Sunteți pe pagina 1din 440

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/321225594

Radio Wave Propagation Fundamentals

Book · November 2017

CITATIONS READS
10 2,874

1 author:

Artem Saakian
Naval Air Systems Command, US Navy, Pax River, MD
5 PUBLICATIONS   12 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Radio Wave Propagation View project

Antennas Analysis and Design View project

All content following this page was uploaded by Artem Saakian on 22 November 2017.

The user has requested enhancement of the downloaded file.


A.S. SAAKIAN

RADIO WAVE PROPAGATION


FUNDAMENTALS

ARTECH HOUSE - 2011


To my wife Arous, and my sons, David and Mark
RADIO WAVE PROPAGATION FUNDAMENTALS
A.S.Saakian
Table of Contents

Preface
Chapter 1. Introduction

1.1. Historical overview


1.2. Classification of radio waves by frequency bands
1.3. The earth’s atmosphere and structure
1.4. Classification of radio waves by propagation mechanisms
1.5. Interferences in RF transmission links
References
Problems

Chapter 2. Basics of electromagnetic waves theory

2.1. Electromagnetic process


2.1.1. Maxwell’s equations of electrodynamics
2.1.2. Boundary conditions of electrodynamics
2.1.3. Time-harmonic electromagnetic process. Classification of media by
conductivity
2.2. Free propagation of uniform plane radio waves
2.2.1. Uniform plane wave in lossless medium
2.2.2. Uniform plane wave in lossy medium
2.2.2.a. Low-loss dielectric medium
2.2.2.b. High-loss conducting medium
2.3. Polarization of the radio waves
2.4. Reflection and refraction of plane radio wave from the boundary of two media
2.4.1. Normal incidence on a plane boundary of two media
2.4.2. Oblique incidence of vertically polarized radio wave
2.4.2.a. Radio wave incident from sparse medium onto the border with
dense medium

i
2.4.2.b. Radio wave incident from dense medium onto the border with
sparse medium
2.4.3. Oblique incidence of horizontally polarized radio wave
2.4.3.a. Radio wave incident from sparse medium onto the border with
dense medium
2.4.3.b. Radio wave incident from dense medium onto the border with
sparse medium
2.4.4. Reflection of the radio wave with arbitrary polarization
2.4.5. Power reflection and transmission
2.4.6. Reflection of the radio wave from the boundary of
non-ideal dielectric medium
2.5. Radiation from infinitesimal electric current source. Spherical waves
2.6. Spatial area significant for radio waves propagation
2.6.1. Principle of Huygens-Kirchhoff
2.6.2. Fresnel zones
2.6.3. Knife-edge diffraction
2.6.4. Practical applications of the Fresnel zones concept
2.6.4.a. Ring-shaped antenna directors
2.6.4.b. Ring-segment diffractors as passive repeaters for radio-relay links
2.6.4.b. Effective area of the radio wave reflection from the flat boundary
References
Problems
Appendix-1
Appendix-2
Appendix-3
Appendix-4

Chapter 3. Basics of antennas for RF radio links

3.1. Basic parameters of antennas


3.1.1. Radiation pattern and directivity
3.1.2. Radiation resistance and loss resistance. Antenna gain and efficiency factor
3.1.3. Antenna effective length and effective area of the aperture
3.2. General relations in radio wave propagation theory

ii
References
Problems

Chapter 4. Impact of the earth surface on propagation of ground waves


4.1. Propagation between antennas elevated above the earth surface. Ray trace approach
4.1.1. Flat earth approximation case study
4.1.2. Propagation over the spherical earth surface
4.1.2.1. Evaluation of the distance to reflection point
4.1.2.2. Divergence of energy of the radio wave while reflecting from the
convex earth surface
4.1.3. Specifics of the propagation over the rough and hilly terrains
4.1.4. Optimal path clearance and choice of the antennas elevations
4.1.5. Propagation prediction models in urban, suburban and rural areas
4.1.5.1. Empirical models
4.1.5.1a. The Okumura-Hata model
4.1.5.1b. Other empirical models
4.1.5.2. Physical models
4.1.5.2a. Non-LOS (Line-Of-Sight) paths
4.1.5.2b. LOS paths
4.2. Propagation between ground-based antennas over the flat earth
4.2.1. Antennas over the infinite, perfect ground plane
4.2.2. Leontovich approximate boundary conditions and structure of radio waves
near the earth’s surface
4.2.3. Propagation over the real homogeneous flat earth
4.2.4. Propagation along the real inhomogeneous flat earth. Coastal refraction
4.3. Asymptotic diffraction theory of propagation over the spherical earth surface
4.3.1. Basic concepts
4.3.2. Propagation between ground-based antennas
4.3.3. Propagation between elevated antennas
4.3.4. Specifics of propagation estimates in penumbra zone
References
Problems
Appendix-5

iii
Appendix-6

Chapter 5. Atmospheric effects in radio waves propagation

5.1. Dielectric permittivity and conductivity of the ionized gas


5.2. Regular refraction of the radio waves in atmosphere
5.3. Standard atmosphere and tropospheric refraction
5.4. Reflection and refraction of the sky waves in ionosphere
5.5. The impact of earth’s magnetic field on propagation of the radio waves in ionosphere
5.5.1. Longitudinal propagation of the radio wave
5.5.2. Transverse propagation of the radio wave
5.5.3. Propagation of the radio wave arbitrary oriented relative to the earth’s
magnetic field
5.5.4. Reflection and refraction of radio waves in magneto-active ionosphere
5.6. Over-the-horizon propagation of the radio waves by tropospheric scatterings
mechanism. Secondary tropospheric radio links
5.6.1. Analytical approaches in description of the random tropospheric scatterings
5.6.2. Physical interpretation of tropospheric scatterings
5.6.3. Effective scattering cross-section of the turbulent troposphere
5.6.4. Statistical models of tropospheric turbulences
5.6.4.1. Gaussian model
5.6.4.2. Kolmogorov-Obukhov model
5.6.5. Propagation factor on secondary tropospheric radio links
5.6.6. The specifics of the secondary tropospheric radio links performance
5.6.6.1. Antennas gain effect on link performance
5.6.6.2. Signal level fluctuations at the receiving point (fading)
5.6.6.3. Limitations to signal transmission bandwidth
5.7. Attenuation of the radio waves in atmosphere
5.7.1. Attenuations in troposphere
5.7.2. Attenuations in ionosphere
References
Problems
Appendix-7

iv
Chapter 6. Receiving of the radio waves: Basic outlines

6.1. Multiplicative interferences (signal fades)


6.1.1. Fluctuation processes and stability of radio links
6.1.2. Fast fading statistical distributions
6.1.2a. Two-Ray random interference
6.1.2b. Random interference of the large number of independent wavelets
6.1.2c. Further generalization of the fast fading statistics
6.1.3. Slow fading statistical distribution
6.1.3a. Normal (Gaussian) distribution of the random variable
6.1.3b. Lognormal distribution of the random variable
6.1.4. Combined distribution of fast and slow fades. Signal stability in long-term
observations
6.2. Additive interferences (noises)
6.2.1. Internal noises of one-, and two-port networks. Noise figure
6.2.2. Noise figure and noise temperature of the cascaded two-port networks
6.2.3. Noise figure of the passive two-port networks
6.2.4. Antenna noise temperature
6.2.5. Receiver sensitivity and signal threshold definition
6.2.6. Environmental (external) noise
6.2.6a. Atmospheric noise
6.2.6b. Thermal noise of the earth’s surface
6.5.6c. Cosmic noise
6.3. Methods of improvement of the radio waves reception performance
6.3.1. Noise-suppressing modems in analog CW-systems
6.3.2. Use of spread-spectrum discrete signals
6.3.3. Diversity reception technique
References
Problems
List of Symbols
List of Abbreviations
Index

v
PREFACE

The main goal of this text is to satisfy the growing demand in propagation study materials that
may be used by a proper audience of students and specialists. The core materials of the
proposed text are developed based on lecture notes that have been offered by author for the
graduate students at Patuxent Graduate Center of Florida Institute of Technology. The
materials included into the text are extended beyond the needs of the single-semester course
and may be used for continuous self-study.
Main objective of this text is to support senior level undergraduate and graduate EE-students by
introduction of the basic principles of electromagnetic waves propagation of radio frequencies
(RF) in real conditions relevant, but not limited to communications and radar systems. It is also
to emphasize the primary role of the antenna-to-antenna propagation path in overall
performance of those systems. Some of the practicing engineers who need quick references to
the basics of propagation mechanisms and principles of the engineering estimates and designs
may use this text in their everyday routine. It may be useful not just for the students and
specialists in the area of radar and communication technologies, but also for the students,
scientists, and engineers of the adjacent areas of science and engineering / technology such as
antenna engineering, astrophysics, geomagnetism, aeronomy, etc.

Chapter-1 is an introductory chapter, which outlines the definitions and classifications that are
commonly used and are adopted by international organizations such as IEEE and ITU. A brief
survey of the structure of the earth's atmosphere is considered in this chapter in support of
atmospheric propagation phenomena that are covered in chapter-5.

Chapter-2 covers the basics of electromagnetic waves theory with the emphasis on those,
which are specific for the RF propagation, such as polarization of radio waves, their reflections
and transmission at the interface of two mediums, as well as diffraction on the knife-edge
obstacles. As a subject of the special interest the Fresnel's zones are analyzed based on
Huygens-Kirchhoff's principles of electromagnetics. This is to clarify the concept of ray forming,
i.e. the concept of the spatial area actively involved in canalization of energy of the radio waves.
Some engineering applications are presented in this chapter to demonstrate the variety of
application of that concept.

1
Chapter-3 is a brief journey to the basic antenna parameters that is needed to evaluate and
analyze antenna-to-antenna propagation path. The end-chapter section presents the main
relations, such as Friis formula, link budget equation etc., as well as introduces propagation
factor, which is a most important measure of the impact of the real conditions on propagation of
the radio waves. Those main relations are of importance to evaluate the radar equation, as well
as for the radio links power budget analysis in communication systems.

Chapter-4 is devoted to propagation of the ground radio waves, i.e. the waves that propagate in
vicinity of the Earth's surface, being affected by that interface. It appears to be methodically
reasonable not to involve any atmospheric effects into consideration within the scope of this
chapter, in order not to confuse the reader by mixing ground effects and atmospheric effects.
The methods of propagation factor calculations are first considered here for the flat-earth
approximation, based on Leontovich’s boundary conditions with further extension for the cases
of convexity of the Earth’s ground. Propagation features over the inhomogeneous paths are
presented within Mandelshtam’s “take-off-landing” concept that is qualitative, rather that
quantitative: it allows reader better understand the behavior of the waves that propagate along
“mixed” type paths. Some quantitative estimates are given to support the engineering
estimates. The coastal refraction effect is analyzed as a particular case along with numerical
estimates discussed. As a particular case of the propagation in mixed, rough terrains the
propagation in urban, suburban and rural areas is considered in subsection 4.1.5.
Last section of the chapter is concluded by introduction of the asymptotic diffraction theory
(V.A. Fock) – theory of diffraction of radio waves over the spherical Earth’s surface.
Engineering approaches for the practical applications are demonstrated by using sample
examples.

Chapter-5 is dedicated to the effects of the atmosphere on propagation of the radio waves.
Smooth refraction in troposphere and reflections from the ionospheric layers are analyzed in
conjunction with the regular inhomogeneties of the refraction index in those atmospheric
regions. Scattering of the radio waves of UHF and higher frequency bands from the random
variations of the tropospheric refraction index (from air turbulences) are considered here by
using the principles of statistical radio-physics. The results are brought to the level of
engineering applications and design of the over-the-horizon troposcatter communication links.

2
Despite the troposcatter radio-links are not widely used nowadays, though the understanding of
the physical mechanisms of the scatterings in troposphere may become a background for the
understanding of scattering phenomenon in general, so there’s no need to discuss them one-
by-one for other scatterings such as from ionospheric meteor trails, raindrops, etc.
The chapter is concluded by analysis of absorptions in atmosphere. Both, absorptions in
tropospheric gases and hydrometeors, as well as in ionospheric layers are introduced to
support signal attenuation estimates on variety of RF radio links.

Chapter-6, the last chapter, that is devoted to the reception of radio waves, which is of the high
importance for the radar and communications systems design. Two types of interferences,
namely multiplicative (fading) and additive (noise) are analyzed in conjunction with the signal-to-
noise ratio (SNR) and communication stability. Statistical distributions of fast and slow fades are
considered, as well as a combined distribution that is to predict a long-term stability of the
communication system's performance. This analysis results in engineering method for power
margin calculation, which is to insure that the objective of the communication stability is met.
Two components of the additive noise are considered: internal noise (receiver noise) and
external noise (environmental noise) that is received from the surrounding areas. The sensitivity
of receiver is discussed in order to define a threshold of the received signal level. The chapter is
concluded by the section that outlines the basic methods of improvement of the radio waves
reception performance: (1) the use of noise-resistive signals, such as analog FM, (2) the use of
the spread-spectrum signals, (3) the diversity reception technique. The main goal of this
chapter is to build the “bridge” between RF-system structure and the propagation conditions
and mechanisms, i.e. to show a direct “coupling” between them.

The scope of the book includes a wide variety of aspects of radio-physics; therefore the
proposed text may not pretend to cover all details of the subject, but rather to encourage a
creative approach amongst the students. A background in mathematics and electromagnetics
that is required in engineering and physics curriculums is assumed. Some of the unique
mathematical techniques and evaluations are either incorporated within proper chapters, or
presented separately in appendixes. Appendix-1 provides some useful mathematical relations
as well as notations adopted for the complex quantities.

3
The numerous of the calculation examples are to support better understanding of the core
materials of the text. The end chapter problems may become highly supportive for the study
process. The problems solution manual is available for teachers / instructors from Artech House
to support the teaching process. This text may be used in senior elective or entry-level graduate
courses.

The list of symbols contains only generalized notations for the quantities (e.g., f for frequency)
in base or derived units. Subscripts are used within the text to denote specific application of the
particular notation (e.g., f c for critical frequency of the ionospheric plasma in Hz); for the

multiple and fractional quantities within the text the units are indicated in symbols by the proper
subscripts (e.g., f c , MHz for the ionospheric plasma frequency in MHz).

First, second, third, etc. derivatives are notated with prime, double prime and triple prime
respectively. Complex numbers and complex vectors are notated with the dot above the
symbol.

Now, I wish to express my deepest gratitude to my reviewers, especially to Dr. D.K. Barton
whose valuable comments and recommendations helped improve the text significantly. I’d like
to thank Prof. V.E. Arustamyan from State Engineering University of Armenia for his revision
and constructive criticism of very first draft of the manuscript in Armenian, as well as Ms.
Svetlana Avanian for help with typing. I’d also like to express my appreciation to Ms. Catherine
Wood for her help with grammatical and syntax corrections of most of the chapters. I’m also
thankful to my colleague, Mr. Frederick Werrell for his review of chapter-3, as well as to my
students for critical remarks during and after the classroom test-studies. My appreciation is
forwarded to Mr. Norm Chlosta, Director of Patuxent Graduate Center of Florida Institute of
Technology for his support and for providing me the opportunities to test the course with smart
audience of graduate students. Finally I’d like to express my thankfulness to my wife Arous:
without her understanding, patience, and support this work would not be possible.

Artem Saakian
December, 2010

4
Chapter 1 INTRODUCTION

1.1 HISTORICAL OVERVIEW


Utilization of radio waves for communication purposes debuted in 1895, when the
Italian-born American engineer G.M. Marconi and the Russian physicist A.S. Popov
independently introduced the first wireless transmission of telegraph signals through the
earth’s atmosphere. This initial effort was particularly notable, as it was conducted
without the support of the then traditional wire guiding line first used by Marconi in 1844
for his wire telegraphy efforts. Instead, a spark-gap was implemented as a transmitting
source for the electromagnetic radiation, and a coherer was utilized as a reception
device. It is difficult to overestimate the importance of that invention for human society,
and for the advancements it heralded for wireless communications. However, that
invention might not have been possible without prior theoretical hypotheses of the
existence of free propagating electromagnetic waves; which were made in 1864 by the
Scottish mathematician and theoretical physicist J. C. Maxwell.

Maxwell’s greatest merit was a theoretical prediction of the displacement currents


in dielectrics and vacuums, which had generalized the concept of current continuity in
Ampere's law [1]. The fundamental equations of electromagnetism (known as Maxwell's
equations) were later updated to achieve complete and symmetric form by the
introduction of magnetic currents. The introduction of displacement electric and magnetic
currents in dielectrics, and in free space, made possible the comprehension of the
nature of electromagnetic waves capable of propagating for long distances independent
of any physical guidance such as wires, waveguides, etc. Those electromagnetic waves
are identified as radio waves, when relative to scientific and commercial applications.
Experimental verification of the existence of electromagnetic waves was achieved by H.
Hertz in the 1880's, when he demonstrated a "propagation" of the spark from a
transmitting Leyden jar to the terminals of remote receiving antenna. The revolutionary
role of Maxwell's equations, which are based on experimental investigations of M.
Faraday and A.M. Ampere, is hard to overemphasize for the immense progress they
have allowed science, engineering, and associated technologies.

5
Soon after Marconi's and Popov's experiments with the transmission of telegraph
signals over the distance of several miles in 1895, Marconi was able to greatly extend
the range of propagation from the UK to Canada, over the Atlantic Ocean, in December
1901. This accomplishment was made possible by the use of a sinusoidal carrier, a
resonant LC-filter at the receiver's input, and a vertical grounded radiator (antenna),
which thereby demonstrated the advantage of vertical polarization for the frequency
ranges of tens and hundreds of kilohertz. The immense success of Marconi’s long-range
signal transmission motivated engineers and research scientists to find a reasonable
explanation for existence of propagation mechanisms [2].
Ionospheric propagation is a mechanism that was initially assumed to be overall
predominant. It was introduced in 1902 after A.E. Kennelly, in the U.S., and O.Heaviside,
in the U.K., postulated the existence of the ionized region in the upper atmosphere,
which seemed to reflect the radio waves and thereby support their long range
propagation.
The second mechanism of propagation was based on the assumption of the
existence of surface waves, which may occur only at the interface of two mediums, such
as the boundary of atmosphere and earth's ground. A detailed theoretical analysis of
surface propagation waves was conducted by the German physicist A. Sommerfeld in
1909, an analysis that was based on Maxwell's equations. A specific approach for the
solution near the flat boundary of two ideal mediums was achieved, and was repeated
ten years later by H. Weyl in a more explicit form. It was shown, that two propagating
modes are existent: (1) A regular TEM mode of a spherical phase front that is specific to
free space propagation; and for this mode the field strength is in inverse proportion to
propagation distance; (2) A surface mode of a cylindrical phase front, which is tightly
bound to the interface of two mediums, and which may exist only in the vicinity of that
interface, whose field strength is in inverse proportion to the square root of the
propagation distance. For moderate and large distances the second component
becomes predominant. In the late 1930s the conceptualization of surface waves,
relevant to the real earth's ground constants, was further developed by J. Zenneck and
K.A. Norton.
The third mechanism of propagation was based on the assumption of the
diffraction of radio waves around the earth surface. These assumptions were proposed
by several mathematicians and scientists prior to and during World War I, and were first
presented by G.N. Watson in 1919 in the form of estimates of the field at the receiving

6
point for the ideal conductive earth surface, as an attempt of direct solution of Maxwell's
equations. Later, in 1937, Van-Der-Pol and Bremmer adopted Watson's approach to
demonstrate arbitrary ground constants. Based on those developments, C.R. Burrows
created an engineering approach that allowed representation of the results in a form that
was more convenient for the applications. Directly following World War-II, in 1945-46,
Soviet physicists V.A. Fock and M.A. Leontovich introduced a Maxwell's equations
solution in parabolic form relevant to the diffraction problem. The complete theory was
developed by taking into account the properties of the earth ground in a wide frequency
range. It was broadly used, until quite recently, for the over-the-horizon propagation
analysis (diffracted field analysis) of radio waves associated with frequencies up to tens
of megahertz.
Both surface wave and diffracted wave approaches result in a single conclusion:
The lower a radio wave's frequency, the more favorable the propagation conditions are.
Prior to and after World War-II, the need to further increase the volume of transmitted
information within a single communication link, as well as the need to advance radar
system performance motivated an increase in carrier frequencies. Thereby higher and
higher frequencies were employed for the support of communication and radar systems
designs, along with newly developed antenna radiating systems and advanced Radio
Frequency (RF) components, together providing increased performance and
effectiveness. Radio waves implementing frequencies of hundreds of MHz and up were
considered to be employed. For those higher frequency ranges, propagation analysis is
based most substantially on geometric optical approaches. The first empirical expression
for the intensity of the current induced in a remote receiving antenna for distances within
Line-of-Sight (LOS) was developed by L.W. Austin in 1911, in the U.S. That expression
was fairly close to the formula developed analytically, in the former USSR, by B.A.
Vvedensky in the late 1920s. Vvedensky’s formula was based on a ray-tracing
(geometrical optics) approach. For higher frequency bands the atmospheric effects
became considerable, and had to be taken into account. Thus, after World War II a
significant amount of attention was focused on issues such as attenuation in
atmospheric gases, as well as refraction, reflection, and scatter of radio waves in the
lower and upper atmosphere. One of those effects, namely the tropospheric scatter of
microwaves by atmospheric turbulence, was discovered in the late 1940s and was later
theoretically validated by H.G. Booker and W.E. Gordon in 1950. Some limited military

7
and commercial applications of the tropospheric scattering effect take place currently, in
nominated troposcatter radio-links within the U.S. and several European countries.
A new era of ionospheric propagation research and investigation initiated after
the launch of the first satellite ("Sputnik"), by the USSR in 1957, to the earth's orbit. This
culminated in our current large network of specialized ionospheric ground-based radars,
sounding stations (ionosondes), and satellite systems worldwide which allow obtainment
of a complete set of data for long-term predictions of the status of ionospheric layers.
Those predictions are widely used for the radio links design and for deployment in HF
and higher frequency bands, which are directly affected by the ionosphere.
In modern times, the radio wave propagation theory and applications still remain
as subjects of extremely high interest to the science and engineering technological
world, and are in consistently in further development and expansion via numerous
worldwide programs.

1.2 CLASSIFICATION OF RADIO WAVES BY


FREQUENCY BANDS
Radio Wave is defined by the Institute of Electrical and Electronics Engineers
(IEEE) as “An electromagnetic wave of radio frequency” [3]. Each particular radio
frequency belongs to the radio spectrum, which is a wide range of frequencies from
several Hertz up to 3 THz. The entire radio spectrum is divided into frequency bands,
shown in Table 1.1, that are based on decimal division. This standard classification is
accepted by the International Telecommunications Union (ITU), which is comprised of
189 member-countries.
A selection of applications pertinent to the radio waves used in engineering, science and
technological efforts includes, but is not limited to:
 Wireless communication systems, including satellite communication systems and
wireless local area networks (WLAN)
 Radar systems
 Telemetry, radio-remote control, and radio-navigation

8
Adhering to ITU-R 1 recommendations, the range for each frequency band extends from
0.3x10N to 3x10N Hz 2, where N is a band number given in the first column of Table 1.1.
There are many subdivisions within each band, reliant upon allocations to services and
the inhabitant worldwide regions [4, 5]. Terminology noted in the last two columns of
Table 1.1 is commonly used, but not officially accepted.
A subdivision of Microwave Bands, shown in Table 1.2, is widely used in radar and
satellite applications [6]. Both, the upper and the lower limits of the radio spectrum are
outlined conventionally, and rely entirely on progressions made in science and
technology. For instance, until the mid-1930’s radio-communications designs were
based on technologies that allowed only utilization of frequencies lower than 100 MHz,
as that represented the upper limit of radio frequencies at that time. In the 1930s and
1940s, overarching progress in the design of a new generation of the radar systems
invoked a claim of involving higher frequencies. Invention of multiple new types of
devices, including magnetrons, klystrons, traveling-wave tubes, and other technologies,
allowed the expansion of the upper limit of radio frequencies to approximately 10 GHz
and higher. In the mid 50-s to early 60-s when the new generation of quantum electronic
devices, such as MASERs, were developed by C.H. Townes, (USA), N.G. Basov, and
A.M. Prokhorov (USSR), they were based on achievements in quantum radio-
spectroscopy. Further broadening of the upper limit of radio frequencies became
possible not just by including CMW, MMW and SMMW, but by also instituting coherent
optical waves to develop Lasers. Implementation of these “new” types of devices made
possible the amplification and generation of coherent radiations even in an optical
domain, and thereby traditional radio technologies and principles became applicable to
the optical frequency domain as well. Optical waves that are used for wireless
information transmission and processing are sometimes called optical radio waves [7].
Table 1.3 shows the classification of optical waves by frequency bands, which is
considered as a non-official classification, and is widely used by specialists in different
areas of engineering and science. In tandem with other items contributing to expansion
of the lower limit of the radio spectrum, there are also rationales of expansion based on
the needs of global military communication services and navigation, as well as scientific
research categories such as geophysics, atmosphere science, and radio astronomy.

1
The Radio Communications section of the ITU
2
Conventionally the upper limit is included into the band, and the lower limit is excluded.

9
Table 1.1

Number, N Wave-

Acronym

Acronym
Frequency Band Frequency Descriptive
Band

length
Name by ITU-R Range, Hz Name
in Meters

Extremely Low 3 5
- ELF < 3x10 > 10 - -
Frequency
Very Low 3 4 5
Miriameter
4 VLF (3 to 30)x10 10 to 10 -
Frequency Waves
Low (30 to 3 4
5 LF 3
10 to 10 Kilometer Waves -
Frequency 300)x10
Medium 6 2 3
Hectometer
6 MF (0.3 to 3)x10 10 to 10 -
Frequency Waves
High (3 to 30) x 2
Decameter
7 HF 6
10 to 10 -
Frequency 10 Waves
Very High (30 to
8 VHF 6
1 to 10 Meter Waves MW
Frequency 300)x10
Ultra High 9 -1
Decimeter
9 UHF (0.3 to 3)x10 10 to 1 DMW
Frequency Waves*
Microwaves *

Super High 9 -2 -1
Centimeter
10 SHF (3 to 30)x10 10 to 10 CMW
Frequency Waves*
Extremely High (30 to -3 -2
Millimeter Waves
11 EHF 9
10 to 10 MMW
Frequency 300)x10 *
(0.3 to -4 -3
Sub-Millimeter
12 - - 12
10 to 10 SMMW
3)x10 Waves *

Table 1.2

Band Name L S C X Ku

Frequency Range, GHz 1-2 2-4 4-8 8 – 12 12 – 18

Band Name K Ka V W mm band

Frequency Range, GHz 18 – 27 27– 40 40 – 75 75 - 110 110 - 300

10
Table 1.3

Wavelength in
Frequency Range,
Optical Band name Atmosphere,
in Hz
in meters
-5 -4 12
Far Infrared (IR) Band 2x10 to 10 (3 to 15)x10
-6 -5 12
Medium IR Band 1.5x10 to 2x10 (15 to 200)x10
-7 -6 14 14
Near IR Band 7x10 to 1.5x10 2x10 to 4.3x10
-7 -7 14 14
Visible Light 4x10 to 7x10 4.3x10 to 7.3x10
-8 -7 14 16
Ultraviolet Rays 10 to 4x10 7.5x10 to 3x10

1.3 THE EARTH’S ATMOSPHERE AND STRUCTURE


The real earth’s atmosphere has a complex structure, which may significantly impact
radio wave propagation, causing such effects as smooth refraction, scatter and energy
absorption of the radio wave. Variations of the electro-magnetic parameters of the
atmospheric air are highly dependent on its gaseous composition, pressure, humidity
and ionization. The vertical profile 1 of distribution of the main composites of atmospheric
air is shown in Figure 1.1.

1
The graph of the elevation dependence of any parameter of the atmosphere is called a "vertical
profile" of that particular parameter. Another example of vertical profiles is free electrons'
distribution, which is shown in Figure 1.2, or elevation dependence of the tropospheric air
refraction index (refractivity) which is shown in Figure 5.4.

11
Figure 1.1 Diagram of the gaseous composition of atmospheric air (by
percentage).

• As one may notice from the diagram, for heights of up to approximately 90 km,
the gaseous composition of the atmosphere is homogeneous as a result of the
continuous mixture caused by ascending, descending and horizontal air streams
that permanently exist in that area. At that range of height, the atmospheric air is
composed of about 78 percent molecular nitrogen and about 20 to 21 percent
molecular oxygen, despite the fact that they have different molecular weights.
The remainder is a mix of carbon dioxide, argon, ozone, and other gases. This
atmospheric area of homogeneous distribution of gases is conventionally divided
into two regions, the troposphere and the stratosphere. The troposphere is the
lowest portion of the earth's atmosphere. It contains approximately 80 percent of
the atmosphere's mass, and 99 percent of its water vapor and aerosols. The
average ceiling of the troposphere is approximately 17 km in its middle latitudes.
It is deeper in tropical regions (up to 20 km more), and is shallower near the
earth’s poles (about 7 km in summer, and an indistinct measurement in the
winter). The remaining portion of the homogeneous atmospheric area is known
as stratosphere. One of the specific features of the troposphere, which
distinguishes these two regions, is a rapid decrease of the concentration of water
vapors reliant on elevation. In fact, the humidity level is highly dependent on
weather conditions. The main characteristics of the troposphere are air pressure,

12
usually measured in millibars, the absolute temperature, measured in kelvins,
and absolute humidity, also measured in millibars. Based on numerous
observations and collaborative measurements carried out worldwide in 1925 the
International Commission for Aeronavigation introduced the “International
Standard Atmosphere”, which was later renamed, and is called now the Standard
Troposphere. It represents a hypothetical troposphere with characteristics that
are averaged from the measurements to portray all locations and seasonal
influences. Those characteristics are noted below [7]:
• A sea-level air pressure of 1013 millibars
• A constant vertical gradient of the air pressure of negative 120 millibars per
kilometer
• A sea-level temperature of 290 K
• A constant vertical gradient of the air temperature equal negative 5.5 K per
kilometer
• Relative humidity of 60 percent, which is assumed to remain elevation-
independent.
The average vertical profile of the temperature is:
T (h ) = T0 − 5.5 ⋅ h . (1.1)

Here T0 = 290 K and elevation h in kilometers. (1.1) may be understood based on the

following rationale: the tropospheric air is transparent to solar thermal radiation; thereby
it does not cumulate the heat directly from solar radiation. The bulk of that thermal
radiation penetrates through the troposphere freely and reaches the earth's surface,
where it is absorbed. The air layers adjacent to the earth’s surface become heated due
to heat transfer and air convection processes. The higher the elevation, the less is the
effect of these processes, resulting in linear decay of the temperature given by
expression (1.1). A similar linear vertical profile is specific for averaged tropospheric air
pressure 1.
As noted later in chapter 5, from the viewpoint of atmospheric propagation problems
analysis, the most important parameter is the dielectric permittivity of the atmospheric

1
As noticed from numerous observations, for the altitudes higher than 10 km the linearity of the
vertical profile of mean temperature and air pressure becomes significantly destroyed. However,
the atmospheric air at those high altitudes is extremely sparse; therefore those distortions do not
play a significant role in propagation mechanisms specifically on radar and communication paths.

13
air. It closely relates to the refraction index, defined as: n = ε . The mean value of
tropospheric air refraction index, being tightened to atmospheric air characteristics,
appears as a smooth, linearly decaying function of elevation (see section 5.3). At the
same time a large number of globally noted experiments and measurements,
undertaken over many decades, have demonstrated existence of seasonal and random
fluctuations of atmospheric air in all atmospheric regions. In the troposphere the main
mechanism of their generation is stipulated by the horizontal and vertical movements of
air masses. Under proper conditions those movements become turbulent, i.e. the air
masses of different refraction indexes are mixed randomly in space and time resulting in
random space-time fluctuations of the refraction index. These turbulent movements may
be observed in the visible region of the spectrum of electromagnetic radiation in multiple
ways to include the twinkling of stars, the wavering appearance of objects seen over the
earth's surface that is heated by sun, and the conversion trails left by the exhaust gases
of aircraft jet engines. The same processes take place in radio frequency bands, and all
of these examples demonstrate that the air in the troposphere is present in a random,
erratic flow.
The stratosphere does not alter the propagation of radio waves significantly, as it
is the atmospheric region containing fairly constant gaseous, whose composition is of a
very low density.
The Ionosphere is the upper part of the earth’s atmosphere which extends from
60 kilometers upwards. At these elevations the atmospheric air becomes ionized, i.e. the
neutral atoms and molecules split into positively charged ions and free electrons. This
state of matter is called plasma. Regarding the latest data obtained from the ionospheric
research, the upper border of the ionosphere is above 20,000 km. Ionization of
atmospheric air is caused by intensive radiation flow, emanating from the outer space
that is sometimes referred to as cosmic rays. Cosmic rays are the intensive flow of a
variety of elementary particles and photons 1 composed of a wide range of energies. A
major contribution to the total intensity of those radiation comes from the sun. Physics
courses will teach that in order to ionize the gas cloud, (i.e. to tear off an electron from

1
Acting as “the envelopes” of electromagnetic waves, photons are often considered as particles,
due to some properties that are specific to the elementary particles. For instance, these particles
are able to eject the electrons from the atom's boundary by bombarding them, or bouncing them
from each other like the balls in billiard game.

14
an atom or a molecule) a quantum of energy greater than a work function, We is to be

applied. This amount of energy must be expended to break the bond between electron
and atom (or electron and molecule), relative to each particular type of atom (or
molecule), and may be acquired from the composite parts of cosmic ray, i.e. from this
radiation that comes from outer space. Forms of this radiation may include:
• Elementary particles (protons, neutrons, electrons etc.), or
• Photons of electromagnetic character, such as ultraviolet, X-rays and
γ − radiation.
Hence, two types of ionization are to be distinguished: (1) Strike-Ionization, caused by
particles, and (2) Photo-Ionization, caused by photons.
For photo-ionization the energy carried by photon must be greater than, or equal to the
work function, i.e.
h f ≥ We , (1.3)

where: h = 6.626 ⋅ 10 − 34 J ⋅ s is a Planck's constant, and f is the frequency of the


photon. The maximum wavelength (threshold wavelength) of the radiation, which is able
to cause the ionization, may be found from (1.3) as
ch
λ ≤ λ max = , (1.4)
We, min

where c = 3 ⋅ 10
8
m / s is the speed of light in free space1, and We, min represents the
minimal energy in joules that a single particle or photon in a cosmic ray may have. That
portion of energy may also be expressed in electron-volts (eV), if the relation
1 J = 1.6 ⋅ 10 −19 eV is applied. Among the composite gases of atmospheric air, nitrogen
−18
oxide has lowest value of the work function, We = We , min = 1.48 ⋅10 J . Thus the
maximum wavelength of radiation that is able to ionize this gas is found from (1.4) as
λ max = 0.134 µ . From (1.4) one also may realize that only the ultraviolet radiation, as
well as the radiation of the shorter wavelengths such as X-rays and γ − radiation, are
able to cause the ionization of atmospheric air. From numerous observations and

1
Note that the ionization process itself does not depend on the intensity of radiation, but instead
on the wavelength of the ionizing radiation.

15
measurements it has been concluded, that the photo-ionization in the real atmosphere is
caused by radiation in the range of wavelengths from 0.03 to 0.14 µm.
The ionization degree of atmospheric air may be expressed by the number of free
electrons per unit volume (predominating per cubic centimeter), which is known as the
concentration of electrons (or plasma concentration), N e . An experimental graph of the

vertical profile of ionospheric plasma concentration is shown in Figure 1.2.

Figure 1.2 Vertical profile of plasma concentration in the real ionosphere

The ionization of atmospheric air starts from the height of about 60 km on upwards. It is
shown analytically [7] that for the hypothetical homogeneous gaseous composition of the
atmosphere and for the exponential model of the vertical profile of air pressure, the
single-layered vertical profile of the plasma concentration will be obtained when the
ionizing radiation is monochromatic. However, in real conditions for a complex structure
of air composition, as well as for the complex mixture of ionizing radiation (multi-particle
and multi-photon cosmic rays), the vertical profile of the ionospheric plasma
concentration becomes multilayered (stratified) as shown in Figure 1.2. As may be noted
from that figure, four layers exist during the day: D-layer (60 to 90 km), E-layer (90 to
120 km), F1-layer (180 to 230 km), and F2-layer (230 km and up). A sporadic ES-layer of
fairly high plasma concentration may appear and disappear randomly. Starting from
elevations of about 400 km and higher, there is no stratification, but only a smooth
decrease of the concentration of ionospheric plasma.

16
Aside from ionization, there are also recombination processes, where randomly
roaming free electrons may collide with positively charged atoms and molecules,
resulting in the recovery of neutral particles. It thereby becomes plain, that the rate of the
recombination process is as great as the number of free electrons and positive charged
particles, i.e. as great as the corresponding plasma concentration. Consider a limited
spatial volume of initially neutral atmospheric air is being ionized. Then, after ionization
is complete, ( i.e. plasma is generated) the considered volume will remain neutral
overall, due to the number of generated negative free electrons remaining equivalent to
the number of positively charged atoms, or to the molecules with an equal amount of
total charge. Initially, when ionizing radiation is applied to neutral gas, the rate of
generation of charged particles is nearly constant, causing an increase of plasma
concentration. When plasma concentration increases, after a certain period of time a
balance between ionization and dissociation will be achieved, so far as the rate of
recombination is proportional to the ionospheric plasma concentration N e . Thus, the

plasma's concentration stabilizes after the transition period ends. Typically the ionization
and recombination processes are in balance at noon and at midnight; whereas during
sunrise (or sunset), when the radiation coming from the sun appears (or disappears) this
balance is destroyed, and smooth changes of the ionization concentration (increasing in
the morning hours, and decreasing in the evening hours) may be observed. This
phenomenon results in the disappearance of two layers during night-time hours: the D-
layer and the F1-layer. Only two of the overall layers, namely the E-layer and F2-layer,
will remain during night-time hours, with a plasma concentration that is much less than it
is during the day (Figure 1.2). Finally, it must be noted that the profile of ionospheric
plasma concentration shown in that figure is simply a graph of the averaged values
of N e . In reality, there are random fluctuations of N e around each point of the graph,

similar to those occurring in a refraction coefficient of troposphere. There are several


factors, which cause the fluctuations of N e :

o Random fluctuations of intensity of the ionizing radiation emanating from


outer space
o Turbulent movements, caused by horizontal and vertical drafts of the
ionospheric plasma

17
o Fast invasion of micro-meteors and cosmic dust, acting as an additional
source of the ionization, causing highly ionized and randomly distributed
prolonged paths of ionizations
o Magneto-hydrodynamic waves originated by the influence of the earth's
magnetic field in the presence of mobile masses of ionized air.
These random fluctuations of ionospheric plasma concentration result in random
scatterings (not reflections) of t radio waves, which are most intensively observed in HF
and VHF bands.

1.4 CLASSIFICATION OF RADIO WAVES BY


PROPAGATION MECHANISMS

Two types of radio waves propagation are: (1) guided propagation, and (2) free
(unguided) propagation. Free (unguided) propagation of radio waves occurs between
corresponding antennas in the earth’s atmosphere, under-water, or in free space 1; in
contrast to guided propagation which occurs in man-made guiding systems, such as
wire-lines, coaxial cables, waveguides, and optical fibers. However, only free
propagating radio waves are subjects for detailed consideration in this textbook. The
following terms are introduced for classification of radio waves by propagation
mechanisms between transmitting and receiving antennas:
1. A direct radio wave (or simply “direct wave”) is a radio wave that propagates from
a transmitting to a receiving point over “an unobstructed ray path” [3], i.e. over
the trajectory that is either a straight line, or close to one. One example of a
direct radio wave is a radio wave that propagates via an earth-to-space (uplink),
space-to-space, or space-to-earth (downlink) path of a satellite communication
system (see Figure 1.3a).

1
Free space is defined as "Space that is free of obstructions and that is characterized by the
constitutive parameters of a vacuum" [3].

18
Figure 1.3. A). Direct radio wave, b). Reflected radio wave, c). Scattered radio
wave, d). Diffracted radio wave

A reflected radio wave (or a reflected wave) is a wave that travels to the receiving point
via a reflection from a boundary of two media, where the boundary of a size that is
much larger than a wavelength and is relatively close to the flat surface 1. The examples
portray radio waves traveling to the receiving point via reflections are: the reflections
from the earth's surface or via structures such as landscapes, metallic bodies placed into
orbits, etc. Near ideal reflection occurs via the ionized layers in ionosphere, when the
radio wave of low frequencies (up to 30 MHz) propagates between corresponding points
A and B, as shown in Figure1.3b.
3. A scattered (or secondary) radio wave is one that appears when the scatterings take
place during propagation. Scatterings may be observed when the radio wave
stochastically reflects from a rough, random surface with the average size of the
roughness comparable or less than the wavelength, or during propagation of the radio
wave through a medium which contains randomly shaped and/or space-time-distributed
irregularities. Typically these volumetric scatterings are observable when the dimensions
of scatterers (or random globules) are comparable or less than the wavelength itself.

1
The IEEE standard definition [3] is as follows: “For two media, separated by a planar interface,
that part of the incident wave that is returned to the first medium.”

19
Each globule then plays the role of a secondary (virtual) radiator of the random origin. A
superposition of the multitude of secondary waves that arrive to the receiving point B
produce the resultant field. Radio waves scattered from the small-scale irregularities of
the refractive index of tropospheric air may be considered as an example of a secondary
radio wave (Figure 1.3c). These random irregularities exist in the lower portion of the
atmosphere, or the troposphere (even in clear atmospheric air), as turbulences caused
by the horizontal and vertical movements of atmospheric air masses. The random
volumes of the irregularities of atmospheric air are able to scatter the microwaves
effectively within wide range of angles. That effect is the main mechanism of long-range
propagation of DMW and CMW, which are able to propagate over the horizon for
distances of many hundreds of kilometers. This phenomenon is known as far (over-the-
horizon) tropospheric scatter propagation of microwaves, or as troposcatter propagation.
The scatter propagation caused by the irregularities of ionization in the ionosphere is
another example of propagation by the mechanism of secondary radio waves. Those
irregularities are mainly generated by small-scale particles (micro-meteors) and dust
coming from outer space. These particles contain highly ionized footprints with an
average length of several meters. Therefore the phenomenon of scatter propagation
through the irregularities of the ionosphere takes place mainly with radio waves within
the VHF frequency band. Note, that both micro meteors and space dust are present in
the ionosphere permanently, so these types of secondary waves may be observed all
day long, regardless of the season.
4. A diffracted radio wave (or simply a diffracted wave) is defined as “an electromagnetic
wave that has been modified by an obstacle or spatial inhomogeneity in the medium by
means other than a reflection or refraction” 1 [3]. As known from the college physics
course, any material body placed across a propagation path may be considered as an
obstacle only if its linear dimensions are comparable or greater than the wavelength.
Otherwise the wave will spill over that material body (i.e. will diffract on it) and will easily
arrive at the observation point placed behind the obstacle. For a rough estimate of
propagation distance one may take into account that the diffraction will take place when
h ≤ λ , where h is shown in Figure 1.3d. The following approximate geometrical

1
Later we will use the term “refraction” to identify the bending of the propagation path in the
stratified troposphere, or ionosphere, and the term “refracted wave” to identify the wave that
penetrates from one medium into the second through their interface.

20
relations may be written based on an expansion of the cos Θ into Taylor's series for
small Θ angles. In fact, for real conditions the propagation distances are much smaller
than the earth’s radius: a = 6370 km. Thus

Θ   1 Θ2  Θ 2 a
h = a − a cos ≈ a 1 − 1 −  = , (1.5)
2   2 4  8

where Θ = r/a (1.6)


represents a geo-central angle between corresponding points A and B, and r indicates a
curvilinear distance (arc) between those points. Taking into account the formula h ≤ λ ,
as well as (1.5) and (1.6) one may define a maximum distance of the propagation of a
diffracted wave as

r≈ 8aλ , (1.7)

or, by expressing r in kilometers and λ in meters, we may obtain

rkilometers ≈ 7 λ meters . (1.8)

The expression (1.8) portrays a rough estimate of the limits of propagation distances
through diffraction mechanism, for radio waves of differing frequency bands. From
expression (1.8) it may be noted that the greater the wavelength and the longer the
propagation distance, the easier the diffraction can occur. This mechanism creates more
favorable conditions for the propagation of the radio waves at frequencies less than 30
kHz, at which the propagation distances may reach up to 1000 km. On the other hand,
for higher frequencies such as HF, the diffraction mechanism of propagation may not be
considered as an essential propagation mechanism as the maximum distances of HF
propagation, caused by the diffraction, become almost equal or even less than the Line-
Of-Sight (LOS) 1 distance, while the real observations portray the distances as much
greater than as noted from (1.8).
The set of terms, “direct”, “reflected”, “scattered”, and “diffracted” relate to the
mechanisms specifying how the radio wave arrives the observation (receiving) point.

1
“Line-of-Sight” is a term that is common to the propagation paths in frequency ranges VHF and
higher. The higher a frequency, the closer propagation properties are to those of the optical
waves.

21
Now we present another set of terms, which allow the classification of radio waves
based on spatial area that the propagation paths are traveling through:
1. A sky wave (or ionospheric wave) is “…a radio wave that propagates obliquely
toward, and is then returned from the ionosphere”. [3].This type of radio wave is
localized in the spatial region between the ionosphere and the earth’s surface,
and is shown in Figure 1.3b. It is thereby evident that it may also be called a
reflected wave, if we intend to specify the mechanism of propagation. Note, that
the long range propagation distances of HF radio waves are stimulated by the
mechanism of subsequent reflections of sky waves from the ionosphere and the
earth's surface. These results in propagation distances of thousands of
kilometers (see Figure 1.4).

Figure 1.4. Illustration of the ionospheric propagation mechanism of the HF radio


waves

2. A ground wave is a radio wave that propagates “From a source in the vicinity of
the surface of the earth, i.e. a wave that would exist in the vicinity of the earth's
surface in the absence of the ionosphere” [3]. Two ground wave modes may be
considered as independently existing:
 A surface wave is a non-TEM mode that propagates along the earth’s
surface and is guided by the air-ground boundary; this type of wave is
1
specific to radio waves generated by “so called” low-elevated antennas

1
Antenna elevation above the earth’s surface that is close to, or less than a wavelength

22
 A space wave is a superposition of direct and ground-reflected TEM-
waves in the vicinity of the earth’s surface; this type of wave is specific to
radio waves generated by so called high-elevated antennas 1, mainly in
frequency ranges of VHF and higher.
Per above definitions, one may conclude that the contribution of each component into
the ground wave depends strictly on the radiating antenna height above the earth's
ground surface. For an antenna with an elevation of several or more wavelengths above
ground level (high-elevated antenna) the space wave component of the ground wave is
predominant. Otherwise, for a low-elevated antenna, i.e. for an antenna with an
elevation that is comparable or less than the wavelength, the surface wave component
will become predominant. These issues will be discussed further in chapter 4.

1.5 INTERFERENCES IN RF TRANSMISSION LINKS


The quality of information transmission, via a radio transmission link between
corresponding points, as well as the quality of radar performance, is highly impacted by
the presence of disturbances to the desired signal’s reception path. The term
interference is commonly used in communications engineering practices to manifest
those disturbances of a desired signal. Meanwhile, in physics and in electromagnetic
theory the same term is used to convey the superposition of the considered
electromagnetic wave intermingling with other electromagnetic wave(s): either of
different origin, or of the same origin, but arriving at the observation point from different
propagation directions. In this section we will amplify the first meaning of the term
interference. However, the second meaning will be examined in the following chapters.
Interferences typically destructively impact the content of information received, vs. the
information actually transmitted. Considered as destructive random processes, which

1
Antenna elevation above the earth’s surface that is greater than several wavelengths. In order
to be considered as a high-elevated antenna, the feeding line must be non-radiating.

23
occur in the receiving mode, these interferences may come into view in two different
forms:
• They may be in the form of random fluctuations of the parameters of the desired
signal. For instance, when a monochromatic signal 1 passes through the
propagation path, then both the amplitude and the phase of the signal will
randomly fluctuate. The rate of these random fluctuations is much less than the
rate of the oscillations of the signal’ carrier; as typically the quasi-period of
random fluctuations is hundreds of milliseconds and up. Thereby, the
fluctuations may be simply interpreted as multiplications of the amplitude of the
signal by slow random variable(s). This is similar to passing the signal through a
linear two-port network with a randomly fluctuating transmission coefficient,
where the input signal and voltage (or current) is multiplied by the transformation
coefficient of a random character to determine the output. Hence, this type of
interference is referred to as multiplicative interference, or simply signal’s fading.
This form of interference is originated in the propagation medium, i.e. along the
propagation path.
• Another form of interference appears in both the propagation medium and in
conjunction with the receiver (including a receiving antenna). It occurs
independently and simultaneously with the desired signal, superimposing
(overlaying) to the desired signal, and is therefore known as additive
interference, or in simple terms: noise. In contrast with fading, this type of
interference may be specific to an extremely wide spectrum, affecting all
applicable RF frequency bands

Based on the above definitions, the output signal of the receiver may be written in time
domain as
s (t ) = κ (t ) ⋅ s (t ) + n(t ) .
~ (1.9)

Here s ( t ) notes the desired signal, κ (t ) is the multiplicative interference, and n (t )

represents the additive interference. Both, κ (t ) and n (t ) indicate random processes.

1
To present a strict approach, the quasi-monochromatic signals are to be discussed, simply
because any information carrying, modulated signal is never purely monochromatic. However, in
practice, RF signals that are transmitted through propagation paths are narrow-banded in most
cases, thus they may be considered as monochromatic for analytical purposes.

24
To present an example of multiplicative interference, the fading of the voice volume of an
HF-broadcast audio signal may be considered. Another example is the “deep fades”
encountered with a receiving antenna output signal, which is indicated by secondary
troposcatter associated with over-the-horizon microwave radio links. Two types of
multiplicative interference, the slow and fast fading of signal level 1, are presented in the
classification chart shown in Figure 1.5. Additionally, a display of seasonal variations of
a received signal is included (conventionally), to indicate a multiplicative interference.
The uniform hum of an HF-broadcasting receiver’s audio output may be considered as
an example of additive interference(s) which permanently exists, regardless of the
existence of a desired signal within a broadcasting channel.
In Figure 1.5 below, the reader will find clarifications for the classification of additive
interference types (noise(s)). Beneath Figure 1.5, interference / noise types are defined
in greater detail.

Figure 1.5 The classification of interferences associated with RF links

1
Occasionally the seasonal variations of signal levels may also be considered as an example of
multiplicative interference. However, they may be excluded if the pre-known character of those
variations is accounted for.

25
1. Receiver noise (or internal noise): A noise type that appears in different areas of the
pre-detection (linear) section of receivers including antennas, feeders, RF and IF
amplifiers, filters, etc. The nature of inner noise is stipulated by several physical
phenomenons to include:
 Random thermal movements of free electrons in resistors, conducting wires,
antennas, and associated elements: The long-term averaged summation of
vector velocities for electrons is zero. Therefore, if no outer forces are
applied, there is no draft of the electron cloud within the circuit. For short-time
periods, the chaotic motions of electrons result in “jumps” (short pulses) of
current/voltage; and the overlay of a large number of these short and
overlapping pulses emanates as a steady hum, which is known as thermal
noise. The spectrum of thermal noise covers the entire RF range, and the
higher the temperature, the more intensive are the movements are and
bigger the power spectral density of thermal noise. However, thermal noise
may exist even without a current flowing through the associated element
 The discrete character of associated particles, i.e. electrons or vacancies,
when flowing through an active electronic component, such as a diode,
transistor, running wave tube, etc. may cause a random sequence of
splashes within the current contained in the circuits. This phenomenon is
known as shot effect. However, it will not occur if the current is sufficiently
large, as the natural averaging effect of a larger current will flatten off the
fluctuations of the current. It is only for small currents flowing through a
component that the shot effect becomes significant. This is particularly
specific to the first conducive stage of RF signal amplifiers (typically placed
right at the receiving antenna output, i.e. before the feed line associated with
receiver).
2. External noise: This noise type usually penetrates into the receiver from the
propagation medium. The three types of external noises are described as follows:
• Man-made noise: Examples of man-made noise include the noise generated by
the ignition system of automobiles, powerful electric motors, power transmission
lines, and high power distribution equipment as well as by other residential and
commercial systems. In these cases, electromagnetic energy will be radiated
into our frequency bands of interest

26
• Atmospheric noise, which is generated by two sources, and therefore may
appear within two types:
 Lightening discharges within the troposphere, and
 A steady noise background, generated by collisions between atoms and
molecules within the tropospheric air layer.
The spectral density of the first type of atmospheric noise will typically be
concentrated in the lower part of the RF spectrum; i.e. in a frequency range
extending up to 30 MHz. For higher frequencies (i.e. VHF, UHF and higher)
the intensities of this first type of noise may be ignored, as they are no
comparable to the effects of the second type of noise. The second type of
noise is characterized by “noise spectral power density”, which increases in
direct proportion to the square of the frequency, and therefore, reigns
significantly within the microwave frequency bands.
• Cosmic noise signifies the RF radiation emanating along the observation
direction border(s) of outer space. The intensity of cosmic noise is reliant upon
the location of where the receiving antenna is directed / aimed. Note that the
ionosphere is impenetrable by radio waves with frequencies of less than 30 MHz,
so this type of noise is applicable only to radio links at operating frequencies
higher than VHF; most typically of those whose receiving antenna(s) is directed
skywards ( i.e. for satellite downlinks). The RF thermal radiations of the earth's
surface may also be considered as cosmic relative to these radio links.
Numerous radio-astrophysical observations have indicated that the most intense
cosmic noise comes either from the sun, or the center of our associated galaxy
i.e. the Milky Way, and/or from several other constellations within the universe.
Additional details about interferences, and their estimates, may be found in chapter 6.

REFERENCES

[1] Maxwell, J.C., “A Dynamical Theory of the Electromagnetic Field,” Proc. Roy. Soc.,
London, Vol. 13, 1864, pp. 531-536
[2] Burrows, C.R., “The History of Radio Wave Propagation Up to the End of WW-I,”
Proceedings of the IRE, Vol. 50, 1962

27
[3] IEEE Standard Definitions of Terms for Radio Wave Propagation. Std 211-1997
[4] Withers, D., Radio Spectrum Management, IEE Telecommunication Series 45, 1999
[5] Derek, M.K., Ah Yo, and Emrick, R., “Frequency Bands for Military and Commercial
Applications,” Ch.2 in Antenna Engineering Handbook, 4-th Ed. McGraw-Hill Co., 2007
[6] IEEE Standard 521-2002 (Revision of IEEE Standard 521-1984), IEEE Standard
Letter Designations for Radar-Frequency Bands, 2002
[7] Dolukhanov, M. P., Propagation of Radio Waves, Moscow, USSR: Mir Publishers,
1971

PROBLEMS

P1.1 Create a chronological timeline and place all the historical events described in
section 1.1 into it. Use that time-line to indicate other events you have found from your
own educational and/or professional career to archive the history of radio wave
propagation theory and practice.

