝑐 − 𝐴3 log10 ℎ𝑡 − 𝑎 ℎ𝑟
+(𝐴4 − 𝐴5 log10 ℎ𝑡 ) log10 𝑑
▪ 𝐴1 , … . , 𝐴5 : numerical parameters
▪ 𝑎(ℎ𝑟 ): correction factor for RX antenna height
▪ Related formulas are available for suburban and rural areas.
Winter 2023 ECSE-610 50

Cost 231 extension to Hata model

▪ The Hata model was extended to 2GHz by European
Cooperation for Scientific and Technical research:

▪ 𝑎(ℎ𝑟 ) is the same correction factor as before

▪ 𝐶𝑀 is 0dB for medium sized cities and suburbs and 3dB
for metropolitan areas.
▪ It is restricted to the following range of parameters:
▪ 1.5GHz < 𝑓𝑐 < 2GHz,
▪ 1km < d < 20km
▪ 30m < ℎ𝑡 < 200m, 1m < ℎ𝑟 < 10m,
Winter 2023 ECSE-610 51

Piecewise linear model

▪ A common empirical method for modeling path loss in outdoor
microcells and indoor channels is a piecewise linear model of
dB loss versus log distance.

▪ A special case of interest is the dual-slope model:

Winter 2023 ECSE-610 52

2.6 Simplified Path-loss Model

Winter 2023 ECSE-610 53

Simplified model
▪ The complexity of radio signal propagation makes it difficult to
obtain a single model that characterizes path loss accurately
across a wide range of conditions.
▪ For general system analysis and design, sometimes best to
use a simplified model that only captures the essence of signal
propagation while leaving out complex details.
▪ The following simplified model for path loss as a function of
distance 𝑑 is commonly used for system design:
𝛾
𝑑0
𝑃𝑟 = 𝐾 𝑃𝑡
𝑑
𝑑
𝑃𝑟,𝑑𝐵 = 𝑃𝑡,𝑑𝐵 + 𝐾𝑑𝐵 − 10𝛾 log10 ( )
𝑑0
▪ 𝐾: constant
▪ 𝑑0 : reference distance
▪ 𝛾: path-loss exponent
Winter 2023 ECSE-610 54

Simplified model (cont’d)

▪ Because of scattering in the antenna near field, the model is
generally valid only for distances 𝑑 > 𝑑0 , where
▪ 𝑑0 from 1 to 10m, indoors
▪ 𝑑0 from 10 to 1000m, outdoors
▪ The value of K is often set to the free-space path gain at
distance 𝑑0 , i.e. (assuming 𝐺𝑟 = 𝐺𝑙 = 1):
𝜆
𝐾 𝑑𝐵 = 20 log10
4𝜋𝑑0
▪ Typical path-loss exponents:
Winter 2023 ECSE-610 55

2.7 Shadow Fading

Winter 2023 ECSE-610 56

Introduction
▪ Shadowing refers to additional attenuation of the propagating
signal caused by:
▪ Obstacles or obstructions along signal path.
▪ Changes in reflecting/scattering objects .
▪ Shadowing cannot be predicted accurately due to the object
complexity (location, geometry, EM properties, etc.)
▪ However, changes in signal power as a function of distance
due to shawdowing exhibit some statistical regularity.
▪ Hence, shadowing is typically modeled as a random path loss:
𝑃𝑡
𝜓= >1
𝑃𝑟
▪ Equivalently, in dB scale:
𝜓𝑑𝐵 = 10 log10 𝜓 = 𝑃𝑡,𝑑𝐵 − 𝑃𝑟,𝑑𝐵 > 0
Winter 2023 ECSE-610 57

Gaussian model for 𝝍𝒅𝑩

▪ Measurements show that 𝜓𝑑𝐵 can be modeled by a
2
normal distribution with mean 𝜇𝑑𝐵 and variance 𝜎𝑑𝐵 .
▪ The corresponding probability density function (PDF) is
1 1 2 ,
𝑓𝑑𝐵 𝜓𝑑𝐵 = 2
exp − 2 𝜓 𝑑𝐵 − 𝜇 𝑑𝐵 𝜓𝑑𝐵 ∈ ℝ
2𝜋𝜎𝑑𝐵 2𝜎𝑑𝐵
▪ In this model:
▪ 𝜓𝑑𝐵 fluctuates around a mean 𝜇𝑑𝐵 with variations on the
order of the standard deviation 𝜎𝑑𝐵 .
▪ 𝜇𝑑𝐵 = 𝐸[𝜓𝑑𝐵 ] is called average dB path-loss and represents
the deterministic component of the path loss.
▪ 𝜇𝑑𝐵 can be obtained from another model (e.g., Section 2.6)
▪ For outdoor channels, empirical studies report values of 𝜎𝑑𝐵
ranging from 4 to 13 dB.
Winter 2023 ECSE-610 58