P1.2. Convert the following free-space wavelengths into frequencies and oscillation
periods utilizing scientific notations and proper engineering prefixes if applicable: 300
km, 1m, 0.5 µm. Determine which frequency bands they are associated with:
Answer
Wavelength Frequency Periods of Oscillation

300 km = 3 ⋅ 10 5 m 1000 Hz = 1 kHz (ELF) 10 −3 s = 1 ms


1m 3 ⋅ 10 8 Hz = 300 MHz (VHF) 3.33 ⋅ 10 −9 s = 3.33 ns
0.5 µ = 5 ⋅ 10 −7 m 6 ⋅ 1014 Hz = 600 THz (visible light) 1.67 ⋅ 10 −15 s = 1.67 fs

28
P1.3. Use expression (1.4) to calculate the ionizing radiation wavelengths ( λ max

thresholds) for the following component-gases of the atmospheric air, when the
ionization work functions We are given in electron-volts (eV):

 Nitrogen oxide – 9.25


 Atomic oxygen – 13.61
 Molecular hydrogen – 15.42
 Atomic hydrogen – 13.6
Answer: 0.134µ, 0.09131 µ, 0.0806 µ, 0.09135 µ
P1.4. Based on the answer from Problem P.1.3, and the information from Table 1.3,
assess whether or not ultraviolet radiation is able to ionize molecular hydrogen. (The
answer is YES or NO; provide the numerical validation of your answer.)

P1.5. Determine the correct information to include, and populate the table below based
on the material addressed in expression (1.8). Show the results for the wavelength limits
provided for each frequency band. The following statement, "The greater the length of
the radio wave, the bigger is the propagation range" may be concluded from this table.
Would that statement prove true or false for all propagation cases? Explain your answer.
Table P1.1
Maximum propagation
Band Wavelength, λ in meters distances, rkm of radio waves,

in kilometers
3 4
LF 10 - 10
MF 102 - 103
HF 10 - 102

P1.6. What are the advantages of the increase in carrying frequencies supporting
communication systems and radars? Additionally, how may that increase change radio
link designs? Explain in your own words.

P1.7. While listening to an AM broadcast (particularly in the HF-band), how may one
identify the presence of multiplicative interference (fading), and additive interference
(noise), and distinguish them from one another? Explain your answer in your own words.

29
CHAPTER 2. BASICS OF ELECTROMAGNETIC
WAVES THEORY

2.1 ELECTROMAGNETIC PROCESS

1
2.1.1 MAXSWELL'S EQUATIONS OF ELECTRODYNAMICS

Dynamic electromagnetic process is considered a unity of two processes, identified as


time-varying electric and magnetic. Analytically, the unification of these processes is
expressed by a system of Maxwell's equations, as follows:
∂D
∇×H = + J tot (Ampere's law) (2.1)
∂t

∂B
∇×E = − (Faraday's law) (2.2)
∂t
∇ ⋅ D = ρ tot (Gauss' law) (2.3)

∇⋅B = 0 (Law of Continuity of Magnetic Field Lines) (2.4)

In this system of equations, E and H represent electric and magnetic field strengths
respectively, whereas D and B represent electric and magnetic induction vectors (or
electric and magnetic flux density vectors) for particular points of space. These vectors
are further coupled by constitutive parameters of medium ε , ε 0 , µ , and µ 0 .

D = ε 0 ε E , and B = µ 0 µ H . (2.5)

Here absolute dielectric permittivity and absolute magnetic permeability of free space
(vacuum) are

1
The term "Electrodynamics" is utilized here to describe the area of "Electromagnetism" relative
to time-varying electromagnetic processes, in contrast to "Electrostatics" and "Magnetostatics"
which are to represent to areas of study relevant to time-constant electric and magnetic fields,
respectively.

30
1
ε0 = 10 −9 F/m, and µ 0 = 4π ⋅ 10 −7 H/m (2.6)
36π
respectively, whereas ε and µ signify relative dielectric permittivity (dielectric constant)
and relative magnetic permeability (magnetic constant) specific to the particular medium.
All media herein are categorized based on ε and µ as follows:
1). Constant and parametric media presenting time dependence
2). Homogeneous and inhomogeneous media presenting spatial dependence
3). Specific character such as isotropic (for scalar ε and/or µ ), and anisotropic (for
tensor ε and/or µ ), media
4). Linear, if ε and/or µ are field intensity independent, and non-linear, otherwise.
If the relations in equations (2.5) are taken into account, then equations (2.1) and (2.2)
may be rewritten as:
∂E
∇ × H = ε 0ε + J tot , (2.1a)
∂t

∂H
∇ × E = −µ 0 µ . (2.2a)
∂t
Per expressions (2.1a), (2.2a), and (2.3), one may realize that electric and magnetic field
vectors E and H interrelate and are coupled to the volumetric total conducting electric
current J tot and volumetric total electric charge ρ tot . They both ( J tot and ρ tot ) are

defined by the presence of the same electric charges (electrons, ions) that are able to
move freely within the considered spatial area 1. Therefore, it is not surprising that they
are coupled by the law of current continuity:
∂ρ tot
∇⋅J = − . (2.7)
∂t
The movements of free electric charges may be stipulated either by external force, or by
internal electromagnetic field, once it's generated so far by the externally forced
movements of charges. Thus, the total conducting current and charge may be presented
as sums:
J tot = J Ext + J (2.8)

ρ tot = ρ Ext + ρ (2.9)

1
Here defined in contrast to the bonded charges, such as those bonded to the crystalline lattice
of solid matter.

31
Here J Ext and ρ Ext represent the components of the current and charge that are

stipulated by external source(s), and J and ρ induced by the electromagnetic field


within the medium that contains free moving charges. It’s well known, that Coulomb's
force, applied to the charged particle, is in direct proportion to the amount of charge and
the electric field intensity. Therefore current J is expected to be in direct proportion to
the electric field as well:
J =σ E (Ohm's Law in differential form) (2.10)
The coefficient σ is called a conductivity of the medium, and is proportional to the
concentration of free charges (i.e. proportional to volumetric density of free electric
charge) in considering point of space.
If equations (2.8) and (2.10) are substituted, then (2.1) may be rewritten as:
∂E
∇ × H = ε 0ε + σ E + J Ext . (2.1b)
∂t
The physical meaning of the first term, in right hand side, translates to a volumetric
current density that exists in space, regardless of the existence of free charges, i.e. it is
stipulated in dielectric medium (or vacuum) by the time variations of the electric field.
This term is conventionally known as a volumetric displacement electric current, or
displacement electric current volumetric density, J dis .

By analogy, the right hand side term in equation (2.2) is known as a displacement
magnetic current volumetric density. The importance of these two displacement currents
is difficult to overestimate. Indeed, only these currents are accountable for keeping the
electromagnetic process running, as they permit continuous energy exchange between
electric and magnetic fields within the united electromagnetic process. The balance of
energy interchange within the spatial area containing the electromagnetic process may
be assessed as follows:
First we multiply both sides of equations (2.1b) and (2.2) by E and H respectively, and
subtract equation (2.1b) from equation (2.2). Next, we will refer to identity (A1.2.4.2),
given in Appendix-1, which may be applied to the left hand side to result in the following
equation:
H ⋅ (∇ × E ) − E ⋅ (∇ × H ) = ∇ ⋅ (E × H ) (2.11)
Now use the transforms that are applied to the right hand side:

32
∂H ∂E
− µ0 µ H − ε 0ε E − σ E ⋅ E − E ⋅ J Ext =
∂t ∂t
∂  µ 0 µ H 2 ε 0ε E 2 
= −  +  − σ E 2 − E ⋅ J Ext . (2.12)
∂t  2 2 
Here, the expression in parenthesis represents the energy per unit volume (volume
density of energy) cumulated by electric and magnetic fields respectively. The second
term represents the power loss per unit volume, due to finite conductivity of the medium.
The higher the conductivity of medium, the greater is the rate of collisions of charge-
carrying free particles within the crystalline lattice, and therefore the rate of
transformation of energy of the electric field into heat is increased.
The last term in equation (2.12) represents solely the power of the external source
implemented into the electromagnetic field.
Now we may integrate both, right and left hand sides, presented by the equations (2.11)
and (2.12) respectively within the volume V that contains an external source J Ext .

V
∫ [∇ ⋅ (E × H )] dV =...
∂  µ0 µ H 2 ε 0ε E 2 
... = − ∫ dV + ∫ dV  − ∫ σ E 2 dV − ∫ E ⋅ J Ext dV . (2.13)

∂ t V 
2 V
2  V V

∫ ∫
Here Wm = (1 / 2) µ 0 µ H 2 dV and We = (1 / 2) ε 0 ε E 2 dV denote total energies
V V


cumulated within volume V by magnetic and electric fields, PL = σ E 2 dV denotes the
V

total thermal loss of power of the electromagnetic field within volume V, and

PExt = − ∫ E ⋅ J Ext dV -- the total power given to the electromagnetic field by the external
V

source 1. The identity (A1.2.2.2), given in Appendix-1 that is known as Gauss theorem,
may be applied to the left hand side of equation (2.13), namely as:

V
∫ [∇ ⋅ (E × H )] dV = ∫ (E × H ) d S ,
S
(2.14)

1
The negative sign indicates a power that is inserted into the electromagnetic field, in contrast to
the positive power that is subtracted from the field.

33
Here the volume integral is replaced by the integration along the closed surface S, which

surrounds volume V. Note that d S illustrates a vector surface element directed outward
of the volume, normally to the surface at the considering point. Therefore, equation
(2.13) may finally be rewritten as:

PExt = (Wm + We ) + PL + ∫ (E × H ) ⋅ d S . (2.15)
∂t S

This is a mathematical formulation of the balance of energy of the electromagnetic field


known as Poynting theorem: the amount of power given to the electromagnetic field
by the external power source within a limited spatial area:
• Is partially consumed to increase the energy stored by electric and
magnetic fields in that spatial area
• Is partially dissipated within the volume as a thermal power loss
• Partially flows away from the volume as a radiated power
The surface integral in the right hand side of equation (2.15) allows introduction of a
vector of power flow density known as Poynting vector, which shows the amount of
power passed through the unit surface that is placed orthogonal 1 to the direction of flow
in a particular point of space.
Π = E×H , (2.16)
It may be seen from (2.16) that the unit for the Poynting vector, i.e. for the unit for the
power flow density (magnitude of vector Π ) is: (V/m)·(A/m)=W/m 2.
Figure 2.1 shows the transformations between time-varying electric and magnetic fields,
within the electromagnetic process, as it follows from Maxwell's Equations.

1
If the unit surface is not orthogonal to vector Π from (2.16), then a scalar product is appropriate.

Namely in equation (2.15), the arbitrary oriented elementary surface dS is represented by the

surface element vector dS , so the scalar product Π ⋅ dS represents an infinitesimal amount of


power that flows through that element.

34
Figure 2.1. Sketch of the main constituents of the electromagnetic process

The external source of electric current J Ext generates the initial magnetic field H . The

time variant magnetic field, at any arbitrary point A, appears as a source of displacement
∂H
electric current µ 0 µ , which forces the generation of the electric field E .
∂t

Consequently, a time varying electric field E , at the point B, appears as a source of


∂E
displacement magnetic current ε 0 ε that initiates the secondary magnetic field H ′ ,
∂t
and so on. This process may remain infinitely long in time, if there are no losses in
considering spatial area, i.e. if the conductivity of the medium is equal to zero.

2.1.2. BOUNDARY CONDITIONS OF ELECTRODYNAMICS


Maxwell's Equations noted in (2.1) – (2.4) represent differential equations in partial
derivatives. In a spatial area free of sources, the electromagnetic field is to be
considered standalone, as an independently existing form of matter. The first two
Maxwell’s equations are utilized here to describe the interrelations between electric and
magnetic fields, and may be rewritten as:

35
∂E
∇ × H = ε 0ε +J, (2.17)
∂t
∂H
∇ × E = −µ 0 µ , (2.18)
∂t
Here J depicts a conducting current from Ohm's Law, defined by equation (2.10).
(2.17) - (2.18) is a system of two first order linear equations with two unknowns being
depicted as E and H . As evident in collegiate mathematics, a general solution of the
system contains arbitrary (undefined) integration constants, creating a multi-valued
(ambiguous) solution. For real conditions, when configuration of boundaries between
mediums is known, the initial conditions may be set up to calculate those integration
constants. In electrodynamics applications, the initial conditions are conventionally called
boundary conditions, because they represent the act of "bonding" (restraining) the
values of electric and magnetic fields to those boundaries. In other words, these
boundary conditions allow transformation of a general solution of Maxwell's equations
into a particular solution that is specific for the given configuration of the boundaries in
which the electromagnetic field is being defined. That particular solution is known as a
boundary value problem.
For more consistency we will now consider those conditions in detail.
First, we take volume integrals of equations (2.3) and (2.4) within a volume V:

∫ (∇ ⋅ D )dV = ∫ D ⋅ dS = ∫ ρ
V S V
tot dV , (2.20)

∫ (∇ ⋅ B )dV = ∫ B ⋅ dS = 0 ,
V S
(2.21)

Within these equations, the divergence theorem is utilized (see (A.2.2.2) in Appendix A).
Here S denotes a closed surface representing the boundary of the volume V. The vector

surface element d S is always directed outbound to volume V. This volume

encompasses a part of the boundary between two mediums via constitutive parameters
( ε 1 , µ1 , σ 1 ) and ( ε 2 , µ 2 , σ 2 ) as shown in Figure 2.2.

36
Figure 2.2. Integration area in (2.20) and (2.21) integrals

Additionally, n indicates a unit vector normal to boundary of mediums and directed from
medium-2 towards medium-1.
As D and B vectors are specific to any particular point of space, then we may choose
to minimize volume V enough to allow uniformity of the field distribution within that
volume as well as on its boundary. Taking that fact into account we may write the
following expression

D ⋅ d S = ( D1 ⋅ n ) d S = Dn1 d S (2.20a)

for the top base of cylinder, and

D ⋅ d S = −( D2 ⋅ n ) d S = − Dn2 d S (2.20b)

for the bottom base of cylinder, wherein D1 , Dn1 and D2 , Dn 2 represent the electric

field inductions vectors and their normal components of the first and second media
respectively. These vectors and their components are considered to be constant along
the top and bottom surfaces ∆S, as mentioned above. Hence, the left hand side of
equation (2.20) indicating the D -vector flow through the closed surface may be
rewritten as:

∫ D ⋅ dS = [( D
S
1 − D2 ) ⋅ n ] ∆S = ( Dn1 − Dn 2 ) ∆S + Ξ D . (2.22)

Here the first term in the right hand side shows a D -vector flow through both bases of
the cylinder, and Ξ D represents a flow through the side surface of the cylinder. Hence it
is evident, that if we shrink the cylinder vertically towards the boundary of media, i.e.
take ∆h → 0 , then Ξ D will disappear.

37
The right hand side of expression (2.20) represents a total charge enclosed within the
volume V. If ρ tot is considered as a volumetric charge density, then obviously for

∆h → 0 the right hand side of expression (2.20) will also disappear. However, for a wide
range of electromagnetic problems, free charge is allocated within a tiny layer directly
atop the boundary surface of two media. With that, it is appropriate to consider a surface
charge ρ S that is distributed in a minute layer of infinitesimal thickness per unit surface

atop the boundary 1. Thus, the right hand side of equation (2.20) may be represented as
ρ S ∆S , and finally that expression may be transformed into the following equation:

( D1 − D2 ) ⋅ n = Dn1 − Dn 2 = ρ S . (2.23)

Similarly expression (2.21) may be transformed into the following:

( B1 − B2 ) ⋅ n = Bn1 − Bn 2 = 0 . (2.24)

Equations (2.23) and (2.24) represent boundary conditions for the components of
electric and magnetic fields that are normal to the boundary of two mediums. As one
may conclude, the normal component of the magnetic field flux density consistently
remains continuous across the boundary between mediums, whereas the normal
components of the electric field flux density may have a discontinuity if a free surface
charge exists on the boundary, i.e. exists within infinitesimal layer that surrounds the
boundary of two media.
Now, the boundary conditions may be evaluated for the tangential components of the
electric and magnetic fields if (2.17) and (2.18) are integrated within a surface S (ABCD)
residing across the boundary and orthogonal to them as shown in Figure 2.3.

1
Such charges do not exist in nature, thus this solution simply represents a mathematical

abstraction.

38
Figure 2.3. Integration area in surface integrals (2.25) and (2.26)

∂E
∫ (∇ × H )⋅ d S = ∫ ε ε
S S
0
∂t
⋅d S + ∫ J ⋅d S ,
S
(2.25)

∂H
∫ (∇ × E )⋅ d S = −∫ µ
S S
0 µ
∂t
⋅d S . (2.26)

Here d S = s0 d S displays a surface vector-element that is directed orthogonal to the

surface ABCD, with s 0 as a unit vector orthogonal to that surface. Now we apply the

Stoke’s theorem (see (A1.2.2.1) from the Appendix-1) to the left hand side of the
equation (2.25):

∫ (∇ × H )⋅ d S =
S Contour ABCD
∫H ⋅d S . (2.27)

It is apparent from Figure 2.3 that the integration path along the rectangular contour
ABCD may be expressed as:

∫H ⋅d S = H
Contour ABCD
1 ⋅ AB + H 2 ⋅ CD + Int , (2.28)

Where Int represents a line-integral along sides BC and DA of height ∆h . That integral
will disappear for the vanishing height of the rectangle, i.e. for ∆h → 0 . If the following

three unit cross-orthogonal vectors s 0 , n , and l 0 are positioned as it is shown in Figure

2.3, then the vectors AB and CD may be replaced by AB = l 0 ⋅ ∆L = (s 0 × n ) ⋅ ∆L , and

CD = − l 0 ∆L = − (s 0 × n ) ∆L . Then, expression (2.28) may be rewritten after simple


transformations as:

39
lim ∆h→0 ∫ H ⋅ d S = [ H ⋅ (s
Contour ABCD
1 0 × n ) − H 2 ⋅ (s 0 × n )] ∆L =

= [n × ( H 1 − H 2 )] ⋅ ( s 0 ∆L) . (2.29)

The first term in the right hand side of (2.25) will vanish to zero when ∆h → 0 , as well as

will a second term, except in the case when the conducting current J has a surface
instead of spatial, distribution (i.e. flows into a layer of infinitesimal thickness along the
boundary of mediums as shown in Figure 2.3). Hence if J S represents a surface

conducting current, then the second term in the right hand side of formula (2.25) may be
replaced by the flow of J S through the line segment MN as shown:

∫J
MN
S ⋅ ( s 0 ⋅ dL) = ( J S ⋅ s 0 ) ∆L . (2.30)

Making (2.29) and (2.30) equal will result in:


[n × ( H 1 − H 2 )] = J S , (2.31)

or, in scalar form:


H 1τ − H 2τ = J S . (2.31a)

The above procedure may be repeated for expression (2.26), and will result in:
[n × ( E1 − E 2 )] = 0 , (2.32)
or, in scalar form:
E1τ − E 2τ = 0 . (2.32a)

Physical meaning of equations (2.31) and (2.32) is defined as follows: Tangential


components of the electric and magnetic field strengths remain continuous across the
boundary of two mediums, unless there exists surface current flowing on the boundary
that results in a jump of the tangential component of the magnetic field strength. That
jump is equal to the value of that surface current density.
In some of applications in antennas and microwave theory a boundary with perfect
electric conductor (PEC) is of interest, and is identified for its infinite conductivity, i.e. for
σ = ∞ . Based on expression (2.1a), for the spatial areas free of sources ( J ext = 0) , it
becomes evident that the electric field must be assumed to be equal to zero due to the
fact that the induced conducting current σ E may not be infinitely large. The absence of
the electric field results in the absence of the magnetic field, thereby indicating an
absence of the entire electromagnetic process within the PEC medium. Hence, the

40
boundary conditions (2.23) (2.24), (2.31), and (2.32) may be rewritten for the PEC
boundary via the following formulations:
ρS
E ⋅ n = En = . (2.33)
ε 0ε
H ⋅n = Hn = 0 . (2.34)

n × H = J S , or Hτ = J S (2.35)

n×E =0 , or Eτ = 0 (2.36)

From formulas (2.33) through (2.36) one may conclude, that the tangential component of
the electric field and the normal component of the magnetic field on the boundary of
PEC are always equal to zero, i.e. the electric field-lines are always perpendicular to the
PEC boundary, whereas the magnetic field-lines are always tangential to it.

----------------------------------------------------------------------------
Example 2.1
The slope angle of electric field lines changes when passing from one medium into
another as shown in Figure E.2.1. Find the angle α 2 if ε 1 = 5 , ε 2 = 1 , angle α 1 = 60 0 ,
and there are no surface charges distributed along the boundary of those media.

Figure E.2.1. Sketch of the electric field lines on the border of two media

Solution
Eτ 1 Eτ 2
• tan α 1 = , tan α 2 = .
E n1 En 2

• Based on (2.32a) E1τ = E 2τ . Thus, E n 1 tan α 1 = E n 2 tan α 2 .

41
• Based on (2.23) for ρ S = 0 we may assume: ε 1 E n 1 = ε 2 E n 2 . By substitution into

the previous expression we may derive finally:

ε2  1 
α 2 = tan −1  tan α 1  = tan −1  3  = 11.30 . (Answer)
 ε1  5 
-------------------------------------------------------------------------------

2.1.3. TIME-HARMONIC ELECTROMAGNETIC PROCESS.


CLASSIFICATION OF MEDIA BY CONDUCTIVITY

If time variations of electric and magnetic fields in dynamic electromagnetic processes


are assumed to be harmonic (sinusoidal), which is of predominant interest in science
and technology, then significant simplifications in Maxwell's Equations may be achieved
by applying a complex variables analysis. If, for example, time-harmonic oscillations of

the electric field are presented analytically as E cos(ω t + ϕ ) , then transformation into

complex form E exp[ i (ω t + ϕ )] allows representation as follows 1:

E cos(ω t + ϕ ) = Re {E exp[i (ω t + ϕ ]} = Re {E exp[iω t ] exp[iϕ ]}. (2.37)


The most attractive feature of complex analysis is the fact that time derivatives (or time
integrals) may simply be replaced by multiplication (or division) by the factor iω , which
transforms differential equation into algebraic. Indeed

{E exp[i (ω t + ϕ )]} = i ω {E exp[i (ω t + ϕ )]} . (2.38)
∂t
Even a time-harmonic multiplier, exp( i ω t ) may be cancelled out of Maxwell's
equations. Maxwell's equations are written now not for the time-varying electric and
magnetic fields but for their vector-phasors, such as:

E = x 0 E X exp(iϕ X ) + y 0 EY exp(iϕ Y ) + z 0 E Z exp(iϕ Z ) , where E X , EY , E Z and ϕ X ,


ϕ Y , ϕ Z represent a coordinate-dependent amplitudes and initial phases of oscillation of

1
Here the Euler's formula exp(iα ) = cos α + isin α is used, where i = −1 .

42
the electric field vector in Cartesian coordinates. A phasor for the magnetic field vector
or any other scalar or vector variable may be introduced similarly.
Taking the above statements into account, Maxwell's equations for the source-free
spatial region may be expressed in complex form as follows:

∇ × H = iω ε 0 ε E + σ E , (2.39)

∇ × E = − i ω µ 0 µ H , (2.40)

ρ
∇ ⋅ E = tot (2.41)
ε ε0

∇ ⋅ H = 0 (2.42)
Then expression (2.39) may be transformed to 1:

 σ 
∇ × H = J tot = J dis + J = iω ε 0 ε E + σ E = iωε 0  ε − i E . (2.43)
 ωε 0 

Here the total volumetric current density J tot is shown as a sum of displacement, J dis

and conducting, J current densities. The same expression (2.43) for the lossless
medium ( σ = 0 ) can be rewritten as:

∇ × H = J dis = iω ε 0 ε E . (2.44)

If equation (2.43) is compared with equation (2.44), then a relative complex permittivity
of the considering medium (or material) may formally be introduced as:
σ
ε = ε − i = ε − i 60λ 0σ . (2.45)
ωε 0
In (2.45) the following substitution is used: ω = 2π c / λ0 , where λ0 represents the

wavelength in free space, and c = 3 ⋅ 10 8 m/s represents the speed of light in free space.

1
Here and below we’ll assume the absence of the magnetic losses, which is specific for the
propagation problems. Moreover, for most of the cases of RF propagation in earth’s atmosphere,
along the earth’s surface, and in space the propagation medium(s) are assumed to be non-
magnetic, i.e. µ = 1.

43
For further analysis we’ll assume both, ε and σ scalar quantities, i.e. the media under
consideration being isotropic 1.
Now, without limitations to generality, consider for simplicity a single-component electric
field (e.g. a field that is directed along x-axis) of zero initial phase shift. Then in (2.43) we

may replace E by scalar E . Thus for the lossy medium expression (2.43) may be
rewritten for the same component in phasor form as

J tot = iω ε 0 ε E = iωε 0 (ε − i 60λ 0σ )E = J dis − i J (2.43a)

As one may notice from (2.43a) the total volumetric current density is a complex quantity
with displacement current as a real part, and conducting current as its imaginary part.
The ratio between imaginary and real parts of the complex dielectric constant (2.45) is
the same as the ratio of conducting and displacement current densities. This ratio may
be used to specify the rate of losses in considering medium or material. In complex
plane it represents tangent of the slope angle δ ε , identified as loss angle (Figure 2.4).

J σ 60 λ 0 σ
tan δ ε = = = . (2.46)
J dis ω ε 0 ε ε

Figure 2.4. Conducting, displacement, and total current densities in a complex


plane

Based on the values of tan δ ε , classification of media by conductivities is introduced

conventionally as follows:
o tan δ ε < 10 −1 – for dielectrics

1
Anisotropic properties of ionospheric plasma interacting with earth’s magnetic field and its
impact on radio waves propagation is considered shortly in chapter 5. All other propagation
mechanisms that are of interest for communications and radar applications relate to propagation
in isotropic media only.

44
o 10 −1 ≤ tan δ ε ≤ 10 - for semiconductors

o tan δ ε > 10 – for conductor

From (2.46) one may notice, that the value of tanδε is dependent on frequency ω. The
same medium may exhibit different behavior in different frequency ranges. For several
media listed in Table 2.1, frequency dependences are presented in Figure 2.5. As one
may notice from the figure, marble and mica may be considered as ideal dielectrics for
the entire radio frequency (RF) spectrum. Quartz, paraffin, glass, atmospheric air,
polyethylene, (etc.) may also be considered as ideal dielectrics for the entire radio-
spectrum. Nearly all metals, including copper, iron, aluminum, mercury, and others
behave as ideal conductors in the entire RF spectrum up to nearly 1010 Hz.
Several media, such as wet and dry soil, sea and fresh water, ice, etc. may display
variant properties in different frequency bands. For example, a dry soil behaves as a
conductor in the LF frequency band, as a semiconductor in MF and HF frequency bands,
and as a lossy dielectric in higher frequency bands.

Figure 2.5. Frequency dependence of tanδε for variant media (see Table 2.1 for
details by reference numbers)

45
Table 2.1

(from figure 2.5)


Reference No
Relative Applicable
Conductivity,
Medium/material dielectric frequency
in S/m
permittivity range (Hz)

1 Sea water ~ 80 1–5 0 – 108


2 Wet soil 10 - 30 3.10 - 3 – 3.10 - 2 0 – 108
3 Fresh water ~ 80 10 - 3 – 2.4.10 - 2 0 – 108
4 Dry soil 3-6 1.1.10 - 5 – 2.10 - 3 0 – 108
5 Marble ~8 10 - 7 – 10 - 9 103 - 108
6 Mica ~7 10 - 11 – 10 - 15 103 - 108

----------------------------------------------------------------------------------------------------------
Example 2.2
Estimate the frequency ranges for sea water ( ε = 80, σ = 1 S / m ) to be considered as
conductor, semiconductor, and dielectric.
Solution
• First transform (2.46) to obtain the frequency:
60 σ c
f = (E-2.2.1)
ε ⋅ tan δ ε
Here c = 3 ⋅ 10 8 m / s is speed of light in free space

• The results of calculations based on expression (E-2.2.1) are shown in the table
below.

tan δ ε < 10 −1 f > 2.25 GHz Dielectric

10 −1 < tan δ ε < 10 22.5 MHz ≤ f ≤ 2.25 GHz Semiconductor

tan δ ε > 10 f < 22.5 MHz Conductor

------------------------------------------------------------------------------------------------------

46
2.2. FREE PROPAGATION OF UNIFORM PLANE RADIO
WAVES

As an electromagnetic process, we'll analyze the propagation of the radio waves based
on Maxwell's equations that are specifically conditioned to free propagation in the media
such as earth's atmosphere, outer space, water of seas and ponds, and the earth's
ground.
Consider a time-harmonic field 1 that is initially generated by an external source. The first
two of Maxwell's equations in a region, which do not contain a source (a source-free
region) may be expressed in complex form as:

∇ × H = iωε 0 ε E , (2.47)

∇ × E = −iωµ 0 µ H . (2.48)

To solve this system of two linear differential equations in partial derivatives we exclude

one of the unknowns ( E or H ) to bring this system to one equation of one unknown.

With that in mind we will take a curl from both sides of (2.48), and then substitute ∇ × H
from (2.47):

∇ × ∇ × E = −iωµ 0 µ ⋅ ∇ × H = k 2 E , (2.49)

where:

k = ω ε 0 ε µ 0 µ = β − iα (2.50)

represents a complex propagation constant with a physical meaning that we will define
later in this text. If the equality (A1.2.3.1) from Appendix-1 is applied to the left hand side

of (2.49), and (2.3) is taken into account, that is ∇ ⋅ E = 0 , for the medium free of electric
charges, then (2.49) may be rewritten as

∇ 2 E + k 2 E = 0 . (2.51)

1
Known also as monochromatic. The term comes from optic, and originated from the Greek
phrase "single colored". "Single colored" radiation in optics means a radiation of a single
frequency. The same meaning is adopted for the RF spectrum.

47
Here ∇ 2 indicates a Laplacian operator applied to E -vector. In Cartesian coordinates,

∇ 2 is expressed as:
∂2 ∂2 ∂2
∇2 = + + , (2.52)
∂x2 ∂ y2 ∂z2

Thus for the vector phasor E ( x, y , z ) = x 0 E x ( x, y, z ) + y 0 E y ( x, y, z ) + z 0 E z ( x, y, z ) the

Laplacian is

∇ 2 E = x0 ∇ 2 E X + y 0 ∇ 2 E Y + z 0 ∇ 2 E Z , (2.53)

where x0 , y 0 , and z 0 are the unit vectors along X, Y, and Z coordinates.

Equation (2.51) displays a particular form of well known wave equation called Helmholtz
equation that is a wave equation written for the time-harmonic process.
The function, given below by (2.54), may be considered as one of the simplest solutions
of Helmholtz equation:

E = x0 E m X exp [i (ω t − k z )] . (2.54)

In this expression, E m X is a complex amplitude (phasor) that includes the real amplitude

E m X and the initial phase shift ϕ 0 , i.e. E m X = E m X e 0 . It’s easy to show by
substitution, that (2.54) satisfies equation (2.51).
Regarding formula (2.54), the electric field contains a harmonic, time and Z-dependence
only, with an amplitude value of E m X . In other words all planes parallel to the XOY are

the planes of constant values of the field for the fixed time instance. As electric and
magnetic fields are coupled to each-other in every point of space, it is reasonable to
expect, for field components of both, electric and magnetic fields are portrayed as:

∂E ∂E ∂H ∂H


= = = = 0. (2.55)
∂x ∂y ∂x ∂y
The value of the magnetic field coupled to solution (2.54) may be found by substituting
the field into (2.47). In Cartesian coordinates, a curl in the left hand side of (2.47) is
conveniently presented in form of determinant, i.e. (2.47) may be presented as:

x0 y0 z0
∂ ∂ ∂
∇ × H = = iωε 0 ε ⋅ E . (2.56)
∂x ∂y ∂z
H X H Y H Z

48
Taking into account (2.55), and observing the fact that the right hand side of (2.56) does

not contain X- and Y-components of E -field, (2.56) may be rewritten as:


dH Y
− = iωε 0 ε E m X exp[ i (ω t − k z )] . (2.57)
dz
Here the partial derivative is replaced by the regular derivative due to single-coordinate
dependence. If we integrate (2.57), then the result will be:

H = y 0 H m Y exp[ i (ω t − k z )] , (2.58)

where
ω ε 0 ε 
H m Y = Em X . (2.59)
k

Both vector-phasors E and H contain the total phase


ϕ = ωt − β z , (2.60)

where β is a real part of the complex number defined by (2.50). Note that it affects a
phase and is called phase coefficient, whereas the imaginary part, α affects the
amplitude and is called attenuation coefficient, or attenuation constant (see subsection
2.2.2 for details). Surfaces represented by ϕ = const in space, namely the surfaces of
constant values of the phase, are called wave fronts. Consider the value of ϕ that
remains unchanged, i.e. we’ll try to find a velocity of movement of the wave front along
Z-axis. Then any increase in time ∆t is to be related to the change in coordinate ∆ z
(Figure 2.6), so thus:
ϕ 1 = ω t − β z = ϕ 2 = ω (t + ∆t ) − β ( z + ∆ z ) = const . (2.61)

Figure 2.6. Movement of the plane wave front. Only a square fragment of the
Infinity wave front is represented here.

49
From this equation (2.61) it is logical to determine the phase velocity, which is the
velocity of the movement of the plane wave front along the Z-Axis:
∆z ω
v= = . (2.62)
∆t β
If, in particular, ∆z represents a distance between two adjacent maximums of the wave
pattern, i.e. ∆ z = λ , then the time interval that is needed to move from one maximum to

the next is ∆t = T = 1 / f = 2π / ω , where T is a period of the oscillations, f is the linear


frequency, and ω is angular frequency. From (2.62) the phase coefficient may be
expressed in terms of the wavelength in considering medium as

β= . (2.62a)
λ
Based on above considerations one may conclude that the electromagnetic process
described by the solution in formulas (2.54) and (2.58) represents a wave that
propagates along Z-Axis with the constant velocity v and contains plane infinite wave
fronts, supporting the uniform distribution of the electric and magnetic fields along those
planes. For the considering particular case of propagation along Z-direction, wave fronts
are flat surfaces parallel to the X0Y plane. These types of waves are known as uniform
plane waves because of the constant distribution of the amplitudes and phases of
electric and magnetic fields along those planes. They do not exist in nature in their pure
form and are introduced simply as a mathematical abstraction to simplify understanding
of the physical concepts and make easier the formal mathematical evaluations. In reality,
the distances, covered by the radio waves, are much greater than the geometric size of
source. It will be shown later, that for those long distances this abstraction is a good
foundation for the radio wave propagation theory, and is capable of outlining all
propagation problems qualitatively and quantitatively.
From equation (2.59) one may realize that if a radio wave contains only an X-component
of the electric field, i.e. that is directed along positive X-axis then the magnetic field may
have only a Y-component that is directed along the positive Y-axis. From the same
equation it may then be confirmed, that for the propagation of time-harmonic radio wave

in lossless mediums ( σ = 0 ) both, ε and k become real numbers, therefore the


Poynting vector (2.16) also becomes real if the cross product is defined as:
~
 = E × H
Π , (2.63)

50
~
i.e. if the phasor H represents a conjugate of the complex vector H . In that particular
case of a lossless medium, the Poynting vector becomes real vector, whereas for any
 is a complex vector. Disposition of E and H vectors satisfies the
lossy medium Π
right-hand rule as shown in Figure 2.7.

Figure 2.7. Disposition of vectors E , H and Π in a free space according the right
hand law

An expression similar to (2.63) may be written for the amplitude phasor of the Poynting
vector as:
~
 = E × H
Π m m m. (2.64)

For practical engineering applications, it is more convenient to introduce a scalar that is


derived from the complex Poynting vector as an RMS value of the instant power flow
density, i.e. power flow averaged within a period of oscillation:

Π=
1
2
( ~
)
Re E m × H m . (2.65)

The factor 1/2 is due to the ratio between amplitude and RMS values of any time-
harmonic oscillations. It’s analogous to the relation between voltage amplitude, current
amplitude, and the power consumption within an electric circuit element.
As one may note from equation (2.59) if the parameters of the propagation medium ε
and µ are constant in time and space, then the ratio between electric and magnetic
fields is time- and space-independent. That ratio may be used to identify a property of
the propagation medium called intrinsic impedance of medium and may be defined from
(2.59), if (2.50) is taken into account, as well as µ = 1 is assumed:

51
E µ0 120 π 377
W = mX = = ≈ , Ohm, (2.66)
H mY ε 0 ε ε ε

The values for ε 0 and µ 0 are utilized from (2.6).

2.2.1. UNIFORM PLANE WAVE IN LOSSLESS MEDIUM

In expression (2.50) for parameter k the value of relative magnetic permeability µ may

be assumed to be µ = 1 , due to the non-magnetic types of mediums the propagation


theory deals with, such as salt and sweet water, ground soil, and atmospheric air.
Another assumption is σ = 0 , due to lossless propagation conditions in the atmospheric

air. Under those two conditions k becomes a real number, i.e. k = β . If this value of k
is substituted into (2.62) then the result for propagation phase velocity will become
1 c
v= = , (2.67)
ε 0ε µ 0 ε
where
1 1
c= = = 3 ⋅ 10 8 m/s (2.68)
ε 0 µ0 (10 / 36π F / m) ⋅ (4π ⋅ 10
−9 −7
H / m)
represents speed of light in vacuum. Expression (2.67) demonstrates that propagation
phase velocity in any real medium is always less than in free space. An exception is the
case of propagation within ionospheric plasma medium, which will be discussed later in
chapter 5.
Recall that the wavelength is a distance that is covered by the radio wave within one
period of oscillation, i.e. within time interval T = 1 / f = 2π / ω , as follows:

1 c 1 2π c λ 0
λ = v ⋅T = = = . (2.69)
ε f ε ω ε
Here:
c 2π c
λ0 = = (2.70)
f ω

52
denotes a wavelength of the radio wave that propagates in free space, i.e. in vacuum 1.

Now consider the propagation constant k defined by (2.50). For this particular case of
propagation in lossless medium:
ω 2π 2π
k =ω ε 0 µ0 ⋅ ε = ε = ε = rad/m. (2.71)
c λ0 λ

Parameter k is usually referred to as a wave number. To be more specific we have to


note that in physics the wave number is defined as a reciprocal of the wavelength, 1 / λ ,
showing the number of the wavelengths on a unit distance along the propagation path
whereas the quantity 2π / λ is referred to as angular wave number. However most of the
authors in electromagnetics are using the term wave number to specify 2π / λ , so we'll
do all the way in this text. To summarize (2.71) we emphasize that the higher the
oscillation frequency, the greater the wave number is.
As mentioned earlier, in the case of the absence of losses, σ = 0 regarding (2.45),

(2.50), and (2.66) three complex quantities, ε , k , and W become real numbers with the
following values:

2π 2π
k=β = = rad/m (2.72)
λ0 / ε λ

120 π 377
W = ≈ Ohms. (2.73)
ε ε

Thus, as is indicated in (2.59), vectors E and H , being cross-perpendicular to each


other, are not phase-shifted, i.e. maximums and minimums of these vectors are
allocated in the same points of space. As follows from (2.54) for this case of lossless
medium ( k is a real number) the amplitude of the wave is Z-independent, i.e. it remains

unchanged along the entire propagation path, i.e. along Z-Axis. If the conjugate of H mY

is substituted from (2.66) and into (2.65), then the effective (RMS) power flow may be
found as:

1
Do not confuse the term “free propagation” with “propagation in free space”. “Free propagation”
is referred to unguided propagation that may occur in any medium such as air, water, soil, etc.,
whereas “propagation in free space" means propagation in ideal conditions (vacuum) that is free
of any material substance.

53
~
1  Em  E 2
Π = Re  E m × = m . (2.74)
2  W  2 W

This expression allows making a conclusion that the Poynting vector also becomes real,
time- and space-independent. The distribution of the electric and magnetic fields along
the propagation path is shown in Figure 2.8.

Figure 2.8. Structure of the radio wave in lossless medium

2.2.2. UNIFORM PLANE WAVE IN LOSSY MEDIUM

In this case the complex relative permittivity both, intrinsic impedance and wave number
are complex, and the expressions (2.45) and (2.66) may be rewritten in following forms:
σ σ
ε = ε − i = ε ′ − i ⋅ ε ′′ , where ε ′′ = (2.75)
ωε 0 ωε0

W = W ⋅ e iΦW . (2.76)
In this case the expression (2.50) is complex and is rewritten here:

k = β − iα . (2.77)
For the plane uniform radio wave that propagates along Z-direction, the expression
(2.54) may be presented in the following form, after (2.77) is substituted into it:

E = x0 E m X e i (ω t −k z ) = x0 E m X e −α z ⋅ e i (ω t − β z ) .

(2.78)

Similarly, based on (2.58), the expression for the magnetic field may be presented as:

H = y 0 H m Y e i (ω t − k z ) = y 0 H m Y e −α z ⋅ e i (ω t − β z ) ,

(2.79)

54
Where:

E m X E m X −i ΦW
H m Y = = e . (2.80)
W W
It is obvious from equations (2.78) and (2.79) that the amplitudes of electric and
−α z
magnetic components of the radio wave decay exponentially as e . Coefficient α 

introduced in (2.50) as attenuation coefficient and β, the phase coefficient, which is a


part of the total phase ω t − β z .
The expression (2.80) demonstrates the fact that there is an initial (permanent) phase
shift, Φ W between electric and magnetic fields oscillations. This phase shift may be

transformed into the distance shift ∆ z = Φ W / β between electric and magnetic waves

as shown in Figure 2.9.

Figure 2.9. Structure of the uniform plane wave in lossy (semi conducting)
medium

The decrease of the intensity of radio wave during propagation may also become
evident if the expression for effective (RMS) value of Poynting vector is recalled. That

expression may be derived if (2.78) and the complex conjugate of H m Y from (2.80) are

used along with (2.65).



1   E m X iΦW  E m X 2 − 2α z
=
Π = Re E m X e e cos Φ W . (2.81)
2  W 
 2W

From equation (2.81) it is evident, that in case of propagation in lossy medium the
average power density flow not just decays exponentially along the propagation path,
but also its initial value becomes smaller than that for propagation in lossless medium,

by the value of the so called power factor, which is equal to cos Φ W .

55
The important parameters of propagation, namely attenuation coefficient α and phase
coefficient β , are considered as composite parts of the complex propagation constant

k . If in equation (2.50) we assume µ = 1 , then taking into account (2.75) it will results in

ω σ
κ = β − iα = ω ε 0 µ 0 ε = ε −i = ...
c ωε 0

ω  σ  2π
... = ε exp − i arctan = ε exp(− iδ ε ) = ... .
c  ωε 0 ε  λ 0

2π  δε δε 
... = ε  cos − i ⋅ sin . (2.82)
λ0  2 2 
Here
2
 σ  ε
ε = ε + 
2
 = (2.83)
 ωε 0  cos δ ε

is the magnitude of complex dielectric permittivity, expressed by its real part, ε and by
angle of losses, δ ε given by (2.46) and shown in Figure 2.4. Thus (2.82) may be

rewritten as:

2π ε  δ δ 
β − iα =  cos ε − i ⋅ sin ε  . (2.84)
λ0 cos δ ε  2 2 
Thereby, the attenuation and phase coefficients may be derived by taking real and
imaginary parts in both sides equal to each other:

2π ε δε
α= sin , (2.84a)
λ0 cos δ ε 2

2π ε δε
β= cos . (2.84b)
λ 0 cos δ ε 2

Here λ 0 is a wavelength in free space.

Expressions (2.84a) and (2.84b) may be simplified for following two extreme cases.

56
2.2.2.a. LOW-LOSS DIELECTRIC MEDIUM
δε 1
In this case tan δ ε << 1 (or δ ε << 1 ) , thus: cos δ ε ≈ 1 , sin ≈ tan δ ε , and therefore
2 2
expressions (2.84a) and (2.84b) may be simplified as follows:
π
α≈ ε tan δ ε , (2.85)
λ0
2π 2π
β≈ εε = . (2.86)
λ0 λ

where λ = λ 0 / ε is a wavelength in the considered particular medium.

2.2.2.b. HIGH LOSS CONDUCTING MEDIUM


π ε σ
In this case tan δ ε >> 1 (or δ ε ≈ ), therefore: = ε ≈ , and
2 cos δ ε ωε 0
δε δε 1
cos ≈ sin ≈ . Thus:
2 2 2

σ ω µ0
α ≈β ≈ . (2.87)
2
In all other cases of semi-conducting mediums, which are more specific for the real
propagation conditions, only the original (2.84a) and (2.84b) formulas are applicable.

The attenuation constant is specified by a Neper-per-meter (Np/m) unit. The physical


meaning of that unit may become clear if (2.78) and (2.79) are recalled. From any of
these equations one may conclude that if α = 1 Np/m, then the amplitude of the radio
wave, which passes a distance of one meter, will decrease e = 2.71.... 1 times. Indeed,
from the equation (2.78) the attenuation constant is found as

1 E m X ( z = 0)
α Np / m = ln (2.88)
z Em X ( z )

Then, for the ratio of the amplitudes equal e = 2.71.... , and for the propagation distance
of z = 1 m, the value of the right hand side in (2.88) is equal to unity. Another unit,
decibel-per-meter (dB/m), which is widely used in engineering applications, is defined as

1
This is the base of the natural logarithm.

57
1  E m X ( z = 0) 
α dB / m = ⋅ 20 log   (2.88a)
z  E ( z) 
 m X 

and is more commonly annotated. The relation between those two units is
α dB / m = 8.68 ⋅ α Nep / m . (2.89)

Now the term attenuation of the radio wave may be introduced as:
A = α z ( AdB = α dB / m ⋅ z ) (2.90)

that displays a total decrease of the radio wave amplitude along the entire homogeneous
propagation path of the length z . In inhomogeneous mediums, where ε and/or σ are
varying in space from point to point, α also becomes variable. Suppose the entire
inhomogeneous propagation track consists on several discrete segments of
homogeneous paths as shown in Figure 2.10. Then it’s apparent, that after passing the
entire z distance the amplitude of the radio wave will become
− ∑ α k ⋅∆zk
E m ( z ) = E m ⋅ e −α1 ⋅∆z1 ⋅ e −α 2 ⋅∆z2 ⋅ ⋅ ⋅ ⋅ ⋅ e −α n ⋅∆zn = E m ⋅ e k
(2.91)

for the total distance of


z = ∑ ∆z k . (2.92)
k

Figure 2.10. Radio wave propagation pattern in a discretely inhomogeneous


medium

If α is a continuous function of the distance, it is apparent that (2.91) may be


transformed into integral form as follows:
− α ( z ′ )⋅dz ′
E m (z ) = E m0 ⋅ e ∫ . (2.93)

Here E m 0 is the initial amplitude. Thus, the expression for total attenuation along the

entire propagation path may be rewritten in an integral from as

58
z
A = ∫ α ( z ′) ⋅ dz ′ . (2.94)
0

-----------------------------------------------------------------------------------------------
Example 2.3
Calculate attenuation coefficient and the phase coefficient for the radio wave
propagating in dry soil ( ε = 4 , and σ = 2 ⋅ 10 − 4 S/m). Calculations are to be performed
for the following three frequencies: f 1 = 6 kHz, f 2 = 600 kHz, f 3 = 6 GHz:

Solution
• For the frequency f 1 = 6 kHz the wavelength in free space is λ 01 = 50000 m.

Then regarding (2.46):


60 λ 01 σ
tan δ ε = = 150 >> 1 . (E-2.3.1)
ε
Hence, expression (2.87) is applicable for both, the attenuation coefficient and
phase coefficient, i.e.

σ ω µ0 2 ⋅ 10 −4 ⋅ 2π ⋅ 6 ⋅ 10 3 ⋅ 4π ⋅ 10 −7
α≈ = = 2.177 ⋅ 10 −3 Np/m
2 2
β ≈ 2.177 ⋅ 10 −3 rad/m
• For the frequency f 2 = 600 kHz the wavelength in free space is λ 02 = 500 m.

60 λ 0 2 σ
Then tan δ ε = = 1.5 ( δ ε = 0.9828 rad) (E-2.3.2)
ε
Hence, neither of the approximate approaches given in sections 2.2.2.a and
2.2.2b is applicable. The general expressions (2.84a) and (2.84b) must be used
for the attenuation coefficient and phase coefficient respectively.

2π ε δε 2π 4  0.9828 
α= ⋅ sin = sin   = 0.0159 Np/m
λ 02 cos δ ε 2 500 cos 0.9828  2 

2π ε δε 2π 4  0.9828 
β= ⋅ cos = cos  = 0.03 rad/m
λ 02 cos δ ε 2 500 cos 0.9828  2 
• For the frequency f 2 = 6 GHz the wavelength in free space is λ 03 = 0.05 m.

60 λ 0 3 σ
Then tan δ ε = = 1.5 ⋅ 10 − 4 << 1 . (E-2.3.3)
ε
Hence, expressions (2.85) and (2.86) may be used for these calculations.

59
π π
α≈ ε ⋅ tan δ ε = 4 ⋅ 1.5 ⋅ 10 − 4 = 0.019 Np/m
λ 03 0.05

2π 2π
and: β ≈ ε = 4 = 251.3 rad/m.
λ 03 0.05
-----------------------------------------------------------------------------------------------

2.3. POLARIZATION OF THE RADIO WAVES


The performance of transmitting and receiving antennas, as well as different
phenomenon accompanying radio wave propagation (such as reflection, diffraction,

scattering, etc.) are highly dependent on the time-space behavior of the E and H
components of the radio wave. When we considered the structure of the uniform plane

radio waves in previous section, it has been noted, that both, E and H vectors reside in
the plane that is perpendicular to the propagation direction. If, in particular, the electric
field is X-directed, then the magnetic field is Y-directed, and they are coupled by
expression (2.66), lying in the XOY-plane, i.e. being perpendicular to propagation Z-
direction. In other words if the direction of one of the vectors is known, then the direction
of the second vector is defined predictably, i.e. both of those vectors are tightly coupled
to each-other.
The term polarization of radio wave is introduced here based on the conventional

approach as a direction of the E vector, or more consistently, as a direction of the

oscillations of the tip of E vector. In section 2.2, a vertically polarized radio wave has

been considered, i.e. vector E , being collinear to the vertical X-Axis, always remains in
the vertical XOZ-plane along the propagation path as shown in Figure 2.7. Therefore, for

vertically polarized wave the proper expression for E may be rewritten as

E = x0 ⋅ E mX ⋅ e i (ωt − kz ) . (2.95)

Similarly, the horizontally polarized radio wave may be introduced here as a wave, which
has a tip of E vector always directed horizontally (collinear to Y-Axis), i.e. always
remaining in YOZ plane.

E = y 0 ⋅ E mY ⋅ e i (ωt − kz ) . (2.96)

60
Generally speaking, if E vector remains oscillating in one fixed, arbitrary oriented plane,
that radio wave is defined as linearly polarized.
For further generalization, consider the case when the tip of the E vector doesn’t remain
in one single plane, but follows a more complex trajectory. Consider the resultant vector

E of an arbitrary polarized radio wave as being composed as a combination of (2.95)


and (2.96), i.e. composed of two cross-orthogonal components, namely vertical and
horizontal.

E = ( x0 ⋅ E mX + y 0 ⋅ E mY ) ⋅ e i (ωt − kz ) . (2.97)

In other words, here any arbitrary polarized radio wave is represented as a superposition
of two, linearly polarized radio waves: one of them is vertically polarized, another one is
polarized horizontally. To clarify, recall the real part of the expression (2.97) and
consider the movement of the tip of the resultant E vector in XOY plane, i.e. in Z = 0
plane (see Figure 2.11).

Figure 2.11. Resultant E vector and its components. (Propagation direction is


shown by using X-sign, meaning the propagation from the observer forward)

The real parts of the components of E vector in particular point of space z = 0 are:
E X (t ) = E mX cos (ω t + Φ 0 X ) , (2.98)

EY (t ) = E mY cos (ω t + Φ 0Y ) , (2.99)

where Φ 0 X and Φ 0Y indicate the initial phases of the proper components, and E m X and

E m Y indicate real amplitudes 1.

1
Note, as a reminder, that Φ 0 X and Φ 0Y initial phases are conventionally included as part of

the amplitude phasors, i.e. E m X = E m X exp ( i Φ 0 X ) , and E m Y = E m Y exp ( i Φ 0Y ) .

61
The magnitude and the angle Ψ with X-axis of the real resultant vector E may be found
from formulas (2.98) and (2.99) as

E (t ) = E X (t ) + EY (t ) = E mX cos 2 (ω t + Φ 0 X ) + E mY cos 2 (ω t + Φ 0Y ) ,
2 2 2 2
(2.100)

EY (t ) E cos(ω t + Φ 0Y )
Ψ (t ) = arctan = arctan mY . (2.101)
E X (t ) E mX cos(ω t + Φ 0 X )
Now we consider the following specific cases:

Case 1: Φ 0 X = Φ 0Y ± 2π n , for n = 0, 1, 2, ……

This is a case of equal-phase oscillations of the components E X and EY . The resultant


vector magnitude from (2.100) may be defined as:

E (t ) = E mX + E mY cos(ω t + Φ 0 ) ,
2 2
(2.102)

where Φ 0 = Φ 0 X . The polarization angle, i.e. the angle between E -vector and XOZ-

plane is defined from (2.101) as follows:


E mY
Ψ = arctan = const (2.103)
E mX
As one may notice angle Ψ is time-independent, thus, in this case, the resultant wave is
linearly polarized with the positive ( + Ψ ) and constant angle of polarization, and the

amplitude of E m = E mX + E mY .
2 2

Case 2: Φ 0 X = Φ 0Y ± (2n + 1)π , where n = 0, 1, 2, ……

This is a case of opposite-phase oscillations of the components E X and EY . The


resultant vector is defined to be the same as in previous case (see (2.102)), and the
angle (2.101) between E and XOZ-plane becomes negative if trigonometric identity
cos [ x ± (2n + 1)π ] = − cos x is recalled, i.e.

E mY
Ψ = − arctan = const . (2.104)
E mX
As in the previous case, the resultant wave became linearly polarized. However a
negative angle of polarization appears, while retaining the same resultant amplitude, as
in previous case.

62
π
Case 3: Φ 0 X = Φ 0Y + , E mX = E mY = E m . In complex plane the horizontal component
2
lags behind the vertical. In this case (2.100) and (2.101) may be simplified as:

 π
E( t ) = E m cos 2 (ω t + Φ 0 X ) + E m cos 2  ω t + Φ 0 X −  = E m = const., (2.105)
2 2

 2
Ψ( t ) = ω t + Φ 0 X . (2.106)

It may be seen from these two expressions that the magnitude of the resultant vector
remains unchanged, but the angle between E -vector and X-Axis (between E and
XOZ-plane) is time-dependent and is increasing linearly. If the observer is facing along
Z-Axis, then E vector will seem to rotate clockwise. This type of polarization is called
right-hand circular polarization, RHCP (Figure 2.12).

Figure 2.12 Structure of the radio wave with right-hand circular polarization
(RHCP)

π
Case 4: Φ 0 X = Φ 0Y − , E mX = E mY = E m . In complex plane the vertical component
2
lags behind the horizontal. In this case (2.100) and (2.101) may be simplified as:

 π
E (t ) = E m cos 2 (ω t + Φ 0 X ) + E m cos 2  ω t + Φ 0 X +  = E m = const ,
2 2
(2.107)
 2
Ψ (t ) = − (ω t + Φ 0 X ) . (2.108)

The magnitude of the resultant vector remains unchanged. However, the angle between
E -vector and the X-Axis (between E and XOZ-plane) is time-dependent and is

63
decreasing linearly, i.e. if the observer is facing along Z-Axis, then the E vector will
seem to rotate counter clockwise. This type of polarization is called left-hand circular
polarization, LHCP (Figure 2.13).

Figure 2.13 Structure of the radio wave with left-hand circular polarization (LHCP)

Case-5 (General case): E mX ≠ E mY , and Φ 0 X ≠ Φ 0Y , or both, amplitudes and initial

phases of the vertical and horizontal components of the resultant radio wave may have
arbitrary values. In this case the resultant radio wave will become elliptically polarized,
i.e. the tip of the resultant E vector will follow along an ellipse in the XOY-plane, either
clockwise or counter clockwise (depending on relationships between Φ 0 X and Φ 0Y initial

phases).
One may therefore conclude that the radio wave with any linear, circular or elliptic
polarization may be decomposed into two cross-perpendicular linearly-polarized radio
waves with the same frequency and propagation direction, and with the proper relations
between amplitudes and initial phases. The opposite statement is also true, i.e. the radio
wave of any linear, circular or elliptic polarization may be decomposed into two mutually
opposite circularly-polarized radio waves of the same frequency and propagation
direction, and with the proper relations between amplitudes and initial phases (see
Figure. 2.14).

64
Figure 2.14. Example of decomposition of linearly polarized radio wave into two
circular, oppositely polarized waves

Detailed mathematical evaluations regarding general cases of polarization may be found


in [1] and [2], as well as in Appendix-2.

2.4. REFLECTION AND REFRACTION OF PLANE RADIO


WAVE FROM THE BOUNDARY OF TWO MEDIA

Consider a radio wave of the plane wave-front that falls onto the flat boundary Y0Z of
two mediums, with the angle of incidence ϕ . The incident ray path is positioned in the
X0Z-plane as shown in Figure 2.15a. The reflected ray path is always positioned in the
same X0Z-plane, and the following two conditions are satisfied at the boundary of
media: phase-matching condition, and boundary condition. The phase-matching
condition is expressed as
β1 sin ϕ = β 2 sinψ , (2.109)
wherein ϕ and ψ represent angles of incidence and refraction respectively, and phase

coefficients β 1 and β 2 are defined by one of the formulas (2.84b), (2.86), or (2.87).
Particularly for the low-loss, or lossless (clear) media, when attenuation is ignorable,
(2.109) may be transformed into

65
ε 1 ⋅ sin ϕ = ε 2 ⋅ sinψ (2.109a)

which is known from optics as Snell's Law, written for the refraction indexes n1 = ε1

and n 2 = ε 2 of both media. Reflection and refraction effects for these types of low-
loss or lossless mediums are consideration below.
-------------------------------------------------------------------------------
Notes:

For a generalized approach, a complex quantity k given by (2.77) is considered as a complex

wave vector k = β − iα , so for the arbitrary oriented distance-vector r expression (2.54) for the

intensity of wave field may be written in generalized form as: E (r ) = E m exp (ω t − k ⋅ r ) . The

dot product in parentheses allows calculation of both, the phase β ⋅ r , and the decay of the
amplitude (i.e. the attenuation) α ⋅r .
It must be noted here, that the real part of the complex wave vector, β is perpendicular to the

phase front at the observation point, and it shows the direction of propagation of the phase front
(unlike the direction of the Poynting vector that shows the direction of canalization of energy).
The imaginary part of the complex wave vector, α shows the direction of the fastest decay of the
amplitude. In isotropic media those vectors support the same direction, i.e. the amplitude decay
takes place in the same direction as a direction of the movement of the phase front.
At the boundary of two media shown in Figure 2.15a the general form of expression (2.109) is

k1 ⋅ zˆ = k z1 = k2 ⋅ zˆ = k z 2 (see Figure 2.15b), (*)

(cont. from previous page)

or k1 sin ϕ = k2 sinψ (**)

which may be shown separately for the real and imaginary parts as

β1 sin ϕ = β 2 sinψ , (2.109)

α 1 sin ϕ = α 2 sinψ . (***)


See [7] for more details.
-----------------------------------------------------------------------------------------

The physical interpretation of the phase-matching condition may be clarified if β 1 and

β 2 in (2.109) are replaced based on (2.72) and (2.86):

66
λ1 λ0 λ2 λ0
= = = , (2.109b)
sin ϕ ε 1 sin ϕ sinψ ε 2 sin ϕ
The meaning of (109b) relation is as follows: as shown in Figure 2.15b on the vertical
XOZ-cut the phase fronts along the interface of two mediums supposed to be "matched",
i.e. points of maximums and minimums supposed to be "stitched" to each-other on that
interface.

Figure 2.15. a). Disposition of incident, reflected and refracted waves at the
boundary of two mediums. b). Vertical pattern of the incident and refracted wave
fronts at the boundary of two media. (1-incident wave, 2-reflected wave, 3-
refracted wave)

The boundary conditions for the tangential components of the total electric, E and total
magnetic, H vectors must also be taken into account. Regarding (2.31a) and (2.32a), in
the absence of conducting currents the tangential components of the electric and
magnetic field strengths are continuous across the boundary (interface) between media,
i.e.
(1) (2 )
Eτ = Eτ , (2.110)
(1) (2 )
Hτ = Hτ , (2.111)

where the superscript in parenthesis indicates the medium under consideration.


Another assumption that is important for further analysis is that the radio wave with the
arbitrary polarization may be decomposed into two waves of cross-orthogonal
polarizations: one with vertical, and the other with horizontal. In other words, the
consideration of reflection and refraction phenomenon at the boundary of two mediums

67
may be limited for just two cases: for vertically polarized (V-pol) and for horizontally
polarized (H-pol) radio waves. For all other cases the incident radio wave must be
decomposed into vertically and horizontally polarized components for the further
separate analysis. In order to retrieve the resultant wave, after the separate analysis is
complete it will be followed by the superposition of the proper components of reflected
(or refracted) waves.

2.4.1. NORMAL INCIDENCE ON A PLANE BOUNDARY OF TWO


MEDIA
In this particular case of normal incidence, the plane wave travels perpendicular to the
flat interface of two ideal dielectric media ( ϕ = 0 ). Part of the total energy of the wave
reflects from the boundary, and the remaining part penetrates (refracts) through the
boundary into the second medium. The relative positioning of electric, magnetic, and
Poynting vectors are interrelated by the right-hand rule (as shown in Figure 2.16) for all
three waves. Conventionally, the ratio of the reflected electric field strength and the total
field is called the electric field reflection coefficient, which generically is represented as a
complex number:

E −i⋅Φ
Γ E = 1 = Γ E ⋅ e Γ E , (2.112)

Ein

where E1 and E in are electric fields of reflected and incident waves respectively.

Similarly a magnetic field reflection coefficient is defined as

H −i⋅Φ
Γ H = 1 = Γ H ⋅ e Γ H . (2.112a)

H in
The magnitudes of both reflection coefficients are always less than one, which means
that the intensity of reflected wave is always less than the intensity of the incident wave.
This statement is called reflection intensity loss. Φ Γ , E and Φ Γ , H are reflection phases

for the vertical and horizontal polarizations respectively. They represent the phase shift
between incident and reflected waves. The negative sign in the exponents of (2.112) and
(2.122a) denotes the fact that the initial phase of the reflected wave always lags behind
the initial phase of the incident wave. Therefore the reflection phase is called sometimes

68
a reflection phase loss. The phase of the reflected wave must never pass ahead of the
phase of the incident wave, as it would contradict the cause-effect principle.

Figure 2.16 The pattern of a normally incident wave onto the


boundary of two media

For this particular case the boundary conditions (2.110) and (2.111) may be written as:
E in + E 1 = E 2 ,
(2.110a)
H in − H 1 = H 2 , (2.111a)
and based on (2.66) the expression (2.111a) may be transformed into

( E in − E 1 ) ε 1 = E 2 ε2 . (2.111b)

By solving the system of equations (2.110a) and (2.111b) for the electric field reflection
coefficient (2.112) for the normally incident wave, one may obtain:

ε1 − ε 2
Γ E norm = . (2.113)
ε1 + ε 2

The reflection coefficient Γ E norm becomes a real number, either positive (if ε 1 > ε 2 ), or

negative (if ε 1 < ε 2 ). Real positive Γ E norm shows the decrease of the amplitude of the
reflected wave relative to the incident, while keeping the initial phase unchanged
( Φ Γ = 0 ). Real negative Γ E norm shows the decrease of the amplitude and inversion of

the initial phase ( Φ Γ = 180 0 ).

69
2.4.2. OBLIQUE INCIDENCE OF VERTICALLY POLARIZED
RADIO WAVE

In this case electric field vectors of all of three rays ( Ein for the incident ray, E1 for the

reflected ray, E 2 for the refracted ray) will lie in the same vertical plane (Figure 2.17).
The cross positioning of all three vectors are chosen in such a way that for the special
case of ϕ = 0 the pattern will transform to that depicted in Figure 2.16.

Figure 2.17. Positions of vectors for the vertically polarized electromagnetic wave
of oblique Incidence to the flat reflection boundary

Then the boundary conditions (2.110) and (2.111) in this case may be written as
(1) (2 )
E τ = E in ,τ + E 1,τ = E in cos ϕ + E 1 cos ϕ = E 2 cosψ = E τ , (2.114)
(1) (2 )
H τ = H in − H 1 = H 2 = H τ . (2.115)

For further evaluations we must define cosψ , which is found from (2.109b) as follows:

ε1
cosψ = 1− sin 2 ϕ . (2.116)
ε2
Now we may rewrite (2.114) and (2.115) with relations (2.73), (2.80), and (2.116) in
mind.

70
ε1
(E in + E1 ) cos ϕ = E 2 1−
ε2
sin 2 ϕ , (2.117)

(E in − E 1 ) ε 1 = E 2 ε 2 . (2.118)

If these two equations are solved for reflection coefficient (2.112), then the result for the
considering case of vertical polarization becomes:

Γ E vert = − ε 2 cos ϕ − ε 1 ε 2 − ε 1 sin ϕ .


2

(2.119)
ε 2 cos ϕ + ε 1 ε 2 − ε 1 sin 2 ϕ
Below we consider two particular cases of reflection of the vertically polarized waves.

2.4.2.a. RADIO WAVE INCIDENT FROM SPARSE MEDIUM ONTO THE


BORDER WITH DENSE MEDIUM : (ε 1 < ε 2 )

For the small angles of incidence ( ϕ ) the value of Γ


 E vert in (2.119) is a real, negative

number. The increase of angle of incidence ϕ from zero results in decrease of the
 E vert | , which, as may be seen from (2.119) by taking numerator to
magnitude Γ E vert = | Γ
zero, tends to zero for the angle of incidence defined as

ε2
sin ϕ 0 = . (2.120)
ε1 + ε 2

There is no occurrence of reflection at the angle ϕ 0 , thus it is called the angle of total

refraction or Brewster's angle. Further increase of ϕ results in real positive values of

Γ E vert within the range of the angles ϕ 0 > ϕ > 0 . Hence the reflection phase Φ Γ E , vert

jumps from 1800 to 00 at ϕ 0 . The patterns of the angular dependencies of the magnitude

and phase of reflection coefficient for this particular case are shown in Figure 2.19
(dotted lines).

71
2.4.2.b. RADIO WAVE INCIDENT FROM DENSE MEDIUM ONTO THE
BORDER WITH SPARSE MEDIUM (ε 1 > ε 2 )

If ϕ = 0 in this particular case, then the reflection coefficient is positive, therefore the
reflection phase is equal to zero. The increase of ϕ from zero will result in same total

reflection phenomenon at the same Brewster's angle ϕ 0 , defined by (2.120), as with the

previous case. When the angle of incidence passes through the value of ϕ 0 , then Γ
 E vert

turns the sign from positive to negative, i.e. the reflection coefficient jumps from zero to
1800. Further increase of the angle of incidence ( ϕ > ϕ 0 ) will result in another

phenomenon that occurs at so called critical angle, ϕ = ϕ cr , when the expression under

the square root in (2.119) becomes equal to zero. Then ϕ cr is defined as

ε2
sin ϕ cr = . (2.121)
ε1
For the values of angle of incidence ϕ > ϕ cr the expression under the square root sign

becomes negative, thus the numerator and denominator of (2.119) become complex
conjugate relative to each-other. Therefore, for all ϕ > ϕ cr the magnitude of Γ
 vert will

remain equal to unity, whereas the reflection phase Φ Γ E , vert will smoothly decrease from

1800 to zero degrees as shown in Figure 2.20 (dotted lines). Hence ϕ cr is called the total

reflection angle.

72
2.4.3. OBLIQUE INCIDENCE OF HORIZONTALLY POLARIZED
RADIO WAVE
In this case electric field vectors of all of three rays lie in the same horizontal plane as
shown in Figure 2.18.

Figure 2.18. Positions of vectors for the horizontally polarized electromagnetic


wave of oblique incidence to the flat reflection boundary

The boundary conditions (2.110) and (2.111), may be rewritten in the following form:
H in cos ϕ − H 1 cos ϕ = H 2 cosψ , (2.122)

E in + E 1 = E 2 . (2.123)

Simplifications similar to those provided for the vertically-polarized incident waves are
applicable to (2.122) and (2.123) as well:
• ψ may be expressed in terms of ϕ using Snell's law (2.109)
• Replace magnetic field strengths by electric, using (2.66)
• After combining (2.122) and (2.123) the ratio (2.112) for electric field reflection
coefficient may be found as

ε 1 cos ϕ − ε 2 − ε 1 sin 2 ϕ
Γ E horiz = . (2.124)
ε 1 cos ϕ + ε 2 − ε 1 sin 2 ϕ
From (2.124) it may be seen that unlike the previous case of the vertical polarization, the
horizontally polarized radio wave can never satisfy the total refraction condition, i.e. the

73
horizontally polarized radio wave will never be totally refracted (penetrate) into the
second medium. Below we consider two particular cases of reflection of the horizontally
polarized waves.

2.4.3.a. RADIO WAVE INCIDENT FROM SPARSE MEDIUM ONTO THE


BORDER WITH DENSE MEDIUM (ε 1 < ε 2 )

For all angles of incidence ϕ the value of Γ


 E horiz represents a real and negative fraction,

which urges to minus one, when ϕ becomes close to 90o, i.e. the magnitude of Γ
 E horiz

smoothly increases, while the reflection phase remains constant and equal to minus
1800 for all values of angle of incidence, ϕ as shown in Figure 2.19 (solid lines).

2.4.3.b. RADIO WAVE INCIDENT FROM DENSE MEDIUM ONTO THE


BORDER WITH SPARSE MEDIUM (ε 1 > ε 2 )
In this case the total reflection occurs at the same condition (2.121). Hence, for the
angles of incidence between zero and ϕ cr , the reflection coefficient Γ
 E horiz remains a

positive fraction and tends to one when ϕ becomes close to ϕ cr . The magnitude of

Γ E horiz tends smoothly to one, while the reflection phase remains equal to zero. For the
values of ϕ between ϕ cr and 90o the magnitude of Γ
 E horiz remains equal to one, while

the value of reflection phase increases smoothly from zero to 1800 as shown in Figure
2.20 (solid lines). Note, that both expressions (2.119) and (2.124) will convert to (2.113)
regardless of sense of polarization if the normal incidence ( ϕ = 0 ) is considered.

74
Figure 2.19. Magnitude and angle of the electric field reflection
coefficients for the boundary of two ideal dielectrics with parameters
ε 1 = 1, ε 2 = 3 .

Figure 2.20. Magnitude and angle of the electric field reflection


coefficients for the boundary of two ideal dielectrics with parameters
ε 1 = 3, ε 2 = 1 .

75
2.4.4. REFLECTION OF THE RADIO WAVE WITH ARBITRARY
POLARIZATION

As was mentioned earlier in this text, any arbitrary polarized radio wave may be
represented as a superposition of two linearly-polarized radio waves: vertically polarized
and horizontally polarized. Each of those two radio waves satisfies their own reflection
properties. Therefore, after reflection at the same angle, the superimposition of those
may result in a radio wave that demonstrates the new polarization properties. If, for
instance, the initial linearly polarized radio wave has a arbitrary polarization angle with
the vertical XOZ-plane (neither zero, nor 90 0), then, before the reflection, it may be
decomposed into two linearly polarized waves (vertically and horizontally polarized) with
properly chosen amplitude-phase relations between them. If after reflection the phase
shift between components becomes Φ vert − Φ horiz = π / 2 , then it may result either in

circular or elliptic polarization reliant on the relationship between their amplitudes. This
phenomenon is utilized in optics to construct the polarization converter.
Additionally, the separation of cross-polarized components may also be achieved.
It is used in optics to separate the cross polarized components of the non-polarized
radiation. To accomplish that, the transparent slab is placed with the proper positioning
of the surface, relative to the direction of primary ray-trace (see Figure 2.21).

Figure 2.21 Spatial decomposition of the non-polarized (or arbitrary


polarized) optical radiation onto two cross-polarized components.

76
The angle of incidence is chosen to be equivalent to Brewster angle ϕ 0 , thus the

vertically-polarized component penetrates completely through the slab, while the


horizontally-polarized component reflects from the slab's surface alone.

2.4.5. POWER REFLECTION AND TRANSMISSION

Electric field reflection coefficients for the vertically and horizontally polarized radio
waves incident to the border of two dielectric mediums are given by (2.119) and (2.124).
Magnetic field reflection coefficients may be derived similarly, based on the systems of
equations represented by (2.117) – (2.118), and (2.122) – (2.123). If in those equations

electric fields are replaced by magnetic fields using W1 / 2 = W0 / ε 1 / 2 (see (2.73)), and
then further evaluate (2.112a) for both polarizations, then it may be shown simply that
Γ H = −Γ E . (2.125)
For the transmission coefficients the definitions are based on the following expressions
for the electric and magnetic fields respectively:

E −i⋅Φ
T E = 2 = T E ⋅ e T E . (2.126)

Ein

H −i⋅Φ
T H = 2 = T H ⋅ e T H . (2.126a)

H in
 E , Γ H , T E , and T H are provided in Table 2.2.
Evaluation results for Γ

To best evaluate reflection and transmission of the power of radio wave, we have to
keep in mind that the power flow must be considered across the boundary in a direction
normal to the border. In other words, the normal components of the incident, reflected,

and transmitted (refracted) waves ( Π in , Π refl , and Π trans


Norm Norm Norm
) shown in Figure

2.22, must be taken into account.

77
Figure 2.22 Power flow directions for the oblique incidence of the radio wave onto
the border of dielectric media

Power flow density at the border of loss-less media may be defined from (2.81). So for
the origin of the coordinate system that is placed on the border ( z = 0) the power flow is

defined as Π = E 2 / 2W , and an expression for the power balance right at the interface
may be written as
Π N in − Π N refl = Π N trans . (2.127)
Then (2.127) is rewritten as:
2 2 2
Ein E E
cos ϕ − 1 cos ϕ = 2 cosψ , (2.128)
2 W1 2 W1 2 W2
where W1 and W2 indicate the intrinsic impedances of first and second media
respectively.
Now we may define the power reflection and transmission coefficients as:

Π N refl
2
E 2
Γ = N = 1 2 = Γ E
P
, (2.129)
Π in Ein

Π N trans E
2
ε 2 cosψ 2 ε 2 − ε 1 sin 2 ϕ
T =
P
= 22 = TE . (2.130)
Π in ε 1 cos ϕ ε 1 cos ϕ
N
Ein

78
Hence, taking into account (2.129) and (2.130), the expression (2.128) may be rewritten
as:
1 − ΓP = T P . (2.131)

Note that (2.131) represents the power balance that may be easily verified by direct
substitution of Γ E and T E for either vertical or horizontal polarizations into (2.129),
(2.130), and further into (2.131).
Table 2.2 summarizes analytical expressions that relate to the cases above.
Table 2.2

Γ E vert = − ε 2 cos ϕ − ε 1 ε 2 − ε 1 sin ϕ


2
Reflection
coefficient ε 2 cos ϕ + ε 1 ε 2 − ε 1 sin 2 ϕ
Electric
Vertical polarization

Transmission 2 ε 1 ε 2 cos ϕ
T E vert =
coefficient ε 2 cos ϕ + ε 1 ε 2 − ε 1 sin 2 ϕ

ε cos ϕ − ε 1 ε 2 − ε 1 sin 2 ϕ
For Electric and Magnetic Fields

Reflection
Γ H vert = 2
ε 2 cos ϕ + ε 1 ε 2 − ε 1 sin 2 ϕ
Magnetic

coefficient

Transmission 2 ε 2 cos ϕ
T H vert =
coefficient ε 2 cos ϕ + ε 1 ε 2 − ε 1 sin 2 ϕ

Reflection ε 1 cos ϕ − ε 2 − ε 1 sin 2 ϕ


Γ E horiz =
coefficient ε 1 cos ϕ + ε 2 − ε 1 sin 2 ϕ
Electric
Horizontal polarization

Transmission 2 ε 1 cos ϕ
T E horiz =
coefficient ε 1 cos ϕ + ε 2 − ε 1 sin 2 ϕ

Reflection ε cos ϕ − ε 2 − ε 1 sin 2 ϕ


Γ E horiz = − 1
ε 1 cos ϕ + ε 2 − ε 1 sin 2 ϕ
Magnetic

coefficient

Transmission 2 ε 2 cos ϕ
T H horiz =
coefficient ε 1 cos ϕ + ε 2 − ε 1 sin 2 ϕ

For Power Reflection and Reflection 2


Γ P = Γ E
coefficient
Transmission
for either vertical or horizontal Transmission 2 ε 2 − ε 1 sin 2 ϕ
T P = T E
polarizations coefficient ε 1 cos ϕ

79
2.4.6. REFLECTION OF THE RADIO WAVE FROM THE
BOUNDARY OF NON-IDEAL DIELECTRIC MEDIUM

This case is specific for radio waves propagation conditions, as for the RF frequencies
the soil, fresh and sea water, as well as the ionosphere behave as semiconductors;
therefore in (2.119) and (2.124) either both, ε1 and , ε2 or at least one of them must be
considered as a complex number, defined by (2.45).
It may be seen, that for non-ideal dielectric boundary a phenomenon such as total
reflection or total refraction will not purely appear, due to the presence of the power
losses. An example of the angular dependence of the reflection coefficient for the real
conditions is shown in Figure 2.23, obtained from (2.119) by using the MATLAB
subroutine.

Figure 2.23. The angular dependence of magnitude and phase of reflection


coefficient for the reflection of vertically polarized radio wave from soil with
the following parameters: ε = 3 , σ = 0.01 , λ = 5 m .

80
2.5. RADIATION FROM INFINITESIMAL ELECTRIC
CURRENT SOURCE. SPHERICAL WAVES

For the spatial area that contains time-harmonic source, the first two Maxwell's
equations may be written as

∇ × H = iωε 0 ε ⋅ E + J Ext , (2.132)

∇ × E = −iωµ 0 µ ⋅ H , (2.133)

where electric current density, J Ext is induced by the external source, and is distributed

within a limited volume V. Regarding Poynting theorem (2.15) the energy of the source is
spent not only to increase the energy stored by electric and magnetic fields, and
dissipated in form of a loss within the area V, but also generates an outgoing
electromagnetic radiation. This radiation may exist if last integral in (2.15) is non-zero 1.
Solution of the systems (2.132) – (2.133) for the radiated field may not be obtained
directly, meaning if we try to turn the two first order linear differential equations for two
unknowns into a single differential equation of the second order for one unknown, then
generically it becomes unsolvable. The main idea of the commonly used indirect
approach for obtaining the result is to introduce “so called” auxiliary magnetic potentials,

i.e. vector potential A , and scalar potential Φ 2. Introduction of these auxiliary functions
allows obtainment solutions for the fair number of the radiation problem. As an example,
the radiation from the linear electric current of the infinitesimal size is considered in
Appendix-3. Results of the solution are presented by expressions (A3.22) and (A3.25).

∫ (E × H ) d S
1  ~
The integral may become equal to zero because of boundary conditions on the
S

closed surface S. If, for instance, the volume V is surrounded by the PEC, then regarding (2.36)
the tangential component of the electric field becomes equal to zero, which results in zero value
of the above integral.
2
Not to be confused with phase shift. For the phase shift notation we always use Φ along with a
subscript.

81
If the time-harmonic multiplier exp (iω t ) is included into those analytical results, then
(A3.22) and (A3.25) may be rewritten as follows:
• For the magnetic field
I l  exp[i (ω t − k r )]
H = ik sin θ ⋅ ϕ 0 (2.134)
4π r
• For the electric field:
Il exp[i (ω t − k r )]
E = θ 0 iω µ 0 µ sin θ (2.135)
4π r
In those two above expressions we assume that the electric current filament, I with
uniform current distribution is placed symmetrically at the origin of the spherical
coordinates, and directed along Z-Axis as shown in Figure 2.24.

Figure 2.24. Hertzian dipole in spherical coordinates

The length of the current filament is assumed to be infinitesimal, i.e. l << r and l << λ .
This type of radiator is known as ideal dipole, or Hertzian dipole. The following
statements can be made based on (2.134) and (2.135):
• Unlike the scalar point source, the intensity of radiation from the electromagnetic
point source such as a Hertzian dipole is non-isotropic, i.e. the radiation is
direction-dependent due to sin θ factor. Generally speaking there are no
electromagnetic radiation sources in nature that contain isotropic (direction-
independent) radiation. Even in this simplest case of electric current point source,
the radiation becomes non-uniform.

• The real part of the propagation constant, namely the phase constant β = Re(k)

defines a phase distribution in time and space, ϕ = ω t − β r . Similar to case of

82
the uniform plane waves, for the fixed time-instance t = const spatial distribution
of the constant phase are presented by the r = const surfaces, which are the
equations of the source-centered spheres. Thereby, the wave fronts for the point
source are synonymous to the spherical surfaces radially expanding from the
source location. Following the procedure same as for the plane waves, it’s easy
to realize, that the propagation velocity becomes equal v = ω / β , i.e. showing
the same as for the plane electromagnetic waves. Further, expression 2.69 for
the wavelength is also applicable to the spherical waves.
• The value for the ratio of electric and magnetic fields (intrinsic impedance), is the
same as for the plane wave, and may be defined by dividing (2.135) to (2.134),

E ω µ 0 µ µ0 120 π 377
W = = = = ≈ , Ohm. (2.136)
H k ε 0 ε ε ε
The result is coordinate-independent, and is specific to the particular propagation
medium.

2.6. SPATIAL AREA SIGNIFICANT FOR RADIO WAVES


PROPAGATION

2.6.1. PRINCIPLE OF HUYGENS-KIRCHHOFF

An important role in radio wave propagation theory and applications plays by virtue of
the principle of Huyhens-Kirchhoff, which was initially formulated in optics [2].
Formulation of this principle played a crucial role in further advancements not just in
optics, but in electromagnetic waves of other frequency ranges, such as RF. It allows the
provision of a quantitative description of the diffraction effects, including estimates of the
field intensity in so called shadow areas that may not be reached by the direct wave.
Assume that there exists an electromagnetic radiation source (an antenna, for
instance) that occupies small volume V0, which is located within a closed surface S. That
surface surrounds the larger volume V that includes V0 as shown in Figure 2.25.

83
Figure 2.25. Illustration of Huygens-Kirchhoff's principle

Regarding Huygens-Kirchhoff principle, the field strength at any point B outside of the
volume V, may be considered as a superposition of secondary waves, radiated from
virtual point-sources, located on the S surface. Each surface element (C, C1, C2, etc.) is
known as a "Huygens source", and contributes to the formation of the electromagnetic
field of the radio wave at the observation point B; i.e. each of those points appears as a
virtual point source of the secondary radiation due to the primary source (antenna)
located in source region V0. In other words, if the distribution of the field of primary
source along S-surface is known, then the field strength at observation point B may be
found as a superposition of secondary, virtual fields (wavelets), emanating from each
surface element located on surface S.
To properly evaluate the field strength in observation B point, the following statements
must be taken into account:
• The field strength in B point caused by each elementary (virtual) radiator located
on closed surface S is proportional to the surface element dS at the point (points
C, C1, C2, …, shown in Figure 2.25) on surface S, and also proportional to the
value of the field in that particular point of the surface
• Considered as a point source of spherical wave-front, each surface element has a
radiation pattern with the maximum radiation along the unit vector n that is
perpendicular to dS surface element. In other words, the field strength in B point is
proportional to the cosine between unit vectors n and r0 that is directed from point

84
C to the observation point B (Figure 2.25). It must be stated for more accuracy, that
here we are discussing the radiating element known as " Huygens Source", which
represents an infinitesimal surface element dS with cross orthogonal electric and
magnetic fields uniformly distributed along that surface element. The spatial
distribution of the radiation from that source is in direct proportion to
[1 + cos(n ^ r0 )]/ 2 (see [4] for reference).

Taking the above statements into account, the analytical expression for the scalar value
of the field strength in observation point B, contributed by single surface-element, dS
may be written in the following form:

exp[ i (ω t − k r )] 1 + cos ( n ^ r0 ) 
dE B = K EC   dS , (2.137)
r  2 
Where:
• EC represents a scalar value of field strength in point C due the primary radiator
• r is a distance from C to B
• K illustrates a constant of proportionality
• n and r0 unit vectors are shown in Figure 2.25.
As mentioned above, the resultant field in B point is due to the contribution from all
elements of surface S (considered as virtual sources of wavelets) and may be found by
taking a sum of their fields. For a strict approach one has to take into account, that
generally speaking, the expression (2.137) nominates a vector quantity, so the vector
summation is to be applied. However, for engineering applications enough precision may
be achieved if the algebraic sum is taken instead, particularly if:
• The volume V0 is much smaller than the volume V , thus the primary radiation
source may be considered as a point source
• The distance between each point of S surface and observation point B is much
greater than the linear dimensions of the surface S.

It is thereby assumed, that all elementary virtual field vectors dE B have the same
directions at the reception point B, in which case the algebraic summation (integration)
may be used as follows:

1 + cos (n ^ r0 ) 
− ikr
E B = K ⋅ e iωt ⋅ ∫ E C (r )
e
r   ⋅ dS . (2.138)
S
2 

85
Here the integration must be performed along the entire S surface, including the areas
that are not "visible" from B point.

2.6.2. FRESNEL ZONES

Now we evaluate the value of E B form (2.138), assuming for simplicity, that V0 → 0 , i.e.

there is a primary harmonic point source placed in the A point, which generates radiation
with the spherical wave fronts (see Figure 2.26). For simplification of further analysis,
assume that S is a combination of plane surface and spherical surface with the radius
that tends to infinity as shown in Figure 2.26.

Figure 2.26. Illustration of (2.138) integral evaluation.

The resultant field induced by the source A placed in the center of the spherical part of
surface S is vanishing to zero if the radius of sphere tends to infinity, so the integration in
(2.138) may be performed only along the flat part of S. Without limitations, and due to
the generality of the approaches, we may consider an ideal electric dipole as a primary
point source placed vertically in point A. Then, regarding (2.135) the field strength in C
point may be written as

Il exp[i (ω t − k r1 )]
E C (r1 ) = iω µ 0 µ cos β1 (2.139)
4π r1

86
where β 1 = π / 2 − θ . If (2.139) is substituted into (2.138), then the following expression
for the integrant may be written as

e − ik (r1 + r2 ) 1 + cos β 2 
dE B = K ⋅ e iωt cos β1   ⋅ dS , (2.140)
r1 ⋅ r2  2 
where
Il
K= iω µ 0 µ (2.141)

is a constant, and the angle β 2 = n ^ r0 is the angle between unit vectors n and r0 , as

shown in Figure 2.25. We also assume that the flat surface S is placed perpendicular to
AB line at the distances r01 and r02 from points A and B respectively, as shown in Figure

2.27a. On the surface S we take an infinitely thin ring (the ringlet) of the radius R and the
area dS . The area of the ringlet may be subdivided equally into smaller, surface
elements dS ′ , dS ′′ , etc. as shown in Figure 2.27b. It’s apparent that in observation point
B each of those elements will produce field intensity of the same amplitude
cos β1 1 + cos(n ^ r0 ) 
K
r1 r2   dS ′ (2.142)
2 
and the same phase, Φ = k (r1 + r2 ) due to the symmetry relative to AB line.

Figure 2.27. The picture of ray traces from point source A to observation point B.
The point source is placed behind the flat diaphragm for the observer in point B

87
Summation of elementary fields from all elements dS ′ , dS ′′ , …. will result in (2.140) as
an elementary field from the whole ringlet of the total area dS = dS ′ + dS ′′ + ... . The
amplitude of the elementary field contributed by the whole tiny ring is the same as
(2.142) if dS ′ is replaced by dS .
Now, assume that S is an opaque infinitely thin flat surface (shield) placed perpendicular
to the AB line with the circular open aperture (diaphragm) of variable radius R, centered
on the AB line. If we gradually open the diaphragm (i.e. gradually increase the radius
from zero) then it will consequently involve more and more ringlets. By keeping the same
dS areas constant from ring to ring, we will force more and more virtual (secondary)
fields to be involved in the summation process at the observation point B. While the
amplitudes of virtual fields are nearly the same 1 from ring to ring, the phases are shifted
due to difference in distances, e.g. ∆Φ = β ∆r = β [∆r1 + ∆r2 ] = β [(r1′′− r1′) + (r2′′ − r2′ )]
between the rays passing through two adjacent rings, as shown in Figure 2.27c. The
summation process of the sequence of elementary fields (wavelets) dE B1 , dE B 2 , dE B 3 ,

… in complex plane is shown in Figure 2.28a.

1
Note that changes in the amplitudes of the elementary fields are much slower than changes in

phases.

88
Figure 2.28. a). Summation of partial fields in complex plane. b). Dependence of
the resultant field strength (amplitude) on the radius of diaphragm. c).The pattern
of the Fresnel zones

As an example, the resultant scalar field at the observation point, E B (R ) is shown in


that figure for seven (which is an arbitrary number) of those tiny ringlets being open. On
the same figure Φ (R) nominates the cumulative phase shifts of the resultant complex
vector. One may note, that while increasing the radius of the aperture, the tip of the
resultant vector (point M) moves along the spiral curve, but not along the circle as it may
seem, because of the slow decrease in amplitudes of wavelets composing the resultant
field. That slow decrease in the amplitudes may be seen from (2.142) due to changes in

r1 , r2 , and β1 , β 2 . Figure 2.28b represents the dependence of the magnitude E B = E B

of the resultant field strength on the radius of the aperture of the diaphragm. The deep
variations of E B are due on change of the radius of the diaphragm. If the radius
increases form zero, then the field intensity at the observation point B will reach its
maximum for R = R1 , when the difference in distances ACB and AOB (Figure 2.27a)
becomes equal to π . For that radius of diaphragm, the resultant field becomes equal to

89
E B , 1 . Further increase of R will result in decrease of field intensity, so that for the radius

of diaphragm R = R2 the first minimum of E B will be achieved when the difference in

distances ACB and AOB becomes equal to 2π , and so on. The n-th extreme value
of E B (maximum or minimum) will appear if the radius of diaphragm becomes equal to

one of its discrete values Rn so the difference in ACB and AOB distances becomes

equal to the integer number on π -s. Namely, the maximums appear for the odd number
of π -s, and minimums – for even number of π -s. Based on above considerations the
following expression may be written for the extreme values (maximums/minimums) of
the field magnitude at the observation point
k [ (r1 + r2 ) − (r01 + r02 )] = nπ (2.143)


where k= , and n = 1, 2 , 3 , ......
λ
The proper value for the radius Rn is calculated below. As seen in Figure 2.27a, the sum

of distances may be defined as

r1 + r2 = Rn + r01 + Rn + r02 .
2 2 2 2
(2.144)

If we take into account the relations Rn / r01 << 1, Rn / r02 << 1 , then (2.144) may be
approximated by using Taylor's series as follows:
2 2
R R
r1 + r2 ≈ r01 + n + r02 + n . (2.145)
2 ⋅ r01 2 ⋅ r02
Now, if (2.145) is substituted into (2.143), then the final result may be found as:

n λ r01 r02
Rn = . (2.146)
r01 + r02
The above consideration allows outlining of several zones on the surface S, each of
them relative to the proper value of Rn : the first one is a circle with the radius R1 , all

others are ring-shaped and of finite dimensions, thereby the n-th zone has the outer
radius Rn and the width ∆R = Rn − Rn −1 . These zones are called Fresnel zones, and, as

one may realize, each zone has its own contribution, E B1 , E B 2 , E B 3 , E B 4 .... into the total

field at the observation point B (see Figure 2.28b). These fields are generated by the
proper zone, and are 1800 out of phase one relative to the other. This fact is represented
in Figure 2.28c via "+" and "-" signs. In other words, the partial waves that are coming to

90
the observation point B from the odd-numbered zones are opposite in phase relative to
those coming from the even-numbered zones; i.e. the even-numbered zones interfere
destructively with the odd-numbered zones at the observation point. If the radius of
diaphragm R is assumed to be infinitely large (i.e. the diaphragm is removed
completely), then we end up with the free-space propagation case. In that case the total
field strength may be represented as:
E B 0 = E B1 − E B 2 + E B 3 − E B 4 + ...... =

E B1  E B1 E  E E 
= + − E B 2 + B 3  +  B 3 − E B 4 + B 5  + .... (2.147)
2  2 3   2 2 
It is therefore evident, that each triple-termed bracket in (2.147) is nearly equal to zero,
and thereby the total field strength at the point B for free propagating conditions
becomes equal to approximately:
E B1
EB0 ≈ . (2.148)
2
Expression (2.148) emphasizes the fact, that the presence of the shielding diaphragm
with the properly set radius of aperture ( R = R 1 ) may even cause the "amplification"

(doubling) of the field at the reception point relative to free space propagation conditions.
This statement may seem surprising. However the reasonable explanation for this
phenomenon is as follows: If only the first Fresnel's zone is left open, then all
destructively affecting spatial areas are shielded by an impenetrable screen, leaving
open just one single spatial area, which affects positively the wave transmission. More
detailed calculations and real observations show that the oscillations of the field intensity
diminish if the radius of aperture of the diaphragm is larger than R5 or R6 (see Figure

2.28b). Further increase in radius R results in nearly steady value of field equal to E B 0

that is specific for the free space, i.e. the oscillations become ignorable.
If two-dimensional shapes of the Fresnel zones are circular rings (the first zone is a
circle), then the 3D-shapes of the Fresnel zones may be found based on (2.143) in the
following form:
nπ − k (r01 + r02 )
r1 + r2 = = const . (2.149)
k
This is the equation of the ellipsoid with the focuses in A and B points. Based on (2.149)
we may conclude that between two corresponding points in the space (i.e. radiation and

91
reception points) only part of the free space, limited by several Fresnel zones, is
effectively involved in the radio wave propagation process. Each Fresnel zone has a
shape of prolonged ellipsoid, as shown in Figure 2.29, with the focuses located at
radiation and reception points, and with the radius of cross-section, positioned between
those two points, defined by expression (2.146).

Figure 2.29. Solid shapes of the Fresnel zones.

From this conclusion, it becomes clear that the term "ray" in optics and in radio waves
propagation theory does not mean an infinitely thin line between corresponding points;
but instead a spatial area that has a finite volume (sometimes significant). It is also
apparent that along the propagation path only those objects must be considered as
disturbing obstacles if they are positioned within that volume. For RF frequencies, when
the wavelength λ is much larger than that for the optical radiations, regarding (2.146),
the cross-sectional dimension of the “ray” becomes much larger.
Now we' will define a minimal cross-sectional dimension of the effectively contributing
area by assuming that the field intensity at the observation point B is the same as in free
space. From Figures 2.28a and 2.28b it may be seen that the minimum radius of
diaphragm R0 , which results in field strength same as in free space, may be found if the

tangential to the spiral line at the point M (Figure 2.28a) has a slope angle of 600, thus
the triangle 010M becomes equilateral. This results in the phase shift of Φ ( R ) = π / 3

between ACB and AOB distances illustrated in Figure 2.27a. Therefore, assigning R0 is

92
the radius of the elliptic zone to be defined, we'll let the right hand side of (2.143) be
equal to π / 3 . Then combining this with (2.145) we may obtain:

2π R0 r01 + r02
2

= π / 3, (2.150)
λ 2 r01 r02

λ r01 r02 R1
R0 = = , (2.151)
3 r01 + r02 3

where R1 is the radius of the first Fresnel zone. This expression conveys the conclusion
that in order to have undisturbed transmission of the radio wave between corresponding
points, at least 60% of the radius of first Fresnel zone along the propagation path must
be free of obstacles.

2.6.3. KNIFE-EDGE DIFFRACTION

Another mechanism that is vital for radio links design is diffraction on the knife-edge type
obstacle, shown in Figure 2.30.

Figure 2.30. Sketch for evaluation of the knife-edge diffraction integral (2.153)

In this case, the integration of the expression (2.140) will more conveniently be
performed in Cartesian coordinates. x0 is a positive height of the knife edge above the

93
horizontal line x = 0 . It becomes negative if the top of the edge is below the point of
origin, i.e. below the Line-Of-Site (LOS) between communicating points.

If (x, y, 0) is any arbitrary point on flat S-surface (S-plane), then approximations similar to

(2.145) may be implemented. Notifying ρ = x 2 + y 2 a distance to any point on the


XOY-plane with the coordinates x and y , then

ρ2 x2 + y2 
r1 ≈ r01 + = r01 + 
2r01 2r01 
, (2.152)
ρ2 x + y2
2

r2 ≈ r02 + = r02 +
2r02 2r02 
The integration range in the expression (2.140) along plane S surface is assumed to be
much smaller than distances, i.e. ρ << r01 , ρ << r02 . Therefore changes in magnitude
of dE B due to differences between distances r1 and r01 , as well as between r2 and r02

are ignorable. However, they may not be ignored in phase exponents. Therefore
quadratic terms in (2.152) are kept below for phase exponents only. Additionally, we
must take into account that the elevation angles of observation from A to C and B to C
points (angles β 1 and β 2 in (1.140)) are small, hence the values of cosines in proper
terms may approximately be assumed equal to unity. Based on these statements and
1
assumptions the integration of the expression (2.140) may be represented as

e − ik (r01 + r02 )  ik  1 1  2 

EB = K
r01 ⋅ r02 ∫∫S exp  − 
 + 
 x + (
y 2
)
 dx dy . (2.153)
 2  r01 r02  
For further evaluation of the double integral of (2.153) we will follow the traditional
approaches given in [3]. The new variables are introduced as:

 1 1  2 
k  +  x = π u 2 
 r01 r02  
, (2.154)
 1 1  2 
k  +  y = π v 2
 r01 r02  

1
The time-harmonic term e iωt is omitted for simplicity.
94
x2 
= u2 
m 
or , (2.155)
y2

= v2 
m 
λ 1
where: m= ⋅ . (2.156)
2 1 / r01 + 1 / r02
From the system (2.155):
dx dy = m ⋅ du dv . (2.157)

Then (2.153) may be rewritten based on Euler’s Identity exp(iα ) = cos α + i sin α :

E B = B (C + i S ) , (2.158)

exp[− ik (r01 + r02 )]


where: B=Km . (2.159)
r01 r02
The surface integral functions in (2.158) are:
π
(
C = ∫∫ cos  u 2 + v 2 ) du dv , (2.160)
S′ 2 

π
(
S = ∫∫ sin  u 2 + v 2 ) du dv . (2.161)
S′ 2 
The integration limits here may be defined based on Figure 2.30, as well as on the
expressions (2.155) for range of changes of variables. Those limits are: u 0 < u < ∞ , and

− ∞ < v < ∞ . The functions (2.160) and (2.161) may be expressed via special functions
called Fresnel integral cosine and Fresnel integral sine respectively, represented as

π 
C F ( u 0 ) = ∫ cos  t 2  dt ,
u0
(2.162)
0
2 
π 
S F ( u 0 ) = ∫ sin  t 2  dt .
u0
(2.163)
0
2 
These functions are widely tabulated (see for example [6]), and also included within
common software applications such as the MATLAB toolbox. To avoid unnecessary
mathematical evaluations that do not contribute to the main text, these special functions
are evaluated in details in Appendix-4. The results of those evaluations are presented by
(A4.12) and (A4.13) as follows:
C ( u0 ) = S F ( u0 ) − C F ( u0 ) , (2.164)

95
S ( u 0 ) = 1 − [ S F ( u 0 ) + C F ( u 0 )] . (2.165)

If these expressions are substituted into (2.158), then it may be rewritten as:
E B = B {( S F − C F ) + i [1 − ( S F + C F )]} . (2.166)

From Figure 2.30 we will note that the free space condition is achieved for x0 = −∞ , and

thereby u 0 = −∞ . For this value of argument, Fresnel's Integrals are shown as

S F = C F = − 0.5 (see Figure A4.1 in Appendix-4). Therefore, from (1.158) the free space
value of the electric field at the observation point is
E B 0 = B ⋅ 2 i . (2.167)

The value of the electric field, due to knife-edge diffraction, may be normalized to that of
the free space as

A (u 0 ) = {( S F − C F ) + i [1 − ( S F + C F )]}.
1
(2.168)
2i

This variable is known as knife-edge diffraction loss. The magnitude of A ( u 0 ) may be

written from (2.168) as:

A ( u0 ) = [ S F ( u 0 ) − C F ( u 0 )] 2 + {1 − [ S F ( u 0 ) + C F ( u 0 ) ]}2 .
1
(2.169)
2
Here the unitless argument u 0 is defined from (2.155) as

1
u0 = x0 . (2.170)
m

( )
The graph depicting the A u 0 function in dB-s is shown in Figure 2.31. Several

specific values are presented in Table 2.3.

96
Figure 2.31. Knife-edge diffraction loss dependence on argument u 0

Table 2.3

Diffraction Loss, A
u0 Comments
in dB

0 -6 Zero gap: half-space is open


Optimal gap: provides the same field strength as in
-0.78 0
free space
-1.22 + 1.37 First maximum (Gain instead of loss)
-1.88 - 1.09 First minimum
-2.35 + 0.8 Second maximum (Gain instead of loss)
-2.74 - 0.74 Second minimum

As will note from this graph and table, the term diffraction loss is not always appropriate
for some specific values of the clearance u 0 (such as u 0 = −1.22 and u 0 = −2.35 ). The

presence of the semi-infinite wall, for those values of the height of knife edge may result
in field gain, rather than in field loss relative to free space propagation.

97
-----------------------------------------------------------------------------------
Example 2.4
Find the knife-edge diffraction loss due to an obstacle shown in Figure E2.4 for the
following initial conditions:
o r0 = 400 m, r01 = 200 m,

o h 1 = 10 m, h 2 = 15 m

o Frequency, f = 3 GHz

o H = 20 m

Figure E2.4

Solution
• First we find the position of point P above the ground from geometrical
relationships:
r01
h0 = ( h2 − h1 ) + h1 = 12.5 m,
r0
then x0 = H − h0 = 7.5 m.

• For the wavelength found as λ 0 = 3 ⋅ 10 8 / 3 ⋅ 10 9 = 0.1 m normalization

parameter m may be calculated from (2.156) as:


λ 1 0.1 1
m= ⋅ = = 5 m2
2 1 / r01 + 1 / r02 2 1 1
+
200 400 − 200
Thus, from (2.170) the normalized height of the obstacle is:
1 7.5
u0 = x0 = = 3.35
m 5

98
• The values for Fresnel's Integrals may be found either from [7], or from the
software applications such as: MATHEMATICA, MATLAB, MATHCAD:
S F = 0.47 , C F = 0.41 .
• Substitute these values into (2.169) for the knife-edge diffraction loss results in

A ( 3.35) = [0.47 − 0.41] 2 + {1 − [0.47 + 0.41 ]}2 = 0.067 (-23.47 dB)


1
2
(Answer)
-----------------------------------------------------------------------------------------

2.6.4. PRACTICAL APPLICATIONS OF THE FRESNEL ZONES


CONCEPT
Historically, the first application of the Fresnel Zones’ concept dates to 1875, when
J.L.Soret [4] introduced an optical ring-shaped diffracting lens, which was composed of
several non-transparent rings concentrically positioned across the optical wave
propagation path. Some of the applications briefly summarized below, may also be
found in [4] and [5] described in further detail.

2.6.4.a. RING-SHAPED ANTENNA DIRECTORS

As noted earlier, the odd and even numbered Fresnel's zones generate fields in the
observation point of opposite phases (Figure 2.28c), which is the cause of oscillating
character of the total field at the reception point (see the graph in Figure 2.28b), when
numerous zones are involved consequently. If evenly numbered zones are covered with
the plane shielding rings, such as shown in Figure 2.32, then in the receiving point B
only the fields contributing from odd numbered zones will superimpose constructively,
and the resultant field will exceed significantly that of the free space, E 0 .

99
Figure 2.32. A sketch of ring-shaped antenna directors

If, for instance, only the second zone is blocked by the shielding ring, then the resultant
field (2.147) may be rewritten as
E B = E B1 + E B 3 − E B 4 + E B 5 − E B 6 + E B 7 − ..... (2.171)

Similar to (2.148) the following relation: E B1 ≈ E B 3 ≈ 2 ⋅ E B 0 may be used to modify the

expression (2.171), i.e.

E E  E E 
E B ≈ 3E B 0 +  B 3 − E B 4 + B 5  +  B 5 − E B 6 + B 7  + .... (2.172)
 2 2   2 2 
Now, if we take into account that the higher order terms in (2.172) shown in parenthesis
are ignorable, then it may be rewritten as
E B ≈ 3E B 0 , (2.173)

For example, when a single ring-shaped director is installed across the antenna's main
radiation direction, then the intensity of the field in observation point becomes three
times greater than in free space, i.e. an additional 4.8 dB antenna gain is achieved. It is
therefore obvious, that the more concentric rings utilized, the greater the antenna gain
that may be achieved.

100
To calculate dimensions of the ring-shaped directors, the expression (2.146) is to be
modified as follows: we may take into account, that the receiving point is placed much
farther from the director, than the radiating point, i.e. r02 >> r01 , r01 + r02 ≈ r02 , thus

(2.146) may be simplified to the form

Rn ≈ n ⋅ λ ⋅ r01 , (2.174)

where n is the number of zones covered by the flat concentric rings (see Figure 2.32).

2.6.4.b. RING-SEGMENT DIFFRACTORS AS PASSIVE REPEATERS FOR


RADIO-RELAY LINKS

Propagation of UHF and higher frequency radio waves occur mainly within LOS
distances, and may cause coverage problems in mountainous regions, as radio waves
of these frequencies are unable to diffract around the hills and mountains. This causes
the occurrence of the shadow zones behind the mountain or hill with unacceptably low
levels of signal intensity (see Figure 2.33a). To avoid this problem, the retransmission of
radio waves is used at the C point on the top of the hill to allow covering the “shadow
area” with a strong enough signal (Figure 2.33b) to retain connectivity. An active
repeater is placed on the top of the hill, when transmitting-receiving electronic equipment
is in use, but is not always effective due to unavailability in the AC or DC power that is
required to be supplied. Therefore, the use of passive re-transmission (a passive
repeater) is sometimes more cost and maintenance efficient. A passive repeater is a
device that is installed on the top of the hill (point C in Figure 2.33b), that allows
redirection of the radio wave in order to reach the reception point in the “shadow region”.
The most straight forward approach is the use of flat metallic sheets (or metallic screens)
which act as a mirror for the incident ray, and allows redirection towards the receiving
point. Another approach, that is considered here to be more interesting, is the use of
barrier-type or a ring-segment diffractor that may serve as an efficient passive repeater
first introduced in 1954 by G.Z. Aizenberg [5], is described below.
Similar to the case of ring-shaped antenna directors, one or several Fresnel zones
(either all odd- or all even-numbered) are covered by shielding rings, or ring segments.
Because the distances r01 and r02 in (2.146) are usually fairly larger than dimensions of

101
Fresnel zones, Rn may become unrealistic insofar as being completely covered by the

shielding ring.

Figure 2.33. a). Shadow zone appearance in a mountainous region. b). Use of the
passive, ring-segment diffractor for VHF / UHF radio link design in a
mountainous/hilly region

Therefore, only a portion of the proper zone may practically be covered by a conducting
sheet or screen in order to achieve a field intensity gain in the shadow area (see Figure
2.34a).

102
Figure 2.34. Ring-type passive diffractor made of metallic screen that
partially covers two Fresnel zones. a). Front view, b). Side view

To find the height (b) of the zone recall, than the difference between distances ACB and
ADB (upper and lower paths on Figure 2.34b) must be equal to π , i.e.
k ⋅ 2∆ r = π , (2.175)

where ∆ r is defined from triangle CDE. Taking into account the fact that α is usually a

small angle, the following expression for ∆ r may be written as

α bα
∆ r = b sin ≈ . (2.176)
2 2
If (2.176) and k = 2π / λ are substituted into (2.175), then b may be found as
λ
b≈ , (2.177)

103
where α = β 1 + β 2 (Figure 2.33b) is typically determined from the geometrical profile of

the landscape. The horizontal size of diffractor L = 2a is usually given a priority based
on the shape of the landscape, and additionally on other construction issues. Therefore,
the size c (Figure 2.34a) may be calculated from the following expression [4]:
2
b L  1 1 
c=   +  . (2.178)
2  2λ  r01 r02 

Note however, that for most cases this dimension may be ignored due to r01 >> L , and

r02 >> L . More detailed analysis of the ring-segment diffractor may be found in [4].

2.6.4.c. EFFECTIVE AREA OF THE RADIO WAVE REFLECTION FROM THE


FLAT BOUNDARY

In section 2.4, the reflection phenomenon was considered based on the ray-concept of
propagation, i.e. the reflection was assumed occurring from an infinitely small point C.
In reality, as is mentioned previously, a finite spatial volume, limited at least by the first
Fresnel zone, is involved in propagation; and thereby the reflection must be considered
not from the C spot, but from the surrounding flat area, defined as an intersection of the
Fresnel zone with a flat boundary of two mediums (see the shaded area in Figure 2.35).

Figure 2.35. Configuration of the area effective for the reflection from a flat
boundary of the earth

104
If Figure 2.28b and expression (2.151) are recalled, then it may be realized, that to
enable the field intensity of the reflected wave to be the same as for the infinite reflecting
boundary, the cross sectional (transversal) size b of the intersection between Fresnel’s
ellipsoid and the reflecting boundary (see Figure 2.35) must be taken as

λ r10 r20
b ≥ R1 / 3= . (2.179)
3 r0
The longitudinal size of the effective area may approximately be evaluated from the
geometrical sketch, shown in Figure 2.35 as
b b r0
a≈ ≈ (2.180)
sin γ h1 + h2
if it is assumed, that the elevation angle is small enough, i.e. γ << 1 .

--------------------------------------------------------------------------------------
Example 2.5
For the given values of wavelength, distance and antenna elevations:
λ = 10 cm = 0.1 m , h1 = 20 m , h2 = 10 m , r0 = 5 km
Find the dimensions of the effective reflection area. Assume that the reflecting
boundary is flat.
Solution
The following proportions may be evaluated from geometrical sketch shown in:
Figure 2.35:
h1 20 h2 10
r10 = r0 = 5⋅ = 3.33 km , r20 = r0 = 5⋅ = 1.67 km
h1 + h2 30 h1 + h2 30

0.1 ⋅ 3.33 ⋅ 1.67


From (2.179) b= = 0.19 m = 19 cm (Answer)
3⋅5
0.19 ⋅ 5000
From (2.180) a= = 32 m . (Answer)
20 + 10
--------------------------------------------------------------------------------------------

105
2.7. REFERENCES

[1] Milligan, T.A., Modern Antenna Design, NJ: IEEE Press, 2005.
[2] Born, M, and Wolf, E., Principles of Optics, 7-th edition, Pergamon Press, 1997.
[3] Saunders, S.R., Antennas and Propagation for Wireless Communication Systems.
Second edition, UK: John Wiley & Sons Ltd., 2007
[4] Hristov, H.D., Fresnel Zones in Wireless Links, Zone Plate Lenses and Antennas,
Norwood, MA: Artech House, Inc., 2000
[5] Айзенберг Г.З. Ямпольский В.Г., Пассивные ретрансляторы для
радиорелейных линий, М. Связь, 1973 (in Russian)
[6] Abramowitz, M, Stegun, I. A. Handbook of Mathematical Functions with Formulas,
Graphs, and Mathematical Tables, New York: Dover Publications, 9-th edition, 1970
[7] Staelin, D.H., Morgenthaler, A.W., Kong, J.A. Electromagnetic waves. Prentice-Hall,
Inc., NJ, 1994.