Motivation for the Gaussian model

▪ For a plane wave signal propagating through an object
with uniform dielectric properties, the signal attenuation is
given by 𝑒 −𝛼𝑑 , where
▪ 𝛼 is an attenuation factor for the medium.
▪ 𝑑 is the length of the propagation path in the medium.
▪ Suppose that shadowing is caused by propagation
through a sequence of 𝑁 objects with (𝑖 = 1, … , 𝑁):
▪ Amplitude attenuation factors 𝛼𝑖 ;
▪ Transmission path lengths 𝑑𝑖 .
▪ The resulting total attenuation is given by
𝑛
σ𝑁
𝐴=ෑ 𝑒 −𝛼𝑖 𝑑𝑖 = 𝑒 𝑖=1 𝛼𝑖 𝑑𝑖
−
𝑖=1
Winter 2023 ECSE-610 59

Motivation for Gaussian model (cont’d)

▪ Hence, according to this model, the corresponding path
loss in dB is given by
𝑁
σ𝑁
𝑖=1 𝛼𝑖 𝑑𝑖
𝜓𝑑𝐵 = 20 log10 (𝑒 ) = 𝑐 ෍ 𝛼𝑖 𝑑𝑖
𝑖=0
where 𝑐 > 0 is a numerical constant.
▪ Let us model the attenuations 𝛼𝑖 and path lengths 𝑑𝑖 as
independent random variables (RV), and assume that the
number 𝑁 of objects is large.
▪ Then, it follows from the central limit theorem that the
distribution of RV 𝜓𝑑𝐵 is approximately Gaussian, i.e.:
Winter 2023 ECSE-610 60

Log-normal model for 𝝍

▪ Since 𝜓𝑑𝐵 ~𝒩(𝜇𝑑𝐵 , 𝜎𝑑𝐵 ), the PDF of the linear path loss,
i.e. 𝜓 = 10𝜓𝑑𝐵 Τ10 can be obtained from the transfomation
theorem of probability.
▪ Specifically, we find (see notes)
𝑑𝜓𝑑𝐵
𝑓𝑙𝑖𝑛 𝜓 = 𝑓𝑑𝐵 𝜓𝑑𝐵
𝑑𝜓
𝜉 1 2
= exp − 2 10 log10 𝜓 − 𝜇𝑑𝐵
2𝜋𝜎𝑑𝐵 𝜓 2𝜎𝑑𝐵
where 𝜉 = 10Τln 10 and 𝜓 > 0.
▪ We say that 𝜓 follows a log-normal distribution.
Winter 2023 ECSE-610X 61

Decorrelation distance
▪ Besides the PDF, several studies have investigated the
autocorrelation function of 𝜓𝑑𝐵 versus separation 𝛿
between two points:
𝑟𝜓𝑑𝐵 𝛿 ≜ 𝐸[(𝜓𝑑𝐵 𝑑 − 𝜇𝑑𝐵 )(𝜓𝑑𝐵 𝑑 + 𝛿 − 𝜇𝑑𝐵 )
▪ The most common model for this autocorrelation function
is based on a 1st order autoregressive process for 𝜓𝑑𝐵 (𝑑)
▪ Specifically, this model yields an exponentially decaying
correlation function, i,e.:
2 −|𝛿|Τ𝑋𝑐
𝑟𝜓𝑑𝐵 𝛿 = 𝜎𝑑𝐵 𝑒
where 𝑋𝑐 is called decorrelation distance.
▪ Indoor: 𝑋𝑐 ~ 1 to 10 m
▪ Outdoor: 𝑋𝑐 ~ 10 to 100 m
Winter 2023 ECSE-610 62

2.8 Path Loss and Shadowing

Winter 2023 ECSE-610 63

Compound model
▪ Models for path loss and shadowing can be super-
imposed to capture power falloff versus distance along
with the random attenuation from shadowing.
▪ In this combined model:
▪ The average dB path loss 𝜇𝜓𝑑𝐵 is given by a simplified path-
loss model such as the one from Section 2.6.
▪ Log-normal shadow fading (with a mean of 0dB), creates
variations about this path loss.
▪ Resultant model (linear scale):
𝛾
𝑑0 1
𝑃𝑟 = 𝑃𝑡 𝐾 ⋅
𝑑 𝜓
Winter 2023 ECSE-610 64

Compound model (cont’d)