2.8. PROBLEMS

P2.1. The slope angle of magnetic field lines changes from α 2 to α 1 when passing from
one medium into another as shown in Figure P.2.1.

Figure P.2.1. Sketch of the magnetic field lines on the border of two mediums
( J S surface current density vector directed towards the observer)

106
Find the angle α 2 if α 1 = 60 0 , µ 1 = 5 , µ 2 = 1 . The magnitude of magnetic field

strength in the first medium is H 1 = 0.2 A/m, and surface current density J = 0.1 A/m.

Answer: 28.7 0
P2.2. Prove, that (2.54) satisfies the Helmholtz equation (2.51).

P2.3. Derive the value of coefficient of proportionality in relation (2.89).

P2.4. Provide the detailed evaluation of expressions (2.102) – (2.108) for cases 1 – 4 of
the radio wave polarization analysis considered in section 2.3.

P2.5. Plot the graph for frequency dependence of magnitude and angle of the complex
dielectric permittivity for wet soil ( ε = 20, σ = 0.02 Sim / m ) as well as for damping factor

α and phase factor β in the frequency range from 102 to 1010 Hz. Outline roughly the
frequency ranges where wet soil may be considered as (1) conductor, (2)
semiconductor, (3) dielectric. Is wide spectrum signal affected when it propagates
through the medium with frequency dependent parameter(s)? Explain.
Hint: you may use MATLAB or similar routine for plotting graphs.

P2.6. Plane radio wave of the frequency 300 MHz penetrates into homogeneous see
water (ε = 80, σ = 1 Sim/m) along Z-axes. The effective field strength at the excitation
point (point of origin) is 0.1 V/m. What is the maximum expected propagation distance if
the least detectable value of the field strength (receiver's threshold) is equal 1 µV/m ?
Answer: 0.579 m

P2.7. Evaluate analytical expressions for the magnetic field reflection coefficients for
both vertical and horizontal polarizations for the radio waves obliquely incident onto the
border of two dielectric mediums.
Note: The answers are given in Table 2.2.

P2.8. Evaluate analytical expressions for the electric and magnetic field transmission
coefficients for both vertical and horizontal polarizations for the radio waves obliquely
incident onto the border of two dielectric mediums.
Note: The answers are given in Table 2.2.

107
P2.9. Confirm the expression (2.131) for the power balance for the reflection /
transmission of the radio wave from the boundary of two mediums for both, vertical and
horizontal polarizations.

P2.10. A horizontally polarized plane radio wave of frequency 80 MHz is incident from
atmospheric air onto flat surface of the wet soil (ε = 10, σ = 3·10 -2 S/m). The angle of
incidence is 30 0. Find the radio wave's penetration depth into the ground if the total
power loss due to both, reflection from the surface, and absorption in the ground is equal
15 dB.
Note: Consider a lossless propagation in atmospheric air.
Answer: 0.712 m

P2.11. Confirm that for the vertically polarized radio wave incident onto the border of two
dielectric mediums the following relation
ϕ 0 + ψ 0 = 90 0
between angle of incidence ϕ 0 and refraction angle ψ 0 is true.

Note: ϕ 0 is Brewster's angle.

P2.12. The inhomogeneous medium of volume V contains two homogeneous regions: V1


and V2 as shown in figure P2.2 (V = V1 + V2). Complex dielectric constants of those
regions are ε1 and ε2 respectively. If the equivalent complex permittivity for the whole
volume is calculated as

Figure P2.2

ε1V1 + ε2V2
εeq = (P2.1)
V1 + V2

108
then based on this expression derive the formulas for real part of the equivalent
permittivity, as well as for equivalent conductivity of this complex medium.
ε 1V1 + ε 2V2 σ 1V1 + σ 2V2
Answer: ε eq = , σ eq =
V1 + V2 V1 + V2

P2.13. In expression (2.81) there’re two factors that result in decrease of the average
power flow density, Π ave : first is exp (−2 α z ) and the second is cos Φ W . Find the

relation between damping factor α and the angle of the second power factor , Φ W for

the low loss medium.


−1 / 2
 λ 0 2 2 
Answer: cos Φ W = 1 + α
 4π 2 ε 
 

P2.14. If circularly polarized radio wave (RHCP) is normally incident on infinite border of
a flat perfectly conducting medium, then the reflected radio wave is also circularly
polarized with opposite direction of E-vector rotation (LHCP). Explain.

P2.15. Linearly polarized electromagnetic wave of arbitrary initial polarization angle is


incident to the border from dense medium ( ε 1 = 5 ) to sparse medium ( ε 2 = 1 ). Both
mediums are lossless dielectrics. Find the angle(s) of incidence, ϕ which result in linear
polarization of the reflected wave.
Answer: 0 < ϕ < 26.6 deg

P2.16. Calculate the values for the reflection coefficients for the vertically polarized radio
wave that is incident from air to dry soil with ε = 6 and σ ≈ 0 (ideal condition). Consider

two cases for the angle of incidence: ϕ1 = 55.56 O , and ϕ 2 = 75.24 O .


What is the difference between the calculated values in those two cases? Explain the
physical meaning.
Answer: Rvert ,1 = 0.191, Rvert , 2 = −0.191

109
P2.17. Horizontally polarized radio wave is incident from air to wet soil with the following
parameters: ε = 20 , σ = 2 ⋅ 10 − 2 S/m. Plot the frequency dependence of the magnitude
and phase of reflection coefficient for the following angles of incidence: 40O, 60O and
85O.
Hint: Use of the MATLAB or similar software is recommended.

P2.18. Find the unknown height H for the radio wave diffracting around the knife-edge
obstacle shown in figure P2.3 for the following initial data: distances r0 = 500 m,

r01 = 100 m, wavelength λ = 3 cm, knife-edge diffraction loss, A = -10 dB, antenna

elevations, h 1 = 10 m, h 2 = 20 m.

Hint: Use the graph in Figure 2.26.

Figure P2.3. Sketch of the geometrical configuration for problem P2.18

Answer: 12.55 m

P2.19. Derive an expression for the ratio of width of the high order Fresnel zone ( n >> 1 )
and first Fresnel zone, ∆ (n) = ( Rn +1 − Rn ) / R1 in the vicinity of the radiating source (

r01 << r02 ). Plot the graph ∆(n) in logarithmic scales for both axes.

Hint: Use Taylor series approximation for lim( Rn +1 − Rn ) , when n tends to infinity.

Answer: ∆(n) ≈ 1 /( 2 n )

P2.20. Derive the functional dependence of the ratio R1 / λ on the relative distance

k = r01 / r0 , where R1 is the radius of first Fresnel zone, r0 = r01 + r02 , and r01 , r02 are

the distances shown in Figure 2.21. Plot the graph of R1 / λ as a function of k, assuming

r0 / λ = const.

110
Answer: R1 / λ = r0 / λ k (1 − k )

P2.21. Show that for a ring-segment passive repeater the gain in field strength relative to
the field strength in free space (i.e. relative to reference conditions) may approximately
be expressed by formula G = 20 log [1 + 2 (∆S / S1 )] , where ∆S is the total area of the

passive reflector, and S1 is the area of the first Fresnel’s zone.


Hint. Consider a direct proportion of electric field on the area dS in expression (2.140), as well as
a nearly uniform distribution of the field along that area.

P2.22. Design the ring-segment passive reflector based on the formula given in problem
P2.21 to obtain the gain of G = 4 dB if h1 = h2 = 10 m, r01 = 3 km, r02 = 5 km, H = 150 m,

λ = 5 cm.
Hint: Take more than one Fresnel’s zone to be covered (Fig.2.34a), if the total length 2a becomes
unrealistically big (e.g. exceeds 50 m).

P2.23. For the knife-edge diffraction considered in problem P2.18 find the smallest value
of x0 (positive or negative – to be defined) which results in the same field strength at the

observation point as for the free space between communicating antennas.


Answer: - 0.854 m

P2.24. A ring-shaped antenna director made of three (3) rings (similar to what is shown
in Figure 2.32) is placed in front of parabolic reflector-antenna to provide additional
antenna gain at a single frequency. The shielding rings are to cover the second, fourth
and sixth Fresnel zones. Estimate the additional antenna gain (in dB) provided by this
antenna director.
Answer: 16.9 dB

111
APPENDIX-1
USEFUL MATHEMATICAL RELATIONS

1. TRIGONOMETRIC EQUALITIES

• sin(α ± β ) = sin α cos β ± cos α sin β (A1.1.1)

• cos(α ± β ) = cos α cos β  sin α sin β (A1.1.2)

π 
• sin  ± α  = cos α (A1.1.3)
2 

π 
• cos ± α  =  cos α (A1.1.4)
2 

α +β α −β
• sin α + sin β = 2 sin cos (A1.1.5)
2 2

α +β α −β
• sin α − sin β = 2 cos sin (A1.1.6)
2 2

α +β α −β
• cos α + cos β = 2 cos cos (A1.1.7)
2 2

α +β α −β
• cos α − cos β = −2 sin sin (A1.1.8)
2 2

• sin 2 α + cos 2 α = 1 (A1.1.9)

A.2. VECTOR ANALYSIS

A.2.1. UNIT VECTOR TRANSFORMS

112
Figure A1.1. Unit vectors: a). in rectangular coordinates, b). in spherical
coordinates

a). Spherical to rectangular transforms

• x0 = r0 sin θ cos ϕ + θ 0 cos θ cos ϕ − ϕ 0 sin ϕ (A1.2.1.1)

• y 0 = r0 sin θ sin ϕ + θ 0 cos θ sin ϕ + ϕ 0 cos ϕ (A1.2.1.2)

• z 0 = r0 cos θ − θ 0 sin θ (A1.2.1.3)

b). Rectangular to spherical transforms

• r0 = x0 sin θ cos ϕ + y 0 sin θ sin ϕ + z 0 cos θ (A1.2.1.4)

• θ 0 = x0 cos θ cos ϕ + y 0 cos θ sin ϕ − z 0 sin θ (A1.2.1.5)

• ϕ 0 = − x0 sin ϕ + y 0 cos ϕ (A1.2.1.6)

A.2.2. VECTOR INTEGRAL RELATIONS

• ∫ A ⋅ dl = ∫∫ (∇ × A ) ⋅ ds
C S
Stoke's theorem (A1.2.2.1)

• ∫∫∫ (∇ ⋅ A) dv = ∫∫ A ⋅ ds
V S
Divergence theorem (A1.2.2.2)

A.2.3. SOME RALATIONS FROM VECTOR ALGEBRA

113
• A × (B × C ) = ( A ⋅ C ) ⋅ B − ( A ⋅ B ) ⋅ C (A1.2.3.1)

• ( A × B ) × C = (C ⋅ A ) ⋅ B − (C ⋅ B ) ⋅ A (A1.2.3.2)

• A ⋅ B ×C = B ⋅C × A = C ⋅ A × B (A1.2.3.3)

A.2.4. VECTOR DIFFERENTIAL OPERATIONS

(Notations: A and B -- vector fields, ψ -- scalar field, ∇ -- Nabla operator)

∂ ∂ ∂
• ∇ = x0 + y0 + z0 (in Cartesian coordinates) (A1.2.4.1)
∂x ∂y ∂z

• ∇ ⋅ ( A × B ) = B ⋅ (∇ × A ) − A ⋅ (∇ × B ) (A1.2.4.2)

• ∇ ⋅ (ψ A ) = A ⋅ ∇ψ + ψ ∇ ⋅ A (A1.2.4.3)

• ∇ × (ψ A ) = ∇ψ × A + ψ ∇ × A (A1.2.4.4)

a). Cartesian (rectangular) coordinates

∂ψ ∂ψ ∂ψ
• ∇ψ = gradψ = x0 + y0 + z0 (A1.2.4.5)
∂x ∂y ∂z

∂AX ∂AY ∂AZ


• ∇ ⋅ A = divA = + + (A1.2.4.6)
∂x ∂y ∂z

x0 y0 z0
• ∇ × A = curl A = ∂ / ∂x ∂ / ∂y ∂ / ∂z =
AX AY AZ

 ∂A ∂A   ∂A ∂A   ∂A ∂A 
... = x 0  Z − Y  + y 0  X − Z  + z 0  Y − X  (A1.2.4.7)
 ∂y ∂z   ∂z ∂x   ∂x ∂y 

∂ 2ψ ∂ 2ψ ∂ 2ψ
• ∇ ⋅ (∇ψ ) = ∇ 2ψ = + + (A1.2.4.8)
∂x 2 ∂y 2 ∂z 2

• ∇ 2 A = x0 ∇ 2 AX + y 0 ∇ 2 AY + z 0 ∇ 2 AZ (A1.2.4.9)

b). Spherical coordinates

114
∂ψ 1 ∂ψ 1 ∂ψ
• ∇ψ = divψ = r0 + θ0 + ϕ0 (A1.2.4.10)
∂r r ∂θ r sin θ ∂ϕ

1 ∂ 2 ∂ ∂Aϕ
• ∇ ⋅ A = div A = (
r Ar + )
1
r sin θ ∂θ
( Aθ sin θ ) + 1
r sin θ ∂ϕ
(A1.2.4.11)
r ∂r
2

∂ ∂A θ  1  1 ∂Ar ∂ 
 ( A ϕ sin θ ) − − ( rAϕ ) = ...
1
• ∇ × A = r0  + θ0 
r sin θ  ∂θ ∂ϕ  r  sin θ ∂ϕ ∂r 

1 ∂ ∂A r 
... + ϕ 0  ( rA θ ) −  (A1.2.4.12)
r ∂ r ∂θ 

1 ∂  2 ∂ψ  1 ∂  ∂ψ  1 ∂ 2ψ
• ∇ 2ψ = r + 2  sin θ  + (A1.2.4.13)
r 2 ∂r  ∂r  r sin θ ∂θ  ∂θ  r sin θ ∂ϕ
2 2 2

• ∇ 2 A = ∇(∇ ⋅ A ) − ∇ × ∇ × A (A1.2.4.14)

APPENDIX-2

POLARIZATION OF RADIO WAVES


GENERAL APPROACH

As mentioned in section 2.3, sense of polarization of the radio wave is defined by the
direction of movement of the tip of electric field vector in X0Y projection plane (Figure
A2.1). And any arbitrary polarized radio wave may be represented as a superposition of
two cross-perpendicular components of linear polarizations (linearly polarized
components), i.e.
E = E X + EY = x0 E m X cos (ω t + Φ 0 X ) + y 0 E m Y cos (ω t + Φ 0 Y ) (A2.1)

The sense of polarization depends on the amplitudes of components, E m X , E m Y and

initial phase shifts Φ 0 X , Φ 0 Y . Strictly saying it depends on ratio of the amplitudes, and

the difference of the initial phase shifts. Hence for further evaluations it's more
convenient to transform (A2.1) to the following form, without limitations to generality:

115
E = E m X [ x0 cos ω t + y 0 M cos (ω t + δ )] , (A2.2)

where δ = Φ 0 Y − Φ 0 X is a difference of initial phases, and M = E m Y / E m X = tan Ψ is

the ratio of the amplitudes (see Figure A2.1).

Figure A2.1. Linear cross-orthogonal components of the total electric field in


Cartesian coordinates.

Now we separate cross-orthogonal components of (A2.2) and present them in form of


system of equations:
X = cos ω t 
 (A2.3)
Y = M ⋅ cos (ω t + δ ) 
It may be seen, that (A2.3) is a parametric form of arbitrary-oriented ellipse with the
center at the origin of the coordinates as shown in Figure A2.2.
τ is a tilt angle relative to the horizontal axes.

Figure A2.2. Polarization ellipse in Cartesian coordinates

116
Figure A2.3. Rotation of coordinate system

For further processing we’ll rotate the X0Y coordinate system by angle τ counter-
clockwise as shown in Figure A2.3. Linear transform of X0Y coordinates to X'0Y' may be
written as
X = X ′ cosτ − Y ′ sin τ 
 (A2.4)
Y = X ′ sin τ + Y ′ cosτ 
Now we solve (A2.4) for X ′ and Y ′ by finding the inverse transform of the matrix of
coefficients in the right hand side, i.e.
X ′ = X cosτ + Y sin τ 
 (A2.5)
Y ′ = − X sin τ + Y cosτ 
Next we substitute X and Y from (A2.3) into (A2.5), and use the following notation:
ωt =ζ .
X ′ = cos ζ cosτ + M cos (ζ + δ )sin τ =
= cos ζ cosτ + M cos ζ cos δ sin τ − M sin ζ sin δ sin τ (A2.6)

Y ′ = − cos ζ sin τ + M cos ζ cos δ cosτ − M sin ζ sin δ cosτ (A2.7)


On the other hand parametric equation for ellipse in the new, X'0Y' coordinate system
may be written as
X ′ = a cos (ζ + ∆ ) 
 (A2.8)
Y ′ = ± b sin (ζ + ∆ ) 
where a and b are the major and minor semi-axes of the ellipse respectively (see Figure
A2.4), and ∆ is the arbitrary initial phase shift.

117
Figure A2.4. Polarization ellipses for right hand elliptical polarization (RHEP) and
left hand elliptical polarization (LHEP).

Note from Figure A2.4, that plus sign results in counter-clockwise rotation of the total E -
vector, which signifies a right hand circular polarization (RHCP), whereas minus sign
represents a clockwise rotation of the total E -vector, i.e. a left hand circular polarization
(LHEP) 1.

The following two polarization parameters are introduced to identify the orientation of the
ellipse, and its shape:
• The angle ε , that shows how stretched the ellipse is
b  b
tan ε e = ± , ε e = tan −1  ±  (A2.9)
a  a
• The axial ratio of the polarization ellipse
a a
AR = ± = cot ε e , ARdB = 20 log . (A2.10)
b b
From this definition it is obvious, that If the magnitude AR changes from 1 to ∞ (from 0

to ∞ in decibels), then the sense of polarization will change from circular (CP) to linear

(LP). For those changes the values of the angle ε e may vary in range − 45 0 ≤ ε ≤ +45 0 .

The system of equations (A2.6) – (A2.7) is actually the same as (A2.8). Therefore we
may take proper parts of those equations equal to each other, as follows:
a cos ζ cos ∆ − a sin ζ sin ∆ = ...

1
Here Z-axis is directed from picture towards the observer, in contrast to that on Figure 2.11.
Therefore we ended up with the counter clockwise rotation of the resultant vector for RHCP in
contrast to the clockwise rotation, when Z-axis has opposite direction, i.e. is directed from
observer towards the picture, like it’s shown in Figure 2.11.

118
... = cos ζ cosτ + M cos ζ cos δ sin τ − M sin ζ sin δ sin τ , (A2.11)

± b sin ζ cos ∆ ± b cos ζ sin ∆ = ...


... = − cos ζ sin τ + M cos ζ cos δ cosτ − M sin ζ sin δ cosτ . (A2.12)
Now we separate like terms in (A2.11) and (A2.12).
a cos ∆ = cosτ + M cos δ sin τ (A2.13)

a sin ∆ = M sin δ sin τ (A2.14)

± b cos ∆ = − M sin δ cosτ (A2.15)

± b sin ∆ = − sin τ + M cos δ cosτ (A2.16)


Next we take the sum of squares of all four expressions (A2.13) – (A2.16). After
simplifications the result is
a2 + b2 = 1+ M 2 , (A2.17)
which stands for a conservation of the power while switching from X0Y to X’0Y’
coordinates.
Now we divide (A2.15) by (A2.13), and (A2.16) by (A2.14)
b M sin δ cosτ M sin δ
± =− =− = ... .
a cosτ + M cos δ sin τ 1 + M cos δ tan τ
sin τ − M cos δ cosτ
... = − (A2.18)
M sin δ sin τ
The following transforms may be applied to (A2.18)
M cos δ
1−
M sin δ tan τ tan τ − M cos δ
= = , (A2.19)
1 + M cos δ tan τ M sin δ M tan τ sin δ

M 2 sin 2 δ tan τ = (1 + M cos δ tan τ )(tan τ − M cos δ ) . (A2.20)


Now we open the parenthesis and combine like terms.
tan τ M tan Ψ
2 =2 =2 cos δ . (A2.21)
1 − tan τ
2
1− M 2
1 − tan 2 Ψ
1
Finally
tan 2τ = tan 2Ψ cos δ (A2.22)

sin x
2
1
Here: 2
tan x
= cos x = 2 sin x cos x = sin 2 x = tan 2 x
1 − tan 2 x sin 2 x cos 2 x − sin 2 x cos 2 x
1−
cos 2 x
119
If Ψ and δ are known a priory, then τ may be found by solving (A2.22), i.e.
π
tan −1 [tan 2Ψ cos δ ] ± n .
1
τ= (A2.23)
2 2
where n = 0, 1, 2, .......
Each value of n in (A2.23) may be related to one position of the polarization ellipse in
space. In other words we have an aggregate of ellipses. But only two of those positions
are different in space. Those are defined by the values n = 0 and n = 1 . All other
positions (for all other values of n) will repeat those two. Hence it is reasonable to
consider only two values of τ :

tan −1 [tan 2Ψ cos δ ] ,


1
τ1 = (A2.24)
2

tan −1 [ tan 2Ψ cos δ ] + 90 0 .


1
and τ2 = (A2.25)
2
( )
It’s easy to show, that (A2.24) associates to the values of tan Ψ < 1 Ψ < 45 0 , whereas

( ) ( )
(A2.25) associates to tan Ψ > 1 Ψ > 45 0 . If tan Ψ = 1 Ψ = 45 0 then τ = ± 45 0 .

In order to obtain ε , first we find the product ± a b by multiplying (A2.13) by (A2.15), and
(A2.14) by (A2.16), and adding the results.
± a b = (cosτ + M cos δ sin τ )(− M sin δ cosτ ) + ...
... + M sin δ sin τ (− sin τ + M cos δ cosτ ) . (A2.26)
After simplifications the following expression may be obtained:
± a b = − M sin δ . (A2.27)
Now we use (A2.9) to evaluate the ratio 1 .
b
2
2ab
± 2 =± a = 2 tan ε e = sin 2ε (A2.28)
a + b2 b 2 1 + tan 2 ε e
e

1+ 2
a
On the other hand we may modify the same expression by using (A2.17) and (A2.27)
2ab 2 M sin δ 2 tan Ψ sin δ
± =− =− = − sin 2Ψ sin δ . (A2.29)
a +b
2 2
1+ M 2
1 + tan 2 Ψ

2 tan x 2 sin x cos x


1
Trigonometric relation = = sin 2 x is used.
1 + tan x sin 2 x + cos 2 x
2

120
Combination of (A2.28) and (A2.29) results in
sin 2ε = − sin 2Ψ sin δ . (A2.30)

Expressions (A2.22) and (A2.30) allow obtaining the direct relations between initial
parameters of the linear cross-polarized components ( M = tan Ψ and δ ) and
parameters of polarization ( τ and ε ). Axial ratio, AR may further be calculated from
(A2.10).

To find the inverse relations one may obtain cos δ from (A2.30), substitute into (A2.22),

and solve for cos 2Ψ . The result is

cos 2Ψ = cos 2τ cos 2 ε (A2.31)

which may be used to obtain Ψ .


If (A2.30) is divided by (A2.22), then
1 sin 2 ε
tan δ = − . (A2.32)
cos 2 Ψ tan 2τ
The final expression may be developed if cos 2Ψ is substituted from (A2.31). Then the
result
tan 2 ε
tan δ = − (A2.33)
sin 2τ
allows finding the phase shift δ between initial LP-components of the elliptically
polarized wave.
--------------------------------------------------------------------------------------------------
Example A2.1.

Find polarization parameters of the radio wave, i.e. sense of polarization, angle of
ellipticity, ε , axial ratio, AR, and tilt angle τ of the polarization ellipse, if parameters of
cross-polarized components (vertical and horizontal) are:

• Amplitude of V-pol component: E1m = 3 mV/m

• Amplitude of H-pol component: E 2 m = 8 mV/m

• Phase shift between components: δ = −60 0 .


Solution
• M = tan Ψ = 8 / 3 = 2.67 , thus Ψ = 69.44 0 , 2 Ψ = 138.89 0

121
• We’ll use (A2.25) because of the condition M = tan Ψ > 1
1
τ = tan −1[tan(2Ψ ) cos δ ] + 90 0 = ...
2
1
... = tan −1[tan 138.89 0 cos(−60 0 )] + 90 0 = 78.210 (Answer)
2
• Now using expression (A2.30) to obtain ε

(
sin 2 ε = − sin 2 Ψ sin δ = − sin 138.89 0 ⋅ sin − 60 0 = 0.5694 , )
sin (0.5694 ) = 17.35 0 .
1 −1
therefore ε = (Answer)
2
• Axial ratio

( )
AR = tan −1 (ε ) = tan −1 17.35 0 = +3.2 (10 dB). (Answer)
The sense of polarization is RHEP for positive ε (see Figure A2.5).

Figure A2.5. Polarization ellipse sketch for the calculated parameters in


Example A-2.1

-----------------------------------------------------------------------------------------------
Example A2.2.

Find parameter M = tan Ψ (the ratio of the amplitudes of initial linear cross-polarized
components) of the radio wave and the phase shift δ , if the following parameters of

polarization ellipse are provided: angle of ellipticity, ε = −15 0 (LHEP), and tilt angle,

τ = 75 0 .
Solution

122
• Using the expression (A2.31)
( )
cos 2Ψ = cos 2τ ⋅ cos 2ε = cos 2 ⋅ 75 0 ⋅ cos (2 ⋅ (−15) ) = −0.75 ,

thus Ψ = 69.3 0 , M = tan Ψ = 2.65 . (Answer)

• Using the expression (A2.33)

tan δ = −
tan 2 ε
=−
(
tan − 30 0 )
= 1.1547 , thus δ = 49.10 .
sin 2τ (
sin 150 0
) (Answer)

APPENDIX-3

A3.1. HELMHOLTZ EQUATION FOR VECTOR POTENTIAL


Recall the Maxwell’s equation (2.4) that testifies the principle of continuity of the
magnetic field lines. Rewrite the equation in form
∇ ⋅ H = 0. (A3.1)
It proves that magnetic fields are always solenoidal. Therefore any magnetic field may
be represented as a curl of the other vector field, i.e.
1
H = ∇× A. (A3.2)
µ0 µ
Here A is called vector potential.
If we substitute (A3.2) into (2.2), then the Faraday's law for time-harmonic process may
be rewritten as

( )
∇ × E + i ω A = 0 . (A3.3)

From (A3.3) one may conclude that because the curl of the vector field E + i ω A is

equal to zero, then the field E + i ω A is purely potential. It is well known from the
Electromagnetics course, that vector gradient of any scalar field Φ is always purely

potential. Thus vector E + i ω A may be represented as a gradient of another scalar field


Φ called scalar potential i.e.

E + i ω A = −∇Φ . (A3.4)

123
Now we substitute (A3.2) into the expression for the Ampere's law (2.1), written for the

lossless spatial area that contains a time-harmonic current source J Ext .

1
∇ × H = ∇ × ∇ × A = iω ε 0 ε E + J Ext . (A3.5)
µ0 µ
In this expression a double cross product (double curl) may be modified by using the
identity (A1.2.3.1) from Appendix -1.

∇ × ∇ × A = ∇ (∇ ⋅ A ) − ∇ 2 A . (A3.6)

Now we substitute (A3.6) and E from (A3.4) into (A3.5).

∇ 2 A + ω 2 ε 0 ε µ 0 µ A − ∇(iω ε 0 ε µ 0 µ Φ + ∇ ⋅ A ) = − µ 0 µ J Ext . (A3.7)

In order to simplify (A3.7) we may apply Lorentz condition or otherwise called calibration
condition that allows make both, vector and scalar potentials interrelate to each-other.
Regarding Lorentz calibration condition

iω ε 0ε µ 0 µ Φ = − ∇ ⋅ A . (A3.8)


Indeed, both potentials, A and Φ has been initially introduced independently. Now an
additional "calibration" request (A3.8) allows making significant simplification of (A3.7)
without loosing the generality of further analysis. Under that condition the equation
(A3.7) may be rewritten in form of Helmholtz equation as

∇ 2 A + k 2 A = − µ 0 µ J Ext 1, (A3.7a)

where k is same complex number as that defined by (2.50) called propagation constant.

Here we're considering the propagation in lossless medium, i.e. we'll assume k = k = β
as being a real number.
In Cartesian coordinates the linear operator ∇ 2 may be applied to each component of

{ }
the vector A = A X , A Y , A Z , thus (A3.7a) may be decomposed into the system of three
linear independent partial differential equations for each scalar components as follows:
∇ 2 A X + β 2 A X = − µ 0 µ J Ext , X , (A3.8a)

∇ 2 A Y + β 2 A Y = − µ 0 µ J Ext , Y , (A3.8b)

1
(A3.7a) looks similar to (2.51). The difference is in right hand side. That's because (A3.7a),
inhomogeneous Helmholtz equation, is written for the source containing spatial region in contrast
to (2.51), homogeneous Helmholtz equation, that is written for free space propagation medium.

124
∇ 2 A Z + β 2 A Z = − µ 0 µ J Ext , Z , (A3.8c)

which may be solved separately. Unfortunately similar decomposition is impossible in


other coordinate systems, such as cylindrical or spherical curvilinear coordinates.
The three equations, (A3.8a) – (A3.8c) are identical in form, therefore it's reasonable to
expect the same approach in their solutions. For the infinitesimal point source located at
the origin the right hand side of those equations become zero except of the point of the
origin. For the source-free spatial region a general form that is common for all three
expressions may be expressed in form of Helmholtz’s homogeneous equation
∇ 2ϑ + β 2ϑ = 0 , (A3.9)

where ϑ is one of the components of the A vector. This problem is spherically

symmetric, hence it’s meaningful to expand the ∇ 2 operator in spherical coordinates


based on (A1.2.4.13) from Appendix-1. The physical interpretation of this case of the
scalar point source allows assuming that the solution will have only the radial
dependence, thus in ∇ 2 derivatives by θ and ϕ are taken equal to zero. Then (A3.9)
becomes single-coordinate ordinary differential equation and may be presented as

1 d  2 dϑ 
r  + βϑ = 0 . (A3.10)
r 2 dr  dr 
It may be shown by substitution, that the following general solution satisfies (A3.10):
exp(± iβ r )
ϑ = C . (A3.11)
r
Here C is an arbitrary integration constant. If we include time dependence into (A3.11),
then it will represent two waves of the phases similar to that given by (2.60), while the
amplitudes are decaying inverse proportional to the radial distance. One may conclude,
that the solution with the negative sign represents the wave that propagates from origin
to infinity, whereas the solution with the positive sign represents the wave that
propagates from infinity to the origin. Hence the positive sign may be eliminated as
meaningless.
The integration constant may be defined based on specific conditions in the origin,
where the point source is located. Hence (A3.11) is to be substituted into the
inhomogeneous Helmholtz equation

1 d  2 dϑ 
r  + βϑ = −δ (r ) (A3.12)
r 2 dr  dr 

125
that contains − δ (r ) function representing a point source at the origin 1. If a volume

integral is taken from both sides of (A3.12), and letting r → 0 the integration constant

may be found as C = (4π ) −1 . This procedure is nothing, but the application of the initial
conditions at the point of origin, which allows turning a general solution (A3.11) into the
particular solution of the following form:
exp(− iβ r )
ϑ = , (A3.13)
4π r
which satisfies the initial conditions. Expression (A3.13) represents a scalar spherical
wave that propagates from the point of origin radially in all directions. Indeed, if time-
harmonic multiplier exp(iω t ) is included into (A3.13), then

exp[i (ω t − β r )]
ϑ (r , t ) = (A3.13a)
4π r
is a harmonic wave process with the constant phase ϕ = ω t − β r that moves along

radial direction with the velocity of v = ω / β (compare with (2.60) – (2.62)), and with the
amplitude that decays, being inverse proportional to the distance. In other words for the
fixed time moment t = const the constant values of the phase ϕ = const will be
associated with r = const distances from the origin. The relation r = const is the
equation of the sphere in spherical coordinate system, thus the constant values of the
phase, ϕ = const (wave fronts) are represented as source-centered spherical surfaces.
Hence the scalar point source radiates a spherical waves radially propagating away from
the point source.
Now suppose that the scalar inhomogeneous Helmholtz equation (A3.8a) contains the
source that has a weighted distribution J Ext , X within a finite volume V, i.e. instead of the

point source, we consider the source of the finite size. Based on principle of linear
superposition we may assume, that the solution will be represented as a sum of the
numerous waves coming from the multiple point sources located within the volume V. In
other words the solution (A3.13a) for that case may be written as

1
δ (r ) is a Dirac's delta function. It’s equal to zero everywhere except of the point of origin. For
that point δ (r ) tends to infinitely, so thus the overall integral along the entire area of coordinates

r = (0, ∞) is equal to unity.

126
exp[i (ω t − β r ′′)]
A X (r , t ) = ∫∫∫ µ 0 µ J Ext , X (r ′, t ) dV , (A3.14)
V
4π r ′′
where the distances r , r ′ and r ′′ are shown in Figure A3.1.

Figure A3.1. Definition of distances in (A3.14) integral

Expressions similar to (A3.14) may be written for Y- and Z-components of the magnetic

vector potential. Hence the total magnetic vector potential A may be written as a vector:
exp[i (ω t − β r ′′)]
A (r , t ) = ∫∫∫ µ 0 µ J Ext (r ′, t ) dV . (A3.15)
V
4π r ′′

A3.2. RADIATION FROM THE ELECTRIC CURRENT POINT


SOURCE
Consider a piece of infinitely thin wire of the length l with an electric current I directed
along Z-axis as shown in Figure A3.2. If the current is time-harmonic, then it is
considered as a source that radiates radio waves. For the current uniformly distributed
within the region − l / 2 < z < l / 2 , and for the wire length l << λ this electric current
source is called an ideal dipole (or Hertzian dipole). It is an electric current point source
because of its infinitesimal size. The right hand side of (A3.7a) may be written as

J Ext ( x, y, z ) = I ⋅ δ ( x) ⋅ δ ( y ) ⋅ z 0 , (A3.16)

where Dirac’s delta-functions represent infinitely small size of the cross-section of the
dipole, and z 0 is a unit-vector along Z-axis. If (A3.16) is substituted into (A3.15), then the

volume integral may be written as

127
+∞ +∞
exp(−iβ r ′′)
l/2
A = z 0 µ 0 µ I ∫ δ ( x′) dx′ ∫ δ ( y ′) dy ′ ∫ d z ′ 1. (A3.17)
−∞ −∞ −l / 2
4π r ′′

Figure A3.2. Sketch for (A3.17) integral evaluation

As may be seen from (A3.17) a triple integral becomes a product of three single
integrals. First and second integrals of the delta-function are equal to unity by definition.
The core of the third integral may be taken out of the integral sign because for the
integration limits the variations in coordinates along the wire are ignorable, i.e. x ′ << r ,
y ′ << r , and z ′ << r , and therefore r ′′ ≈ r . This means the core remains constant and
independent within the integration limits. Hence (A3.17) may be simplified to
µ µ I exp(−iβ r )
A = A Z z 0 = z 0 l 0 . (A3.18)
4π r
This expression demonstrates that the orientation for the magnetic vector potential in
any point P of space is the same as the orientation of a dipole itself.
Magnetic field of the radiated radio wave may be found if (A3.18) is substituted into
(A3.2) and the identity (A1.2.4.4) from Appendix A-1 is applied.

H =
1
(
∇ × A Z z 0 =
µ0 µ
1
)
(∇A Z ) × z 0 +
µ0 µ
1 
AZ ⋅ (∇ × z 0 ) .
µ0 µ
(A3.19)

The second term in right hand side is equal to zero because the curl of the constant
vector z 0 is identically equal to zero. The first term in (A3.19) contains gradient of A Z ,

which, as may be seen from (A3.18), is only distance-dependent. Therefore the relation
(A1.2.4.10) from Appendix A-1 may be applied in the following form

1
The time-harmonic multiplier exp(iω t ) is not included for simplicity.

128
µ 0 µ I l d  exp(−iβ r ) 
∇A Z =  r0 = .
4π d r  r 
µ 0 µ I l  − iβ exp(−iβ r ) exp(−iβ r ) 
= −  r0 (A3.20)
4π  r r2 
Then magnetic field strength may be found from (A3.19) if (A3.20) is substituted into it.
I l  − iβ exp(−iβ r ) exp(−iβ r ) 
H = −  (r0 × z 0 ) . (A3.21)
4π  r r2 
For further modifications of (A3.21) we'll apply the following two approaches:
• Second term in brackets diminishes much faster than the first one, therefore
may be ignored, when so called "far zone" is considered, i.e. when r >> l ,
• The cross-product r0 × z 0 may be replaced by − sin θ ⋅ ϕ 0 based on cross-

relations between Figures A1.1.a and A1.1.b of the Appendix-1.


For more generic approach we may consider the case, when the space that surrounds
the source includes some losses. It’s easy to show, that in order to take them into

account the real value of β must be replaced by the complex propagation constant k .
Taking above statements into account the final the expression for magnetic field strength
may be written from (A3.21) as

I l  exp(−ik r )
H = ik sin θ ϕ 0 . (A3.22)
4π r
Electric field strength may be defined by substituting (A3.22) into Ampere's law in
complex from, (2.47), i.e.
1 1
E = ∇ × H = ∇ × ( H ⋅ ϕ 0 ) (A3.23)
iω ε 0ε iω ε 0ε

I l  exp(−i k r )
where H = ik sin θ is a scalar phasor.
4π r
Here we may apply the identity (A1.2.4.4) from Appendix-1 to the cross product in right
hand side, i.e.
∇ × ( H ⋅ ϕ 0 ) = ∇H × ϕ 0 + H (∇ × ϕ 0 ) . (A3.24)

Second term in right hand side is identically equal to zero, because of the absence of
curl for the unit vector ϕ 0 . We may expand the first term based on formula (A1.2.4.10)

given in Appendix-1 as follows

129
 ∂H 1 ∂H 1 ∂ψ 
∇H × ϕ 0 = r0 + θ0 + ϕ0 × ϕ 0 = ...
 ∂r r ∂θ r sin θ ∂ϕ 

... =
I l    exp − ikr
(ik ) − ik
( )
sin θ (r0 × ϕ 0 ) −
(
exp − ikr )
sin θ (r0 × ϕ 0 ) + ...
4π  r r2

... +
(
exp − ikr) 
cosθ (θ 0 × ϕ 0 ) . (A3.25)
2
r 
As one may notice from (A2.25) the second and third terms in brackets are inverse
proportional to square of distance and may be ignored, compared to the first, when the
far field region ( r >> λ ) is considered. Thus (A2.25) may approximately be rewritten as

∇H × ϕ 0 = −θ 0
I l  2 exp − ikr
k
(sin θ .
) (A3.26)
4π r
Here, the cross product r0 × ϕ 0 is replaced by − θ 0 based on disposition of those unit

vectors (see Figure A1.1b from Appendix-1). Now we can substitute ∇ × ( H ⋅ ϕ 0 ) by

(A3.26) in expression (A3.23), then the final result for the electric field phasor will
become

Il exp(−ik r )
E = iωµ 0 µ sin θ ⋅ θ 0 , (A3.27)
4π r
where k is defined as ω ε 0ε µ 0 µ . It may be seen from the last expression that the

maximum value of the electric field appears at θ = 90 0 , i.e. in direction perpendicular to


dipole’s axis. For non-magnetic medium ( µ = 1 ) the magnitude of that electric field may
be written as
I lωµ 0 60π I l
Em = = . (A3.28)
4π r λr
For some applications it is more convenient to express (A3.27) in terms of the dipole
moment. The Hertzian dipole is his original form is shown in Figure A3.3, where the
spheres on both ends of the dipole are to cumulate charges that are moving due to
current I .

Figure A3.3. Hertzian dipole

130
The general definition for the vector of dipole moment is
p = ql , (A3.29)

where q is the charge that cumulates on both ends of the dipole shoulders, and l is a
vector of the length of the dipole, and direction along its axes. For the time-harmonic
source we may write
dp dq
= iω p = l = Il . (A3.30)
dt dt
Now if we substitute the magnitude of (A3.30) into (A3.27) it will result in

k 2 p exp(−ik r )
E = θ 0 sin θ . (A3.31)
4π ε 0ε r
Note: minus sign is omitted in the last expression.

APPENDIX-4
FRESNEL'S INTEGRALS

Expression (2.153) and (2.154) for C and S surface integrals may be modified as based
on trigonometric identities (A1.1.1) and (A1.1.2) from Appendix A-1.
π π  π 
(
C = ∫∫ cos  u 2 + v 2 ) du dv = ∫ ∞ ∞
cos  u 2 du ⋅ ∫ cos  v 2 dv −
S′ 2  u0
2  −∞
2 
∞ π  ∞ π 
− ∫ sin  u 2 du ⋅ ∫ sin  v 2  dv , (A4.1)
u0
2  −∞
2 
π π  π 
(
S = ∫∫ sin  u 2 + v 2 ) du dv = ∫ ∞ ∞
sin  u 2 du ⋅ ∫ cos  v 2 dv +
S′ 2  u0
2  −∞
2 
∞ π  ∞ π 
+ ∫ cos  u 2 du ⋅ ∫ sin  v 2  dv . (A4.2)
u0
2  − ∞
2 
As mentioned in section 2.6.3 the limits for variables in (A4.1) and (A4.2) are
u 0 < u < ∞ , and − ∞ < v < ∞ .

131
Now consider the integral

 π 
I = ∫ exp i t 2  dt . (A4.3)
0  2 
To evaluate (A4.3) the following change of the variable is applied:
π
ζ 2 = −i t2 . (A4.4)
2
1+ i
Then t= ζ. (A4.5)
π
If (A4.5) is substituted into (A4.3), then
1+ i
I= ∫

(
exp − ζ 2 dζ . ) (A4.6)
π 0

The integral in this expression is well known Gaussian error-integral, which is equal to

π / 2 . With that in mind if we recall (A4.6), we may obtain

I=
1
(1 + i ) . (A4.7)
2
On the other hand (A4.3) may be rewritten as
∞ ∞ ∞
 π  π  π 
I = ∫ exp i t 2  dt = ∫ [cos t 2  + i ∫ sin  t 2  ] dt =C F (∞ ) + i S F (∞ ) . (A4.8)
0  2  0 2  0 2 
From the comparison of (A4.7) and (A4.8) we may define
∞ π  ∞ π 
C F ( ∞ ) = ∫ cos  t 2  dt = S F (∞ ) = ∫ sin  t 2  dt = .
1
(A4.9)
0
2  0
2  2
Based on those values, the expressions (A4.1) and (A4.2) may be transformed to
∞ π  ∞ π 
C ( u 0 ) = ∫ cos  u 2  du − ∫ sin  u 2  du , (A4.10)
u0
2  u0
2 
∞ π  ∞ π 
S ( u 0 ) = ∫ cos  u 2  du + ∫ sin  u 2  du , (A4.11)
u0
2  u 0
2 
where the integrals in (A4.1) and (A4.2) with the integration limits (−∞, ∞) are twice

larger, than those with the limits (0, ∞) . Then (A4.10) may be rewritten as

π  1 u0  π 
C ( u0 ) =
1 u0
− ∫ cos  u 2  du − + ∫ sin  u 2  du =
2 0 2  2 0 2 
= S F ( u0 ) − C F ( u0 ) (A4.12)

Similar transforms may be applied to (A4.11). Then it may be rewritten as

132
S ( u 0 ) = 1 − [ S F ( u 0 ) + C F ( u 0 )] . (A4.13)

In (A4.12) and (A4.13) we used the following notations

π 
C F ( u 0 ) = ∫ cos  t 2  dt ,
u0
(A4.14)
0
2 
π 
S F ( u 0 ) = ∫ sin  t 2  dt .
u0
and (A4.15)
0
2 
(A4.14) and (A4.15) are called Fresnel's cosine and Fresnel's sine respectively.
Originally this two functions of u 0 had been presented in form of so called Cornu spiral

that is shown in Figure A4.1. It's easy to see from (A4.14) and (A4.15) that both, C F ( u 0 )

and S F ( u 0 ) are odd functions, i.e.

C F (− u 0 ) = − C F ( u 0 ) , and S F (− u 0 ) = − S F ( u 0 ) . (A4.16)

Figure A4.1. Cornu spiral

133
Chapter 3. BASICS OF ANTENNAS FOR RF RADIO
LINKS

3.1. BASIC PARAMETERS OF ANTENNAS

Antenna is defined by IEEE as “…part of the transmitting or receiving system that is


designed to radiate or receive electromagnetic waves” [1]. There are a variety of
antennas designed for a wide range of applications, which are being characterized by
the common set of parameters. Those are as follows:
• Radiation pattern/diagram
• Directivity and gain
• Input impedance
• Radiation resistance
• Effective area of aperture / Effective antenna-length
• Polarization
• Noise temperature
Regarding to the principle of reciprocity in electromagnetics the same antenna may be
used in both regimes, namely in transmitting (radiating) and receiving. The basic
parameters are the same in both regimes besides of the noise temperature, which is
specific just for receiving mode of the antennas and discussed in Chapter 6.

3.1.1. RADIATION PATTERN AND DIRECTIVITY


Assume a hypothetical isotropic radiating antenna of infinitely small size is placed in
point A of space. Then the parameters of radiated field of harmonic excitation will only
be distant-dependent. The isotropic antenna is just a useful abstraction in antennas and
propagation theory. For non-isotropic antennas the radiated electromagnetic field
strength at the observation point B depends not just on the distance between
communicating A and B points, but also on antenna's spatial orientation relative to

134
orientation of the straight line between them. In other words, if the radiating antenna is
placed in the origin of the spherical coordinate system (Figure 3.1a), then the field
strength at B point will depend not just on radial distance r but on azimuth angle ϕ and

zenith angle θ . For large distances 1, when antenna's relative dimensions are ignorable,
and the antenna is considered as a point in space, this statement may be expressed in
the following form:

E B (r , θ , ϕ ) = Ψ (θ , ϕ ) ,
K 
(3.1)
r
i.e. electric field is inverse proportional to distance. For the particular type of the radiating
element such as infinitesimal line current (Hertzian dipole) electric field is given by
(2.135), so from the comparison with (3.1) one may define a constant of proportionality

K=
Il  (θ , ϕ ) = sin θ . The phase exponent, exp[i (ω t − kr )] is omitted
iω µ 0 µ , and Ψ

in (3.1) for simplicity.
 (ϕ , θ ) is a function that presents dependence on the angular coordinates. In general,
Ψ
 (ϕ , θ ) is usually more complicated than that for Hertzian dipole and
the expression for Ψ
becomes a complex function of angular coordinates.

Figure 3.1. a). Spherical coordinate system: θ – zenith angle, ϕ – azimuth angle, γ –
angle of elevation. b). Surface element dS in spherical coordinates (dΩ –
elementary solid angle).

1
Observation point is assumed to be allocated in so called “far zone”, i.e. when the distance

r ≥ 2 D 2 / λ . Here D is a maximum dimension of the antenna.

135
For the lossless propagation medium the power flow density (magnitude of Poynting
vector) in any point of space may be defined by substitution of (3.1) into (2.74), i.e.

1 K2  2
Π (r , ϕ ,θ ) = 2
Ψ (ϕ , θ ) . (3.2)
2W r
Now recall, that at the observation point for any arbitrary direction (ϕ , θ ) the portion of

the total radiated power that flows through the surface element dS = r d Ω , which is
2

placed across the observation direction, is

K2  2
dPΣ = Π dS = Ψ (ϕ , θ ) dΩ , (3.3)
2W
where dΩ = sin θ dϕ dθ is solid angle subtended by the surface element dS from the
origin at the observation point as shown in Figure 3.1b. The factor that represents the
2
 (ϕ ,θ ) ≤ 1 . Based on (3.3) the term
angular dependence is a limited function, i.e. Ψ

radiation intensity may be introduced now as


dPΣ K 2  2
 (ϕ ,θ ) 2 ,
I Σ (ϕ ,θ ) = = Ψ (ϕ ,θ ) = I Σ , max Ψ (3.4)
dΩ 2W
where I Σ , max = K 2 /( 2W ) is the maximum value of the radiation intensity. It may be seen

from (3.4) that the radiation intensity, I Σ (ϕ , θ ) has physical meaning of the power that is
radiated within the unit solid angle (within the angle of one steradian) along the
observation direction. On the other hand, from (3.4), the function

 (ϕ , θ ) 2 = I Σ (ϕ , θ ) ,
Ψ (3.5)
I Σ , max
is solid geometrical figure that represents the angular distribution of the normalized
radiation intensity. In antenna theory and practice its known as a power radiation pattern
 (ϕ , θ ) is called amplitude radiation
of the antenna. The real part of complex quantity Ψ

pattern, Ψ (ϕ , θ ) (or just simply radiation pattern), whereas the imaginary part of it,

arg [Ψ (ϕ , θ )] is called phase pattern. All three, power pattern, amplitude pattern, and
phase pattern are solid figures, which may be presented either in spherical or in
Cartesian (rectangular) coordinates. An example of three-dimensional (3D) amplitude
radiation pattern is shown in Figure 3.2a for the vertically placed electric current point
source (elementary electric dipole) that is analyzed in Appendix-3. As may be noticed

136
from (A3.25) the electric field has a Z-axial symmetry, thus the radiation intensity is only
θ -dependent, which, in this case, is expressed in a simple form: Ψ ( θ ) = sin θ . In
spherical coordinates Ψ ( θ ) is toroidally shaped figure as shown in Figure 3.2a. It must
be noted that in engineering practice the use of spatial 3D radiation patterns is not
always convenient. The use of two-dimensional (2D) cuts of that solid figure is more
preferable in most of applications. They may be derived from Ψ (ϕ , θ ) by keeping either

θ = const, which results in so called conical cuts 1, or by keeping ϕ = const, which


results in vertical cuts. Examples of those cuts are given in Figures 3.2b and 3.2c for the
same electric current point source.
Directional properties of the antenna may be identified not only by radiation pattern, but
also by a numerical quantity called directivity, (marked D). It is a quantity that shows how
many times the total radiated power of the hypothetical isotropic antenna placed at the
transmission point must be increased in order to achieve the same field intensity at the
receiving point if antenna of directed radiation were be placed in the same transmission
point under the same conditions. In other words, if the antenna with directed radiation is
placed in point of origin and the radiated field is observed at point B, then the observer,
who assumes the radiating antenna to be isotropic, will attribute to it the total
hypothetical radiated power of

PΣ = PΣ ,hypothetical = D ⋅ PΣ . (3.6)

Here PΣ is the total power radiated by the real antenna.

1
Horizontal cut for θ = 90 0 is a particular case of it.

137
Figure 3.2. Examples of amplitude radiation patterns of the vertically placed
symmetric electric dipole: a). Three-dimensional pattern, b). Vertical cut,
c). Horizontal cut

In order to express directivity, D in terms of radiation pattern, the expression (3.6) may
be modified as follows:
′ 4π I Σ (ϕ 0 ,θ 0 ) 4π [Ψ (ϕ 0 ,θ 0 )]2

D(ϕ 0 ,θ 0 ) = = 2π π
= 2π π
, (3.7)

∫ dϕ ∫ I (ϕ ,θ ) sin θ dϕ dθ ∫ dϕ ∫ [Ψ (ϕ ,θ )] sin θ dϕ dθ
2
Σ
0 0 0 0

where ϕ 0 and θ 0 are angular coordinates of the observation point, D(ϕ 0 , θ 0 ) is the

value of directivity for the direction from antenna towards the observation point, ϕ and

θ are independent variables. Total power PΣ in denominator of (3.7) is found as an


integral of (3.3) taken for the entire space. The expression for the elementary solid angle
is dΩ = sin θ dϕ dθ . In the numerator we assume, that the radiation intensity along any

given direction (ϕ 0 , θ 0 ) for the isotropic hypothetical antenna, I Σ (ϕ 0 , θ 0 ) is constantly


spread all around the full solid angle 4π . Thus for the total power PΣ the integration is

simply replaced by the product 4π I Σ (ϕ 0 ,θ 0 ) .

If we take into account that the average radiation intensity is defined as the actual
radiated power PΣ divided by 4π , i.e.

138
2π π
1
I Σ , ave =
4π ∫ dϕ ∫ I Σ (ϕ ,θ ) sin θ dϕ dθ
0 0
(3.8)

then another definition for directivity may be formulated from (3.7) as


I Σ (ϕ 0 , θ 0 )
D(ϕ 0 , θ 0 ) = , (3.9)
I Σ , ave
namely, the directivity shows how much more the antenna’s radiation intensity in any
particular direction (ϕ 0 , θ 0 ) is greater than antenna’s average radiation intensity.

In some cases in engineering practices the term radiation pattern refers to the angular
dependence of directivity and is presented as D (ϕ , θ ) function. When the term “radiation

pattern” is used for the function D (ϕ , θ ) , then the term “directivity” is used to identify the

maximum value of that function, Dmax = D ϕ max , θ max . ( )


Depending on the type of antenna, its frequency range, working conditions, etc. the
directivity may vary in a wide range: from 1.5 to 106 (unitless), and even more.
 (ϕ , θ ) | 2 , Ψ (ϕ , θ ) , and arg [Ψ
Similar to | Ψ  (ϕ , θ )] the function D(ϕ , θ ) is also a 3D-solid

figure and, as said before, for the practical applications its sometimes useful to refer to
both, conical (elevation) and vertical (azimuthal) cuts as follows:
Dcon (ϕ ) = D ( ϕ , θ = const ) , Dvert (θ ) = D ( ϕ = 0, θ ) . (3.10)

If antenna has a predominant direction of radiation, then another parameter is applicable


to specify the directivity indirectly: that is the radiation beam width (or antenna’s angular
aperture). This parameter is defined as a range of angular coordinates that surrounds
the direction of maximum radiation. For those angles the value of D remains either
• greater than zero, which defines a first null beam width, FNBW = 2∆ϕ 0 , or

• greater than 0.5 Dmax , which defines a half power beam width, HPBW = 2∆ϕ 0.5 .

An example of sketch for the angular apertures 2∆ϕ 0.5 and 2∆ϕ 0 is shown on Figure

3.3. This approach is equally applicable for both, conical and vertical 2D cuts.

139
Figure 3.3. Example of 2D cut of the antenna radiation pattern in polar
coordinates: 2 ∆ϕ 0.5 is FNBW, and 2 ∆ϕ 0 is HPBW

Several lobes may occur in the radiation pattern of the directive antenna as shown in
Figure 3.3. The biggest that associates to the most intensive radiation direction is called
the main lobe, others are called side lobes, or back lobe.
For some applications the real radiation pattern (the example is shown in Figure 3.4a) is
idealized, i.e. replaced with the hypothetic pattern, where the total radiated power is
concentrated within the main lobe of the uniform intensity Dmax = D (ϕ max ,θ max ) = const,

as shown in Figure 3.4b.

Figure 3.4. a). Real 3D radiation pattern of the antenna, b). Equivalent replacement
(idealized) radiation pattern in form of 3D spherical sector

140
Recall that denominator of (3.7) is equal to total power radiated by the real antenna. The
replacement by the idealized pattern assumes that the same power is radiated within the
solid angle Ω , where the directivity remains constant and equal Dmax . From

normalization condition (3.5) we may have Ψmax = 1 , thus for the radiation solid angle
2

Ω the following expression may be written for denominator of (3.7).


2π π

∫ dϕ ∫ [Ψ (ϕ ,θ )] sin θ dϕ dθ = Ψmax Ω = Ω
2 2
(3.11)
0 0

It's easy to see, that in right hand side of (3.7) the numerator becomes equal 4π . Then
the expression (3.7) may be rewritten as

Dmax = D(ϕ max , θ max ) = , (3.12)

This formula defines the directivity as a ratio of the whole solid angle 4π and the solid
beam width Ω . It confirms the fact that the narrower the antenna’s main beam Ω is, the
greater the directivity is.
If the main beam of the antenna is single-lobed solid figure, then the solid angle Ω may
be approximately calculated as a product Ω ≈ (2∆ϕ 0.5 ) ⋅ (2∆θ 0.5 ) of both 2∆ϕ 0.5 , and

2∆θ 0.5 , if they’re small enough, so the expression (3.12) becomes [2,3]
4π 41253
Dmax ≈ = . (3.13)
(2∆ϕ 0.5 ) rad ⋅ (2∆θ 0.5 ) rad (2∆ϕ 0.5 ) deg ⋅ (2∆θ 0.5 ) deg
Note that If the radiation pattern is ϕ -independent, i.e. it has a shape of body of rotation
about Z-axes (see example in Figure 3.2), then antenna is called omnidirectional 1.
------------------------------------------------------------------------------------------------------
Example 3.1
Find the directivity of the electric current point source (Hertzian dipole).
Solution
• The amplitude radiation pattern may be defined from the expression for the
electric field (A3.27) given in Appendix-3.
Ψ (ϕ , θ ) = sin θ (E3.1.1)

• The solid angle of the main lobe is defined by (3.11).

1
Do not confuse with isotropic, which has a radiation pattern independent on both, ϕ and θ.
141
2π π 2π π

Ω = ∫ dϕ ∫ [Ψ (ϕ , θ )]2 sin θ dϕ dθ = ∫ dϕ ∫ sin 3 dθ = . (E3.1.2)
0 0 0 0
3
• The directivity may be found from (3.12), i.e.

Dmax = = 1.5 (1.76 dB) (Answer)

------------------------------------------------------------------------------------------------

3.1.2. RADIATION RESISTANCE AND LOSS RESISTANCE.


ANTENNA GAIN AND EFFICIENCY FACTOR

Generally speaking, an antenna may be considered as a complex load for the feeding
line (feeder) that has a complex impedance called the input impedance of the antenna
(see Figure 3.5). It is defined as
Z ant = Rant + i X ant = (Rant , 0 + Rant ,Σ ) + i X ant , (3.14)

where Rant , 0 is the one of the components of real part of the input impedance

representing thermal losses from metallic parts of the antenna, and X ant is a reactive

component of the input impedance.

Figure 3.5. Equivalent schematic representation of real antenna

Another component of the antenna's input impedance that is specific for the radiating
systems is radiation resistance, Rant , Σ . The physical meaning of Rant , Σ is a virtual

142
resistance, which would dissipate the radiated power PΣ irreversibly, if antenna is
replaced by the equivalent load at the feed’s terminal.
In other words, the power PΣ radiated by the antenna into free space is represented as

thermal loss on resistor R ant,Σ .


If antenna with the input impedance Z ant = Rant + i X ant is connected to a feed line with

characteristic impedance Zf , then from circuit theory, the maximum power transfers from
the feed line to the load (antenna) will be achieved if
X ant = 0 
 (3.15)
Rant = Z f 
The first of these two conditions means that antenna is tuned to feeding line (adjusted),
while the second condition means that antenna is matched with the feeding line. Even if
both conditions are satisfied, just part of the total power PTx from the transmitter, guided
by feeding line towards the antenna’s input, is utilized as a radiated power PΣ . Another

part of it will be wasted as an irreversible thermal loss P0 in metallic body of the

antenna. In other words, P0 is dissipated on equivalent resistor Rant , 0 as heat. Taking

into account, that the active power is in direct proportion to the resistance, the
expression for antenna efficiency is given as
PΣ PΣ Rant , Σ
η ant = = = , (3.16)
PTx P0 + PΣ Rant , 0 + Rant , Σ
which shows what part of the total power PTx, that inputs antenna from the feeder, is
radiated into free space. The meaning of antenna efficiency may be generalized by
including several types of losses other than just thermal losses that occur in antenna. In
addition to thermal losses, those are:
• Losses caused by the mismatch between the antenna input impedance and the
feeding line
• Polarization losses (for the receiving antennas only), caused by the mismatch
between antenna polarization and polarization of the radio wave received
• Losses caused by the non-efficient use of the antenna’s radiating aperture (for
the aperture antennas only), such as non-uniform distribution of the amplitude
and phase of the near field along the aperture
• Spillover losses of the energy for the aperture illuminating radiation

143
Now if PΣ is found from (3.16) as
PΣ = η ant ⋅ PTx (3.17)

and substituted into (3.6), then another important parameter, antenna gain may be
introduced as


G = η ant ⋅D = . (3.18)
PTx
From (3.18) the physical meaning of antenna gain may be stated as follows: it shows
how many times the power delivered to the input of hypothetical isotropic ideal (lossless)
antenna must be greater than the power delivered to the input of real antenna under
consideration to get the same field strength at the observation point in the same
conditions. In other words, G shows what the gain of power is when the real antenna
with directive radiation is used instead of hypothetic ideal isotropic antenna in the same
conditions.

3.1.3. ANTENNA EFFECTIVE LENGTH AND EFFECTIVE AREA OF


THE APERTURE
For practical applications its sometimes useful to introduce two closely related antenna
parameters such as effective length and effective area of the aperture. Note that these
parameters may have a real physical meaning just for specific types of antennas.
In order to introduce the antenna effective length, consider a particular case of lossless,
twin-wire open-end transmission line with the harmonic voltage source at the input
(Figure 3.6a). If we move the open endpoints of the wires away from each-other as
shown in Figure 3.6b, then the electric field lines will spread in the surrounding area.
Time variations of the electric field will produce a time-varying magnetic field and vice
versa, so this continuous electromagnetic process will result in generation of the radio
waves that freely propagate away from the point of origin. This is a way how to come up
with the simplest type of antenna, namely a dipole that is widely used for many
applications.
The distribution of currents along the wires in dipole is shown in Figure 3.6c with dashed
line. To define the effective length l eff of this antenna, the area under the current

distribution curve I (l ) is replaced by the rectangular (shaded region in Figure 3.6c) with

144
the same value of maximum current I max and the same area, so the height of this

rectangular, l eff represents the effective length of antenna 1. Mathematically it may be

expressed as
l/2
1
l eff =
I max ∫ I ( x)dx ,
−l / 2
(3.19)

where x is the integration variable, and the origin of coordinates is assumed to be at the
center-point of dipole. It is obvious, that one may get different values of the ratio l eff / l

for different current distributions I ( x ) along the wire. It also may be concluded from the
geometrical picture that the more evenly the current distribution is, the closer this ratio is
to unity. Some calculations using (3.19) are recommended to reader in problem P3.4.

Figure 3.6. a). The distribution of current I (l ) and voltage U (l ) along twin-wire
open-end line, b). Radiation illustration of the symmetric electric dipole, c).
Definition of the effective radiating length of the dipole, d). Equivalent schematic
diagram of the dipole as a symmetric two-port network.

1
General definition of the antenna effective length may be found in [2] or [3].

145
For a wire antenna with any current distribution the amplitude value of electric field
strength may be defined analogous to that for the ideal (Hertzian) dipole given by
expression (A3.28) of Appendix-3, i.e.
60π I max l eff
E max = , V/m, (3.20)
λr
Recall the wave impedance of the free space, W0 = 120π Ohms, then if (3.20) is

substituted into (2.74) the following expression may be obtained for the power flow
density at the considering distance r from the antenna.

 I max ⋅ l eff
2

Π = 30π   . (3.21)
 λr 

If it is assumed that the total power is radiated isotropically, i.e. PΣ = I max ⋅ Rant , Σ , then
2

power flow density for the distance r and for any direction may be defined as
′ 2
I max Rant , Σ

Π Isotr = = . (3.22)
4π r 2 4π r 2
Then based on definition (3.7) the directivity may be found as the ratio of (3.21) and

(3.22), which is the same as the ratio PΣ / PΣ :
2
 I max l eff 
30π  
D=  λr  . (3.23)
2
I max RΣ
4π r 2
Finally, (3.23) may be solved for the effective antenna length as

λ D Rant , Σ
l eff = . (3.24)
π 120
It is easy to show, that instead of (3.24) the expression

λ G Rant
l eff = (3.25)
π 120
may be used, where Rant is an input impedance of antenna in optimal regime (tuned and

matched). Generally speaking the expressions (3.24) and (3.25) may be applicable to
any type of antenna, and not just to dipoles or even not only to wire antennas. In other

146
words l eff may be considered as a universal parameter and may be used to define the
1
RMS of electromotive force inducted by radio wave in receiving antenna
emf = E ⋅ l eff , (3.26)

where E is the RMS value of electric field strength of the radio wave in close vicinity of
receiving antenna. For the analysis of reception of the radio waves, the antenna is
replaced by the Thevenin’s equivalent voltage source at the receiver's input with the
equivalent parameters that are emf voltage and Rant as the inner resistance of the

source (see Figure 3.7).

Figure 3.7. Equivalent replacement of the receiving antenna by Thevenin’s


voltage source. Rant is the antenna input impedance, emf is the electromotive force
induced by the radio wave, Z input is the input impedance of the receiver

As it is known from circuits theory, maximum power is transferred from source to load
(i.e. maximum efficiency of the antenna-to-input is achieved), if Z input = Rant . Then the
2
power that inputted to the receiver is equal to

1
This expression is true only if antenna polarization is the same as polarization of the received
radio wave, i.e. if there is no polarization loss. Generally effective length is introduced as a vector
with the magnitude defined by (3.26) and direction the same as polarization. Hence (3.26) must
be replaced by the dot product of two vectors.
2
When Z ant = Rant + iX ant and Z input = Rinput + iX input are complex, then optimum conditions

means both, matching Rant = Rinput and tuning X ant = − X input are satisfied.

147
2
e2 E 2 l eff
PRx = = . (3.27)
4 Rant 4 Rant

An alternate to the effective length of antenna is the antenna’s effective are of the
aperture. Recall, that the magnitude of Pointing vector at the observation point is the
power flow density, which is expressed by (2.74), so at this point in free space
2
E E2 E2
Π= m = = , (3.28)
2W W 120 π

where E = E m / 2 is the RMS value of the electric field. The effective area of the

antenna's aperture across the propagation direction, S eff is defined as a parameter that

allows converting power flow density of the radio wave into received signal power, and
backwards in transmitting antennas. Thus the total power absorbed by receiving antenna
may be found as

E2
PRx = Π ⋅ S eff = ⋅ S eff . (3.29)
120 π
If (3.25) and (3.27) are substituted into (3.29), then the effective area of the aperture
(effective aperture) may be found as
G λ2
S eff = . (3.30)

Only for several types of antennas the term effective area of the aperture has a physical
meaning. For the antennas, such as reflector type antennas, horn antennas, lens
antennas, etc. where radio wave radiation/reception occurs from the geometrical
aperture, the term antenna's effective aperture relates to the antenna's geometrical
aperture, i.e. to the surface, where electromagnetic field is distributed during radiation or
reception. For those types of antennas the effective aperture S eff relates to real

geometrical aperture S as
S eff = ν ⋅ S , (3.31)

where ν is a aperture utilization factor called aperture efficiency. It shows what part of
the geometrical aperture is being used effectively for radiation / reception. It may be
realized, that ν is always less than one, and depends on field distribution of radio wave
along geometrical aperture: the closer to a uniform distribution the field is across the

148
aperture, closer the value of ν is to unity. In practical applications the value of ν may
vary approximately from 0.5 to 0.8.

3.2. GENERAL RELATIONS IN RADIO WAVE


PROPAGATION THEORY

Suppose a hypothetical isotropic transmitting antenna is placed at a point A in free


space. Recall the expression (3.22) for the power flow density at reception point B at the
distance r


Π Isotr = (3.32)
4π r 2
that is found as a surface density of the power uniformly distributed along the spherical
surface of the radius r. Now we use (3.28) for that isotropic radiator in free space and
combine with (3.32). Then the effective field strength at the receiving point B for this
ideal case may be defined as

30 PΣ′
E Isotr = . (3.33)
r
For the same free space, when a real transmitting antenna with the gain GT and with


the input power PT is used, then PΣ must be replaced by the product PT GT in (3.32)
and (3.33). Then field induced in ideal conditions (free space) by the real antenna at the
observation point B may be written as

30 PT GT
E0 = , (3.34)
r
and the power flow density is
PT GT
Π0 = . (3.35)
4π r 2
At the observation point, the power generated by the incoming radio wave at the output
of the receiving antenna may be found if Π 0 is multiplied by the receiving antenna

effective aperture (3.30), i.e.

149
PT GT G R λ2
PR , 0 = , (3.36)
(4π r )2
where G R is the gain of receiving antenna.
As mentioned, both expression (3.34) and (3.36) are derived for so called reference
condition, i.e. for ideal propagation path, that is known as a reference path 1, when the
propagation in free space (vacuum) is considered. In order to take into account the real
conditions, when the propagation is affected by a number of factors, such as
attenuations and scatterings in atmosphere, the impact of the lossy earth’s ground, etc,
the propagation factor, F is introduced as a multiplier for E 0 . Propagation factor is

officially defined as follows: “For the time-harmonic wave propagation from one point to
another, the ratio of the complex electric field strength at the second point to that value
which would exist at the second point if propagation took place in a vacuum” [4]. It is
apparent that in general, the propagation factor is a complex quantity, i.e. it affects both,
the amplitude and phase of the electric field strength at the reception point. It is apparent
also, that for the received power on real propagation path the value PR , 0 must be

multiplied by the square of propagation factor, F 2 . On the other hand it is to be noted


that the real natural propagation conditions are randomly variable, especially if we take
into account variations of the atmospheric refractive index, variations of attenuations in
atmospheric precipitations and in ionosphere, etc., which results in random variations of
the propagation factor around its median value. Thus the evaluations of the propagation
factor are expected to be done in statistical sense.
It may be concluded from the above definition, that for the real propagation conditions
the reference values E 0 and PR , 0 must be replaced by the following 2:

30 PT GT
E = E0 F = F, (3.37)
r
PT GT G R λ 2
and PR = PR , 0 F =2
F2. (3.38)
(4π r ) 2

1
Here and below reference conditions will be denoted by “0” subscript.
2
The effectiveness of feeding lines are not considered in (3.37) and (3.38). They may be
′ ′
emerged into the values of GT and G R as GT = GT ⋅ η F , T and G R = G R ⋅ η F , R , where η F , T
and ηF,R are transmission coefficients of the feeders on transmitting and receiving sites
respectively.

150
Expression (3.38) is referred to as Friis transmission formula. Both, (3.37) and (3.38) are
general expressions for engineering calculations in propagation problems. The overall
goal of the radio wave propagation theory is to define the value of F, by taking into
account all factors, affecting propagation between transmitting and receiving antennas
such as the impact of the earth’s ground and atmospheric effects, except of the free
space losses that are included so far in (3.34) and (3.36).
Equation (3.37) is more conveniently used for the frequency bands lower than HF, where
wire-type antennas are mostly used, and therefore the EMF induced by radio wave at
the terminals of the receiving antenna: in this case EMF is calculated by (3.26). For other
cases, when aperture-type antennas are in use (usually for frequency bands VHF and
higher) the expression (3.38) is more preferable. This expression may be written in
logarithmic units (decibels) in the following form:
PR , dB = PT , dB + GT , dB + G R , dB − L0, dB − L A, dB , (3.38a)

where PR , dB is the power received (called power level of the received radio wave), PT , dB

is transmitted power, G R , dB and GT , dB are the gains for receiving and transmitting

antennas respectively,
 4π r 
L 0, dB = 20 log  , (dB) (3.39)
 λ 
is the free space loss or the reference loss, and is common for any propagation path.
The last term in right hand side of (3.38a)

1
L A, dB = 20 log   , (dB) (3.40)
F
is the propagation factor in logarithmic form that physically expresses the additional
losses relative to the reference conditions.
Formula (3.38a) is a logarithmic form of Friis formula, is sometimes referred to as an
equation of the RF communication link budget, which is widely used for the engineering
estimates in RF communication systems design. For the practical applications a term
total propagation loss is introduced as a ratio between transmitted and received powers.
It may be defined based on (3.38) as

 4π r 
2
P 1 1
Ltot = T =  ⋅ ⋅ 2, (3.41)
PR  λ  GT G R F

151
which shows how many times the received power will decrease while propagating from
transmitting antenna's input to receiving antenna's output. It may be seen from (3.41),
that the total loss on propagation path may be reduced by proper choices of GT and G R
antenna gains. However, for the generalized approaches, the antennas' properties must
be excluded from (3.41). Having that in mind the more convenient form of the overall
propagation loss between corresponding points caused only by the propagation path
may be defined as

 4π r  1
2
P EIRPT
L = T GT G R = =  , (3.42)
PR ( PR / G R )  λ  F 2
where a new useful term is introduced here as
EIRPT = PT GT (3.43)
called effective isotropic radiated power (EIRP) that shows how much power must be
radiated by a hypothetical isotropic antenna, placed in the same transmitting point, in
order to achieve the same field strength at the reception point. Analogous to EIRPT in

(3.42) a physical meaning of denominator, PR / G R is the power at the output of the


hypothetical isotropic receiving antenna, considered as predicted effective isotropic
received power. Note, that this term is not widely used in engineering practices.
Friis formula as well as the expression for communication link budget is useful tool for
the communication radio links engineering analysis and assessments.
The propagation phenomena impact considerably not just communication, but also the
RADAR 1 systems performance. In traditional radars the radiated signal in form of a
series of short pulses with a low duty cycle 2 is used. For the target ranging, i.e. for the
measurement of the distance from radar station to target a time delay between sounding
(direct) pulse and its echo replica, t R is defined and then is further transformed into the
value of range by using the following formula [5 - 6]
ct R
r= , (3.44)
2
where c is the velocity of electromagnetic waves in free space.

1
Abbreviation of the “RADio Detection And Ranging”.
2
Pulse duty cycle is a ratio between pulse “active” state and period of repetition, i.e. is the
duration of the pulse relative to period of its sequence.

152
For finding the direction of target the antennas with a sharp “pencil” shape beam, or
vertically positioned “fan” shape beams are used. Both, ranging and direction finding
may be considerably affected by the parasitic reflections from the earth’s surface (Figure
3.8), as well as by refractions and absorptions in the atmosphere. The free space
propagation losses, as well as the disturbances caused by the earth and the atmosphere
relative to the reference path may be taken into account in the radar equation while
estimating the power budget for the radar links, similarly to what Friis formula allows for
the communication links.

Figure 3.8. Signal propagation pattern on radar targeting track. r – target range
(real distance), d – horizontal distance

In order to obtain the radar equation, we first introduce a term radar cross section (RCS)
that is used as a “…measure of reflective strength of a radar target…” [4]. The following
explanations are to clarify this statement:
• Consider Π 0 is power flow density of the radio wave incident on a target with

specified direction of incidence


• For any direction of the scattered radio wave a hypothetic flat PEC surface of the
area σ RCS is used, so that the total power scattered along that direction may be

found as a product Π 0σ RCS

• In other words RCS is the area of hypothetical PEC surface, σ RCS that is used for

the equivalent replacement of target, and creates the same mirror-reflected


power in direction of observation, as the power of the wave that is scattered from
the real target

153
• Apparently RCS is highly dependent on signal frequency, shape of the target, its
orientation, and materials used
Based on above explanations, the power flow density of the echo signal that comes back
from target to the radar station point of location may be found if Π 0σ RCS is divided on the

area of spherical surface with the radius r, i.e.


PT GT 1
Π 0, radar , received = σ RCS . (3.45)
4π r 2
4π r 2
In order to obtain the amount of total power of the echo signal that is received by the
radar antenna, this value must be multiplied by the effective area of the antenna, 1 which
is defined by expression (3.30). Hence the power received in reference condition is
PT G 1 Gλ2 PT G 2 λ 2
PR , 0 = Π 0, radar , received S eff = σ RCS = σ RCS . (3.46)
4π r 2 4π r 2 4π (4π ) 3 r 4
This expression does not count the impacts caused by the real propagation conditions,
such as earth’s and atmosphere’s impacts, and is valid for free space only. As in case of
Friis formula all propagation phenomena may be included into consideration by involving
the relative loss factor, so thus (3.46) may be replaced by
PT G 2 λ 2
PR = PR , 0 F 4 = σ RCS F 4 . (3.47)
(4π ) r
3 4

In contrast to (3.38), where F 2 is implemented to count for the losses in real condition
on communication links, in (3.47) the propagation factor is counted to the power of four
because the signal is affected twice: for direct propagation, and then for the backwards
propagation of the echo. The radar range equation may then be found from (3.47) by
solving it for the distance, i.e. (see also [5])
1/ 4
 P G 2λ 2 
r =  T 3 σ RCS F 4  . (3.48)
 (4π ) PR 
If the sensitivity of the radar receiver is denoted as PR , min 2, then the maximum

detectable range of the radar is

1
In most widely used so called "mono-static" radars the same antenna is used for sounding and
reception of the echoes, i.e. GT = G R = G is assumed.
2
PR , min is the lowest signal power the receiver is able to “sense”.

154
1/ 4
 P G 2λ 2 
rmax = T 3 σ RCS F 4  . (3.48a)
 (4π ) PR , min 
The values of F may be estimated based on the materials provided in chapters 4 and 5,
and some estimates for PR , min may be found in Chapter 6.

REFERENCES

[1] IEEE Std 145-1983. IEEE Standard Definitions of Terms for Antennas
[2] Stutzman, W.L., Thiele, G.A. Antenna Theory and Design. John Wiley & Sons, Inc.
1998.
[3] Balanis, C.A. Antenna Theory. Analysis and Design. John Wiley & Sons, Inc. 2005.
[4] IEEE 100. The Authoritative Dictionary of IEEE Standard Terms. Seventh Edition.
IEEE Press 2000
[5] Collin, R.E., Antennas and Radio Wave Propagation, NY: McGraw-Hill Book
Co.,1985
[6] Skolnik, M.I., Introduction to RADAR systems, 3-rd edition, NY: McGraw-Hill Co.
2001.

PROBLEMS

P3.1. Derive the expression (3.25) based on expression (3.24).


P3.2. Confirm the expression (3.13).
P3.3. Based on results of problem P2.24 calculate how many times the diameter of
parabolic reflector antenna must be increased in order to achieve the same gain
increase, as in case of using ring-type director. What the advantages and disadvantages
of those two methods are for different applications. Explain.
Answer: 7
P3.4. Use (3.19) to calculate the effective length l eff for the following wire antennas:

a) Ideal dipole (Hertzian dipole) with uniform current distribution: I ( x) = I max

 2 x 
b) Short dipole ( l << λ ) with linear current distribution: I ( x) = I max 1 − 
 l 

155
λ π x 
c) Half-wave dipole ( l = ) with sinusoidal current distribution: I ( x) = I max cos 
2  l 
 2π x 
d) Full-wave dipole ( l = λ ) with sinusoidal current distribution: I ( x) = I max sin  
 l 
l is the length of dipole, I max is the maximum current.
Tabulate the results. Are there any differences between the 3rd and the 4th cases?
Explain.
Ans: a). l eff = l , b). l eff = 0.5 l , c). l eff = 2λ / π , d). l eff = 2λ / π

P3.5. Calculate the diameter of the parabolic reflector antenna of the gain G = 50 dB for
the frequency 10 GHz. Ignore all types of losses, and assume that the aperture
efficiency is equalν = 0.65 .
Answer: 3.74 m
P3.6. Find the whole solid angle by integrating the expression dΩ = sin θ dϕ dθ along
angular coordinates.
Answer: Ω = 4π
P3.7. Normalized power pattern of the omnidirectional antenna is approximated by the
2
 (ϕ , θ )
expression Ψ = cos n θ for 0 ≤ θ ≤ 90 deg. For the values of n = 0, 1, 2, and 3

1. Sketch the graph of the power radiation patterns in Cartesian coordinates


2. Define HPBW for all values of n by calculation based on the formula, and from
the graph. Compare the results
3. Calculate directivity for the above values of n by using formulas (3.7) and (3.12).
Compare the results
Answer: n HPBW, deg D, unitless D, dB
0 - 2 0
1 120 4 6
2 90 6 7.8
3 75 8 9

P3.8. A monostatic reconnaissance radar system radiates monochrome pulses with the
duration 0.1 µs and duty cycle of 10 -4 . Calculate the smallest RCS area of the target if

156
transmitter’s peak power is PT = 1 MW, frequency is 10 GHz, sensitivity of the receiver’s
is -100 dB/W, antenna gain is 50 dB. Perform the calculation for the reference condition.
Answer: 11.16 m2
P3.9. A geostationary satellite broadcasts direct TV program at the frequency 12 GHz.
The onboard antenna has EIRP = 70 dB / W. Estimate approximate area of effective
aperture of the receiving ground dish antenna without taking into account additional
losses such as polarization loss, atmospheric absorptions, etc. Take the distance from
earth to satellite equal 36000 km, and the receiver's input power -90 dB / W.
Answer: 1.63 m2.

157
Chapter 4. IMPACT OF THE EARTH SURFACE ON
PROPAGATION OF GROUND WAVES

As mentioned above, the main goal of the radio wave propagation theory is to estimate
field strength at the observation point, if both, transmitting antenna at the point A, and
receiving antenna at the point B are placed apart as end points of the real propagation
path in natural environment. In this chapter we’ll discuss only the impact of the earth’s
surface, i.e. no atmospheric effects are taken into account.
For the communicating antennas placed above the earth’s ground surface the initial details
to be specified are:
• Transmitted power, PT

• Frequency, f (or free space wavelength, λ 0 )

• Antennas elevations, h1 and h2


• The distance, r between corresponding points A and B
• Electromagnetic parameters of the earth’s ground and specifics of the landscape
profile along the propagation track from A to B
Only the median value of the field strength (or median value of the received power) will be
targeted, as no atmospheric fluctuation processes are taken into account at this step of
analysis.

4.1. PROPAGATION BETWEEN ANTENNAS ELEVATED


ABOVE THE EARTH SURFACE: RAY TRACE APPROACH

By elevated antennas we mean that the heights h1 and h2 of the two antennas are
significantly greater than the wavelength, and that the feeding lines are non-radiating.
Practically, antennas may be considered as elevated in most cases for frequency bands
VHF and higher. For those frequency bands the cross-sectional dimension of the spatial
area that is actively involved into the canalization of the energy of radio wave (first Fresnel
zone) is much smaller than the antenna heights and propagation path dimensions.

158
Therefore the ray-tracing approach, which is specific for optical applications, may be
applied here for further analysis.

4.1.1. FLAT EARTH APPROXIMATION CASE STUDY

Now we’ll start with the simplest case, namely:


• The earth’s surface is considered to be smooth and flat for the limited distances
• Transmitting and receiving antennas are considered to be elevated above the
ground level
Figure 4.1a illustrates the elevated antennas placed in the A and B points above the flat
and smooth earth’s surface. Amongst the infinite number of rays that are emitting from
radiating antenna A, just two rays are capable of reaching the receiving point B: a direct
Ray-1 (AB path) and a reflected ray-2 (ACB path). The effective value of the field strength,
caused by a direct ray-1 is defined by (3.34), which is written for free space. Therefore the
field oscillations at the receiving point may be represented as follows:

60 PT GT i (ω t −k r ′ )
E1 (t ) = e , (4.1)
r′
Here the amplitude is found by multiplication of the effective (RMS value) by 2 . r ′ is the
distance AB.

Figure 4.1 Ray tracing patterns for the antennas above the flat earth‘s
surface. a). Between antennas on LOS radio communication link
b). For ground-based radar

159
The field strength induced by ray-2 at the receiving point may be written by taking into
account the reflection from point C, thereby according to general expression (2.112), it may
be written as:

60 PT GT  i (ωt −k r ′′ ) 60 PT GT
E 2 (t ) = Γe = Γ e i (ωt −kr ′′−Φ Γ ) , (4.2)
r ′′ r ′′
where r ′′ represents ACB distance and Γ
 = Γ e − iΦ Γ is the reflection coefficient at the

reflection point C.
In general both E 1 and E 2 are nominated as complex vectors with different directions in
space, so their superposition at the receiving point must be taken as a vector sum.
However, in reality one may apply simplifications, based on the fact, that h1 and h2 are

usually much smaller, than distance r, thus vectors E 1 and E 2 become nearly collinear in
space and the vector sum may be replaced by algebraic sum of two complex quantities:
E = E 1 + E 2 . (4.3)

Another approximation is r ≈ r ′ ≈ r ′′ (see Figure 4.1a), whereby after substitutions of (4.1)


and (4.2) the expression (4.3) may be written as:

E (t ) =
60 PT GT
r
[ ]
1 + Γ e − i (k∆r + Φ Γ ) ei (ω t − k r ) . (4.4)

Here
∆ r = r ′′ − r ′ (4.5)
denotes a difference in paths between direct and reflected rays.
As is evident, the approximation r0 ≈ r ′ ≈ r ′′ is not applied to the expression in parenthesis

for the phase as the total field of E (t ) is much more "sensitive" to phase changes than to

the variations of the amplitudes of direct or reflected waves. Therefore, eliminating ∆ r


from the phase term may cause significant misrepresentation in further analysis. The
magnitude of the complex quantity E (t ) may simply be determined from (4.4), and
converted into the effective value, which results in:

30 PT GT
E= 1 + 2Γ cos(k ∆ r + Φ Γ ) + Γ 2 . (4.6)
r
If (4.6) is compared with (3.34), then an additional factor (multiplier) may be determined as:

F = 1 + 2Γ cos (k ∆ r + Φ Γ ) + Γ 2 . (4.7)

160
This multiplier occurs as a result of including the ground, i.e. expression (4.7) takes into
account the impact of the earth's surface on radio wave propagation. It becomes obvious,
that if earth's impact disappears ( Γ = 0 ), then F = 1 and propagation conditions are back
to ideal (3.34). In other words, (4.7) represents nothing but an additional "loss" relative to
the ideal (reference) condition, i.e. it is the propagation factor for this simple case.
Another approximation may be used if the elevation angle γ and distance ∆ r are

expressed in terms of the heights h1 , h2 and the distance r .

h1 + h2
tan γ = ≈ γ << 1 . (4.8)
r
In Figure 4.1a consider point A" a mirror image of the transmitting antenna A. Then, from
the ADB and A"DB triangles, the distances AB ( r ′ ) and A”B ( r ′′ ) may be defined as:

 h − h2 
2

r′ = r 2 + ( h1 − h2 ) = r 1+  1  ,
2
(4.9)
 r 

 h + h2 
2

r ′′ = r + ( h1 + h2 ) = r 1+  1  .
2 2
(4.10)
 r 
We can expand these two expressions into the Taylor series, with only the first two terms
left for simplicity, i.e.

 1  h − h2  2 
r ′ ≈ r 1 +  1
(h − h2 )2
 =r+ 1 (4.11)
 2  r   2r

 1  h + h2  2 
r ′′ ≈ r 1 +  1
(h1 + h2 )2 .
and:   = r + (4.12)
 2  r   2r

From (4.11) and (4.12) the difference of the ray paths may be defined as:
2 h1 h2
∆ r = r ′′ − r ′ ≈ . (4.13)
r
Note then, that Expression (4.7) contains also a reflection phase Φ R , which was studied in
chapter 2 in greater detail. Particularly from Figures 2.19 and 2.20, one may conclude, that
in case of horizontal polarization, for a very small elevation angle γ << 1 (i.e. when the
angle of incidence is close to 900) the reflection phase tends to become 1800,
(i.e. Φ R ⇒ 180 0 ). Thus after simple trigonometric transforms, and taking (4.13) into
account, (4.7) may be presented as

161
4π h1h2
F≈ 1 + Γ 2 − 2Γ cos . (4.14)
λr
Reliance of F on distance r is represented in Figure 4.2a. It may be noted that the graph is
limited from the top and bottom by the envelopes 1 + Γ and 1 − Γ . The shapes of those
two envelopes are defined by the reflection coefficient Γ which is distance-dependent
indirectly, as variations of r are causing variations in γ and ϕ . Γ is in direct dependence
on either of those. It must be emphasized, that the magnitude of the reflection coefficient
relies on polarization of the radio wave, and on electromagnetic parameters of the ground
at the reflection point C. As one may note from Figure 4.2a deep oscillations are quite
specific to the F (r ) graph.
The maximum values of F are obtained when
4π h1h2
cos = −1 . (4.15)
λ rn , max
Therefore the distances rn , max for the maximums may be derived from (4.15) by

establishing proper values of cosine argument, as


4π h1h2
= (2n − 1) π , n = 1, 2, ... (4.16)
λ rn , max
Solution of (4.16) for rn , max results in

4 h1 h2
rn , max = . (4.17)
λ (2n − 1)

Figure 4.2. a). Distance dependence of the propagation factor for


communicating antennas elevated above the earth’s surface. b). Angular
dependence of the radiated field above the earth’s surface.

162
From (4.14) it may be noted that the greatest values of F are achieved for the cosine (4.15)
equal to negative unity, then the propagation factor becomes
Fmax = 1 + Γ , (4.18)

From (4.17) we may define the distance r1, max of the first maximum of propagation factor

by assuming n = 1 . This is the largest distance from the radiating antenna for the
sequence of the maximums of the propagation factor found as
4 h1h2
r1, max = . (4.19)
λ
Least values of F = Fmin may be obtained, when the cosine is equal to positive unity, i.e.:

4π h 1 h2
cos = +1 . (4.20)
λ rn , min
Thus the argument sustains the following discrete sequence:
4π h1h2
= 2π n , n = 1, 2, ... (4.21)
λ rn , min
Solution of (4.21) for rn , min results in

4h1h2
rn , min = . (4.22)
2λ n
At those distances the propagation factor achieves its minimum value
F min = 1 − Γ . (4.23)

Actually, the first minimum in (4.22) may appear for n = 0 , but it is not counted, as rmin
tends to infinity for that particular value of n . Thus we’ll start counting distances of the
minimums for the propagation factor from n = 1 (the furthest distance) and higher.
The physical meaning of the results obtained is as follows: The sequence of interchanging
maximums and minimums on the graph for propagation factor F (r ) is a result of
superposition between direct ray (1) and the ray reflected from the earth's surface (2) 1 .
Maximums of the propagation factor (constructive interferences) appear when the phase
shift between those rays, caused by difference in path lengths ∆ r becomes equal to the
odd number of π -s, whereas minimums (destructive interference) appear when it becomes

1
This is particular case of so called multipath propagation.

163
equal to the even numbers of π -s. As mentioned above, an additional inversion of the

phase for reflected wave comes from the reflection point ( Φ Γ = 180 0 ).
For vertically polarized radio waves, the evaluation of the propagation factor may be
performed similarly, and proposed to the reader as end-chapter problem P4.3. Note that
the reflected wave may not always exist on ground-to-ground, or ground-to-air (air-to-
1
ground) radio links. Existence of the reflected wave depends on the profile of landscape
as well as on the size of reflecting area surrounding the reflection point (see section
2.6.4c).
For radar applications it is useful to study not just distance-dependence but also γ -
dependence (elevation-dependence) of the field strength. Let’s move the observation point
B to infinity as shown in Figure 4.1b to insure consideration of long ranges from the
transmission point, which is specific for the radar applications. Direct and reflected waves
will become parallel. Assume that Ψ (γ ) represents the vertical cut of the normalized
amplitude radiation pattern for the transmitting antenna in free space (see Figure 4.2b).
Apparently if the highly directive transmitting antenna is pointed at elevation angle γ , as
shown in Figure 4.1b, then greater is γ , less the illumination of the ground is, therefore
less is the intensity of the reflected wave. It becomes evident, that in expression (4.7) for
the propagation factor the value of Γ must be replaced by Γ ⋅ ΨΓ to take into account both
causes of the intensity decrease: due to reflection loss ( Γ ), and due to less radiation
intensity for the Ray-2 ( ΨΓ ). Then (4.7) may be rewritten as

F = 1 − 2 Γ ΨΓ cos (k ∆ r ) + [ Γ ΨΓ ] ,
2
(4.24)

which, regarding IEEE terminology, is called pattern-propagation factor [24]. Note that for
the ground-based surveillance radars with pencil-shaped highly directive antennas only
targeting of the objects that are close to horizon line will suffer from the destructive
interference between direct and reflected waves. Indeed, for those targets γ ≈ 0 , thus

ΨΓ ≈ Ψ0 = 1 . Hence for the high reflective earth’s ground ( Γ ≈ 1 ) propagation factor


diminishes, i.e. target becomes “invisible” for the radar. Meanwhile for the high-elevated
targets ( γ ≠ 0 ) the value of ΨΓ sharply decays (along with the increase of ∆r ), which

results in F ≈ 1 , i.e. the performance of the radar becomes close to that in reference
conditions (free space). The same condition may exist for the satcom radio links, i.e. for

1
See Section 4.4 below.

164
earth-to-space, or space-to-earth communication paths. The impact of the atmosphere will
remain as the only predominant issue in those radio-link performances (see chapter 5).
In reconnaissance radars with fan-beam antennas, namely with narrow horizontal and wide
vertical beam widths, intensities of direct and reflected waves become comparable in wide
range of angles γ , thus expression (4.25) may approximately be kept in his original form

(4.7) by assuming Ψ0 ≈ ΨΓ . Then the expression for the difference between path lengths

of direct and reflected waves, ∆ r may be found based on Figure 4.1b to display it as a
function of γ :

∆ r ≈ 2h sin γ . (4.25)
For further estimates we undertake the same assumptions for the horizontally polarized
wave: Φ Γ ≈ 180 0 , and Γ ≈ 1 . Then (4.24) may easily be simplified to

30 PT GT   2π h  
E≈ 2 sin  sin γ   . (4.24a)
r0   λ  
It’s easy to realize, that the expression 2 sin [(2π h / λ ) sin γ ] is nothing but the propagation
factor that represents the earth’s impact on angular distribution of the total field. Note that
the coefficient 2 confirms the energy conservation principle, i.e. the radiated energy is
distributed in the upper hemisphere only, in comparison to reference conditions, where the
energy is distributed along the entire space. Therefore, the field strength is doubled for
some directions 1 compared to free space. The angular distribution of the radiated field is
shown in Figure 4.2b.
For the terrestrial radio links antenna beams are positioned fairly close to horizontal,
hence, as pointed above, the intensities of direct and reflected waves are comparable
(even if antennas are directive, and if reflection condition exists: see below). Then
propagation factor may approximately be found from (4.24a) if ∆ r is defined from (4.13)
to express it as a function of distance:

2π h1h2
F (r ) ≈ 2 sin . (4.26)
λr

1
Note that the spatial average of the field strength is not doubled. It may be shown however, that
the average power flow density will be doubled.

165
Here the sine may be replaced by the argument if conventionally it is not greater than 20
degrees (i.e. less than π / 9 radians). In other words, if the following condition is taken into
account:
2π h1 h2 π
≤ (4.27)
λr 9
Then (4.26) may be further simplified down to
4π h1h2
F (r ) ≈ . (4.28)
λr
From (4.27) one may realize, that (4.28) is applicable if the distance is limited to
18 h1h2
r ≥ rlim = = 4.5 r1, max , (4.29)
λ
where rlim is the lower limit of the distances for expression (4.28) to be applicable, and

r1, max is defined by (4.19). In other words (4.28) is applicable if the distance from the
transmission to the reception point is farther than about five times the distance of the first
maximum of the propagation factor F (r ) .
Now if we substitute (4.28) into (3.37) and (3.38), the following expressions for effective
electric field strength and received power may be written as
4π h1h2 4π h1h2
E ≈ E0 = 30 PT GT , (4.30)
λr λ r2
and

PR = PT GT G R
( h1 h2 )2 . (4.31)
r4
Some discrepancies may seem to appear if (4.30) and (4.31) compared to each other.
Namely there is a λ -dependence (or frequency-dependence) in (4.30) that isn't exposed
in (4.31) in directly. However the issue may be clarified if (3.30) is rewritten in the form
4π S eff
GR = (4.32)
λ2
for the receiving antenna gain. Formula (4.32) exposes the inverse λ2 dependence for the
received power when substituted into (4.31). If we now consider field strength E as being
proportional to the square root of power flow density at the reception point (i.e. PR / S eff ),

then first degree inverse λ -dependence for the field strength will occur as it does in (4.30).

166
Electric field in (4.30) is expressed in V / m unit. However, in engineering applications
more convenient units for the variables may be implemented by transforming expression
(4.30) into appropriate form:

2.17 PT , kW GT h1, m h2, m mV


E ≈ , . (4.33)
λm rkm 2
m

Here the use of the common units is: E is in millivolts per meter, PT , kW is in kilowatts, GT

is unitless, rkm is in kilometers, λ m , h 1, m and h 2, m are in meters. The inverse square

dependence of the field strength on distance is in contrast to the reference path (see
(3.34)), where it has a first order inverse dependence on distance. It is evident that the
earth's impact provokes decrease of field intensity with distance faster than it would occur
in free space. It may also be seen from a sketch of both, direct and reflected wave phasors
in a complex plane shown in Figure 4.3. Indeed if E 1 nominates a direct wave phasor

arbitrarily placed in a complex plane 1, then the position of the reflected wave phasor, E 2
may be defined by considering a reflection phase of 1800 at the reflection point, as well as
an additional phase shift, caused the difference between ray traces of direct and reflected
waves
4π h 1 h 2
∆Φ = k ∆ r = . (4.34)
λ r0

Figure 4.3 Electric field phasors for direct ( E 1 ), reflected ( E 2 ) and resultant ( E )
waves in complex plane

Due to condition (4.27), i.e. due to ∆Φ << 1 , the following approximate relation may be
defined from one of the triangles shown in Figure 4.3

1
Absolute initial phase of this complex vector is not important for the two-ray interference stand
point, therefore it is ignored here.

167
∆Φ
E = 2 E1 . (4.35)
2
While developing (4.35) it was assumed initially, that E 1 = E 2 due to R ≈ 1 , i.e. the

intensities of direct and reflected waves are nearly equal. Then (4.30) may easily be
achieved, if (4.34) is substituted into (4.35).
As one may realize, formula (4.30) and (4.34) are applicable only in case of propagation
over the flat and smooth earth's surface. As a demonstration of this approach, propagation
along the sea surface for limited distances (and at frequency ranges higher than VHF) is
considered below in Example 4.1. In this case sea surface produces almost ideal reflection
of the incident wave, which is one of the requirements for application of the proposed
formulas. Results thereby become relatively close to those predicted.

-------------------------------------------------------------------------------------------
Example 4.1
Find the value of the field strength at the receiving point at the distance r = 20 km for
the following initial conditions:
• Antenna elevations: h1, m = h2, m = 10 m

• Wavelength: λ m = 0.1 m

• Transmitting power: PT , kW = 1 kW , and

• Transmitting antenna gain: GT = 1000 (30 dB )

• The propagation path is to be considered aligned to the ocean’s surface,


thus ideal reflections may be considered.

Solution
18 ⋅10 ⋅10
From (4.29) rlim = = 18000 m = 18 km .
0.1
Comparison of the given distance with its limit, r = 20 km > rlim = 18 km shows that
both expressions (4.30) and (4.34) are applicable. Therefore, if we apply (4.33) the
result is:

2.17 1kW ⋅ 1000 ⋅ 10 m ⋅ 10 m mV


E≈ = 171 (-15.3 dB per V/m) (Answer)
0.1m ⋅ (20 km )
2
m
------------------------------------------------------------------------------------------------

168
4.1.2. PROPAGATION OVER THE SPHERICAL EARTH SURFACE

Further review of the results achieved in the previous two sections can be made by
counting the spherical curvature of the earth's surface, which is ignorable for small
distances, but likely to be counted for greater distances.
As a first step we determine the horizon distance ( r 0 ) for observation point A, which is

placed at the height ∆ h over the earth’s ground (see Figure 4.4.).

Figure 4.4. Definition of the horizon distance

The geocentric angle Θ is defined as


r0
Θ= , (4.36)
a
where a = 6370 km represents the earth's average geometric radius 1. From ACO triangle
−1
a  ∆h
cos Θ = = 1 +  . (4.37)
a + ∆h  a 
If both sides of (4.37) are expanded into Taylor series, then the following relation may be
written for the first order approximation:

Θ2 ∆h
1− ≈ 1− . (4.38)
2 a
Now we substitute Θ from (4.36) into (4.38) then the horizon distance may be found as:

1
To count the impact of the atmospheric effect in engineering designs geometric earth’s radius

MUST be replaced (from here below) with the equivalent one, as it is introduced in section 5.3

169
r0 ≈ 2 a ∆h . (4.39)

Now we refer to Figure 4.5 which demonstrates a ray tracing pattern for real propagation
conditions when both antennas are placed over the spherically convex earth's surface. The
common background between the "flat" approach presented in Figure 4.1 and the
"spherical" approach shown in Figure 4.5 may be developed if the MN plane tangential to
the reflection point C is considered, and the apparent (seeming) antenna heights h1′ and

h2′ are introduced instead of the real antenna heights h1 and h2 .

Figure 4.5. Ray tracing pattern for propagation between elevated antennas above the
convex earth's surface in support of the antennas’ apparent heights calculations

In real propagation conditions when


r01 << a , r02 << a 
, (4.40)
Θ1 << 1, Θ 2 << 1 
the apparent antenna heights may geometrically be approximated as:
′ 
h1 ≈ h1 − ∆ h 1
′  . (4.41)
h2 ≈ h2 − ∆ h 2 

If ∆ h 1 and ∆ h 2 are derived from (4.39) and substituted into (4.41) then result is as

follows:

170
′ r01
2

h1 ≈ h1 − 
2a 
2 . (4.42)
′ r0 2 
h2 ≈ h2 −

2a 
From (4.42) one may see, that the apparent values of the antenna heights h1′ and h2′ are
less than the real antenna heights. This will cause a lesser value of the calculated field
strength at the reception point, when they are substituted into (4.33), i.e.

2.17 PT , kW GT h ′1, m h ′2, m mV


E ≈ , , (4.43)
λm rkm 2
m
Where:
rkm = r01 + r02 (4.44)

illustrates a total horizontal distance between corresponding points in km.

4.1.2.1. EVALUATION OF THE DISTANCE TO REFLECTION POINT

The distances r01 and r0 2 may be defined based on the geometrical relations shown in

Figure 4.5. If we apply the Law of Sines to triangles AOC and BOC then the following two
equalities may be written.
h1 + a a
= (4.45)
sin (γ + 90 ) 0
sin [180 − (γ + 90 0 + Θ1 )]
0

h2 + a a
= (4.45a)
sin (γ + 90 ) sin [180 − (γ + 90 0 + Θ 2 )]
0 0

After simplification, (4.45) and (4.45a) may be rewritten as:


(h1 + a ) cos (γ + Θ1 ) = a cos γ (4.46a)

(h2 + a ) cos (γ + Θ 2 ) = a cos γ (4.46b)


After exclusion of γ from the systems 4.46 we may end up with

a a
cos Θ1 − cos Θ 2 −
a + h1 a + h2
= . (4.47)
sin Θ1 sin Θ 2
Now, the approximation (4.38) may be applied to (4.47) as follows:

171
h1 / a − Θ1 / 2 h2 / a − Θ 2 / 2
2 2

= . (4.48)
Θ1 Θ2
If geocentric angles Θ1 and Θ 2 are now expressed in terms of the earth’s radius and

proper curvilinear partial horizontal distances, ( Θ1 = r01 / a and Θ 2 = r02 / a ), then we may

rewrite (4.48) as

h1 − r01 /(2a ) h2 − r02 /(2a )


2 2

= (4.49)
r01 r02
along with r0 = r01 + r02 . (4.50)

Taking into account (4.42), the expression (4.49) may be presented as


′ ′
h1 h2
= (4.49a)
r01 r02
For the fairly small distance r0 , when the "flat earth approximation" is applicable, the

second terms in the numerators in (4.49) become negligible. Therefore (4.49) may be
1
simplified to
h1 h2
≈ . (4.51)
r01 r02
Thereby, the simple case of "flat earth", (4.51) demonstrates a direct proportion between
heights and partial distances r01 and r02 , which may be seen from the geometry of the

path, shown in Figure 4.1a. With that proportion in mind, expressions for r01 (or r02 ) may

be evaluated if r02 = r0 − r01 (or r01 = r0 − r02 are substituted into (4.51). The result is as

follows:
h1
r01 ≈ r0 , (4.52a)
h1 + h2
h2
r02 ≈ r0 . (4.52b)
h1 + h2
Unfortunately this simple direct proportion becomes useless for the case of greater values
of the horizontal distance r 0 , when "flat earth" approximation is no longer applicable. For

those cases of convex earth’s surface a graphical estimates are provided in [2, p.387].

1
The same relation may easily be found geometrically from Figure 4.1a.

172
The expressions (4.43) for electric field, (4.8) for grazing angle, and (4.13) for ∆ r contain

′ ′ ′ ′
only the product h1 h2 and the sum h1 + h2 but not the values of the apparent heights
themselves. For practical engineering applications in this case of the large enough
distance r0 , we may follow [1] by introducing the variables

r0
k = r01 / r0 , 1 − k = r02 / r0 , x = h2 / h1 , and p = . (4.53)
2ah1
1
Then (4.49) may be transformed into

1 − k 2 p 2 x − (1 − k ) 2 p 2
= . (4.54)
k 1− k
Based on the MATLAB numerical solution of the equation (4.54), the correction factors m
and n are evaluated in graphical forms in order to support the replacement of the product
′ ′ ′ ′
h1 h2 and the sum h1 + h2 by their apparent values h1 h2 and h1 + h2 respectively. In
other words, h1 h2 and h1 + h2 are replaced by using the following relations:

′ ′ ′ ′
h1 h2 = m (h1h2 ) , h1 + h2 = n (h1 + h2 ) . (4.55)
Families of graphs for correction factors m and n are shown in Figure 4.6 for various
discrete values of the parameter p within practically most convenient range of its changes.

Figure 4.6 Families of graphs of the correction factors m and n for calculation of the
apparent quantities in (4.55)

1
x = h2 / h1 or x = h1 / h 2 is applicable
Note that because of geometrical symmetry either
depending on which one satisfies the condition 0 ≤ x ≤ 1 .

173
For the practical calculations it is more convenient using a closed form equations for
precise estimates of the distance to reflection point given in [25], rather than numerical
evaluation of the expression (4.54). A system of those equations is given below:
r0  Φ +π 
r01 = − q cos  , (4.54a)
2  3 

where q=
2
( )
a (h1 + h2 ) + r0 / 2) 2 , (4.54b)
3

 2a (h 2 − h 1 ) r0 
and Φ = cos −1  . (4.54c)
 q3 
Note that:

1. The system (4.54a) – (4.54c) is applicable for the case h 2 > h 1 . For the opposite

case, h 2 < h 1 the heights h 2 and h 1 in (4.54c) must be flipped, and (4.54a) is to

be written for r02

2. For the practical engineering applications the value of the earth’s geometrical
radius a = 6370 km must be replaced by its equivalent value ae = 8500 km, as

shown in section 5.3 of chapter 5.

--------------------------------------------------------------------------------------------------
Example 4.2
Find the partial distances r01 and r02 as well as correction factors m and n for a radio link

with elevated antenna heights of h1 = 160 m, and h2 = 50 m. Make the calculations for the
following two values of total horizontal (curvilinear) distance between the corresponding
points: (1) r0 = 5 km, and (2) r0 = 50 km.

Solution
• x = h2 / h1 = 160 / 50 = 0.313
• For the first distance r0 = 5 km:

r0 5000
*** p1 = = = 0.11
2ah1 2 ⋅ 6370 ⋅ 10 3 ⋅ 160
*** Expression (4.54) becomes

174
1 − 0.112 k 2 0.313 − 0.112 (1 − k ) 2
=
k 1− k
*** Numerical MATLAB solution results in k = 0.761 , which is close to that
found from "flat earth” approximation (compare with k = 0.7619 found from (4.52a))
*** Then r01 = k r0 = 0.761 ⋅ 5 = 3.8 km, r02 = (1 − k ) r0 = 0.239 ⋅ 5 = 1.2 km


h1 = h1 − r01 /(2a ) = 160 − 3800 2 /(2 ⋅ 6370000) = 158.87 m
2
***


h2 = h2 − r02 /(2a ) = 50 − 1200 2 /(2 ⋅ 6370000) = 49.89 m
2
***

*** Then correction factors are found as:


′ ′ ′ ′
m = (h1 h2 ) /(h1 h2 ) = 0.99 (Ans), n = (h1 + h2 ) /(h1 + h2 ) = 0.994 (Ans)
• For the second distance r0 = 50 km:

r0 50000
*** p2 = = = 1.1
2ah1 2 ⋅ 6370 ⋅ 10 3 ⋅ 160
*** Expression (4.54) becomes
1 − 1.1 k 2 0.313 − 1.1 (1 − k ) 2
=
k 1− k
*** Numerical solution of this equation results in k = 0.687 . Then
*** r01 = k r0 = 0.687 ⋅ 50 = 34.35 km, r02 = (1 − k ) r0 = 0.313 ⋅ 50 = 15.65 km

h1 = h1 − r01 /(2a ) = 160 − 34350 2 /(2 ⋅ 6370000) = 67.38 m
2
***


h2 = h2 − r02 /(2a ) = 50 − 15650 2 /(2 ⋅ 6370000) = 30.78 m
2
***

*** Correction factors are found as


′ ′
m = (h1 h2 ) /(h1 h2 ) = 0.26 (Answer)

′ ′
n = (h1 + h2 ) /(h1 + h2 ) = 0.47 (Answer)
--------------------------------------------------------------------------------------------------------

4.1.2.2. DIVERGENCE OF ENERGY OF THE RADIO WAVE WHILE


REFLECTING FROM THE CONVEX EARTH SURFACE

Another phenomenon that relates to the earth's convex surface is the divergence of
reflected radio wave shown in Figure 4.7.

175
Figure 4.7 Illustration of divergence of the wave front of the radio wave while
reflecting from the spherically curved earth's surface

As illustrated in this figure, even if the wave front of the incident radio wave is considered
as a plane (1), after reflection it becomes curved (2). In simple terms, an additional
divergence of energy of the radio wave takes place which affects the value of the reflection
coefficient. The apparent (effective) value of the reflection coefficient is introduced as a
product of the real reflection coefficient R , and the quantity Ddiv , i.e.