▪ Corresponding model on dB scale:
𝑑
𝑃𝑟,𝑑𝐵 = 𝑃𝑡,𝑑𝐵 + 10 log10 𝐾 − 10𝛾 log10 − 𝜓𝑑𝐵
𝑑0
2
𝜓𝑑𝐵 = 10 log10 𝜓 ~𝒩 0, 𝜎𝑑𝐵
▪ Graphical representation:
Winter 2023 ECSE-610 65

2.9 Outage probability

Winter 2023 ECSE-610 66

Definition of outage
▪ In wireless systems there is typically a target minimum
received power level 𝑃𝑚𝑖𝑛 below which performance
becomes unacceptable (e.g., voice quality too poor)
▪ However, with shadowing the RX power 𝑃𝑟 (𝑑) at any
distance 𝑑 from the TX is log-normally distributed, with
some non-zero probability of falling below 𝑃𝑚𝑖𝑛 .
▪ We define outage probability 𝑃𝑜𝑢𝑡 (𝑃𝑚𝑖𝑛 , 𝑑) under path loss
and shadowing to be the probability that the RX power
𝑃𝑟 (𝑑) falls below 𝑃𝑚𝑖𝑛 :
𝑃𝑜𝑢𝑡 (𝑃𝑚𝑖𝑛 , 𝑑) = Pr[𝑃𝑟 𝑑 < 𝑃𝑚𝑖𝑛 ]
Winter 2023 ECSE-610 67

Calculation of outage probability

▪ For combined path loss and shadowing,
𝑑
𝑃𝑟,𝑑𝐵 = 𝑃𝑡,𝑑𝐵 + 𝐾𝑑𝐵 − 10𝛾 log10 − 𝜓𝑑𝐵
𝑑0
▪ Hence, the condition 𝑃𝑟 < 𝑃𝑚𝑖𝑛 for outage is equivalent to
𝜓𝑑𝐵
𝑍≜ > 𝑍0 (𝑑)
𝜎𝑑𝐵
𝑑
𝑍0 (𝑑) = (𝑃𝑡,𝑑𝐵 + 10𝐾𝑑𝐵 − 10𝛾 log10 − 𝑃𝑚𝑖𝑛,𝑑𝐵 )/𝜎𝑑𝐵
𝑑0
▪ For log-normal shadowing, 𝑍~𝒩(0,1) and the outage
probability is computed as
𝑃𝑜𝑢𝑡 𝑃𝑚𝑖𝑛 , 𝑑 = Pr 𝑍 > 𝑍0 𝑑 = 𝑄(𝑍0 𝑑 )
∞
1 − 𝑦 2 Τ2
𝑄 𝑥 = න 𝑒 𝑑𝑦
2𝜋 𝑥
Winter 2023 ECSE-610 68

2.10 Cell Coverage Area

Winter 2023 ECSE-610 69

Cell coverage area

▪ area in a cellular system is defined as the expecteThe cell
coverage d percentage of locations within a cell where the
RX power is above a given minimum 𝑃𝑚𝑖𝑛 .
▪ In practice, the TX power adjusted for an average RX
power 𝑃ത𝑟 at the cell boundary (at a distance 𝑅).
▪ However, shadowing will cause some locations within the
cell to have RX power below 𝑃ത𝑟 .
▪ In practice, a fraction of
the locations will even
experience an RX power
below the preset
minimum 𝑃𝑚𝑖𝑛 .
Winter 2023 ECSE-610 70

Mathmatical formulation
▪ The cell coverage area can be formulated as
1
𝐶=𝐸 2
ඵ Ind 𝑃𝑟 𝑟 > 𝑃𝑚𝑖𝑛 𝑟𝑑𝑟𝑑𝜃
𝜋𝑅 𝑐𝑒𝑙𝑙
where
▪ 𝐸{} denotes mathematical expectation.
▪ Ind(. ) is the indicator function.
▪ Using the previously derived expression of the outage
probability in Section 2.10, we find that
2 − 2𝑎𝑏 2 − 𝑎𝑏
𝐶 = 𝑄 𝑎 + exp 2
𝑄( )
𝑏 𝑏
𝑃𝑚𝑖𝑛 − 𝑃ത𝑟 (𝑅) 10𝛾 log10 𝑒
𝑎= , 𝑏=
𝜎𝑑𝐵 𝜎𝑑𝐵
Winter 2023 ECSE-610 71

Selected sources
▪ A. F. Molish, Wireless Communications, 2nd Ed., IEEE Wiley, 2011.
▪ A. Goldsmith, Wireless Communications, Cambridge, 2005.