Γ′ = Γ Ddiv , (4.56)

where Ddiv ≤ 1 is called the divergence factor and is defined by the following expression

[3, p.351]:
1

  2
2 ′ ′
 2 r0 h1 h 2 
Ddiv = 1 + 3
. (4.57)
 a  h ′ + h ′  
  1 2
 
In most of the engineering calculations this type of divergence factor is usually taken fairly
close to unity, i.e. for most of the practical cases the divergence of the reflected rays
caused by the convexity of the earth surface may be ignored.
--------------------------------------------------------------------------------------------------------
Example 4.3
Calculate and plot the distance-dependence of the effective (RMS) value of the electric
field strength in observation point for the following initial data:
• Range of distances: 1 km and up
• Frequency: 150 MHz (free space wavelength = 2 m)
• Antenna elevations: 10 m (transmitting), 50 m (receiving)

176
• Transmitting antenna gain: 31.6 (15 dB)
• Transmitter power: 1 kW
• Polarization: horizontal
• Propagation path is located over the sea water, thus the reflections may be
assumed as being close to ideal
Solution
1. First we define the maximal line-of-sight (LOS) distance as a sum of two horizon
distances (see Figure E4.3.1a):

rLOS , max = 2ah1 + 2ah2 = 2a ( h1 + )


h2 =

= 2 ⋅ 6370 ⋅10 3 ( 10 + )
50 = 36000 m = 36.5 km

Figure E4.3.1 Propagation ray patterns over the convex earth in Example E4.3.

2. Now we define criteria to evaluate the upper limit for the distance, where “flat earth”
approximation is applicable. That is
∆ h1 ∆ h2
≤ 0.1 , or ≤ 0.1 , whichever is appropriate (see Figure E4.3.1b)
h1 h2
Recall (4.42) and (4.52a). Then
2
∆ h1 r flat earth
h1
=
h1
2a h1 + h2
≤ 0.1 ; and r flat earth ≤
0.2a
h1
(h1 + h 2 ) = 21.4 km
Similarly

∆ h2
2

h2
=
r flat earth
2a
h2
h1 + h2
≤ 0.1 ; and r flat earth ≤
0.2a
h2
(h1 + h 2 ) = 9.6 km
The second of those two conditions is stronger, thus we’ll choose the final upper limit of
distances that is appropriate for the “flat earth” approximation as follows:
r flat earth, max = 9.6 km.

177
3. From inequality (4.29) we may define the lower limit for distances, when approximate
formula (4.33) is applicable.
18 h1h2 18 ⋅10 ⋅ 50
r ≥ rlim = = = 4500 m = 4.5 km
λ 2
4. Finally the distance limits are outlined as:
• Range-1: 1 < r ≤ 4.5 km use of the expression (4.26)
• Range-2: 4.5 < r ≤ 9.6 km use of the expression (4.33)
• Range-3: 9.6 < r ≤ 36.5 km use of the expression (4.43)
5. Calculation results are summarized in form of the graph shown in Figure E4.3.2 below.

Figure E4.3.2 Distance dependence of effective (RMS) electric field


strength in Example 4.3

Notes:
a). Range-1 contains maximums and nulls of the electric field as a result of constructive
and destructive interferences between direct and reflected waves
b). Distances for maximums of the electric field are consistent to those calculated from
formula (4.17) approximately because it exposes maximums of the propagation factor, but
not maximums of the electric field itself
c). Distances for minimums of the electric field are consistent to those calculated from
formula (4.22) precisely

178
d). Transition from Range-2 to Range-3 results in faster decay of the electric field caused
by the convexity of the earth’s surface
---------------------------------------------------------------------------------------------------

4.1.3. SPECIFICS OF PROPAGATION OVER A ROUGH AND HILLY


TERRAIN
The analysis of ground radio wave propagation near the earth, in Sections 4.1 and 4.2,
was based on the assumption that the earth's surface is ideally smooth (either flat or just
spherically convex). In reality, there are several types and different levels of random
roughness such as grass, stones, bushes, hills, residential and commercial buildings, sea-
waves, etc. All of these may affect propagation conditions, and are to be further
considered.
Suppose the plane wave front (1) is incidental to a random, rough surface as shown in
Figure 4.8.

Figure 4.8 Reflection of a radio wave from a rough surface


(small scale randomness).

After reflection the plane wave front will be disturbed and, referring to Hyugens-Kirchhoff's
principle, each infinitesimal element of the front may be considered as a virtual radiator of
spherical waves. Thus the randomly shaped reflected wave front (2) will cause a multitude
of scattered waves surrounding the main direction of reflection (direction of the specular
reflection). The random fluctuations of the reflected wave front are caused by the
difference in distances covered by deviant ray paths of incident wave. Two extreme rays,

179
EBG and DAF have the greatest variance in propagation distances. That variance may be
defined from triangle ABC as:
∆ r = 2 CB = 2 ∆ sin γ , (4.58)

where ∆ is the average height of the irregularities. Then, the fluctuations of the phase shift
are equal to:
2π 4π ∆ sin γ
∆Φ = k ∆ r = ∆r = . (4.59)
λ λ
It is widely recognized from observations, and conventionally agreed, that if those phase
fluctuations are less than π / 2 , then the roughness may be ignored and the reflection may
be considered as specular; otherwise it becomes diffuse. Therefore the condition for the
specular (mirror) reflection will be obtained from (4.59) if
4π ∆ sin γ π
≤ . (4.60)
λ 2
Thus the tolerable average height of the roughness may be found from (4.60) as
λ
∆≤ . (4.61)
8 sin γ
This expression is known as the Rayleigh criterion, which allows estimation of the
permissible value of average height of the roughness in order to consider reflections as
specular. As seen from (4.61) the allowable average height of the surface roughness
depends not just on wavelength, but also on slant angle γ to consider reflection as
specular. For instance, in microwave frequency bands, down to wavelengths of several
centimeters, even bushes and trees do not disturb specular reflections, for proper defined
value of γ . If (4.61) is satisfied then the formulas for reflection coefficients are applicable.
Otherwise the reflection is diffuse, and reflection coefficient may not be defined by the
approaches given in section 2.4. Part of the radio wave’s energy will scatter away from
direction of specular reflection, and the calculated value of R may highly distinguish from
predicted estimates. In that case, only approximate estimates may be performed based, for
instance, on data demonstrated in Table 4.1. Those data are provided for antenna heights
from 10 to 50 meters and for distances from 5 to 50 km.

180
Table 4.1
Wavelength, in cm
1.5 - 3 5 7-8 15 - 18
Type of surface
See/ocean water surface 0.2 – 0.45 0.65 – 0.85 0.8 – 0.95 0.9 – 0.99
Flat country, wet meadows - - 0.6 – 0.95 0.8 – 0.99
Flat forests 0.1 – 0.3 0.3 – 0.5 0.4 – 0.6 0.6 – 0.8
Hilly forests - - 0.2 – 0.3 0.3 – 0.5

Another particular issue is a divergence, which can occur when reflection from the top of
hills / elevated areas takes place. This would represent a case of a large-scale obstacle in
the path of a reflected ray trace, when the average size of the reflecting top of the obstacle
is much larger than the wavelength. A simplified sketch of this case is presented in Figure
4.9. Similar to formula (4.57) that represents the divergence caused by the regular ideal
spherical earth’s surface, the divergence factor for this case of a large-scale obstacle is
presented as
1

 2
r10 r20
2
∆h  2
Ddiv = 1 + 32  , (4.62)
 r0
2
(∆ r ) 2 H (0)
where sizes and distances are shown in Figure 4.9. H (0) denotes the clearance between
the top of obstacle and the direct ray trajectory AB when no atmospheric effects are taken
into account at this point (zero in parenthesis for H (0) is to call attention to this fact).
Later, in chapter 5, the tropospheric refraction will be discussed in detail, and additional
path clearance ∆H will be introduced, so the resultant path clearance is to be considered
as H (0) + ∆H . Here we’ll ignore atmospheric effects, and leave H (0) to the rest of this
chapter.

181
Figure 4.9 Reflection of a radio wave from large-scale obstacle

To clarify the meaning of the term "large-scale" it must be noted, that (4.62) would prove
applicable if ∆r is comparable or greater than horizontal dimension of the Fresnel zone as
defined by (2.179) and (2.180). Additionally, ∆h is to be much greater that the wavelength.
For a further reflection coefficient adjustment, the value of Ddiv from (4.62) must be

substituted into (4.56).


To utilize these formulas for engineering applications, it is practically more convenient to
express antenna heights h1 , h2 , and grazing angle γ in terms of horizontal distances r10

and r20 , shown in Figure 4.9, as well as H (0 ) , which was defined above as a direct ray

trace clearance. We will initiate these replacements by recalling Figure 4.1a, and using
similarity between triangles A'AC and B'BC.
h1 − h2 H (0 ) − h2 
= , 
r0 r20
h1 h2  (4.63)
= . 
r10 r20 
If the relative coordinate of the reflection point is denoted as
k = r10 / r0 (4.64)

then from the system (4.63) the following expressions for h1 , h2 and γ may be derived:

H (0 )
h1 = , (4.65)
2k
H (0 )
h2 = . (4.66)
2 (1 − k )
H (0 )
γ≈ , (4.67)
2 r0 k ( 1 − k )

182
The convenience of using (4.65), (4.66) and (4.67) is that H (0 ) and k are usually obtained
from the practical draft of the landscape profile, as shown in Figure 4.10. That is an
example of a drawing of the terrain profile for a 50 km single-hop LOS radio link along a
diverse and hilly landscape.

Figure 4.10 Longitudinal terrain profile of a single-hop LOS radio link

The reference level is sketched, to account for the curvature of the earth's surface
(represented by the dashed line). It may be shown that by switching from geocentric polar
coordinates to rectangular coordinates a spherical earth curvature will be transformed to a
parabolic, which is expressed as:

y=
x
( r0 − x ) . (4.68)
2a
Here a = 6370 km represents the radius of earth, x is an independent horizontal

coordinate (variable horizontal distance), and r0 is total horizontal distance between

corresponding points. The data table of distances and heights, which is typically known
from land-surveying investigations, is used to construct the path profile above the
reference level. Note that it is not necessary to use a sea level as a reference. Ordinates of
the path profile are shown along the vertical axis relative to the reference level.
Dealing with the clearance H (0 ) that is found from the terrain profile as well as with

unitless quantity k allows automatic accountability for the earth’s spherical curvature by
calculating the apparent heights h1′ and h2′ from (4.65) and (4.66) as:

183
′ H (0 )
h1 = , (4.65a)
2k

′ H (0 )
h2 = . (4.66a)
2 (1 − k )
Based on (4.65a) and (4.66a), Expressions (4.14), (4.30), and (4.61) may be transformed
to the following:

 4π H 2 (0) 
F ≈ 1 + Γ 2 − 2Γ cos  , (4.69)
 λ r0 4k (1 − k ) 
where Γ is a ground reflection coefficient,

30 PT GT 4π H 2 (0 )
E≈ , (4.70)
r0 λ r0 4k (1 − k )
λ r0 k (1 − k )
∆≤ . (4.71)
4 H (0 )
------------------------------------------------------------------------------------------------------------
Example 4.4
Calculate apparent heights h1′ and h2′ for the terrain profile given in Figure 4.10 for the

path clearance H (0 ) = 27,5 m , and horizontal distances r10 = 20 km , r20 = 30 km between

the hill and transmitting and receiving points respectively.


Solution
r10 20
• k= = = 0.4
r0 20 + 30

′ H (0 ) 27.5
• h1 = = = 34.38 m
2k 2 x0.4

′ H (0 ) 27.5
• h2 = = = 22.9 m
2 (1 − k ) 2 x(1 − 0.4)
Note: compare the results with real antenna heights shown in Figure 4.10.
----------------------------------------------------------------------------------------------------

184
4.1.4. OPTIMAL PATH CLEARANCE AND CHOICE OF THE
ANTENNAS ELEVATIONS
For the LOS propagation path (Figure 4.10) we will consider a case when field strength at
the receiving point becomes the same as in free space, even if the impact of the reflection
from the earth's surface is significant. In other words the case when the interference of
direct and reflected waves results in same field strength at the reception point as in free
space. Recall Figure 4.3, where the phasors of direct wave E 1 and reflected wave E 2 are

added in a complex plane. It becomes visible that the resultant phasor E will be equal to
the reference (free space) value of the field, E 1 if the phase angle is ∆Φ = 60 0 1. In this
case the gap between direct ray and the top of the hill, the path clearance becomes
optimal, and it is denoted as H opt (0 ) . In order to determine the value of that optimal path

clearance, H opt (0 ) the expression (4.34) must be considered equal to π / 3 . Now if we

replace h1 and h2 in accordance to (4.65) and (4.66), then the following expression may
be written:
4π h1 h2 4π H 2 opt (0 ) π
= = . (4.72)
λ r0 λ r0 4k1 (1 − k1 ) 3
After solving (4.72) for H opt (0 ) the result occurs as

1 λ r10 r20
H opt (0) = λ r 0 k1 (1 − k1 ) =
1
. (4.73)
3 3 r10 + r20
Expression (4.73) is derived based on the geometrical optics concept, i.e. via the ray-
tracing approach. The same expression may be developed based on diffraction theory, i.e.
by using the Huygens-Kirchhoff's Principle as it follows from (2.151), which is the same as
(4.73). The resultant field becomes equal to E B 0 shown in Figure 2.28b as field strength in

free space (reference field strength). For the real propagation paths the intensity of direct
wave is usually greater than that of the reflected wave. Therefore, Expression (4.73) is not
thoroughly correct. To achieve the same resultant E B 0 , the phase shift between direct and

1
Magnitudes of the direct and reflected waves are assumed equal to each other, i.e. E1 = E2 ,
meaning that the reflection is assumed to be ideal ( R = 1) . As shown in section 2.4, this assumption
is true for small values of the slant angle γ .

185
reflected waves must be greater than π / 3 , which results in approximately a 10% margin
for path clearance, i.e.
H (0 ) ≈ 1.1 H opt (0 ) . (4.74)

In some cases the landscape profile is such that the antenna elevations need to become
unrealistically high to obtain the desired optimal path clearance. For instance, if the hill
blocks the ray trace significantly, causing the antenna elevations to increase too
considerably, then either an active repeater, or an additional reflector or diffractor on the
hill top is needed to overcome the hill. The reader may refer to Section 2.6.4.b, where
passive ring-sector diffractor is considered as one of the possible means for overcoming
the blockage caused by the hill.

------------------------------------------------------------------------------------------------------------
Example 4.5
Find the power received at the observation point B of the single-hop LOS for the landscape
profile is shown in Figure 4.10, and for the following initial data-points provided:
• Transmitting power: PT = 0.5 W

• Carrier frequency: f = 4 GHz ( λ 0 = 7.5 cm )

• Transmitting/receiving antenna gains: G1 = G 2 = 10 4 ( 40 dB ) (losses in


feeding lines are ignored).
• Total distance: r0 = 50 km

• Geometrical path clearance: H (0 ) = 27,5 m (from Figure 4.10)

• Distance from the obstacle to transmitting point: r10 = 20 km

( k1 = 20 / 50 = 0.4 )
(Note: Regular atmospheric refraction is not to be taken into consideration)
Solution
1. The value of the optimal clearance may be found from (4.73) as:

H opt (0) = λ r0 k1 (1 − k1 ) = ⋅ 0.075 ⋅ 50000 ⋅ 0.4 ⋅ (1 − 0.4) = 17.3 m


1 1
3 3
which is less than the real gap, so the reflection is in effect.

2. To check if the Rayleigh criterion (4.61) is satisfied, the elevation angle is calculated
from the Formula (4.67):

186
H (0 ) 27.5
γ = = = 1.15 ⋅ 10 −3 rad .
2 r k (1 − k ) 2 ⋅ 50 ⋅ 10 ⋅ 0.4 ⋅ (1 − 0.4 )
3

Then the acceptable average height of the roughness at the reflection point is

λ 7.5 ⋅10 −2
∆≤ ≈ ≈ 8.2 m
8 sin γ 8 ⋅1.15 ⋅10 −3
It may be noted from Figure 4.10, that the average height of the roughness around C-point,
(i.e. the height of trees in forest area) is greater than 8.2 m, i.e. the Rayleigh criterion is
violated at the reflection point C. Therefore, the reflection is considered to be diffuse and
the reflection coefficient is found from Table 4.1 approximately as Γ ≈ 0.25 .
3. The propagation factor is found from (4.69):

 4π H 2 (0 ) 
F ≈ 1 + Γ 2 − 2Γ cos   ≈ 0.9 .
 λ r0 4k1 (1 − k1 )
4. Finally, the received power may be derived from (3.38):

PR = 0.5 ⋅
(10 ) 0.075
4 2 2
⋅ 0.9 2 = 0.58 µW (- 62.4 dBW) (Answer)
(4 ⋅ 3.14 ⋅ 50 ⋅10 )
3 2

----------------------------------------------------------------------------------------------------

4.1.5. PROPAGATION PREDICTION MODELS IN URBAN,


SUBURBAN AND RURAL AREAS
VHF, UHF, and higher frequency bands are widely used in urban and suburban areas for
radio and television broadcasting, personal and business communications, police and fire
service communications, emergency medical and other services, etc., as well as to
organize highly developed cellular phone networks. For these and other services, the
communication systems are utilized mainly to provide radio links between a base station
and applicable users 1 within the coverage area; i.e. to provide initial network coverage
over an area surrounding the base station.

1
The term “Mobile Stations” is annotated below to denote not only moving transmitters/receivers
(such as cell phones) but also, other services to include, consumer televisions with fixed antennas
placed on roof tops.

187
The accompanying terrain is typically populated with randomly orientated streets,
buildings, trees, and other fixtures at random orientation, positioning, and heights.
Thereby, the associated propagation path is typically oriented along a randomly shaped
terrain profile. Path analysis becomes difficult, as at least one of the corresponding
antennas will typically be "buried" between top and bottom levels of buildings, trees, etc.
Thereby, the received signal becomes randomly distributed in time and space. In that
case, when either the receiver or the transmitter is moving within the area of coverage, the
random spatial variations of field will be transformed into time variations and precise
predictions to define field strength at the receiving point may become impossible. Certainly,
it would require a detailed knowledge and analysis of the path profile for each position of
the corresponding stations including locations, distances, and electromagnetic parameters
of every tree, building, and obstacle within the area of coverage (i.e. the accumulation of a
huge amount of initial data that in most cases is not easily obtained and counted for). At
the same time, from designer's point of view, an analysis need not necessarily be precisely
exact by representing a predicted field for each point of the covered area; but simply
provide the designer of the radio-link or network with the averaged values (median values
in statistical terms) of the field around the base station. A straight forward, deterministic
approach (such as given in Example 4.3) is not an effective tool for a wireless network
designer as the amount of pre-processing and post-processing of data necessary would
prove excessive. Instead a "statistical sense" is more practically used for field strength
predictions, which are based on mathematical models with parameters adjusted to be
relevant to the particular area of coverage. Such adjustments are based on a multitude of
long-term observations and measurements in different types of areas such as urban,
suburban, and rural. Two standard classes of propagation prediction models for outdoor
transmitting and receiving antennas are typically considered: empirical models and
physical models. Analysis of potential propagation prediction models for cases where both
antennas are placed inside a building is not the subject of this current section, but may be
referred to [4], [5], and [15].
The empirical models are mathematical expressions with parameters/constants adjusted to
fit results of measurement obtained for particular types of the landscape such as
metropolitan, suburban, and/or rural areas. Despite the fact that empirical models are
recommended by ITU-R, and widely used for wireless network design and planning, they
do not provide a clear physical insight to propagation mechanisms as they are not based

188
on the analysis of propagation effects such as reflections, transmissions, and diffractions
caused by specifics of urban structures and structure placement allocations.
In contrast, physical models are introduced based on "deterministic" approaches; (i.e. by
analysis of “knife-edge” diffractions on roof tops of the buildings, reflections and
transmissions from/through the walls, "waveguide-like" propagation along the streets, etc.).
However, the element of "randomness" that is specific to that propagation environment still
results in field prediction difficulties. The basic goal here in this section is to demonstrate
main concepts and applications, rather than to provide the reader with a complete survey
of the variety of existing prediction models. The reader may be referred to [8] for more
details and [15] for the complete survey.

4.1.5.1. EMPIRICAL MODELS

The ideal propagation model, aligned along the flat and smooth earth's surface is
considered as a background for most empirical models. Recall that on terrestrial
propagation tracks, when a flat-earth approximation is considered, the path loss is
proportional to the fourth degree of distance. However, equation (4.31) may be rewritten in
the generalized form for the overall losses between transmitting and receiving points as
L = K rn , (4.75)

which is typical to all empirical models. In this equation K and n represent empirical
parameters adjusted to fit the large volumes of data experimentally obtained from long-
term collective observations of multiple and varying urban and rural terrain types spanning
the globe.
A simplified sketch of the propagation path for urban/suburban/rural areas is shown in
Figure 4.11 in its conventional form. Here A and B nominate the antennas of the base and
mobile stations respectively; h b and h m are the heights of the base and the mobile
stations, and h roof is the average height of the roof tops.

Figure 4.11 Simplified sketch of the propagation path representing


"averaged" conditions.

189
Most typical values of n are fairly close to that of an ideal model ( n = 4 ), i.e. close to the
case, when the ground wave propagates over the flat, smooth earth’s surface (see section
4.1). Some variations of n are reliant on particular conditions. The shadowed area on the
figure, below the roof tops, may be considered as an equivalent for the "lossy region"
between average height of the roof tops and the ground. It is a hypothetical region with
absorptions that are unevenly distributed, and are caused by propagative effects such as
diffractions, reflections, and absorptions in buildings, trees, and other terrain-effective
items.
4.1.5.1a. THE OKUMURA-HATA MODEL
This model [4, 5] is considered as one which is highly quoted by international organizations
[6]. It was primarily developed by Okumura, based on an extensive volume of
measurements in the form of a series of graphs. Later Hata transformed these into a set of
approximating formulas. As mentioned earlier, parametric sets and relations are adapted to
the different categories of landscapes: urban, suburban, and open (rural) areas. The
following parameters are common for all three types of areas:
A = 69.55 + 26.16 log f MHz − 13.82 log hb (4.76)

B = 44.9 − 6.55 log hb . (4.77)

Here hb represents the height of the base station antenna in meters, frequency f MHz is in

MHz. Another set of parameters is specific for the each type of the area. Those are listed
below.
1). Urban areas
The median value of the total loss between corresponding points is defined in decibels as:
LdB = 10 log [ EIRPT /( PR / G R )] = A + B log r0, km − E , (4.78)

where EIRPT = PT GT is effective isotropic radiated power of the transmitter, and PR / G R is


effective isotropic received power that was introduced in section 3.2. Parameter E may be
calculated from the expressions given in Table 4.2:
Table 4.2

f MHz ≥ 300 MHz E = 3.2 (log (11.75 hm )) − 4.97


2

Large cities
f MHz < 300 MHz E = 8.29 (log (1.54 hm )) − 1.1
2

190
E = ( 1.1 log f MHz − 0.7 ) hm − (1.56 log f MHz − 0.8)
Medium and small
cities

r0, km indicates the horizontal distance between base and mobile stations in km, and hm is a
height of the antenna of the mobile station in meters.
2). Suburban areas
The median value of the total loss between corresponding points is defined as:
LdB = A + B log r0, km − C . (4.79)

Here, parameter C is defined by the following expression:

C = 2 ( log ( f MHz / 28)) + 5.4 .


2
(4.80)
3). Rural / open areas
The median value of the total loss between corresponding points is defined as:
LdB = A + B log r0, km − D . (4.81)

Here parameter D is defined by the expression:

D = 4.78 ( log f MHz ) + 18.33 log f MHz + 40.99 .


2
(4.82)
The Okumura-Hata model is applicable for the following limited conditions:
o Carrier frequencies: 150 MHz ≤ f c ≤ 1500 MHz ,

o Antenna elevation of the base-station: 30 m ≤ hb ≤ 200 m ,

o Antenna elevation of the mobile-station: 1 m ≤ hm ≤ 10 m ,

o Horizontal distances: r0 > 1 km .

If (4.78) is expressed in non-logarithmic (linear) form, then one may note that
B / 10 equates to the same meaning as parameter n in (4.75), being fairly close to the
value B / 10 ≈ 4 , which is specific for the ground wave propagating over the flat earth. The
base station antenna elevation must be defined relative to the average ground level within
the distances of 3 – 10 km from the base station along the propagation path. Therefore hb

may vary slightly when mobile station moves.

191
4.1.5.1.b OTHER EMPIRICAL MODELS

The COST 231-Hata 1 model allows extension of the Okumura-Hata model to the
frequency band 1500 – 2000 MHz, applicable for small and medium cities [7, 8].
LdB = F + B log r0, km − E + G , (4.83)

where: F = 46.3 + 33.9 log f MHz − 13..82 log hb . (4.84)

B is defined by (4.77), E is defined from Table 4.2 for small and medium cities, and
 0 dB for medium − sized cities and suburban areas
G= (4.85)
 3 dB for uburban (metropoli tan) areas
Parameter G here is not to be confused with the antenna gain. Other restrictions to this
method are the same as for the Okumura-Hata method.

The Lee model [10] is expressed as a power of distance:


LdB = 10 n log r0, km − 20 log hb , eff − M − 10 log hm + 29 , (4.86)

where parameters M and n are obtained from the measurements, and are displayed in
Table 4.3.
Here, in this method, a new term, base antenna effective height hb , eff is introduced as

follows: a line that is tangent to the landscape profile at the mobile antenna location is
extended towards the vertical line at the base station antenna; hb , eff is found as a distance

from the base antenna positioning point to that intersection as shown in Figure 4.12.

Table 4.3

Propagation Area n M

Free space 2 - 45

Ideal flat earth 4 -45

Open (rural) area 4.35 - 49

Suburban 3.84 - 61.7

1
COST stands for Committee on Science and Technology. This committee programs and coordinates
European Community collaborative studies in the areas of science and technology.

192
Urban (averaged) 3.96 - 73.75

Figure 4.12 Demonstration of the base antenna effective heights in the Lee model.

-----------------------------------------------------------------------------------------------
Example 4.6
Use the Okumura-Hata model to obtain the spatial distribution of the median values of the
total power loss at 800 MHz around the base station of the cellular phone network, located
in an urban area of a large city, utilizing the initial data given below:
• Base station antenna average height: hb = 30 m

• Average height of mobile phones above the ground level: hm = 3 m.

Solution
Based on (4.76), (4.77), and formulas provided in Table 4.2 proper parameters are
calculated as follows:
• A = 69.55 + 26.16 log 800 − 13.82 log 30 = 125.1 dB
• B = 44.9 − 6.55 log 30 = 35.22 dB

• E = 3.2 ( log (11.75 ⋅ 3)) − 4.97 = 2.69 dB


2

Results of calculations of the total loss LdB from (4.78) are shown in Table E-4.6 below.

Table E-4.6
Distance from the base station, r0, km 1 1.5 2 2.5 3 3.5

Total power loss, dB 122.41 128.6 133 136.4 139.2 141.57

193
4.1.5.2. PHYSICAL MODELS

Amongst the numerous types of physical models created during last several decades, the
Walfish-Ikegami Model is the one most widely recommended by ITU-R. This method is
based upon the combination of two: one proposed by Walfish and Bertoni [9], and the
other proposed by Ikegami, et al [10]. This combined methodology allows the prediction of
overall losses (in dB) between corresponding points. It is based on fairly precise analytical
counts of the propagation phenomenon. However, some instituted corrections may permit
better matching of predicted losses with those obtained from experimentally observed
data. Note that a complete description of propagation physical models and mechanisms
may be found in [4], [5] and [8].

4.1.5.2a. Non-LOS (LINE-OF-SIGHT) PATHS

In this case the total loss (3.42) is defined as a combination of three composite terms:

LdB = 10 log [ EIRPT /( PR / G R )] = L 0 + Lmsd + Lrts , (4.87)

where L0 represents a free space loss (reference path loss) defined by (3.39) for the ideal

conditions and isotropic antenna. Lmsd acts as the “multiple screen diffraction loss”, i.e.

the total loss due to diffraction from multiple roof tops along the propagation path (see
Figure 4.13)), and Lrst represents the loss caused by diffraction from the roof of the last

building down to street level.

194
Figure 4.13 Sketch of the simplified urban/suburban terrain and
propagation path for the Walfish-Ikegami "flat edge" model.

Note that a simplified "flat edge" approach for the buildings is employed in this model, i.e.
all buildings are considered of the same height (average value of the roof top heights) and
equally spaced.
On Figure 4.13 w denotes the width of a street, which is usually assumed to be equal to
the wall-to-wall distance between buildings, thus w = b / 2 , where b is the average
distance between buildings, and ϕ (in degrees) is the street orientation relative to the
direct propagation path.
Lrst term introduced in (4.87) is dependent on ϕ angle.

(hroof − hm ) 2
Lrts = −16.9 + 10 log f MHz + 10 log + L (ϕ ) , (4.88)
w

 − 10 + 0.354 ϕ , 0 0 < ϕ < 35 0



where L (ϕ ) =  2.5 + 0.075 (ϕ − 35), 35 0 < ϕ < 55 0 (4.89)
4.0 − 0.114 (ϕ − 55 0 ), 55 0 < ϕ < 90 0

Multiple screen diffraction loss: Lmsd is defined as

Lmsd = Lbsh + k a + k d log r0 + k f log f MHz − 9 log w (4.90)

with the following values for the Parameter k f :

195
  f MHz 
− 4 + 0.7  925 − 1, suburban areas
kf =    (4.91)
 − 4 + 1.5  MHz − 1,
 f
urban areas
  925 
The other parameters in (4.90) are dependent on positioning of the antenna of the base
station relative to the average roof tops level as follows:

Antenna of the base station is above the average roof tops level ( hb > hroof ).

Lbsh = −18 log [1 + (hb − hroof )] (4.92a)

k a = 54 , (4.93a)

k d = 18 , (4.94a)

Antenna of the base station is below the average roof tops level ( hb < hroof ).

Lbsh = 0 (4.92b)


 54 − 0.8 (hb − hroof ) r0 ≥ 0.5 km

ka =  , (4.93b)
 (hb − hroof ) ⋅ r0
54 − 0.8 r0 < 0.5 km
 0.5

hb − hroof
k d = 18 − 15 , (4.94b)
hroof

Note that for the LOS condition between corresponding points ( ϕ = 0 0 ) formula (4.87) is

simplified to L ≈ L0 , which means in this case the term Lmsd + Lrts ≤ 0 , and must be

excluded from consideration. A more accurate approach is given by the Walfish-Ikegami


model in (4.96) below. It must be noted here, that this considering model is restricted to the
following limits: 800 < f MHz < 2000 , 4 m ≤ hb ≤ 50 m , 1 m ≤ hm ≤ 3 m , 0.2 km ≤ r0 ≤ 5 km .

------------------------------------------------------------------------------------------------------
Example 4.7
Use the Walfish-Ikegami model to define the spatial distribution of the median values of the
total power loss around the base station of the cellular phone network located in the urban

196
area of a large city at 800 MHz, with the initial data given in Example 4.6 if the average
height of the roof tops is hroof = 20 m, and streets with the width of w = 30 m, are

positioned orthogonal to propagation path ( ϕ = 90 0 ). Compare the results with those


given in Table E-4.6.
Solution
• Combination of (4.88) and (4.89) is as follows:

(hroof − hm ) 2
Lrts = −16.9 + 10 log f MHz + 10 log + [4 − 0.114 (ϕ − 55)] . (E4.7.1)
w
After proper substitutions into (E4.7.1) the result is Lrts = 21.98 dB

• For this case of the base antenna above the average roof tops ( hb > hroof ) the

combination of (4.90) with (4.92a), (4.93a), and (4.94a) is


Lmsd = −18 log[1 + (hb − hroof )] + 54 + 18 log r0 + .....

 f 
..... + − 4 + 1.5  MHz − 1 log f MHz − 9 log w = 9.76 + 18 log r0 (E4.7.2)
  925 
• Calculation results with formula (E4.7.2) are given in Table E-4.7
• Total power loss is found from (4.87), and is presented in the table below
• The differences between two considered prediction models are shown in the last
row of the same table.
Table E-4.7

Distance from the base station, r0, km 1 1.5 2 2.5 3 3.5

Free space loss, L 0 dB (formula (3.39)) 90.5 94 96.5 98.5 100 101.4

Lmsd , dB 9.76 12.93 15.18 16.92 18.35 19.55

Total power loss, dB


122.24 128.9 133.66 137.4 140.33 141.9
( from Walfish-Ikegami model)
Difference between Walfish-Ikegami and
Okumura-Hata predicted data, dB -0.17 + 0.3 + 0.66 +1 +1.13 + 0.33
(compared with data from Table E4.6)

197
4.1.5.2b. LOS PATHS
In the presence of the LOS between communication end points, when streets are radially
directed from mobile station towards the base station ( ϕ = 0 0 ) 1, a two-ray interference
approach that is presented in sections 4.1 and 4.2 for flat-earth approximation may be
applied. Generally speaking, that approach is applicable if at least the first Fresnel's zone
is clear. In other words, based on (2.179) the width of the street must exceed the maximum
cross-sectional dimension of the first Fresnel's zone in order for LOS approach to be
applied, i.e.
1
w ≥ R1 / 3 = λ r0 = 0.3 λ r0 . (4.95)
2 3

Here we assume r10 = r20 = r0 / 2 . For most practical combinations of w , f , and r0 the

expression (4.95) is valid. Indeed, in the worse case of the lowest usable frequency
f = 100 MHz, and a large enough distance of r0 = 50 km, the inequality (4.95) becomes
w ≥ 116 m, which is practically reasonable.
The Walfish-Ikegami model provides an expression the total loss for LOS in the following
form that is adjusted to experimental results [4]:
LdB = 42.6 + 26 log r0 + 20 log f MHz . (4.96)

However the ITU-R Recommendation P.1411 provides the following set of formulas that
allows estimation of the upper and lower borders for variations of the total loss, and may
be used to check the accuracy of the range for the total loss calculated from (4.96):
 r 
 20 log  0  r0 ≤ rbp
 r 
LdB , lower = Lbp +   bp  (4.97)
 r 
40 log  0  r0 > rbp
 r 
  bp 

 r 
 25 log  0  r0 ≤ rbp
 r 
LdB , upper = Lbp + 20 +   bp  (4.98)
 r 
40 log  0  r0 > rbp
 r 
  bp 

1
Sometimes refers to as street canyons

198
Here rbp is so called breaking point distance, which is simply the distance of the first

maximum of the field defined by (4.19), i.e.


4 hb hm
rbp = r1, max = . (4.99)
λ
Then

 8π h b h m 
Lbp = 20 log   (4.100)
 λ
2

is the total loss at the breaking point distance. Expression (4.100) is developed based on
the fact that at the breaking point distance r0 = rbp = r1, max the unitless (linear) value of the

free space loss becomes equal to


L 0 = 4π r0 / λ = 16π h b h m / λ 2 (4.101)

after we substituted (4.99) into (4.100). The difference between expressions (4.100) and
(4.101), as noted in Section 4.2, is relative to the real field demonstrating twice the strength
of the reference field in free space, when constructive interference occurs between direct
and reflected waves.
-----------------------------------------------------------------------------------------------
Example 4.8
Based on the general approaches given in Sections 4.1 and 4.2, calculate predictable
losses if a LOS condition exists between corresponding points ( ϕ = 0 0 ) for the initial data
provided in Example 4.6. Compare with results from the Walfish-Ikegami Model, and verify
that they satisfy the upper and lower boundaries given by (4.97) and (4.98).
Solution
Table E-4.8 (Example 4.8 calculation results)

Distance from the base station, r0, km 1 1.5 2 2.5 3 3.5

LLOS , lower , dB 84.84 91.88 96.88 100.25 103.92 106.6


Boundaries
outlined by
ITU-R

LLOS , upper , dB 104.84 111.88 116.88 120.25 123.92 126.6

Free space (reference) loss, L 0 dB –


90.5 94 96.5 98.5 100 101.4
expression (3.39)

199
Total power loss in dB for two-ray
96.51 98.58 99.25 99.56 99.72 99.82
interference (sections 4.1 and 4.2)
Total power loss, dB (Walfish-Ikegami
100.66 105.24 108.49 111 113.07 114.81
model, formula (4.96))

From Table E-4.8, one may confirm the fact that the Walfish-Ikegami model (4.96) for LOS
condition provides field predictions for urban propagation that are closest to the mid-values
outlined by ITU-R boundaries.
-----------------------------------------------------------------------

4.2. PROPAGATION BETWEEN GROUND-BASED


ANTENNAS OVER THE FLAT EARTH

Recall that in the case of elevated antennas of horizontal polarization a general expression
for the propagation factor is (4.14), and an approximate formula for the field strength at the
reception point over the “flat earth” is (4.30). From those expressions we may conclude
that the decrease of antenna heights, h 1 and h 2 , or increase of the wavelength will result

in the decrease of the field strength. It tend to zero much faster than it does in free space.
As mentioned in sections 4.1 and 4.2, in the case of a horizontally-polarized transmitting
antenna, the resultant field tends to zero due to cancellation between direct and reflected
waves of opposite phases, (i.e. the smaller is h 1 , the closer is the phase shift between

those waves to 1800). Indeed, the smaller the antennas heights are, and the closer the
angle of incidence to 900, then the closer the phase shift ∆Φ to zero, thus the overall
phase shift between direct and reflected waves becomes equal to 1800 as shown in Figure
4.3.
In the case of a vertically polarized transmitting antenna. lowering the antenna heights
does not result in complete vanishing of the total field. However, in reality if the antenna
heights (either h 1 or h 2 , or both) are vanishing, the intensity of the field in the vicinity of

air-to-ground interface may differ significantly from the field intensity predicted by two-ray

200
interference 1 as demonstrated in sections 4.1 and 4.2. In reality both, structure of the field
and the propagation mechanisms of the radio wave near that interface, are significantly
distinct from what is predicted by the ray-tracing approach. Below, a surface wave
propagation mechanism that is specific to ground-based antennas is considered. We will
use the term ground-based not only to the antennas that are placed on the earth’s surface,
but also near it, i.e. we will assume that the antenna elevation heights over the ground are
less or comparable to the wavelength.
Generally speaking, we will consider not the absolute heights of the antenna elevations,
but rather their relative values: h1 / λ and h2 / λ are to be considered as key terms for
distinctions between elevated and ground-based antennas, and also to distinguish the
propagation mechanisms associated with them. From this viewpoint the antennas are
considered as elevated when both of those ratios are greater than several unities.
Alternatively, the antennas are considered as ground-based, if one (or both) of those ratios
is (are) less than a unity or even close to zero. It is evident that for real antenna heights
(from several meters up to hundreds of meters) antennas are considered mainly as
ground-based in the frequency range is limited to less than approximately 100 MHz. In
other words, the physical concepts described in this section are reasonably applicable for
relatively low frequencies such as VHF, HF and lower.

4.2.1. ANTENNAS OVER THE INFINITE, PERFECT GROUND PLANE


As a reference (ideal) case for real propagation conditions, we now consider a vertical
ground-based electric line current that is placed on infinite ground plane constructed of the
perfect electric conductor (PEC) as shown in Figure 4.14a. We will assume the length of
the line current, l small enough, i.e. l / λ << 1 , and that the current is distributed uniformly
along the line. A non-symmetric excitation of this radiator is shown in Figure 4.14c, and this
case is known as a monopole above the PEC flat ground plane.

1
This approach is also referred as a "Ray-Tracing" or "Geometrical Optics" approach, meaning that
the distances and antenna heights are much greater than the wavelength. This allows disregard any
diffraction effects and apply methods of analysis that are specific to geometrical optics.

201
Figure 4.14. a). Field structure of the ground based monopole over the
infinite PEC surface. b). Field structure of the dipole in free space. c).
Excitation of the ground based monopole. d). Excitation of the symmetric
dipole

Based on image theory, the structure of electric and magnetic field lines in the upper
hemisphere will remain the same if the ground plane is replaced by the identical mirror-
image of the monopole with an opposite excitation phase. In other words, to keep the field
structure unchanged we’ll turn hypothetically the monopole into a symmetric dipole without
ground plane as shown in Figure 4.14b. In both cases, namely for non-symmetric
monopole over the ground, and for its symmetric replacement, we will assume the source
of power is ideally matched with the antenna's input impedance as shown in Figures 4.14c
and 4.14d.
As may be noted from Figure 4.14a, in close vicinity of the PEC surface electric field lines
of the monopole are normal to the ground surface, whereas the magnetic field lines are
tangential to them. Hence the direction of the vector of the power flow Π = E × H is

202
parallel to the surface. This means there no power flow that is penetrating into the ground,
as there are no losses caused by the PEC-to-air interface.
The RMS values of the electric field for the monopole above the PEC ground plane may be
defined based on evaluations similar to those found in Section 3.2. Namely, the power flow
density at the distance r is:
PT GT
Π0 =
MPG
, (4.102)
2π r 2
where equivalent isotropic radiated power ( EIRP = PT GT ) is distributed uniformly along the

semi-spherical surface 2π r 2 of the upper hemisphere only, unlike the expression (3.35),

where EIRP is assumed to be distributed along the whole area 4π r 2 of the spherical
surface centric to the radiation point 1. The radius of the sphere r is a distance to the
reception point. As one may note from (4.102), in this ideal case of no losses and for the

given power PT of the source the field strength will be 2 times greater than it is in free
space defined by (3.34). In other words, in a case of monopole above the ideal PEC
ground, the electric field strength at the distance r is defined as:

30 PT GT 60 PT GT
= = =
MPG
E0 2 2 E0 . (4.103)
r r
Another expression is derived for the electric field strength based on the given value of the
current I uniformly distributed along its length l . Based on (2.135), the RMS value of the
electric field, generated by the small line radiator with evenly distributed current in free-
space, along direction of maximum radiation ( θ = 90 0 ) is:
Il I l W0 k 0
E0 = ω µ0 = , (4.104)
4π r 4π r

where W0 = µ 0 / ε 0 = 120π Ohms shows the intrinsic impedance, and k 0 = 2π / λ0


depicts a wave number in free space. Now, if the monopole is placed on PEC ground we
may replace it with hypothetical dipole, as shown in Figure 4.14b, by maintaining the
current of the same intensity and same direction in its both shoulders. Thus the
replacement of the monopole over the PEC ground by the equivalent dipole in free space
results in doubling the length of the radiating filament, thereby, doubling the field strength
at the reception point, i.e.

1
MPG superscript in (4.102) stands for Monopole above the Perfect Ground.

203
(2 l ) I W0 k 0
= = 2 E0 .
MPG
E0 (4.105)
4π r
Discrepancy between the viewpoints given by (4.103) and (4.105) may be resolved if the
input impedance of the monopole is analyzed in comparison to its replacement dipole.
Appendix 5 provides the analytical expressions for the input impedances for vertical and
horizontal line-currents placed on height h above the PEC ground plane. Those results
MPG
are shown in graphical form in Figure 4.15 for the input impedance Z ant of the line

current source above the PEC ground plane relative to its free space value, Z ant , 0

(reference case) as a function of the relative height, h / λ of the source above the ground.

Figure 4.15 Dependence of input impedance of the line current radiator


on a distance to PEC ground plane.

As one may realize from the graphs, the input impedance of the vertical line current source
doubles when it is moved from infinite height (i.e. from free space) to air-to-PEC interface,
whereas for the horizontal current element it tends to zero (i.e. the horizontal line current
placed on the top of the PEC surface becomes shorted). If now we express the transmitted
power as a function of input current of the monopole as

PT = Z ant
MPG
I2. (4.106)

then the explanation of the above discrepancy is as follows: for the given power PT ,
generated by transmitter and radiated into the upper hemisphere by vertical monopole on
MPG
PEC ground, the doubling of the input impedance Z ant will results the decrease of the

excitation current by a factor of 2 , to keep PT unchanged. Hence (4.105) becomes

204
=
MPG
E0 2 E 0 , which is the same as (4.103). Note that if the vertical monopole above the
PEC ground is equivalently replaced by a hypothetical dipole in free space, then the non-
symmetric (unbalanced) excitation shown in Figure 4.15c should also be replaced by the
symmetric excitation shown in Figure 4.15d.

4.2.2. LEONTOVICH APPROXIMATE BOUNDARY CONDITIONS


AND STRUCTURE OF RADIO WAVES NEAR THE EARTH’S
SURFACE
For most of the electromagnetic problem solutions, which consider a field structure in the
vicinity of the interface between two mediums, the system of Maxwell's Equations are to be
solved for each medium separately, and then “interwoven” by utilizing boundary conditions
for electric and magnetic field components, provided by (2.23), (2.24), (2.31), and (2.32).
For practical engineering applications this procedure becomes cumbersome, and
occasionally even deadlocked. For a radio wave propagating along the flat earth surface,
those boundary conditions may be simplified by using an approximate approach,
introduced by M.A. Leontovich in 1944 [21]. This approximation is based on the fact that in
a relatively low-frequency range, of up to about hundreds of MHz, the ground of the earth
is considered as a fairly dense medium, and thereby the refraction index, hence the
magnitude of its complex dielectric constant is always much greater than unity, i.e.

n= ε2 = ε 2 2 + (60λσ )2 >> 1 , (4.107)

whereas for the propagation in the atmospheric air it is assumed as ε1 = 1 . As


demonstrated below, condition (4.107) allows definition of relations between components
of the electromagnetic field in the area above the ground by using the electromagnetic
parameters of the ground, without “interweaving” fields into upper and lower mediums, i.e.
without taking into account the values of electromagnetic field components below the air-
to-Ground interface.
Recall section 2.4, and consider the radio wave that falls onto the boundary of two
mediums, as shown in Figure 2.15 of Section 2.4. As illustrated in that section, the angles
of incident, ϕ and refraction, ψ are bounded by the Snell's Law that may be presented
here in general form for the lossy mediums as:

205
ε1 sin ϕ = ε2 sinψ , (4.108)

where ε1 and ε2 represent the magnitudes of the dielectric constants of the upper and

lower mediums respectively. Then, from (4.107) and (4.108) it becomes evident that the
angle of refraction ψ is fairly close to zero, regardless of what the angle ϕ is, even when

it becomes close to its maximum value, ϕ ≈ 90 o , (i.e. when radio wave propagation occurs
along the horizontal axis). Thereby, part of the radio wave that penetrates into the second
medium (refracted wave) will propagate in the direction close to perpendicular to the
interface. As shown in Figure 4.16a, a vertically polarized, horizontally propagating wave
that is initially generated by the vertical monopole above the earth’s ground will induce a
downwards propagating plane wave within the ground. The electric and magnetic fields are
tangential to the interface, and relate to each other via the intrinsic impedance of the
ground expressed by formula (2.66).
120π
E Y 2 = W 2 H X 2 = H X 2 . (4.109)
ε 2 − i (60λσ )
Now we rewrite the boundary conditions (2.31a) and (2.32a) for tangential components of
1
electric and magnetic fields as

E Y 1 = E Y 2 
. (4.110)
H X 1 = H X 2 
Regarding (4.110) E Y 2 and H X 2 in (4.109), they may be replaced by E Y 1 and H X 1 . Then
(4.109) may be rewritten as:
120π
E Y 1 = W 2 H X 1 = H X 1 . (4.111)
ε 2 − i (60λσ )
On the other hand, if (4.111) is taken into account, for the total electric field above the
ground we may write

E 1 = W1 H X 1 = 120π H X 1 = ε2 E Y 1 , (4.112)

and

E Z 1 = E 1 − E Y 1 = (ε2 − 1) E Y 1 .
2 2 2 2
(4.113)

1
Note that the condition for the magnetic field in (4.110) does not include J S , as it is shown in

(2.31a), as the presence of the conducive currents in the earth's ground are taken into account by
introducing the complex form for the magnetic fields.

206
Figure 4.16 a) Disposition of the field vectors on the air-earth ground
interface. b). Power flux density (Pointing vector) and its components

Thus, referring to (4.107), expression (4.113) may be rewritten in approximate from as


E
E Y 1 = W0 H X 1 ≈ Z 1 . (4.114)
ε2
This expression is one of the forms of the Leontovich approximate boundary condition.
Another form of Leontovich approximate boundary condition may be derived by following
[22], based on the Gauss Law (2.3), which may be rewritten for both, upper and lower
mediums as
∂E X 1 ∂E Y 1 ∂E Z 1 ∂E X 2 ∂E Y 2 ∂E Z 2
+ + = + + = 0. (4.115)
∂x ∂y ∂z ∂x ∂y ∂z
The volume charge, ρ tot , which is responsible for carrying conductive currents, is already

included in the expression for complex dielectric constant indirectly, therefore it is omitted
here. Below we will make the same assumption, namely that within the second medium the
refracted plane wave is propagating vertically, along the negative Z-axis. Hence, the field
components that are tangential to the interface remain continuous when crossing the air-
to-ground border, i.e.
∂E X 1 ∂E Y 1 ∂E X 2 ∂E Y 2
+ ≈ + . (4.116)
∂x ∂y ∂x ∂y
Then, taking (4.116) into account from (4.115) we may conclude:

207
∂E Z 1 ∂E Z 2
≈ . (4.117)
∂z ∂z
Despite the E Z 2 component of the electric field below the interface is vanishingly small,
however its propagation along the negative Z-axis may be expressed analytically as

E Z 2 = E Z 2, max exp (ik2 Z ) , (4.118)

Therefore:

∂E Z 2 / ∂ z = ik2 E Z 2 , (4.119)


and (4.117) may be rewritten as
∂E Z 1
≈ ik2 E Z 2 = ik 0 ε2 E Z 2 , (4.120)
∂z
where k 0 is a wave number in free space.

Now we will utilize the boundary condition (2.23), which in our case of ε1 = 1 is presented
as
E Z 1 = ε2 E Z 2 . (4.121)

If we use this expression to replace E Z 2 by E Z 1 in (4.120), then we will ultimately end up


with
∂E Z 1 ik
≈ 0 E Z 1 , (4.122)
∂z ε2
which is another form of the Leontovich approximate boundary condition.

Compared to exact boundary conditions, the advantage of expressions (4.114) and (4.122)
for the approximate conditions at the air-to-ground boundary is obviously clear: they allow
expressing the relations of the field components in the air through the parameters of the
medium (ground) beneath the interface. It allows achievement of the solution for the
ground wave propagation problems without referring to the field in the second medium.
In the next section we will consider analysis of the propagating field strength, namely the
propagation factor that calculates for the losses obtained along the propagation path.
Here we will consider the structure of a radio wave that propagates along the earth's
surface and within its close vicinity.
Assume a vertical monopole is used to generate a primary electric field E Z 1 at the

transmission point. In the presence of the lossy ground a tangential component E Y 1 will

208
occur that closely relates to E Z 1 through expression (4.114). E Y 1 literally represents the
voltage drop on unit length of a real, non-perfect interface along the propagation path,
which does not occur while propagation along the PEC ground. The occurrence of E Y 1
component results in the tilting of the resultant electric field vector by angle χ , as shown in
~
 = E × H
Figure 4.16a. Hence the Poynting vector Π will also be tilted by the same angle
towards the ground (Figure 4.16b), resulting in the splitting of the total power flow into two
portions: one relates to the propagation along the interface, and the other that is directed
normally downwards, perpendicular to the interface, and relates to the penetration into the
ground, thereby identifying the power losses along the propagation path.
Due to the complex character of the dielectric constant ε2 there is a phase shift between

field components E Y 1 and E Z 1 that may be found from (4.114). Based on the complex
forms (2.45) and (2.46), that phase shift may be expressed as

1 1  60 λ 0 σ 
δ ground = δ ε 2 = tan − 1   , (4.123)
2 2  ε2 
where δ ε 2 indicates the argument of the complex dielectric constant of the earth’s ground.

As may be seen from (4.123), in the low-frequency range, when λ 0 is large enough,

δ ground ≈ 45 0 , whereas for the high frequencies, when λ 0 is small, δ ground ≈ 0 0 . It becomes

obvious that for the given amplitude-phase relations between E Y 1 and E Z 1 components,

the total field E 1 will become elliptically polarized. Unlike the elliptically polarized waves
considered in section 2.3, where the plane of polarization ellipse is positioned
perpendicular to the propagation direction, here we have the case where the plane of the
polarization ellipse is vertically coincident to the propagation direction, as shown in Figure
4.17. Figure 2.5 shows that for frequencies above 100 MHz both, the fresh water, as well
as ground soil are considered as dielectrics, thus δ ε 2 ≈ δ ground ≈ 0 0 . Hence, regarding

section 2.3, the elliptic polarization will become linear with the tilt angle

χ = tan −1 ( EY 1 / E Z 1 ) = tan −1 (1 / ε2 ) << 1 . (4.124)

209
Figure 4.17 Elliptically polarized radio wave excited by a vertical monopole on the
semi-conducive earth's surface

The appearance of the horizontal component in electric field of the vertically polarized
radio wave, as a result of non-ideal properties of the ground surface, may allow reception
of the signal via the horizontal wire antenna. Indeed, if one of the components of the
electric field of the radio wave is directed along the linear wire antenna, then as mentioned
in section 3.1.3, an electromotive force will be induced in that wire (see (3.26)):
emf = EY 1 leff . (4.125)

Here l eff is the effective length of the wire. The induced electromotive force simply

indicates the desired signal to be received. In other words, to receive the signal carried by
a vertically polarized radio wave propagating along the real earth's surface it is not
necessary to use only the receiving antenna such as a monopole in the form of a vertical
wire. In LF and MF frequency bands the height of that vertical antenna may become
unrealistically large. Instead, a horizontally placed wire antenna may be used that is easier
to handle. Despite the fact that: EY 1 << E Z 1 , the large enough value of the electromotive

force may be achieved by increasing the horizontal length l 2 , as a trade off to constructing
high vertical towers.
An interesting application of the above propagation mechanisms is the use of the specific
structure of the radio waves propagating along the non-ideal earth's ground surface for
designing directive receiving antennas in low-frequency bands, such as LF or MF.
Consider an L-shaped wire antenna as shown in Figure 4.18. If the radio wave propagates
from left-to-right (Figure 4.18a), then the total EMF is:
emf1 = E Z 1 l1 + EY 1 l 2 (4.126)
whereas, for right-to-left propagation (Figure 4.18b):

210
emf 2 = E Z 1 l1 − EY 1 l 2 . (4.127)

Here l1 and l 2 are to be considered as effective lengths of the horizontal and vertical parts
of the antenna respectively.

Figure 4.18 Reception of the vertically polarized radio waves by an L-


shaped directive wire antenna in low-frequency bands in the presence
of the real earth's surface.

As one may note from the expressions (4.126) and (4.127), the electromotive force
induced in the receiving antenna by the radio wave from one direction becomes smaller,
than from the other. In other words, a significant directivity may be achieved by the proper
choice of effective lengths: l1 and l 2 . If, for instance, E Z 1 l1 = EY 1 l 2 then no signal will be
induced at the receiver's input by the radio wave coming from right to left, whereas the
signal induced by the radio wave coming from left to right will be doubled. Note finally that
1
if the horizontally polarized ground radio waves are excited along the earth's surface
instead of vertically polarized, then based on the principle of electromagnetic reciprocity
the vertical component may be found similar to (4.114) as follows:
E
E Z 1 = Y 1 . (4.128)
ε2

1
As mentioned above, the horizontally polarized radio wave may not exist above the ideal, PEC
ground plane. However it may exsit along the real, non-perfect earth's ground by using a vertical
magnetic monopole, such as a closed, horizontally placed current loop above the interface.

211
4.2.3. PROPAGATION OVER THE REAL HOMOGENEOUS FLAT
EARTH
As mentioned in section 4.6.1 for the vertical radiating line-current above the flat PEC
ground, on or close to its surface, the power flux density of the radio wave doubles due to
distribution of the radiated power in the upper hemisphere only, in contrast to the wave
propagating in free space. The doubling in power results in the increase of the electric field

strength by the factor of 2 . This statement is expressed by (4.103), which we'll rewrite
here as

60 PT GT
= 2 E0 =
MPG
E0 . (4.129)
r
MPG
Here E 0 represents the electric field of the line-current in free space. If E 0 is

considered as a reference field, i.e. the field strength in ideal propagation conditions, i.e.
over the PEC ground, then for the propagation in real conditions the losses along the
propagation path must be taken into account by imbedding into (4.129) the propagation
factor, F as follows:

60 PT GT
E = E0 F=
MPG
F. (4.130)
r
The first successful attempt of solving the problem for the vertical monopole over perfect
flat ground was undertaken by Sommerfeld in 1909, and later updated by himself and
others (Norton, Weyl, Van-der-Pol, Burrows) to account for the real propagation conditions
existing above the flat earth's ground. Theoretical analyses [3, 20, 22] demonstrate a
presence of two types of so called surface waves: Zenneck wave and Norton wave. The
Zenneck wave is not considered to be a dominant surface mode. It instead demonstrates a
cylindrical wave front with electric field strength that is inversely proportional to square-root
of distance, and an extremely fast, exponential decay in the vertical direction. In other
words, it exists only in close vicinity of the excitation point, is tightly coupled to the
interface, and has negligibly small intensity, compared to that of the Norton wave.
The Norton wave was evaluated asymptotically, for large distances, and is considered as a
dominant surface propagation mode, with the intensity that decays inversely proportional to
distance. Propagation factor for this type of surface wave is introduced by the expression
known as Weyl & Van-der-Pol (W&VdP) formula [3]:

F = 1 − i π x exp(− x ) erfc ( i x ) , (4.131)

212
where:
x = r / s = x exp (−ib) (4.132)
generically is a complex number called numerical distance, and

2 ε2
2
λ 0 ε2 2
s = i =i (for vertical polarization) (4.133)
k 0 (ε2 − 1) π ε2 − 1

2 λ0 1
s = i =i (for horizontal polarization) (4.133a)
k 0 (ε2 − 1) π ε2 − 1

denotes a distance scale. k 0 and λ 0 are wave number and wavelength in free space

respectively. In (4.131) erfc (i x ) is a complementary error function of the complex

argument, i x , that is provided by well known mathematical reference handbooks, as


well as by the library of MATLAB subroutines, or similar software.

2
x ) = ∫ exp (−t ) dt = 1 −erf (i
2
erfc (i x ) . (4.134)
x i x

It belongs to the class of special functions, and relates to the widely tabulated regular error
function as follows:
i x
2
x ) = ∫ exp (−t
2
erf (i ) dt . (4.135)
x 0

Figure 4.19 shows the reliance of the magnitude of propagation factor on the magnitude of
numerical distance, x for several values of its argument, b . Those curves are widely used
in radio engineering applications.
For the practical engineering estimates (4.131) may be replaced by the following
approximate formula may be used for the magnitude of propagation factor calculation:

2 + 0.3 x x
F≈ − [exp (−0.6 x)] sin b , (4.136)
2 + x + 0.6 x 2 2
where x is the magnitude of the numerical distance.
The complex expression (4.131) for large values of the magnitude of numerical distance,
x ≥ 25 asymptotically tends to
1
F ≈ , (4.137)
2 x

213
which may also be seen from (4.136). For those distances the curves on Figure 4.19
merge and become a straight line.

Figure 4.19. Propagation factor dependence on magnitude, x and


argument, b of the numerical distance for the ground wave
propagating along the real homogeneous earth's surface

Note finally, that the applicability of the flat-earth approximation, which is considered here
for the ground surface-waves propagation along the homogeneous real earth's surface, is
limited by the following condition [22]:

λ0 125
r << 3
a2 ≈ km, (4.138)
2π 3
f MHz

where a = 6370 km is the radius of earth, λ 0 is the free space wavelength, and f MHz is the

frequency in MHz. In practice, as it follows from (4.138), applicable distances are limited to:

r ≤ 80 / f MHz
1/ 3
km, (4.139)
which is shown in Figure 4.20 in the form of the graph. For the distances greater than
those restricted by the “flat earth” approximation the convexity of earth surface may be
taken into account by introducing so called “shadow factor” as presented in section 4.3.2.

214
Figure 4.20 Graph defining the “flat earth” approximation applicability
for the surface wave propagation estimates

-------------------------------------------------------------------------------------------------------
Example 4.9

Find the field strength induced at the receiving point by the ground-placed vertical
monopole antenna of the AM-broadcast station at the distance 60 km for the following
initial data:
o Transmitting power: PT = 1 kW

o Transmitting antenna gain: GT = 1.5 ( ≈ 1.76 dB )

o Frequency (free space wavelength): f = 1 MHz ( λ 0 = 300 m )

o Propagation over homogeneous wet soil: ε = 40 , σ = 3 ⋅10 − 2 S/m,


Solution

1. Flat-earth approximation applicability check (4.139):


80
= 80 km >60 km (“flat earth” approximation is applicable)
31
2. Complex dielectric constant from (2.45) is
ε = ε − i 60λ 0σ = 40 − i ⋅ 60 ⋅ 300 ⋅ 3 ⋅10 −2 = 40 − 540 i = 541.5 ∠ − 85.76 0
3. Distance scale from (4.133), for | ε |>> 1 , is approximately equal

λ0 200
s ≈ i ⋅ ε = i (40 − 540 i ) = 34377 + 2546 i , meters
π 3.14
4. Numerical distance from (4.132) is:

215
r 60 ⋅10 3
x = = = 60 ⋅10 3 (2.893 − 0.214 i ) ⋅10 −5 =
s 34377 + 2546 i

= 1.7358 − 0.12856 i = 1.74 ∠ − 4.24 0

Thus x = 1.74 , b = 4.24 0 .


5. Taking into account x < 10 , the propagation factor may be calculated from
(4.136), or from the graph given in Figure 4.19.

2 + 0.3 x x
F≈ − [exp (−0.6 x)] sin b = ....
2 + x + 0.6 x 2 2

2 + 0.3 ⋅1.74 1.74


... = − [exp (−0.6 ⋅1.74)] sin 4.24 0 = 0.43
2 + 1.74 + 0.6 ⋅1.74 2 2
6. Magnitude of the vertical component for the field strength at the reception
point form (4.130):

60 ⋅10 3 ⋅1.5 V
EZ1 = ⋅ 0.43 = 2.15 ⋅10 −3 = 2150 µV/m (66.65 dBµV/m)
60 ⋅10 3
m
(Answer: compare with the result found from Figure 4.30)
7. Magnitude of the horizontal component of field strength at the reception
point form (4.114):
EZ1 2.15 ⋅10 −3 V µV
EY 1 = = = 9.24 ⋅10 −5 = 92.4 (Answer)
| ε | 541.5 m m

---------------------------------------------------------------------------------------------------

4.2.4. PROPAGATION ALONG THE REAL INHOMOGENEOUS FLAT


EARTH. COASTAL REFRACTION
In real conditions the surface radio wave mainly propagates from the transmitter to the
receiver along the inhomogeneous path, by passing through areas with different
electromagnetic parameters of the ground. We will omit the cases of gradual changes of
those parameters along the propagation path, and consider only the cases where the path
consists on two or more sections with constant electromagnetic parameters within each
section, and the abrupt changes of them from section to section. Those types of
propagation paths are called "mixed paths", and are the subject of further analysis.

216
For the frequency ranges VHF and higher, when the antennas are considered mainly as
high elevated (see Sections 4.1 and 4.2), the only impact of the electromagnetic
parameters of earth's surface appears at the reflection point C (Figure 4.1), namely the flat
area on the surface that is limited by the first Fresnel zone around the reflection point. All
other areas along the propagation path are not actively involved in forming the field
strength at the receiving point. Unlike the case of the elevated antennas, in the event of
ground-based antennas, when the surface-propagation mechanism dominates, each point
of the entire propagation path as well as the closely surrounding areas of the earth's
surface are involved in the forming of the field intensity at the reception point. The question
arises: does each part of the entire propagation path have the same contribution on
forming the field at the reception point? To answer that question we refer to Figure 4.21. It
presents a pattern of the constant phases of the radio wave in a plane of propagation
along the earth's surface. As one may see from the figure, the Pointing vector is almost
parallel to the ground surface only in region (1), i.e. above the dash-dotted line. In the
region (2) the Pointing vector is tilted towards the ground because of existence of the
tangential component of electric field in vicinity of the ground. As mentioned above, the
tilting of the Pointing vector results in an outflow of energy of the radio wave into the
ground, with further dissipating in it in the form of thermal losses. Therefore, the region (2)
becomes energy-impoverished. Hence, the energy of the radio wave flows mainly through
the region (1). In region (3), which is specified by high density ( ε >> 1 ) the wavelength

becomes much smaller than in the upper hemisphere, (i.e. λ 2 = λ 1 / ε << λ 1 ), so from
Figure 4.21 it may be seen once again, that the Pointing vector in region (3) is close to
perpendicular to the air-to-ground interface.

Figure 4.21. A pattern of the equal-phase lines in vertical plane for


a radio wave propagating near the earth's surface

217
The conclusion drawn from the above considerations is that the canalization of the energy
of the radio wave between ground-based transmitting and receiving antennas takes place
in the upper hemisphere through the spatial area that is separated from the earth's surface
by the energy-impoverished region. While propagating along the earth's surface, the radio
wave does not "touch" that surface, therefore it is not impacted significantly by absorbing
surface of the earth. In other words, the radio wave, radiated by the transmitting antenna,
"flies up" at the transmission point, and then further propagates horizontally towards the
point of reception. Taking into account the principle of reciprocity, we may emphasize that
at the reception point an opposite process occurs, i.e. a radio wave that propagates along
the earth's surface "touches down" to the receiving antenna while approaching the
destination. This statement was first declared by L.M. Mandelstamm, of the former Soviet
Union as a "takeoff-landing" concept, demonstrated in Figure 4.22 [1].

Figure 4.22. Pattern of "takeoff" and "landing" for the ground waves
propagating along the earth's surface between ground-based
antennas

From the above statements it may be realized, that if the radio wave propagates along the
mixed path AB (Figure 4.22) over the earth's surface, then regardless of the distribution of
the electromagnetic parameters of the earth ( ε and σ ) along the path the resultant
attenuation of the energy is mainly defined by the electromagnetic parameters of "takeoff"
and "landing" areas, which surround the transmitting A and receiving B points respectively.

In most of the practical cases, when the numerical distances are large enough, the
resultant propagation factor for the mixed path may be calculated based on Millington's
approximate formula [23] as a mean-geometrical of two quantities:

F ( r) ≈ FA (r ) FB (r ) . (4.140)

218
Here FA (r ) is the propagation factor if assumed, that the entire propagation path is
homogeneous, with the constant electromagnetic parameters of the ground at A-point;
FB (r ) is the attenuation factor if assumed that the entire propagation path is
homogeneous, with the constant electromagnetic parameters of the ground at B-point.
FA (r ) and FB (r ) may then be defined from Figure 4.19 or from the expressions (4.131),
(4.136), or (4.137), whichever is appropriate.
Consider (4.140) for the mixed path that contains sections with significantly different
electromagnetic parameters, such as Sea-Land-Sea, or Land-Sea-Land. One may note
that on the border between the two ground segments, Sea-Land, or Land-Sea with
different parameters, the abrupt increase or decrease of the resultant propagation factor
occurs (Figure 4.23). Particularly it may be noted from both, Figure 4.23a and Figure 4.23b
that when radio wave crosses the Land –Sea border a small raise in the field strength may
occur instead of the smooth decrease that occurs on regular homogeneous propagation
paths.

Figure 4.23. Observed pattern of the field strength as a function of the distance for
two different inhomogeneous propagation paths
1 – Graph for the homogeneous path over the sea surface
2 – Resultant graph for the inhomogeneous path (solid line)
3 – Graph for the homogeneous path over the land surface

Finally, we will consider a phenomenon, which is known as the coastal refraction effect that
was first observed during World War-I. This effect occurs when a ground radio wave of MF

219
or lower frequency band crosses the border of sea-to-land, or land-to-sea with a non-zero
angle of incidence on the border line: propagation direction changes while crossing a sea-
to-land or land-to-sea border.
To fully understand the physical mechanism of this phenomenon recall (4.130), written in
complex from as:

E = E0
MPG 
F. (4.141)

Here F = F ⋅ exp ( i Φ F ) represents a complex propagation factor that impacts both, the

magnitude and phase of the resultant electric field E . The impact is caused by the real
conditions relative to the propagation in reference condition, i.e. propagation over the PEC
ground. Hence, the time variations of the resultant electric field (4.141) at the observation
point may be expressed in complex form as

120 PT GT
E ( t ) = F exp [ i (ω t − k 0 r − Φ F )], (4.142)
r
where k 0 represents a wave number in free space, and Φ F is the argument of the

propagation factor, which is an additional phase shift caused by the presence of the real
earth's surface. In order to derive the propagation phase velocity note, that if at the time
moment t the phase front is located at the distance r, then at the moment t + dt the phase

front will move to the distance r + dr . Taking into account the fact, that Φ F is also
distance-dependent, the following expression may be written for those two time moments,
if assumed the phase remains unchanged:

ω t − k 0 r − Φ F (r ) = ω (t + dt ) − k 0 (r + dr ) − Φ F (r ) + Φ F ′ (r ) ⋅ d r  , (4.143)
 
From (4.143) the phase velocity may be determined as:
dr ω c
vp = = = , (4.144)
d t k + Φ ′ (r ) ′
1 + Φ F (r )
1
0 F
k0

′ ∂ Φ F (r )
where Φ F (r ) = , is a distance-derivative of the argument of propagation factor, c
∂r

is propagation velocity in free space. As one may notice from (4.144) the phase velocity v p


is less than in free space and depends on the Φ F (r ) derivative. For rough estimates of

vp we’ll consider the following two extreme cases:

220
1. If numerical distance is large enough, i.e. its magnitude x >> 1 , then the
combination of (4.137) with (4.132) and (4.133) will result in:

 1  π 60λσ
Φ F (∞ ) ≈ arg  = arg[ i (ε − i 60λσ )] = − tan −1 =
 2 x  2 ε
π  1 
= − tan −1  . (4.145)
2  tan δ 
For the most conducive type of the earth surface, such as sea water, we have
π
tan δ >> 1 , and therefore Φ F (∞ ) ≈ . For the land-area such as dry soil that is
2
close to dielectric, we have tan δ << 1 , and therefore Φ F (∞ ) ≈ 0 .

2. If small numerical distances x << 1 F ≈ 1 , thus Φ F (0 ) ≈ 0 .

Figure 4.24 represents the behavior of Φ F (r ) qualitatively rather than quantitatively, not
just for the above two extreme cases, but for intermediate distances as well.

Figure 4.24. Distance dependence of the argument of propagation factor,


Φ F (r ) for the surface wave propagating along the earth ground of various
conductivities


From the given graphs one may realize, that the derivative Φ F (r ) is a finite number,

which for any value of tan δ becomes zero if r → ∞ , so the phase velocity tends to speed
of light.
Now consider, for example, the surface wave that crosses Sea-Land border as shown in
Figure 4.25.

221
Figure 4.25. The pattern of coastal refraction

If we take into account the fact that the magnitude ε for sea water is much greater than

that for dry soil, then for limited propagation distances the value of the numerical distance
x along the sea area is much smaller than unity, i.e. x << 1 , whereas for the land area

it is large enough, x >> 1 . Hence, the radio wave faces an abrupt change (a decrease) in

argument Φ F while crossing the sea-to-land border. From (4.144) this will result an abrupt

decrease in v p along the path near the border. The wavelength λ = v p / f along the

propagation path will drop in the area close to shoreline, resulting in change of the
direction of propagation as illustrated in Figure 4.25 graphically. We may see that the
direction of propagation is shifted from AOB’ to AOB after crossing the sea-to-land border
at the Point O. The shift angle ∆ϕ , known as the angle of coastal refraction, may be found
from the following approximate expression [22] for this particular case of sea-to-land
propagation

tan ϕ λ
∆ϕ = ± , (4.146)
2π ε Land r2

222
where r2 is the closest distance from reception point to coastal line. ∆ϕ , and ϕ are
shown in Figure 4.25.The distance of the transmitting point to the shore-line is not critical,
and here is considered as infinity. Note that coastal refraction may have a significant
impact on LF or VLF radio-navigation systems resulting in direction finding error ∆ϕ .
Expression (4.146) is derived by assuming the sea surface as a perfect conductor. If
conventionally assumed that for sea-to-land propagation ∆ϕ is positive, then based on the
principle of the reciprocity we may conclude that by exchanging the positions of
transmitting and receiving antennas the sign of ∆ϕ will change from positive to negative.

Note also that ∆ϕ is in inverse proportion to r2 , which means that coastal refraction will
have a significant occurrence only if the receiving point is close to the shoreline. This issue
may be understood by referring again to Mandelstamm's "takeoff-landing" concept, namely
the surface wave propagation is affected significantly by the surface areas that are closely
surrounding the transmitting and receiving points. Along the remaining propagation path
the radio wave propagates being separated from the earth's surface by lifting up. So any
irregularities (including the coastal line) have minimal, or almost no impact on propagation
conditions if the distance between transmitting and receiving points is large enough.

4.3. ASYMPTOTIC DIFFRACTION THEORY OF


PROPAGATION OVER THE SPHERICAL EARTH SURFACE
The accuracy of the radio wave field strength calculation methods at the receiving point
considered until this point were limited by certain ranges of distance, frequency, and
antenna height. For example, the method of geometrical optics is applicable only to
elevated antennas, and is further limited by the range of direct LOS. Corrections presented
in section 4.3 are used to estimate the curvature of the earth's surface. A more rigorous
approach is apparently based on Maxwell's equations solution for real parameters of the
convex, spherical shape of the earth's surface. That approach which was first presented by
V.A. Fock [17] in 1945 is most generic, and is free of limitations such as the “flat earth"
approximation, LOS distances, or "low / high" antenna elevations, and most importantly is
able to support the radio wave field analysis in the diffraction, (or shadow) zone shown in
Figure 4.26 behind the horizontal distance of the horizon, R 0 . In other words, the

223
previously considered methods allowed estimation of the field strength in the range of
distances less, or near to the border between illuminated (1) and shadow (2) zones, shown
in Figure 4.26.

Figure 4.26. Zoning of the earth’s spherical surface: 1. Illuminated


zone, 2. Semi-shadow (penumbra) zone, 3. Shadow (diffraction) zone
( R 0 is the horizon distance)

4.3.1. BASIC CONCEPTS


Now we consider a dipole, either electric or magnetic, placed over the spherical earth’s
surface of finite impedance as shown in Figure 4.26. Mathematical evaluation for the
solutions of this boundary-value problem is given in Appendix-6, where the wave equation
for the H ϕ component of the vertically-polarized wave (vertical electric dipole), or E ϕ

component for the horizontally-polarized wave (vertical magnetic dipole) are analyzed in
spherical coordinates. Simplification is applied based on the fact that the curvature of the
earth's surface is fairly small compared to propagation distances. As a result, the general
solution is obtained by using the separation of the variables technique. By following [17] or
[22] in order to proceed further, the so called "large parameter" is introduced as
1/ 3
k a 
1/ 3
 π ae 
M = 0 e = 
 2   λ0  , (4.147)
 

224
where k 0 = 2π / λ 0 denotes the free space wave number, and ae is the equivalent radius

of the earth: it is introduced in section 5.3 of the Chapter 5 as an equivalent replacement of


the average geometrical earth’s radius, a = 6370 km in presence of the “standard”
tropospheric refraction. The relation between those two radii is ae = (4 / 3) a ≈ 8500 km.

The distance between transmitting and receiving points is counted along the arc of the
great circle as (see Figure A.6.1 in Appendix-6)
R = θ ae = X Lr , (4.148)

R
where X =Mθ = (4.149)
Lr
is called normalized distance (a unitless quantity).
Parameter
1/ 3
ae  λ 0 ae 
2

Lr = = (4.150)
M  π 

denotes a natural unit of distance.


The elevations of antennas are transformed into normalized heights (unitless, vs. real
elevations that are usually expressed in meters) as
h h
y= = . (4.151)
H h M / k0
Here
1/ 3
M 1  ae λ 0 
2

Hh = = (4.152)
k 0 2  π 2 

denotes the natural unit of heights.


Regarding Fock's theory [17], the separation of variables technique results in the splitting
of the wave equation in parabolic form into two ordinary differential equations: one for the
normalized distance X , another for the antennas' normalized heights, y1 and y 2 for the
transmitting and receiving antennas respectively. Then, the resultant magnitude of
propagation factor is evaluated in Appendix-6 in the following form of infinite series
(A.6.65) and is rewritten here:

exp (i X z n ) w (z n − y1 ) w (z n − y 2 )
F ( X , y1 , y 2 ) = 2 π X ∑
n =1 z n − q 2 w (z n ) w ( z n )
. (4.153)

225
Here w belongs to class of special functions called Airy-function, and basically is a
combination of the Airy functions of first and second kinds (see (A.6.43) in Appendix-6).
They may be found in mathematical tables [18, 19], in the MATLAB library of subroutines,
or expressed by second kind Hankel's function of the fractional order, as follows:

π t 
[ 2 / 3 (t ) ] .
1/ 2

w ( t ) = exp (− i 2π / 3) 
( 2) 3/ 2
H1/ 3 (4.154)
 3 
Note, that expression (4.153) contains parameter q introduced by V.A. Fock (see [17, 22])
that was developed based on Leontovich boundary conditions. It takes into account the
electromagnetic properties of the earth's ground and is defined as follows:
• For the vertical polarization:

π ae 1
q = q vert = −i 3 , (4.155)
λ0 ε − i 60λ 0σ
• For the horizontal polarization:

π ae
q = q horiz = −i 3 ε − i 60λ 0σ . (4.156)
λ0
Another parameter z n , noted in (4.153) represents a sequence of the complex roots of the

following differential equation that also relates to the boundary conditions for proper
electric and magnetic field components (see Appendix-6):
w′( z ) − q w ( z ) = 0 , (4.157)
The roots of this equation are given in Table 4.4 for two extreme values of parameter q ,

namely for q = 0 and q = ∞ . It is easy to realize from (4.157), that for q = 0 the roots are

calculated from the equation w′( z ) = 0 , whereas for q = ∞ , from the equation w ( z ) = 0 .

To define z n for any other intermediate value of q ,the following approximations may be

applied:

q q
• If < 1 , then z n , q ≈ z n , q =0 + , (4.158)
z n z n , q =0

q 1
• If > 1 , then z n , q ≈ z n , q =∞ + . (4.159)
z n q

226
As follows from the right-hand-side of (4.159), if q > 30 , then for practical calculations the

value of z n may be chosen approximately the same as for q = ∞ , i.e. z n , q ≈ z n , ∞ . Based

on data provided in Table 2.1 (see chapter 2), we may confirm by simple calculations, that
for any type of real earth's ground the statement q horiz = ∞ is always true for the

horizontally polarized wave of any RF frequency.


Table 4.4

Magnitude z n = z n = z n exp − i π( 3
)*
n, root order
q=∞ q=0
1 2.338 1.019
2 4.088 3.248
3 5.521 4.820
For n = 4, 5, 6,
≈ [1.5 ( n − 0.25)π ] ≈ [1.5 ( n − 0.75)π ]
2/3 2/3

…(and higher)
* Note: the complex number z n is defined by taking a product of the numerical value from proper

column, and exponential multiplier exp ( i π / 3) . For example, z 2, ∞ = 4.088 exp ( i π / 3) .

On the other hand, if q < 0.5 , then we may choose the value of z n , q ≈ z n , 0 , which directly

relates to q = 0 . It is thereby evident from (4.155) and (4.158) that this case may occur for
vertically polarized waves only. Figure 4.27 demonstrates the dependence of parameter q
on frequency for two polarizations, and for radio waves propagating along different types of
earth surfaces.

227
Figure 4.27. Frequency dependence of parameter q for propagation along sea

water ( ε = 80 , σ = 3 S / m ), wet soil ( ε = 30 , σ = 0.03 S / m ), and dry soil ( ε = 3 ,

σ = 10 −5 S / m )

As one may conclude from the above graphs, for horizontally polarized radio waves the
values of z n may always be chosen from the second column of Table 4.4 regardless of the

frequency and type of the earth's ground along the propagation path, i.e. z n = z n , q =∞ .

For the vertically polarized radio wave some caution is needed, i.e.
• The assumption z n = z n , q =∞ is applicable for frequencies higher than 500 MHz

• In the case of sea water and wet soil, the assumption z n = z n , q =0 (last column of

Table 4.4) is applicable for frequencies less than 50 kHz,


• For intermediate frequencies expressions (4.158) and (4.159) must be used for
z n estimates
Within the illuminated zone-1 (see Figure 4.26) the series (4.153) converges very slowly;
hence the calculation of the propagation factor becomes either highly time/effort
consuming, or even impossible. Therefore, it is meaningless to use it instead of the
interference formulas given in sections 4.1 to 4.3. However, in the shadow-zone it
converges fairly fast, thus any term higher than the first one may easily be ignored and a
"single-term" (sometimes called “single-mode”) formula may be adopted as follows:

exp (i X z1 ) w ( z1 − y1 ) w ( z1 − y 2 )


F =2 π X . (4.160)
z1 − q 2 w ( z1 ) w ( z1 )
This expression represents a main achievement of Fock's approach known as the outcome
of the asymptotic diffraction theory (ADT).

228
Expression (4.160) contains three separate terms. First term,

exp ( i X z1 )
U (x ) = 2 π X (4.161)
z1 − q 2
we’ll call a attenuation factor. It depends only on normalized distance X . The second and
third terms are symmetric and may be written in general form (4.162) given below. They
depend only on the transmitting and receiving antennas' elevations y1 and y 2 respectively,
i.e.
w (z1 − y1, 2 )
V ( y1, 2 ) = . (4.162)
w ( z1 )
Those terms are called height-gain function. So, in general, for the given value of q, the
“single-term” propagation factor may be expressed as a product:
F ( X , y1 , y 2 , q ) = U ( X , q ) V ( y1 ) V ( y 2 ) . (4.163)
The fact that both height-gain functions are counted in (4.163) symmetrically comes from
the principle of reciprocity in electromagnetic, and confirms the correctness of all
statements of the Fock's asymptotic diffraction theory.

4.3.2. PROPAGATION BETWEEN GROUND-BASED ANTENNAS

As noted previously, the term "ground-based antenna" signifies y1, 2 << 1 . If we take into

account that in reality the elevations of the ground-based antennas rarely exceed 100
meters, then it is easy to define the frequency ranges where the "ground-based antenna"
concept becomes applicable. Condition y1, 2 << 1 may be rewritten if (4.151) is taken into

account, i.e.
h1, 2 100
= 1/ 3
<< 1 . (4.164)
1  ae λ 0 
Hh 2

2  π 2 

For earth's equivalent radius of ae = 8500 km, condition (4.164) results in λ >> 3 m. In

other words, the "ground-based antenna" approach is conventionally applicable for


frequencies lower than 30 MHz (lower than HF), i.e. when the wavelength is greater than
10 meters. Assuming for those frequencies y1 << 1 and y 2 << 1 , in general expression

229
(4.153) the height-gain functions containing ratio of the Airy functions become equal to
unity, hence the “multi-term” expression (4.153) may be rewritten as

exp (i X z n )
F ≈2 π X ∑
n =1 z n − q 2
. (4.165)

Then in this particular case the proper “single-term” expression is written as

exp ( i X z1 )
F =U ( X )= 2 π X (4.166)
z1 − q 2
by neglecting the contributions from the higher order terms when the distances are large
enough. However for the distances close to or less than those limited by (4.139) an
approximation that is applied to surface wave propagation along the “flat earth” (W&V-d-P
approach: see section 4.2.3) may be considered as an alternative.
Now let’s compare those two approaches for the border distance that is defined by (4.139):

Rkm = 80 / 3 f MHz . (4.167)

As pointed above, this is a distance that puts the limitation on the applicability of Weyl &
Van-der-Pol approach. Note that here we assume rkm ≈ Rkm , i.e. the direct distance from

radiating antenna to observation point is approximately the same as the horizontal distance
along the earth’s surface. Then if (4.167) is combined with (4.149) and (4.150), the
following value for the normalized distance may be found: X = 0.42023 . The propagation
factors calculated based on both, W&V-d-P and ADT approaches are given in Table 4.5.
Table 4.5

f, PROPAGATION FACTOR, dB E, dB per µV/m


R, km
MHz
W&V-d-P ADT W&V-d-P ADT

0.01 371.33 -0.17 4.3131 57.9787 62.4604


0.03 257.46 -0.44 4.9896 60.8931 66.3177
0.1 172.35 -1.30 6.6919 63.5187 71.5059
0.3 119.50 -3.90 8.5128 64.0933 76.5075
1 80.00 -16.60 -17.5965 54.8789 53.8841
3 55.47 -23.47 -28.6953 51.1956 45.9661
10 37.13 -40.49 -43.3703 37.6529 34.7770
30 25.75 -50.92 -53.2175 30.4085 28.1106

230
As one can realize from the above table the results calculated from “single-term” ADT
approach are not trustworthy for the frequencies below 1 MHz, because they violate the
physical meaning, namely because the propagation factor may not be positive. For those
frequencies the calculations must be performed based either on W&V-d-P-method, or by
using “multi-term” ADT approach, which is much more efforts consuming, than W&V-d-P
approach. Electric field calculation results for the distances same as from (4.167) are
shown in Figure 4.28 along with data provided by ITU-R [12].

Figure 4.28. Comparison of the electric field strength calculation results (in dB per
µV/m) obtained from W&V-d-P and ADT methods with those provided by ITU-R for

data shown in Figure 4.30 for ε = 40 and σ = 3 ⋅ 10 − 2 S/m (wet soil).

Comparison between results obtained from the calculations based on W&VdP and ADT
methods for propagation along the wet soil confirms that for the frequencies above 3 MHz
both approaches are fairly applicable. However for the frequencies below 3 MHz the W&V-
d-P calculation method that is usually considered as the one preferred.
The contributions from the higher order terms in ADT-approach may be assessed from
Figure 4.29 where the ratios of the magnitudes of the second and third terms to the
magnitude of the first term are shown in graphical from for three types of the ground: dry
soil, wet soil, and sea water. The graphs represent vertically polarized radio wave, which is
most common for the ground-based antennas; those graphs are obtained by using the
MATLAB programming tool.

231
Figure 4.29. Contribution of the second and third terms relative to the first term (in
percent) of the multi-term propagation factor (formula 4.165) for the vertical
polarization
• Contributions from the second term: (1) – 300 kHz, (2) – 3 MHz, (3) – 30 MHz
• Contributions from the third term: (4) – 300 kHz, (5) – 3 MHz, (6) – 30 MHz

One may realize from this figure that larger is the distance from the transmission to
observation point, less is the contribution from the higher order terms in (4.165), thus
higher is the accuracy of the single-term formula (4.166), especially for the propagation
above the sea water. It becomes clear, that depending on calculation tolerances, a single-
term formula is applicable for the distances greater than hundreds of kilometers.

-----------------------------------------------------------------------------------------
Example 4.10
Calculate and plot distance-dependence of the electric field strength for the initial data
given in Example 4.9, and for distances 80 km and up, based on two approaches:
• Weyl and Van-der-Pol (W&V-d-P) approach (“flat earth” surface wave approach)
• ADT - Fock’s asymptotic diffraction theory approach (single-term formula)
Compare the results with those provided by ITU-R (see Figure 4.30)
Solution

232
1. W&V-d-P - method:
• Natural unit of distance is defined by (4.133) as
λ 0 ε2 2 300 (40 − i 60 ⋅ 300 ⋅ 3 ⋅10 −2 ) 2
s = i =i = 5.17 ⋅10 4 ∠4.34 0
π ε2 − 1 3.14 (40 − i 60 ⋅ 300 ⋅ 3 ⋅10 − 1)
−2

• propagation factor is defined by (4.137), electric field strength is defined by (4.130).

3. ADT – method
• The normalized distance, x = R / Lr , and the natural unit of distance is defined by
1/ 3
 λ 0 ae 2   300 ⋅ 8500000 2 
1/ 3

(4.150) Lr =   =   = 190372 m
 π   3 . 14 
 
• q-parameter is defined by (4.155)

π ae 1 3.14 ⋅ 8500000 1
q vert = −i 3 = −i 3
λ0 ε − i 60λ 0σ 300 40 − i 60 ⋅ 300 ⋅ 3 ⋅10 −2
q vert = 1.92∠ − 47 0

• It may be shown that for the given q vert the condition (4.159) is applicable, thus

1
z1, q ≈ z1, q =∞ + = 1.5236 − 1.6429 i = 2.2407∠ − 47.2 0
q
• Then the magnitude of denominator in (4.166) may be found as follows:

z1 − q 2 = 2.7093
• The magnitude of numerator in (4.166) may be transformed to:

exp (i X z1 ) = exp (− X Im z1, q ) = exp (−1.6429 X )


• Then finally, for this particular case, (4.166) may be written as

πX
F =U ( X )= exp (− 1.6429 X )
1.3546
• All calculation results are shown in Table E4.10, and on Figure E4.10

Table E4.10

E, dBµV/m
R, km
W&V-d-P ADT ITU-R

233
80 61.7 64.1 64
120 54.6 59.3 57.7
160 49.6 55.0 52.5
200 45.8 51.1 47.9
240 42.6 47.3 43.8
280 39.9 43.6 40
320 37.6 40.0 36.2
360 35.5 36.5 32.6
400 33.7 33.1 28.9
440 32.1 29.7 25.5
480 30.5 26.3 22.3

Figure E4.10. Distance dependence of the electric field strength in Example E4.10

Both, Table E4.10 and Figure E4.10 demonstrate the comparison of the calculation results
with the data recommended by ITU-R [12] as a reference (Figure 4.30). As one may
realize from the considered examples E4.9 and E4.10 the field strength induced at the
observation point by the ground-based antenna in close vicinity of the earth surface at
relatively low frequency, may be accurately calculated based on Weyl & Van-der-Pol
approach, for distances limited by (4.139) condition. Otherwise a single-term formula
provided by asymptotic diffraction theory may be used as a fairly good propagation
prediction tool. It is apparent that more terms are included in multi-term formula (4.165)
better accuracy may be achieved.
---------------------------------------------------------------------------------

234
As stated in subsection 4.3.1, for vertically polarized radio waves of frequencies up to 10
kHz, parameter q vert ≈ 0 . Hence, the expression (4.166) for the magnitude of the

propagation factor may further be simplified. Namely, based on (4.158) the value of first
order z-parameter is defined as z1 ≈ z1, 0 = 1.019∠π / 3 = 0.51 + i 0.88 . The exponential

term in numerator of (4.166) may be written as exp ( i X z1 ) = exp ( −0.88 X ) . Hence we

will rewrite the expression for the propagation factor (4.166) in the following approximate
1
form

2 π
F≈ X exp (−0.88 X ) = 3.48 X exp (−0.88 X ) . (4.168)
z1
For the remainder of the considered frequencies, i.e. form 10 kHz to 30 MHz, the
expression (4.166) must be used for large enough distances, and for the properly defined
value of the q form (4.158) or (4.159), as well as for proper z1, q , based on Table 4.4.

To best support engineering estimates a complete set of graphs of the effective field
strength (RMS values) as a function of distance and frequency is introduced by ITU-R in
[12]. As an example, one of those graphs is shown in Figure 4.30 that illustrates predicted
field strengths of the radio wave generated by a quarter-wavelength vertical monopole over
wet soil.

1
Note that expression (4.168) is based on ground waves propagation mechanism only. No
atmospheric effects are taken into account. In reality the intensity of the field in VLF frequency band
and lower is highly impacted by ionosphere, which is outside of the scope of this text.

235
Figure 4.30. A family of graphs of the electric field strength of radio wave from
vertical quarter-wavelength ground-based monopole as a function of frequency
and distance. Parameters of the ground are: ε = 40 and σ = 3 ⋅ 10 − 2 S/m [12,
Figure 4, p.8].

The family of graphs given in Figure 4.30 was designed based on expression (4.130),
which we will rewrite here:

60 PTx GTx V
E = E0 F=
MPG
F, . (4.169)
r m
MPG
It includes reference electric field E 0 , i.e. the field strength generated by the vertical

monopole over the ideal, perfectly conducting spherical ground surface., and the
propagation factor F , which takes into account real losses caused by the semiconducting
earth's ground. The graphs are designed for nominal radiated power PTx = 1 kW, and

quarter-wavelength monopole antenna of the gain GTx = 1.5 (see Chapter 3). If those

nominal values of PTx and GTx are substituted into (4.169), and, in addition, the distance r

is expressed in kilometers, and E in µV/m, then (4.169) will easily be transformed to

236
3 ⋅10 5 µV
E′ = F, . (4.170)
rkm m
The plots in Figure 4.30 are designed based on (4.170). For the power PT and the
antenna gain other than nominal values the following conversion may be applied:

GTx µV
E = E′ PTx , kW , . (4.171)
1.5 m

4.3.3. PROPAGATION BETWEEN ELEVATED ANTENNAS


It was noted in section 4.3.1 that for the frequency bands of VHF and higher antennas are
considered as elevated. Single-term propagation factor in its general form (4.160) may be
modified for q = ∞ . This value of q-parameter is specific for VHF and higher, namely from

(4.159) z1, q ≈ z1, ∞ and from the Table 4.4 z1, q ≈ 2.34 exp ( i π / 3) = 1.17 + i 2.02 . If this

value is substituted into (4.160), then, taking into account expression (4.157), it may be
transformed to

F =2 π X
[
exp X ⋅ Im ( z1, ∞ ) ] w ( z1, ∞ − y1 ) w ( z1, ∞ − y 2 ) ≈
z1, ∞ w′ ( z1, ∞ ) w′ (z1, ∞ )
1− 2
q

w ( z1, ∞ − y1 ) w ( z1, ∞ − y 2 )
≈ 2 π X exp (− 2.02 X )
w′( z1, ∞ ) w′( z1, ∞ )
, (4.172)

where w′ is a derivative of the Airy function. Note that (4.172) is another “single-term”
expression, similar to (4.163). It contains three composite parts:
F ( X , y1 , y 2 ) = U EL ( X ) VEL ( y1 ) VEL ( y 2 ) . (4.173)

It’s obvious however that in case of elevated antennas the distance factor, U EL and

elevation factors, VEL are not the same as in (4.163). They are:

U EL ( X ) = 2 π X exp (− 2.02 X ) (4.174)

and

237
w ( z1, ∞ − y1, 2 )
VEL ( y 1, 2 ) =
w′ ( z1, ∞ )
. (4.175)

The elevation factor depends solely on the normalized height of antenna (either y1 or y 2 ),
as in (4.160). The numerator in (4.175) is defined as a combination of the Airy functions
(see (A.6.43)), whereas the denominator is the same for derivatives.
It is evident that if the normalized heights of the antennas are close to zero, then w′
( )
remains a finite number for q → ∞ only if w z1, ∞ → 0 in order to satisfy (4.157). In other

words, if the antennas' elevations are decreasing to zero, then the elevation factors will
decrease to zero as well, unlike those in (4.160), where height-gain functions tend to unity.
The U EL ( X ) and VEL ( y ) dependencies are represented in Figure 4.31, which may be
used for the practical calculations.

Figure 4.31. Graph of the distance factor U EL ( x ) , and elevation factor VEL ( y ) for
“elevated” antennas’ field strength calculations in diffraction zone

----------------------------------------------------------------------------------------------------------

Example 4.11
The transmitting antenna with 35 dB ( GT = 3162 unitless) gain is placed on the height

h1 = 30 m above the dry soil of parameters: ε = 4 and σ = 10 − 4 . For a total radiated


power of PT = 1 kW, and a frequency of 1 GHz ( λ 0 = 0.3 m), calculate and plot the

238
distance-dependence of the field strength at the h 2 = 30 reception point above the ground

within the distance range from 50 to 500 kilometers.

Solution

• Natural unit of heights is calculated based on (4.152):

1/ 3
1  ae λ 0
2
 1  8500 ⋅10 3 ⋅ 0.32 
1/ 3

Hh =   =   ≈ 21.32 m
2  π 2  2 3.14 2 

• Transmitting antenna and reception point normalized heights are:
30
y1 = y 2 = = 1.41
21.32
• Elevation factors may be found from Figure 4.31, or calculated based on (4.180).

o From Table 4.4 z1, ∞ = 2.338 exp( i π / 3) , then z1, ∞ − y1, 2 = -0.241 + 2.0248 i

o Using the MATLAB special function utility the elevation factor is found as

w ( z1, ∞ − y1, 2 )
VEL ( y1, 2 ) = = 1.76 (4.9 dB)
w′ ( z1, ∞ )
• Natural unit of distance is calculated based on (4.150):
1/ 3
 λ 0 ae 2   0.3 ⋅ (8500 ⋅10 3 ) 2 
1/ 3

Lr =   =   = 19037 m
 π   3.14 
 

• Numerical distance: X = R / Lr

• Distance factor is defined as U EL ( X ) = 2 π X exp (− 2.02 X )


• Propagation factor: F ( X , y1 , y 2 ) = U EL ( X ) VEL
2
(y )
1, 2

• Electric field strength is defined as E = E0 F , where E0 = 30 PT GT / r is a free


space electric field
• Calculation results are shown in Table E-4.11 and on Figure E-4.11.
Table E-4.11
R, km 50.00 100.00 200.00 300.00 400.00 500.00
F, dB -21.15 -64.22 -153.37 -243.77 -334.69 -425.88
E, dBmV/m 84.65 35.55 -59.62 -153.55 -246.96 -340.09

239
Figure E-4.11. Distance dependence of the electric field strength in Example 4.11.

-----------------------------------------------------------------------------------------------

4.3.4. SPECIFICS OF PROPAGATION ESTIMATES IN PENUMBRA


ZONE
To properly assess the application limits of the “single-term” formula (4.172) for VHF and
higher frequency bands, we’ll apply the law of cosines for the triangle AOC in Figure 4.26

(ae + h) 2 = ae + r 2 + 2ae r cos γ ,


2
(4.176)

where ae = 8500 km is the earth's equivalent radius, γ is the elevation angle at the

observation point. Taking into account the assumption h 2 << ae , expression (4.176) may
2

be simplified to
2 ae h − r 2
cos γ ≈ . (4.177)
2 r ae
After multiplication by the "large parameter" M from (4.147) and performing simple
algebraic transforms, (4.177) may be rewritten as

y−X2
p = M cos γ ≈ . (4.178)
2X

240
Here the approximation sign is used because the direct distance AC is replaced by the
horizontal distance, i.e. AC = r ≈ R . We may further see from Figure 4.26, that the horizon

distance is defined for γ = 90 0 , and therefore p = 0 . Then the numerical value of the
horizon distance is

X0 = y, (4.179)

which is a “normalized” form of (4.39). Unfortunately at the distance x0 and in its vicinity

the single-term ADT-approach does not provide the accuracy that is enough for the
practical applications. As a good compromise between Weyl & Van-der-Pol (W&V-d-P)
approach and asymptotic diffraction theory (ADT) approach is a calculation method
evaluated in MIT radiation lab and presented after WW-2 by Burroughs and Atwood (B&A-
method) in [2, pp. 404-421]. The idea of B&A-method is the use of so called “shadow
factor” for counting for the convexity of the earth surface concurrently with “flat earth”
formulas, as well as including the elevation factors similar to those from ADT approach.
Like in ADT, the “single-term” formula, presented by B&A, is a function of normalized
distance x (small letter) and normalized heights h1 and h2 as follows:

FBA = 2 FWVdP ( x) FS H BA (h1 ) H BA (h2 ) . (4.180)

The composite parts of (4.180) are given below without strict mathematical evaluations:
• FWVdP (x) is a propagation factor defined from the “flat earth” approach given in
section 4.2.3 that is solely dependent on normalized distance x for “flat earth” 1.
The expression (4.137) for magnitude of FWVdP and x is applicable here

• FS is a “shadow factor” given as

FS = 2π X 3 / 2 exp(− 1.607 X ) , (4.181)

where the capital letter X


1/ 3
 λ 0 ae 2 
X = R/  (4.182)
 2π 
 
is used here for the normalized distance for the spherically convex earth’s surface
• H BA (h) is a high-gain factor used in (4.180) for the antenna elevations h1 and h2 :

1
Do not confuse normalized distance x introduced in W&V-d-P method with X in ADT method
241
2
2h h 
H BA (h) = g (h) 1 + +  
(
hs 4Q + 1  hs
2
) 
(4.183)

  2hc 
0.904


g (h) = = 0.1356  h  × 10 for h > hc
0.948 h / ( 2 hc )
(4.184)
= 1 for h ≤ hc

1 ε
Q= = (4.185)
tan δ ε 60λ 0σ

 λ0 ε
= for vertical polarization
 2π ε − 1
hs =  (4.186)
λ
= 0 1
for horizontal polarization
 2π ε − 1

1/ 3
 λ 0 2 ae 
hc = 0.5   (4.187)
 4π 2 
 
It must be noted that a single-term B&A-method was initially designed for the distance
ranges beyond the LOS, i.e. on and beyond the horizon line. However it is fairly accurate
for the distances much shorter. The use of this method is demonstrated below as an
example.
---------------------------------------------------------------------------------
Example 4.12

Utilize both B&A and ADT methods to calculate propagation factor on the sample path with
the following initial data:
• Path length: R = 100 km
• Antenna heights: h1 = 10 m, h2 = 100 m

• Frequency: 300 MHz ( λ 0 = 1 m)

• Parameters of the earth’s surface: ε = 75 , and σ = 5 S/m,

ε = ε − i60λ 0σ = 75 − i 300 , ε = 309.2

• Horizon distance: R0 = 2a e ( h1 + )
h 2 = 54.3 km

242
Perform the calculations for both, vertical and horizontal polarizations. Extend calculations
for the distance range from 0.8 R 0 to 4R 0 , where R 0 is the horizon distance, and plot the

distance dependences of the propagation factors to compare calculation methods.


Solution

1. B&A-method

• Natural unit of distance is defined from (4.133) and (4.133a): s = 98.36


(vertical polarization), s = 0.00103 (horizontal polarization)

• Normalized distance is defined from (4.132). We’re interested in


magnitudes: x = 1016.67 (vertical polarization), x = 97.09 ⋅10 6 (horizontal
polarization)

• “Flat earth” propagation factor is defined from (4.137): FWVdP = 4.926 ⋅10 −4

(-66.15 dB, vertical polarization), FWVdP = 5.1523 ⋅10 −9 (-165.76 dB,

horizontal polarization)

1/ 3
 λ 0 ae 2 
• Normalized distance from (4.182) X = R /   = 4.43
 2π 
 

• Shadow factor is defined from (4.181):

FS = 2π X 3 / 2 exp(− 1.607 X ) = 0.01892 (-34.45 dB)

• Calculations of the high-gain factors are based on (4.183) – (4.187):


Q = 0.25 , hs = 2.8 m (vertical polarization), hs = 0.0091 m (horizontal

polarization), hc = 29.97 m, g (h1 ) = 1 , g (h2 ) = 1.4315 , H BA (h1 ) = 4.41

(12.89 dB, vertical polarization), H BA (h1 ) = 1105.35 (60.78 dB, horizontal

polarization), H BA (h2 ) = 52.3 (34.37 dB, vertical polarization),

H BA (h2 ) = 15649.5 (83.98 dB, horizontal polarization),

• Propagation factor is defined from (4.180) as a product, or as a sum of dB


values: FBA = 4.305 ⋅10 −3 (-47.32 dB, vertical polarization),

FBA = 3.412 ⋅10 −3 (-49.34 dB, vertical polarization)

243
ADT-method

1/ 3
 λ 0 ae 2 
• Natural unit of distance defined from (4.150) Lr =   = 28437.8 m
 π 
 

• Antenna natural unit of heights defined from (4.152)


1/ 3
1  ae λ 0
2

Hh =   = 47.57 m
2  π 2 

• Normalized distance defined from (4.149) as X = 3.5164

• Antennas normalized heights defined from (4.151) as y1 = 0.2102 , and

y 2 = 2.102

• Distance factor defined by (4.174):

U EL ( X ) = 2 π X exp (− 2.02 X ) = 0.0054 (-45.4 dB)

• Elevation factor for the transmitting antenna defined by (4.175):

VEL ( y 1 ) = w ( z1, ∞ − y1 ) / w′ ( z1, ∞ ) = 0.212 (-13.48 dB)

• Elevation factor at the reception point defined by (4.175):

VEL ( y 2 ) = w ( z1, ∞ − y 2 ) / w′ ( z1, ∞ ) = 3.108 (+9.85 dB)

• Propagation factor: FADT = U EL ( X )VEL ( y1 )VEL ( y 2 ) = 0.003538 (-49.025 dB)

Comparison between B&A- and ADT-methods demonstrates fairly close results, which
may be noticed either from the above calculation, or from the Figure E4.12 below. The
reason that ADT method do not distinguish between polarizations is because of the
parameter q that is extremely large for the frequencies VHF and higher for both, vertical
and horizontal. In ADT-method it is taken as infinitely big, which allows significant
simplifications. However that difference is counted in B&A-method, which results in slight
differences in the outcomes.

244
Figure E4.12. Comparison between calculation methods in diffraction zone

Another surprising fact may be emphasized from this example is that the “single-term”
approach seems to be applicable not just in shadow zone, but even in wide vicinity of the
horizon. We may assume, that as far as this statement is true for 300 MHz, then it will be
true for higher frequencies, because, as pointed above, higher is the frequency better the
series (4.153) converges to a “single-term” case.

---------------------------------------------------------------------------------

REFERENCES

[1] Dolukhanov, M. P., Propagation of Radio Waves, Moscow, USSR: Mir Publishers, 1971
[2] Burroughs, C.R., and Atwood, S.S. (Editors), Radio Wave Propagation. Consolidated
Summary Technical Report of the Committee on Propagation of the National Defense
Research Committee, NY: Academic Press, 1949
[3] Collin, R.E., Antennas and Radio Wave Propagation, NY: McGraw-Hill Book Co.,1985
[4] Saunders, S.R., Antennas and Propagation for Wireless Communication Systems.
Second edition, UK: John Wiley & Sons Ltd., 2007
[5] Doble, J., Introduction to Radio Propagation for Fixed and Mobile Communications,
Norwood, MA: Artech House, 1996
[6] International Telecommunication Union, ITU-R Recommendation P.529-2: Prediction
methods for the terrestrial land mobile service in the VHF and UHF bands. Geneva, 1997

245
[7] COST 231 Final report, Digital Mobile Radio Towards Future Generation Systems,
Commission of the European Communities and COST Telecommunications,
Brussels,1999
[8] Lee, W.C.Y., Mobile Design Fundamentals, NY: John Wiley & Sons, 1993
[9] Walfish, J., and Bertoni, H. L., “A Theoretical Model of UHF Propagation in Urban
Environments,” IEEE Transactions In Antennas and Propagation, Vol.AP-38, 1988,
p.p.1788-1796
[10] Ikegami, F., et al., “Propagation Factors Controlling Mean Field Strength on Urban
Streets,” IEEE Transactions on Antennas and Propagations, Vol. AP-32, 1984, p.p. 822-
829
[11] Barclay, L., Propagation of Radiowaves, (2-nd Edition), The Institution of Electrical
Engineers, London, 2003
[12] Rec. ITU-R P.368-7: “Ground-wave propagation curves for frequencies between 10
kHz and 30 MHz”, 1992
[13] Vaughan, R., Andersen, J.B., Channels, Propagation and Antennas for Mobile
Communications. IEEE, 2003
[14] Hall, M.P.M., and Barclay, L.W., Radio wave propagation, London, UK: Peter
Peregrinus, Ltd., 1989
[15] Sarkar, T.K., et al., “Survey of Various Propagation Models for Mobile
Communications,” IEEE A&P Magazine, Vol. 45, No.3, June 2003
[16] Al’pert, Ya.L., Radio Wave Propagation and Ionosphere. Vol 2, Propagation of
Electromagnetic Waves Near the Earth, NY: Consultants Bureau, 1974
[17] Fock, V.A., Electromagnetic Diffraction and Propagation Problems, UK, Oxford:
Pergamon Press, 1965
[18] Gradshtein, I.S., Ryzhik, I.M., Tables of Integrals, Series, and Products, Seventh Ed.,
Elsevier, Inc., 2007
[19] Abramovitz, M, and Stegun, I.A., Handbook of Mathematical Functions with Formulas,
Graphs, and Mathematical Tables, Washington DC: NBS, 1972
[20] Maclean, T.S.M., and Wu, Z., Radiowave Propagation Over Ground, London, UK:
Chapman & Hall, 1993
[21] Леонтович, М.А., “Об одном методе решения задач распространения радиоволн
по земной поверхности,” Известия АН СССР, Серия Физика, т.8, 1944 (in Russian).
[22] Feinberg, Ye.L., The Propagation of Radio Waves along the Surface of the Earth,
Wright-Patterson AFB, Ohio, 1967 (Translated from Russian)

246
[23] Millington, G., “Ground Wave Propagation over an Inhomogeneous Smooth Earth,”
Journal of IEE, Vol. 96, part III, 1949, p.53
[24] IEEE 100. The Authoritative Dictionary of IEEE Standard Terms. Seventh Edition.
Standard Information Network. IEEE Press, NY, 2000
[25] Blake, V.L., Radar Range-Performance Analysis, Norwood, MA: Artech House, 1996

PROBLEMS

P4.1. A Line-Of-Sight communication link operates at 100 MHz, with a transmitting


antenna of horizontal polarization installed at the height of 60 m above the flat earth. The
distance between transmitting and receiving antennas is 4 km. Determine the height of the
receiving antenna to achieve maximum of received signal.
Answer: 50 m
P4.2. An omnidirectional FM broadcast antenna of horizontal polarization, at the height h1
= 80 m has a gain of GTx = 5 (unitless) at the frequency f = 100 MHz. Calculate the field
strength of the radio wave for the following distances, if the receiving antenna is placed at
the height h2 = 10 m, and the transmitter’s output power is PTx = 500 W: (1) r = 1 km, (2) r =
6 km, (3) r = 25 km.
Note: 1). Take into account the divergence factor (earth’s surface convexity) if needed

2). Consider the propagation over the dry soil ( ε = 3, σ = 10 −4 S/m), without any roughness
on ground surface
Answer: 512.2 mV/m, 25.5 mV/m, 1.46 mV/m
P4.3. Evaluate the expression for the propagation factor, for vertically polarized radio
waves similar to those in Section 4.1. Plot the graph and provide physical explanations of
the graph pattern.

P4.4. Base Station (BS) of the cellular network radiates the radio wave of power of 10 W at
a frequency of f = 850 MHz. Transmitting antenna has a gain of 10 (unitless), and is placed
at the elevation hb = 80 m. Determine the threshold (minimum) value of the power at the
receiving point if the coverage area of the BS has a radius of r = 3 km and is located in an
urban area. Assume that the average height of the mobile receiving antenna is hm = 2 m,
and gain Gr = 1.5 (unitless). Use following two approaches for calculations:

247
1. Okumura-Hata empirical model,
2. Walfish-Ikegami physical model for hroof = 50 m, w = 100 m, and two extreme

orientations of propagation paths relative to the street directions: (a). ϕ = 0 0 (LOS

between corresponding antennas), and (b). ϕ = 90 0

Answer: 1). -124.3 dBW, 2a). -121.86 dBW, 2b). -90.6 dBW

P4.5. Calculate a field strength of a vertical monopole placed on homogeneous wet soil
with the following parameters: ε = 15, σ = 10-2 S/m. The initial data is as follows:
o Power at the input of transmitting antenna – 10 W, antenna gain – 20
o Frequency – 20 MHz
o Propagation distance – 12 km
Answer: 33.4 µV/m

P4.6. A coastal navigation system operates at the frequency of f = 1 MHz. The system is
placed at a distance of r2 = 300 m from the sea - wet soil border. Plot the correction chart
for the ∆ϕ error as a function of the azimuthal angles of arrival ϕ.

P4.7. Validate the values of propagation factors given in Table 4.5 based on both
approaches: W&VdP and ADT. Parameters of the earth’s surface are: ε = 40 ,

σ = 3 ⋅10 −2 S/m.

P4.8. Repeat problem P4.8 for the parameters of the earth’s surface as follows:
• ε = 3 , σ = 3 ⋅10 − 5 S/m (dry soil)
• ε = 80 , σ = 3 S/m (sea water)

P4.9. Calculate a field strength of a vertically polarized antenna of h1 = 30 m elevation over


the homogeneous sea surface, for the frequency f = 50 MHz. The receiving antenna of h2
= 20 m elevation is placed at a distance of r = 100 km. The transmitted power is PTx = 2
kW, and transmitting antenna gain is GTx = 15 dB.
Answer: 33.5 µV/m

248
P4.10. A transmitting antenna of horizontal polarization is placed at a height of 50 m above
the earth’s ground. Calculate the field strength at the receiving point,100 km away from the
transmitting antenna, with the following initial data:
• Frequency: 3 GHz
• Power transmitted: 1 kW
• Transmitting antenna gain: 40 dB
• Elevation of the receiving point above the earth’s ground: 20 m
Answer: 6.28 µV/m

APPENDIX-5
INPUT IMPEDANCE OF THE RADIATING CURRENT
ELEMENT ABOVE THE PEC GROUND PLANE

A current filament of infinitesimal length with uniformly distributed current in free space has
been analyzed in section 2.5 and in Appendix-3. The expressions (2.134) and (2.135)
allow determining electric and magnetic field strengths at far distances from the filament
location. Assume PΣ is the power that is transferred from the source to the filament, to be

radiated into free space. Then excitation of the current I in that filament must be
associated with the impedance Z ant , 0 , which is considered as a load for the power source,

and actually is an input impedance of the radiating filament. Free space radiation means
no PEC ground is considered at this moment. Then the total radiated power PΣ = Z ant , 0 I 2
2
may be defined if the power flow density (Pointing vector) Π 0 = E 0 / 2 W0 is integrated

along the whole spherical surface, centered at the radiator.


2
2π π E 0
PΣ = Z ant , 0 I 2 = ∫Π 0dS =
Whole Sphere
∫ ∫ 2W
0 0 0
r 2 sin θ dϕ dθ , (A5.1)

where W0 = 120π Ohms is the intrinsic impedance of free space, and

Il
E 0 = k 0W0 sin θ (A5.2)
4π r

249
is the magnitude of the electric field defined from (2.135). After proper substitutions we'll
rewrite (A5.1) as
2 π
( I l ) 2 k 0 W0 I 2l 2 2
PΣ = Z ant , 0 I 2 = π ∫0 θ θ =
3
2 sin d k 0 W0 . (A5.3)
16π 2 2 12π
Hence, for the radiator in free space the input impedance becomes
2
l 2 k 0 W0
Z ant , 0 = . (A5.4)
12π
Now consider current elements over the PEC ground plane: one placed vertically, another
– horizontally, at the height h as shown in Figure A5.1a. For further analysis
image method is employed, i.e. the PEC ground plane is replaced by the mirror image of
those elements. Note that for the vertical current element the image has the same direction
of the current, but for the horizontal current element it has the opposite direction (see
Figure A5.1a).

Figure A5.1. a). Positions of images induced by vertical and horizontal current
elements in PEC ground plane, b). Ray tracing pattern for the vertical radiating
current elements above PEC ground plane

At the far enough observation point the overlay of the fields from both, real current element
and its image is to be considered. Electric field strength from the real current element in
any point of the free space above the ground is presented by (2.135), which we denote as
E 1 with the magnitude given in (A5.2). From Figure A5.5 we may see that the wave
coming from the image is shifted in phase because of the difference in propagation
distances ∆r , and may be written as

250
E 2 = E 1 exp (−ik∆r ) . (A.5.5)

The resultant field is written as a sum of the phasors in complex plane

E 0 = E 1 + E 2 = E 1 [1 + exp(−ik 0 ∆r )] .
MPG
(A5.6)

This expression is written under traditional assumption that electric fields of both waves are
close to collinear therefore vector sum is replaced by the algebraic sum of two complex
phasors. The magnitude of the expression in brackets is easy to express as

 k ∆r 
1 + exp(−ik 0 ∆r ) = 1 + cos(k 0 ∆r ) + i sin( k 0 ∆r ) = 2 cos  0  . (A5.7)
 2 
The argument of the cosine in the RHS may be derived based on the geometric sketch in
Figure A5.1b: ∆r = 2h cos θ . Then we may rewrite (A5.7) in the following from:

1 + exp(−ik 0 ∆r ) = 2 cos (k 0 h cosθ ) . (A5.7a)

Based on (A5.6), and taking into account (A5.2) the magnitude of the resultant field E 0
MPG

induced at the reception point by the vertical current element may finally be written as
Il
E 0 = E 1 1 + exp(−ik 0 ∆r ) = k 0W0 ⋅ sin θ ⋅ cos(k 0 h cos θ )
MPG
(A5.8)
2π r

Similar approach may be used to develop the magnitude of the resultant field E 0
MPG

induced at the reception point by the horizontal current element. Just keep in mind that the
image current is 180 degrees out of phase relative to the current in original element.
Il
E 0 = k 0W0 ⋅ sin θ ⋅ sin (k 0 h cos θ )
MPG
(A5.9)
2π r
Power radiated by monopole may be defined by substitution of (A5.8) and/or (A5.9) into
(A5.1), and the integration must be performed for the upper semispherical surface only, i.e.
the integration by θ must be performed within 0 ≤ θ ≤ π / 2 limits. Then the power radiated
by the vertical current element at the height h above the PEC ground may be found after
some transforms as
π /2
(I l) 2 2
= k 0 W0 ⋅ ∫ sin θ ⋅ cos 2 (k 0 h cos θ ) dθ .
MPG 3
PΣ (A5.10)
4π 0

(I l) 2 2
Now, if we use the following notations: B = k 0 W0 , b = k 0 h , and cos θ = x , then

(A5.10) may be simplified to

251
( )
1
= B ⋅ ∫ 1 − x 2 ⋅ cos 2 (bx) dx .
MPG
PΣ (A5.11)
0

The close form for this integral may be found from the table [1, p.186] and the result is

( I l ) 2 2  1 cos(2k 0 h) sin (2k 0 h) 


= k 0 W0  − +
MPG
PΣ . (A5.12)
4π  3 4(k 0 h)
2
8(k 0 h) 3 

= Z ant
MPG MPG
Taking into account a relation PΣ I 2 we'll end up with the following expression

 3 cos(2k 0 h) 3 sin (2k 0 h) 


= Z ant , 0 1 − +
MPG
3 
Z ant 2
. (A5.13)
 4 ( k 0 h ) 8 ( k 0 h ) 
In case of the horizontal current element at the height h from the PEC ground
similar procedures may be applied. However, in equivalent replacement of the PEC ground
plane by its image, both radiating filaments become parallel to each-other, thus a strong
coupling between them is to be taken into account. Original evaluation is performed by
Schelkunoff and Friis [3], and the final expression for the input impedance is given in
following form [2]:

 3 sin (2k 0 h) 3 cos(2k 0 h) 3 sin (2k 0 h) 


= Z ant , 0 1 − − +
MPG
3 
Z ant 2
. (A5.14)
 4 ( k 0 h ) 8 ( k 0 h ) 16 ( k 0 h ) 
Expressions (A5.13) and (A5.14) are presented graphically in Figure 4.15.

REFERENCES
1. Gradshteyn, I.S., Ryzhik, I.M. Table of Integrals, Series, and Products. Academic
Press, 1980
2. Maclean, T.S.M., Wu, Z. Radiowave Propagation Over Ground. Chapman & Hall,
London, 1993
3. Schelkunoff, S.A. Electromagnetic Waves. D.Van Nostrand Co., Inc., NY, 1943

252
APPENDIX-6

DIFFRACTION OF RADIO WAVE AROUND EARTH'S


SURFACE: BASIC THEORY

A.6.1.a. GENERAL SOLUTION OF THE WAVE EQUATION


RELEVANT TO PROPAGATION FACTOR

Rewrite first (2.1) and second (2.2) Maxwell's equations for the source-free region and
harmonic variations of the field as

∇ × H = iω ε 0 ε E , (A.6.1)

∇ × E = −iω µ 0 H , (A.6.2)

where ε is a complex dielectric constant of the propagation medium, which here is


considered as a non-magnetic ( µ = 1 ). For the problems that relates to propagation along

the spherical boundary of the earth the spherical coordinate system ( r , θ , ϕ ) is to be


adopted as the most appropriate (Figure A.6.1).

Formula for the curl of vector H in spherical coordinates may be expressed by its
components as provided by (A.2.4.9) in Appendix-1.

 1  ∂H θ   1  ∂ 1 ∂H r  

 
∇×H =   −
∂ 
(
H ϕ sin θ ) 
 r 0 +   ( )
rHϕ −  θ 0
 r sin θ  ∂ϕ ∂θ  

r
  ∂ r sin θ ∂ϕ  

1  ∂H r 

+ 
 ∂θ


∂r
( )ϕ
rH θ 0. (A.6.3)
 r  

253
Figure A.6.1. Transmission point (A) and observation point (B) over the convex
earth's surface in spherical coordinates: r = a + h is the radial coordinate, θ is
zenith angle, ϕ is azimuth angle of the observation point B, a = 6370 km is the
radius of earth.

Here r 0 , θ 0 , ϕ 0 are the unit vectors, and H r , H θ , H ϕ are the proper vector

components. Based on (A.6.3) equation (A.6.1) may be rewritten separately for the field
components as follows:

 ∂E θ ∂ 
− iωµ 0 H r =
1
r sin θ
 − ( E ϕ sin θ ) , (A.6.4)
 ∂ϕ ∂θ 
1 ∂ 1 ∂E r 
− iωµ 0 H θ = ⋅  ( r E ϕ ) − , (A.6.5)
r ∂ r sin θ ∂ϕ 

1  ∂E r ∂ 
− iωµ 0 H ϕ = ⋅ −
r  ∂θ ∂ r
( rE θ ) . (A.6.6)

Similarly the following system of equations may be written based on (A.6.2):

 ∂H θ ∂ 
iωε 0 ε E r =
1
r sin θ
 − ( H ϕ sin θ ) , (A.6.7)
 ∂ϕ ∂θ 

1 ∂ 1 ∂H r 
iωε 0 ε Eθ = ⋅  ( r H ϕ ) − , (A.6.8)
r ∂ r sin θ ∂ϕ 

254
1  ∂H r ∂ 
iωε 0 ε Eϕ = ⋅
r  ∂θ

∂r
( rH θ ) . (A.6.9)

If vertical electric dipole is considered as a source of the electromagnetic wave, then we
may take into account symmetry of the problem (symmetry of the field pattern shown in
Figure 4.14). Hence it may be assumed Eϕ = 0 as well as ∂ / ∂ϕ = 0 , therefore we may

conclude H r = H θ = 0 from (A.6.4) and (A.6.5). Under those assumptions only E r , E θ ,

and H ϕ components exist. Then from the above system of six equations only three

nontrivial equations will remain:


∂ 
E r =
i 1
⋅ ⋅
ωε 0 ε r sin θ ∂θ
( H ϕ sin θ ), (A.6.10)

1 ∂
⋅ ( r H ϕ ) ,
i
E θ = − ⋅ ⋅ (A.6.11)
ωε 0 ε r ∂ r

1  ∂ E ∂ 
⋅ ⋅  r − ( r E θ ) .
i
H ϕ = (A.6.12)
ω µ 0 r  ∂θ ∂r 
For ε = 1 equations (A.6.10) through (A.6.12) specify the electromagnetic field in air
region, and for ε = ε2 specify the electromagnetic filed within the earth's ground.

Now we substitute E r and E θ from (A.6.10) and (A.6.11) into (A.6.12). After some

transforms we may obtain a single equation for H ϕ component.

∂  1 ∂  ∂2

∂θ  sin θ ∂θ
(H ϕ sin θ ) + r 2 ( rH ϕ ) + k 0 r 2 H ϕ = 0 .
 2
(A.6.13)
 ∂r
This is a general form of the wave equation for the electric dipole in spherical coordinates.
Similarly if we assume that to excite a horizontally polarized radio wave a magnetic dipole
is radially directed above the earth's surface. Then H r becomes predominant component
of the magnetic field, and from the symmetry of the pattern the following field components
are vanishing, i.e. H ϕ = E r = E ϕ = 0 . From (A.6.4) through (A.6.9) the following system of

equations may be derived, similar to those of (A.6.10) – (A.6.12).


∂ 
H r =
i
ωµ 0

1

r sin θ ∂θ
( E ϕ sin θ ), (A.6.14)

i 1 ∂
H θ = − ⋅ ⋅ ⋅ ( r E ϕ ) , (A.6.15)
ωµ 0 r ∂ r

255
1  ∂ H r ∂ 
− ( r H θ ) .
i
E ϕ = ⋅ ⋅ (A.6.16)
ω ε 0 ε r  ∂θ ∂r 
Combination of these three equations will result in

∂  1 ∂  ∂2

∂θ  sin θ ∂θ
(Eϕ sin θ ) + r 2 ( rE ϕ ) + k 0 r 2 E ϕ = 0 ,
 2
(A.6.17)
 ∂r

which is a wave equation for E ϕ -component for the radial magnetic dipole in spherical

coordinates. Equations (A.6.13) and (A.6.17) are identical, which follows form the principle
of reciprocity in electromagnetics. Hence we may expect the same form for the general
solutions for both. It'll be shown later that the difference occurs when the boundary
conditions are applied.
For now we'll start with (A.6.13), trying to find the solution in specific form of spherical
wave as follows:
exp(i k 0 R)  exp(i k 0 aeθ ) 
H ϕ (r ,θ ) = C F ( r ,θ ) = C F ( r ,θ ) . (A.6.18)
R aeθ
Here ae is the equivalent radius of the earth, which is to replace the average geometrical

radius of earth, a = 6370 km in presence of the “standard” troposphere as shown in section


5.3 of the Chapter 5. The reader may verify from that section that ae = (4 / 3)a ≈ 8500 km.

In (A.6.18) R = aeθ is the horizontal distance from radiation to reception point, θ is the

geo-central angle between radiation and reception points, k 0 = 2π / λ 0 is the free space

wave number, F (r ,θ ) is the complex propagation factor 1. Now we can create a general
expression for the attenuation function by substitution of (A.6.18) into (A.6.13), and
performing some mathematical transforms. First term in (A.6.13) may be simplified by
assuming sin θ ≈ θ because the propagation distances are much smaller that the earth's
radius, i.e.

∂ 1 ∂ 1 ∂H ϕ ∂ H ϕ
2


∂θ θ ∂θ
(θ ⋅ H ϕ ) = − 2 H ϕ + θ ∂θ + 2 .
  1 
(A.6.19)
 θ ∂θ
If (A.6.18) is substituted into (A.6.19), then the first term will become

1
Do not confuse radial distance r as one of the spherical coordinates with the distance from
antenna to observation point, which was in use before, in Chapter 4: the same notation is used in
both.

256
1  b 1 ∂ F ∂ F ∂ 2 F 
C exp (bθ ) b 2 F − F − + 2b + , (A.6.20)
θ  θ θ ∂θ ∂θ ∂θ 2 

where b = i k 0 ae . If similar substitutions are performed with second and third terms of

(A.6.13), then after combining with (A.6.20), and discarding the term C (1 / θ ) exp (bθ ) we
may finally end up with the differential equation for attenuation function as follows:

∂ 2 F  1  ∂F k a
+  2ik 0 ae −  − i 0 e F + ...
∂θ 2
 θ  ∂θ θ
∂F
( )
2 
2 ∂ F
... + 2 r +r + k 02 r 2 − ae F = 0 .
2
(A.6.21)
∂r ∂r 2

While development of the expression (A.6.21) the only assumption was sin θ ≈ θ . Now for
further simplifications we'll introduce following parameters:
• Large parameter
1/ 3
k a 
1/ 3
 π ae 
M = 0 e = 
 2   λ0  . (A.6.22)
 
As may be demonstrated by simple calculations, M is greater than 10 in common RF
frequency ranges.
• Numerical distance of propagation along the arc of the great circle
R
X = M ⋅θ = . (A.6.23)
Lr
If we assume θ = 1 radian, then by definition R = ae , therefore, from (A.6.23)
1/ 3
ae  λ ae 
2
Lr = =  . (A.6.24)
M  π 

Here Lr is called a distance scale.

• Antenna(s) numerical elevation(s) above the earth’s ground are


h h
y= = , (A.6.25)
H M / k0
1/ 3
M 1  ae λ 2 
where H= =   (A.6.26)
k 0 2  π 2 
is the elevation scale.

257
Now we convert independent variables r and θ in equation (A.6.21) x and y , i.e. instead

of considering F (r ,θ ) as a function of radial coordinate r , and elevation angle θ , we'll

replace it by F ( X , y ) as a function of the numerical distance and height. To do this


conversion the following supportive relations may be written based on (A.6.22) – (A.6.26):
• Radial coordinate of the observation point B is
r =a+h (A.6.27)
then y = h / H = (r − a ) / H , therefore

∂F ∂F ∂y 1 ∂F


= = , (A.6.28)
∂r ∂ y ∂r H ∂ y

∂ 2 F 1 ∂ 2 F
and = . (A.6.29)
∂ r2 H 2 ∂ y2
• The angular derivative of the attenuation function may be found as

∂F ∂F ∂X ∂F


= =M (A.6.27)
∂θ ∂X ∂θ ∂X
where ∂X / ∂θ = M is defined from (A.6.23).

Then the first term in equation (A.6.21) becomes

∂ 2 F 2 
2 ∂ F
= M (A.6.28)
∂ 2θ ∂2 X
The second term is

 1  ∂F  2π ae M  ∂F  1  ∂F


 2ik 0 ae −  = 2i − M = 4M 4  i −  . (A.6.29)
 θ  ∂θ   λ0 
X  ∂X  4M X
2
 ∂X
The third term is
k 0 ae  2π ae M 
−i F = −i F = −i 2 M 4 F . (A.6.30)
θ λ X
The fourth term may be defined if ae / H = 2 M 2 is substituted, i.e.

∂F 1 ∂F a  y  ∂F


2r = 2 (H y + a e ) = 2 e 1 +  = ...
∂r H ∂y H  ae / H  ∂y

 y  ∂F
… = 4 M 2 1 +  . (A.6.31)
 2M 2  ∂y
The fifth term may be found similarly, i.e.

258
2
∂ 2 F 2 
 ae    ∂ 2 F
2
2 1 ∂ F
r2
(
= H y + ae ) =   1 +
y
 2 = ...
∂r 2
H ∂y
2 2
 H   ae / H  ∂ y

y  ∂ 2 F
2

= 4 M 1 +
4
2 
. (A.6.32)
 2M  ∂ y
2

Finally the last term is found as

( ) 4π 2 4π 2
k 02 r 2 − ae F = 2 (r − ae )(r + ae )F = 2 (H y )(H y + ae )F = ...
2

λ λ
 π 2a2 1   y  4  y 
... = 8 2  y 1 +  = 4M y 1 + F. (A.6.33)
 λ (ae / H )   2(ae / H )   4M 
2

Now if we substitute all terms from (A.28) – (A.6.33) into (A.6.21) and divide both sides by
4M 4 , then the following differential equation in partial derivatives may be written for the
propagation function:

1 ∂ 2 F  1  ∂F i 
+ i −  − F + ...
4M ∂ X  4M X  ∂X 2 X
2 2 2

 y  ∂ 2 F 1  y  ∂F  y 
... + 1 +  +  1 + 2 
+ y 1 + 2 
F = 0. (A.6.34)
 2M  ∂ y
2 2
M  2M  ∂ y
2
 4M 
The next simplifications may be applied under the following assumptions: y /( 2 M 2 ) << 1

for all practical cases of the "near-the-ground" corresponding points, and 1 /( 4 M 2 X ) << 1 .
Then (A.6.34) may be rewritten in following approximate from:

∂F i  ∂ 2 F
i − F+ + y F = 0 . (A.6.35)
∂X 2 X ∂ y2
Now we can apply a method of separation of the variables, which means that the unknown
function F ( X , y ) is assumed to be presented as a product of two functions of single
variable each, i.e.
F ( X , y ) = FX ( X ) F y ( y ) . (A.6.36)

Now we substitute (A.6.36) into (A.6.35), then divide by product F X F y , and separate

variables. After separation, we end up with the equation with the left side being dependent
only on one of the variables, whereas the right side is dependent on the other variable
only. It appears to be true if, and only if both sides of that equation are equal to so called
separation constant z , and generally is expected to be an arbitrary complex number, i.e.

259
i dFX 1 d F y
2
i
− + = + y = z . (A.6.37)
FX dX 2 X F y d y 2
Let's rewrite (A.6.37) as two separate equations for each independent variable, X and
y as follows:

dFX  1 
−  i z +  FX = 0 , (A.6.38)
dX  2X 
d 2 F y
+ ( y − z ) F y = 0 . (A.6.39)
d y2

If (A.6.38) is modified and represented as dFX / FX = [ i z + 1 /( 2 X )] dX , then, after


integration of both sides, the solution may be obtained in following form:

log FX = i z X + log X + const . (A.6.38a)

The integration constant is assumed to be equal to zero: as one may notice from below,
this assumption does not restrict generality of the considered approach. Then from
(A.6.38a) we may define

FX ( X ) = X exp ( i z X ) . (A.6.40)

The second equation, (A.6.39) is known as Airy differential equation that has a general
solution in form of combination of Airy functions of the first, Ai ( z − y ) and the second,

Bi ( z − y ) kinds, i.e.

F y ( y ) = C [ Ai ( z − y ) − i B ( z − y )] . (A.6.41)

Finally we may obtain the general solution of (A.6.35) by substitution of (A.6.40) and
(A.6.41) into (A.6.36), i.e.

F ( X , y ) = C X exp ( i z X ) w ( z − y ) , (A.6.42)

where we denoted C as the integration constant, and


w( z − y ) = Ai ( z − y ) − i B ( z − y ) . (A.6.43)

A.6.1.b. PARTIAL SOLUTIONS OF THE WAVE EQUATION FOR


HORIZONTALLY AND VERTICALLY POLARIZED RADIO WAVES
As known from the theory of differential equations, the particular solution of the system of
differential equations may be obtained from the general solution, by using the initial
(boundary) conditions, i.e. one have to "stitch" the fields on the boundary of the mediums.

260
For the components of the field radiated by the vertical electric dipole shown in Figure
A.6.1 in spherical coordinates the boundary condition (4.111) may be rewritten here as
W
E θ = 0 H ϕ . (A.6.44)
ε2

On the other hand the relation between H ϕ and E θ components in the area above the

Earth’s ground is defined by (A.6.11), which may be modified and rewritten for that spatial
area ( ε = 1 ) in the following form:

i 1  ∂H ϕ 
E θ = −  H ϕ + r . (A.6.45)
ωε 0 r  ∂r 
 
Combination of (A.6.44) and (A.6.45) at the Earth’s ground, r = ae results in

W0  i i ∂H ϕ
Hϕ = − H ϕ − , (A.6.46)
ε2 ω ε 0 ae ωε 0 ∂ r
which we rewrite as

  ∂H ϕ
 1 − i k 0 ae  H ϕ + ae = 0. (A.6.47)
 ε2  ∂ r
 
Now we will acquire H ϕ by combining (A.6.18) and (A.6.42). One may realize that the r-

dependence is represented only by the last term of (A.6.42). Therefore, after combining
(A.6.18) and (A.6.42), substitution them into (A.6.47), and after simple transforms, it may
be written as

 w ( z − y ) + ae ∂ w ( z − y )
 
 1 − i k 0 ae = 0. (A.6.48)
 ε2  ∂r
  y =0

The first assumption here, y = 0 verifies that field behavior is considered near the earth’s
ground. The second assumption is
k 0 ae >> 1 , (A.6.49)

which is true for all RF frequency ranges, and all types of earth’s ground.
The second term in (A.6.48) may be evaluated if y is recalled from (A.6.25), i.e.

h h k
y= = = 0 (r − a) . (A.6.50)
H M / k0 M
Then

261
d w ( z − y ) d w ( z − y ) dy  k  d w (z )
= = − 0  . (A.6.51)
dr y =0
d ( z − y ) dr y =0  M  d z
Taking into account (A.6.49), and (A.6.51) the equation (A.6.48) may finally be written as
dw ( z )
− q vert w ( z ) = 0 , (A.6.52)
dz
where

M π ae 1
q vert = −i = −i 3 (A.6.53)
ε2 λ0 ε 2 − i 60λ 0σ
is parameter introduced by V.A. Fock [2] that takes into account electromagnetic properties
of the earth’s ground.
Evaluations similar to those provided above for the vertically polarized wave can be
performed for the diffraction of the horizontally polarized wave radiated by the vertical
magnetic dipole, i.e. for the horizontal loop current shown in Figure A.6.1b. The step-by-
step procedure is as follows:
• Boundary condition for the field components that are tangential to the earth’s
ground in its vicinity are

ε2 
H θ = Eϕ . (A.6.54)
W0

• This expression for H θ is taken equal to the right hand side of (A.6.15). After

algebraic transforms the following equation may be written for the field component
E ϕ in close vicinity of the earth’s surface:

∂E
(1 − i k a
0 e )
ε2 E ϕ + ae ϕ = 0 .
∂r
(A.6.55)

• If the general solution (A.6.42) along with


exp(i k 0 aeθ ) 
E ϕ (r ,θ ) = C F ( r ,θ ) (A.6.56)
aeθ
is substituted into (A.6.55) then it may easily be transformed into

∂ w ( z − y )
(1 − i k a
0 e )
ε2 w( z − y ) + ae
∂r
=0 (A.6.57)
y =0

• Taking into account the assumption (A.6.49), as well as the expression (A.6.51) an
equation similar to (A.6.57) may finally be written as

262
dw ( z )
− q horiz w ( z ) = 0 (A.6.58)
d z
where

π ae
q horiz = −i M ε2 = −i 3 ε 2 − i 60λ 0σ . (A.6.59)
λ0

Equations (A.6.52) and (A.6.58) are identical except of the polarization constants q vert and

q horiz , so the solutions of both are expected to show up as an infinite set of eigenvalues

(roots) of the variable z n . Apparently those two sets are different for vertical and

polarizations. They are given in Table 4.4 for two extreme values of either q horiz or q vert .

For the intermediate values of q parameter the approximate expressions (4.158) and
(4.159) may be applied.
For the given single eigenvalue (root), z n one of the particular solutions of the equation

(A.6.35) may be written based on the general expression (A.6.42) as

F n ( X , y ) = C n X exp ( i z n X ) w ( z n − y ) . (A.6.60)

There’s an infinite number of those solutions, thus the final expression for the propagation
function for numerical (normalized) distance X and numerical (normalized) height y = y 2
of the reception point B, may be presented as a linear combination of (A.6.60)
components, i.e.

Fn ( X , y ) = X ∑C
n =1
n exp ( i z n X ) w ( z n − y 2 ) . (A.6.61)

Evaluation of the integration constants C n is the most cumbersome part of the diffraction

theory and is outside of the scope of this text. The reader can find them in [2]. We’ll just
mention here, that those integration constants are developed by “matching” solutions
obtained for the small distances (illuminated zone) either from interference formulas, if
antennas are elevated, or from Weyl and Van-der-Pol’s approach, if antennas are
considered as ground-based. After that “matching” technique is applied, the values of
integration constant are found

2 π w (z n − y1 ) exp ( iπ / 4 )
Cn =
( )
, (A.6.62)
z n − q 2 w 2 ( z n )

263
where y1 is the numerical (normalized) height of the transmitting antenna. Final
expression for the complex attenuation function may be defined by substitution of (A.6.62)
into (A.6.61).

 π  ∞ exp (i X z n ) w (z n − y1 ) w (z n − y 2 )


F ( X , y1 , y 2 ) = 2 π X exp  i  ∑ . (A.6.65)
 4  n =1 z n − q
2
w (z n ) w (z n )
REFERENCES
1. Abramowitz, M, Stegun, I. A. Handbook of Mathematical Functions with Formulas,
Graphs, and Mathematical Tables, New York: Dover Publications, 9-th edition,
1970
2. Fock, V.A. Electromagnetic Diffraction and Propagation Problems. Oxford, NY,
Pergamon Press, 1965

264
Chapter 5. ATMOSPHERIC EFFECTS IN RADIO
WAVES PROPAGATION

5.1. DIELECTRIC PERMITTIVITY AND CONDUCTIVITY OF


THE IONIZED GAS

Earth’s atmosphere is a complex gaseous media with electromagnetic parameters that


expose regular and random variations in space and time. Large-scale regular spatial variations,
as well as small-scale random fluctuations are specific for troposphere and ionosphere, - the
two atmospheric regions that are of significant impact on propagation of RF different frequency
bands. Ionosphere, as the most complex and diverse part of the atmosphere, is nothing but
sparse, gaseous plasma where concentration of free electrons controls the electromagnetic
properties of that medium, such as dielectric permittivity and conductivity. As noted in
introductory chapter, the regular changes of dielectric permittivity are causing reflections and
refractions, whereas the random, small-scale fluctuations are causing the scatterings of the
radio waves. The finite value of conductivity results in absorptions of energy of the radio waves
while propagation. We’ll start our study with the simple case of the impact of free electrons’
cloud on electromagnetic properties of propagation medium.
Consider a free electron in free space in the absence of external forces and fields. It has only a
kinetic energy, which remains constant, resulting in constant speed and direction of the
movement. In real ionospheric plasma, in presence of a numbers of other electrons, neutral
atoms, charged particles (ions) as well as the propagating radio wave, the uniform movement of
the electron will be disturbed. In particular, if the free electron is placed into the sinusoidal
electric field of the propagating radio wave then it will make sinusoidal oscillations, and may be
considered as a harmonic oscillator (see Figure 5.1).

265
Figure 5.1. Oscillating movement of free electron imbedded into the electric field of
propagating radio wave (T is a period of the radio wave oscillations)

Each electron under the forced movement may be considered as an independent oscillator that
generates its own radiated field of the same frequency, polarization and direction of propagation
as those of the initial radio wave. Being originated by the large number of free electrons in
plasma, this secondary radiation overlay with the primary radio wave and enforces its
propagation. In other words the amount of energy spent by the radio wave to excite the
electron-oscillator is returning back to the radio wave as an energy of the secondary radiation:
the plasma's electron-cloud doesn't consume the energy from the radio wave, so thus no
attenuation of the radio wave is taking place if the collisions with the neutral atoms or molecules
are eliminated. Collisions are the cause of losses of energy of the radio wave while propagation
in the plasma medium: they result in a decrease of the amplitude of the radio wave. The
physical mechanism of those losses is as follows: part of energy of the radio wave transforms
primarily into the energy of free electrons; then free electrons pass that energy to the neutral
atoms/molecules while colliding with them, which results in increase of kinetic energy of those
neutral particles; the extra kinetic energy of the neutral particles due to collisions with free
electrons is not reversible back to the energy of the primary radio wave because the neutral
particles are unable to generate their own secondary radiation, therefore that extra kinetic
energy appears as an energy of thermal losses.
Now, after above qualitative consideration, we’ll make some quantitative evaluations.

266
Suppose a linearly polarized radio wave propagates in a plasma medium along z-direction with
E -vector, directed along x-axes. A force that is applied by electric field of the radio wave onto
electron, e E is compensated by two other forces:

d 2x
• inertial force, Finertial = me , (5.1)
d t2

where me = 9.1 ⋅10 −31 kg is a mass of the electron, and

• F friction friction force between electron and surrounding neutral atoms/molecules.


Recall from the Physics course, that the impulse of the force is equal to the momentum of the
movement, i.e.
dx
F friction ⋅ t = me , (5.2)
dt
where t is a time period, which may be chosen as a time interval between two subsequent
collisions, so that t = 1 / ξ , where ξ is a number of collisions within a unit time interval (one
second). Thus (5.2) may be rewritten as
dx
F friction = ξ me . (5.2a)
dt
Then the balance of forces may be presented as

d 2x dx
e E = Finertial + F friction = me 2
+ ξ me , (5.3)
dt dt

where e = 1.6 ⋅10 −19 C is electron's charge. Here we assume all three vector-forces being
collinear, thus the vector sum is replaced by algebraic.
The second order linear differential equation (5.3) may be simplified, if recall that both, electric
field E and x -coordinate of the electron are harmonic functions, and may be represented in
complex form as follows:

E (t ) = E m exp (iω t ) , x (t ) = xm exp [i (ω t − ϕ ) ] , 



d x
= i ω x ,
d 2 x
= − ω 2 x .  (5.4)
d t2 
dt 
After expressions (5.4) are substituted into (5.3) it's easy to define velocity of movement of the
electron as
d x e E
v = = . (5.5)
d t me (ξ + i ω )

267
It’s easy to realize, that the magnitude of volumetric current density caused by the movement of
the electrons cloud may be found as a product
J e = e N e v , (5.6)

where N e is a concentration of the free electrons, i.e. the number of free electrons in unit

volume (e.g. in a cubic centimeter). In other words J e is found as a total charge of free-

electrons, within the prism of the length v and cross-sectional surface S = 1 sq. m, which
crosses that surface in one second (see Figure 5.2).

Figure 5.2. The definition of current density due to electrons-cloud movement

If we consider ionospheric plasma as a lossy dielectric, then by recalling (2.43), we may assume

that the total volumetric current density J tot is being composed of two components: one is

iω ε 0 E , the displacement current specific for the free space (vacuum) 1, and the second is the

current caused by the electrons-cloud motion, J e . Hence (2.43) may be rewritten in scalar form

as
Jtot = iω ε 0 E + J e . (5.7)

Now we substitute (5.5) and (5.6) into (5.7). After simple algebraic transforms we’ll obtain

  e2 Ne 
J tot = iω ε 0 1 + E . (5.7a)
 iωε 0 me (ξ + iω )
Regarding the procedure given in subsection 2.1.3 the complex dielectric permittivity of plasma
may be found as an expression in brackets (see (2.43a)) in the following form:

1
Note that there’re no other types of displacement currents, such as displacement current caused by the
polarization effects in conventional dielectric materials.

268
N ee2 N ee2 N e e 2ξ
ε = 1 + = 1− −i
i ω ε 0 me (ξ + i ω ) (
ε 0 me ω 2 + ξ 2 ) (
ω ε 0 me ω 2 + ξ 2
.
) (5.8)

If (5.8) is compared with (2.45) for the general definition of the complex permittivity, then real
part of dielectric permittivity as well as the conductivity may be separated from (5.8). If me and

e are substituted form above along with ε 0 from (2.6), then real part of dielectric constant and
conductivity of plasma may be extracted from (5.8) in following forms:
Ne e2 N N
ε = 1− = 1 − 3182 electrons / m3
= 1 − 3182 electrons / cm3
10 6 ,
(
ε 0 me ω + ξ
2 2
) ω +ξ
2 2
ω +ξ 2
2
(5.9a)

Ne e2 ξ N electrons / m3 ξ −8 N electrons / cm3 ξ −2


σ= = =
(
me ω 2 + ξ 2
2 .
)8
ω2 +ξ 2
10 2 . 8
ω2 +ξ 2
10 , S/m, (5.9b)

where N e = N electrons / m3 = N electrons / cm3 ⋅10 6 is used. Note that the quantity N electrons / cm3 is more

commonly used in geophysics and engineering applications.


As reader may see, compared to the other type of dielectric media such as solids, liquids, gases
that always have a permittivity greater than unity, the unique property of the plasma medium is
that dielectric constant is less than unity. The above expressions are evaluated for the plasma
without the impact of the external permanent magnetic field 1. Table 5.1 represents some
quantitative values of N e and ξ , as well as heights above the earth’s surface that are typical for

the real ionospheric layers. As one may realize from the data provided in table, for the HF and
partially for MF frequency bands the collision rate is much less than the radian frequency of the
radio wave for all layers, i.e. ξ << ω , except of D-layer, which disappears in the night-time and
never disturbs the shape of the ray paths; it causes absorptions only in the day-time.
Table 5.1
Ne, electrons ξ, number of
Ionospheric
per cubic collisions per h, height in km Notes
Layer
centimeter second
Disappears at
D 102 to 103 10 7 60 to 90
night

1
In presence of the external magnetic field plasma is called “magneto-active plasma”. It exhibits more
complex properties than non-magneto active plasma that is considered here. We’ll discuss some of the
properties of the ionospheric plasma under the impact of the earth’s magnetic field later in section 5.5.

269
(1 to 4.5) ⋅10 5 Day
5
E 10 100 to 140
5 ⋅10 3 to 10 4 Night

Disappears at
F1 (2 to 4.5) ⋅10 5 10 4 180 to 240
night
Maximum at day
2 ⋅10 6
230 to 400 time
F2 103 to 104
300 (standard) Maximum at
3⋅10 5
night time

Note: in the night-time the F1- and F2-layers converge into a single F-layer as one may notice
from the Figure 1.2.

If the relation ξ << ω is taken into account, then (5.9a) may be rewritten as

N electrons / cm3 e 2 N electrons / cm3


ε ≈ 1− = 1 − 3182 10 6 , (5.9c)
ε 0 me ω 2
ω 2

for e = 1.6 ⋅10 −19 C, me = 9.1 ⋅10 −31 kg, ε 0 = (1 / 36π ) ⋅10 − 9 F/m. Hence, the ionospheric layers

with the dielectric constant represented by (5.9c) may be considered as low-lossy dielectric
media with strong frequency dependence of electromagnetic properties, especially near the so
called “cutoff” frequency. That is a frequency, also known as plasma frequency or Langmuir
frequency, which turns dielectric constant down to zero. Those media demonstrating frequency
dependence of the propagation properties are called dispersive media. From (5.9c) the circular
plasma frequency may be defined by solving ε = 0 equation. The result is:

ω c 2 ≈ 3182 N elctrons / cm ⋅10 6 .3 (5.10)

This expression may easily be transformed to give the plasma frequency in kHz, which in most
cases is simply called plasma frequency:

f c , kHz ≈ 80.8 N electrons / cm3 . (5.10a)

If we plugin (5.10) and (5.10a) into ((5.9c) then the expression for dielectric constant may be
written in a simple form:
2
80.8 N electrons / cm3  f c , kHz 
ε ≈1− =1−   . (5.11)
2  f kHz 
f kHz  

270
Some confusion may occur if we recall (2.67) for the phase velocity of the radio wave, which is
rewritten here as
ω c
vp = = . (5.12)
β ε
For ε < 1 the phase velocity of the radio wave traveling in ionospheric plasma becomes greater
that a speed of light in free space that seems to contradict the physical concepts. To resolve
that confusion just recall that phase velocity is not a velocity of the energy transfer: the velocity
of the energy transfer is given by [1, p.98]
1
vg = =c ε , (5.13)
dβ dω
thus the energy always “propagates” with the velocity less than a speed of light in free space.
Being highly dispersive, especially if the carrier frequency of the radio wave is close to cutoff,
ionospheric plasma negatively affects propagation of the wideband signal causing distortions of
the waveform. Indeed, different frequency components of a wideband signal will travel to the
reception point with different velocities hence will arrive with different delays. Being combined at
the reception point they will result in the waveform different from the original one.
It may be seen from (5.11), that for the frequencies f < f c the propagation phase velocity

v p = c / ε becomes imaginary number, which means that the cut-off condition takes place, i.e.
the radio wave cannot propagate. From Figure 1.2 it may also be seen, that the ionospheric
plasma concentration N e (h ) tends to increase for the heights up to 300 km, thus, regarding

(5.11), for those heights, the elevation dependence of the dielectric constant ε ( h ) is a decaying
function.

5.2. REGULAR REFRACTION OF THE RADIO WAVES IN THE


ATMOSPHERE

In both atmospheric regions, troposphere and ionosphere the averaged dielectric permittivity is
a decaying function of the height: in the troposphere it decreases slowly to unity, whereas in the
ionosphere it decreases slowly and becomes even less than a unity, as it was shown above.

271
Apparently the similar behavior takes place for the refraction index n = ε , which results in
curvilinear propagation paths of the radio waves.
For further analysis, in order to define the shape of ray path of the radio wave, we’ll replace
smooth, gradual change of the refraction index in vertical direction by its tiny jumps, i.e. replace
hypothetically the real atmospheric air by planar 1 stratified medium arranged of the layers of
infinitesimal thickness. In other words each layer of the height dh has a constant value of
refraction index as shown in Figure 5.3, and the jump of refraction index dn on the border from
layer to layer.

Figure 5.3. Radio wave refraction pattern in planar stratified model of the real atmosphere

Then the Snell's law (2.109a) may be applied for each single border between two adjacent
layers. At the point A, for example, it may be written as
n0 sin ϕ = (n0 + d n )sin (ϕ + dϕ ) . (5.14)

To define the radius of curvature A from Figure 5.3 first we find AB-distance from the triangle
ABE as
dh dh
AB = ≈ . (5.15)
cos (ϕ + dϕ ) cos ϕ
From triangle AOB an expression for radius RR may be defined if AB is used from (5.15), i.e.

1
Planar stratification is acceptable for most of applications. However some corrections due to convexity

of the atmospheric layers may be found in [1, pp. 121-126].

272
AB dh
RR = = . (5.16)
dϕ cos ϕ dϕ
Now cos ϕ dϕ may be found from (5.14) as follows:

n0 sin ϕ = n0 sin ϕ cos dϕ + n0 cos ϕ sin dϕ +


+ d n sin ϕ cos dϕ + dn cos ϕ sin dϕ . (5.17)

Here, we take into account dn << 1 , and dϕ << 1 , then by assuming sin dϕ ≈ dϕ , cos dϕ ≈ 1 ,

and dn cos ϕ sin dϕ ≈ 0 , we may rewrite (5.17) as

d n sin ϕ
cos ϕ dϕ = − , (5.18)
n0
and substitute into (5.16).
1
RR = − . (5.19)
dn
sin ϕ
dh
We also assumed n0 ≈ 1 near and below the border between troposphere and ionosphere.

In (5.19) ϕ is the initial angle of incidence, and dn / dh is the vertical gradient of refractive index
in the proper region of the atmosphere (either in troposphere or ionosphere).
Particular applications of the expression (5.19) are considered below.

5.3. STANDARD ATMPSPHERE AND TROPOSPHERIC


REFRACTION
The numerical value of the tropospheric air refraction index is fairly close to unity. Since for the
troposphere it usually doesn't exceed unity by more than about 0.05 to 0.1 percent, so called
refractivity has been introduced for convenience as follows :
N = (n − 1) ⋅ 10 6 . (5.20)
Numerical values of the refractivity (not to be confused with the electrons concentration in
plasma N e ) are more convenient for further mathematical evaluations and engineering

assessments. If for instance n = 1.00006, then N = 60, which is much more convenient to have
deal with.
As mentioned above the refraction index, as well as the refractivity of the troposphere is in direct
dependence on air temperature, pressure and humidity. They vary regularly and randomly in

273
time and space. In order to support the communication efforts between researchers and
engineers worldwide, ITU introduced a standard (or reference) atmosphere (meaning its
tropospheric region only) that is specified by the refractivity near the earth's surface equal
N 0 = 324 (n = 1.000324). This value is a result obtained statistically from the enormous number
of observations for many years and for all geographic regions around the globe. For the
standard atmosphere the regular (deterministic) variations of N are introduced as a linear
function of elevation N (h) shown in Figure 5.4 as a dashed line. The value of standard
refractivity decays linearly to zero up to the heights of 8km (which is considered as a ceiling for
the standard atmosphere), with the vertical gradient:
dN
grad N = − = − 40 , km-1. (5.21)
dh

Figure 5.4. Elevation dependence (vertical profile) of refractivity, N in troposphere

As one may notice from Figure 5.4 the observed values of N-index are randomly fluctuating.
The nature of those random fluctuations is the horizontal and vertical movements of the air
masses, which result in turbulences (globules) of the air with random distribution of their sizes
and positions in space and time. Being transparent for the visible light, the globule has slightly
different value of refractivity, compared to the surrounding area, due to fluctuations of the
temperatures, pressures and humidity from the inside to outside of the globule.
In order to count the effect of troposphere on the ground radio waves propagation we introduce
now the earth’s equivalent radius. First we define the term curvature of the path as a reciprocal
of the radius of curvature at the considering point on the curved line. Particularly the curvature
of the earth's convex surface is defined for any great circle as 1 / a , where a = 6370 km is the

274
radius of earth. Similarly for the ray path with the radius of curvature RR the absolute curvature
is defined as 1 / RR . Along with absolute curvature another term, relative curvature is used in
radio wave propagation theory. An example of relative curvature is the curvature of the ray
relative to the curvature of the earth's convex surface, which is defined as a difference
1 / a − 1 / RR (see Figure 5.5a).

Figure 5.5. Illustration of the definition of the Earth's equivalent radius in presence of the
regular tropospheric refraction: a). Shape of the propagation path above the earth’s
surface in real conditions, b). Straight propagation path above the equivalent earth’s
surface

Now we may account for the tropospheric regular refraction by replacing the earth’s real radius
with the equivalent one, ae the way, which results in straight ray path RR , e = ∞ as shown in

Figure 5.5b. Here 1 / RR , e = 0 is the “curvature” of the straight ray path. “Equivalent “

replacement means the hypothetical replacement: (1) of the earth’s real radius by equivalent
one, (2) replacement of the curved ray path by the straight line, (3) keeping the relative
curvature between ray path and the earth’s surface the same as in real model, (4) consider no
stratification of the atmospheric air in equivalent model. This is demonstrated on Figure 5.5.
Then, based on the above statements, the following expression may be written:
1 1 1 1 1
− = − = . (5.22)
a RR ae RR , e a e

From (5.22) the equivalent earth’s radius, ae may be found as

a
ae = . (5.23)
1 − a / RR
To make (5.23) practically usable, first define the vertical gradient of the refractive index of
troposphere from (5.20) as

275
dn d N
= ⋅10 −6 , 1/m (5.24)
dh dh
and substitute in (5.19):

10 6
RR = − ,m. (5.25)
dN
dh

While transforming (5.19) to (5.25) we’ve assumed ϕ ≈ 90 0 for the angle of incidence. That is
true for the terrestrial radio paths and most of the ground-based radars. For any given value of
the vertical gradient of refractivity dN / dh the radius of curvature of the ray path may easily be
found from (5.25), which allows taking into account the effect of regular refraction in troposphere
by inserting into (5.23) for earth’s equivalent radius.
If the particular value of vertical gradient of refractivity dN / dh is not provided, then the radio
link design is usually being performed for the standard atmosphere (troposphere), by
substituting dN / dh = −0.04 m-1 into (5.25) and (5.23) respectively. Then the result is
4
RR = 25,000 km, ae = a = 8,500 km. (5.26)
3
The conclusion made of the above considerations is as follows: in all analysis and calculations,
where the atmospheric effects were ignored (see Chapter 4), the regular refraction in
troposphere may be counted by simply replacing the real radius of the earth by its equivalent
value form (5.26). After the replacement, the ray trajectories may then be considered as straight
lines, as we would have it in case of absence of troposphere.
The positive role of smooth atmospheric refraction may be realized, for instance, from the
expressions (4.42) and (4.43), for propagation within LOS distance. If a is replaced by ae for

refraction to be counted, then the apparent antenna heights will be increased, thus resulting in
increase of the field strength in (4.43), compared to the case of the absence of refraction. It may
also be seen from the Figure 5.6 for the VHF/Microwave LOS terrestrial link, where an extra gap
∆ H arises between obstacle and ray trace because of curvature of the ray in presence of the
refraction. This extra gap will certainly increase the field strength at the receiving point.
The path profile on Figure 5.6a will transform to one shown in Figure 5.6b if in expression for the
variable y defined by (4.68) for no-refraction condition

y = y1 =
x
(r − x ) (5.27a)
2a
the real earth's radius is replaced by the equivalent one, i.e.

276
y = y2 =
x
(r − x ) . (5.27b)
2ae

Figure 5.6. a). real path profile of the LOS link in presence of tropospheric refraction, b).
path profile of the LOS link after equivalent replacement of the earth’s radius

Then the extra gap, ∆ H due to tropospheric refraction, may be found as a difference

r2
∆H = y1 − y 2 = k (1 − k ) , (5.28)
2 RR
where (5.22) has been used. In above formula k = x / r denotes the relative coordinate of the
obstacle. If RR is substituted from (5.25), then (5.28) may be rewritten for the magnitude of
∆H as
r2
k (1 − k )
dN
∆H = ⋅10 −6 m (5.29)
2 dh
Thus the total path clearance H (N ) between ray and the obstacle may be found by adding an

extra gap ∆H to H (0 ) defined by (4.74) in absence of the refraction:

H ( N ) = H (0) + ∆H ≈ 1.1 ⋅ H opt (0) . (5.30)

In order to satisfy (5.30) for the given H opt (0 ) the value H (0 ) becomes less, hence the antenna

heights may be taken less to get the same field strength at the receiving point, when compared
to the case of absence of tropospheric refraction.
Another positive role of regular refraction may be noticed for the radio wave intensity in
diffraction zone. Indeed if equivalent radius of the earth is substituted into (4.147), then the
numerical distance (4.149) becomes less, resulting in the higher value of the propagation factor.

277
It may be verified from expression (4.168) for the ground-based antennas and from Figure 4.31
for the elevated antennas.

------------------------------------------------------------------------------------
Example 5.1
Calculate the power received at reception point B for the initial conditions, provided in Example
4.5, taking into account the smooth tropospheric refraction with the vertical gradient of
refractivity d N / d h = -0.03 1/m = -30 1/km.
Solution

• An additional path clearance due to tropospheric refraction is defined from (5.29):

∆H = −
(50 ⋅10 ) 3 2
⋅ 0.4 ⋅ (1 − 0.4 ) ⋅ 0.03 ⋅10 −6 = 9 m.
2
• Total path clearance in presence of the tropospheric refraction is defined from (5.30),
where H (0 ) = 27,5 m is defined in Example 4.5:

H ( N ) = H (0 ) + ∆H = 27,5 + 9 = 36.5 m
• Propagation factor may be found from (4.69) by using H(N) instead of H(0), i.e.

 4 ⋅ 3.14 36.5 2 
F= 1 + 0.25 2 − 2 ⋅ 0.25 ⋅ cos  ⋅  = 1.046
 7.5 ⋅10 ⋅ 5 ⋅10 4 ⋅ 0.4 ⋅ (1 − 0.4 ) 
−2 4

• Then the received power may be defined from (3.38) as

PRx = 0.5 ⋅
(
10 4 ⋅10 4 ⋅ 7.5 ⋅10 −2 )
2

⋅1.046 2 ≈ 0.78 µW = - 61.1dBW (Answer)


(4 ⋅ 3.14 ⋅ 50 ⋅10 )3 2

Comparison of this received power with the proper result from Example 4.5 shows an increase
of the received power by 1.3 dB due to the positive impact of tropospheric refraction.

---------------------------------------------------------------------------------------

Depending on the values of the vertical gradient of refractivity, dN / dh different types of


refraction may occur while tropospheric propagation of the radio waves. Table 5.2 illustrates the
classification of the refraction types. For the given values of gradients of refractivity the relations
for ae and RR are assessed based on (5.23) and (5.25), and presented in right column.

278
Table 5.2

dN
= 0 , a = a e , RR = ∞
dh

dN
− 40 < < 0 , RR > 25000 km
dh
a < ae < 8500 km,

dN 4
= −40 , ae = a ,
dh 3
RR = 25000 km

dN 4
− 157 < < −40 , ae > a ,
dh 3
RR < 25000 km

dN
= −157 , ae = ∞ ,
dh
RR = a

dN
< −157 , ae < 0 ,
dh
RR < a

279
5.4. REFLECTION AND REFRACTION OF THE SKY WAVES IN
IONOSPHERE
As follows from (5.8) dielectric permittivity of the ionospheric plasma depends on the frequency
of the radio wave and concentration of free electrons N e , which increases up to the elevations

of 300 km (see Figure 1.2) causing a continuous decrease of ε , and hence the a decrease of

the refraction index n = ε . In section 5.2 we’ve provided a brief survey of the refraction
pattern applicable to any stratified medium with smoothly changing refraction index; then we’ve
applied it to the refraction in troposphere. In order to apply that survey to the refraction of the
radio wave in ionosphere let’s consider an ideal propagation ray path in ionospheric layer shown
in Figure 5.7.

Figure 5.7. A pattern of the sky-wave reflection from the ionosphere

A portion AP of that ray path is similar to the path shown in Figure 5.3, which we recall for the
further analysis. Now let’s move up gradually from layer-to-layer. The Snell's law for each border
between those tiny layers on Figure 5.3 may be written as:
o At the point A n0 sin ϕ = (n0 + d n ) sin (ϕ + dϕ ) ,

o At the point B (n0 + dn) sin (ϕ + dϕ ) = (n0 + 2d n ) sin (ϕ + 2dϕ ) ,

o At the point C (n0 + 2dn) sin (ϕ + 2dϕ ) = (n0 + 3 d n ) sin (ϕ + 3dϕ )


o …………………………………………………………………..
For the highest point on trajectory (the turning point P on Figure 5.7) the angle of incidence
tends to 900, thus following expression may be written:

n0 sin ϕ = n (h ) = ε (h ) . (5.31)

280
Assume n0 ≈ 1 , which is true on and below the bottom of the ionosphere. Then after

substitution of (5.11) into (5.31) it converts to

N e (h )
sin ϕ = 1 − 80.8 2
. (5.32)
f kHz
For the given angle of incidence this formula allows finding the elevation of the highest, turning
point on ray path if the vertical profile of the plasma concentration N e (h) is known a priory. In

fact, we may see from the Figure 5.7 that the elevation of the point P (vertex of the ray
trajectory) is defined by the numerical value of plasma concentration at the particular height h
to satisfy the equation (5.32).
Now we transform (5.32) to

80.8 N e (h )
cos ϕ = . (5.33)
f kHz
The following statements can be made from (5.33) for reflection from ionospheric layer:
 because the concentration of ionospheric plasma is limited to its maximal value,
( N e (h ) ≤ N e , max ), then for the given frequency f there's a lower limit of the angles of

oblique incidence,

 80.8 N e , max 
ϕ min = cos −1  , (5.34)
 f 

so the rays with angles of incidence less than ϕ min will not reflect, but penetrate trough
the ionosphere into the free space, whereas the rays with the angles of incidence equal
or greater than ϕ min will reflect and return back to the earth (see figure 5.8);

 vertically incident wave ( ϕ = 0 ) will reflect from the Ionosphere only if the frequency is
less, than plasma frequency, defined from (5.10a) as

fc = 80.8 N e , max . (5.35)

So it's easy to realize, that if the frequency of the radio wave is greater that its critical value f c ,

then reflections are limited to angles of incidences defined as ϕ > ϕ min . As a result for those
frequencies so called silence zone will surround the transmitting point A as shown in Figure 5.8,
which is unattainable for the sky-waves, backing from the ionosphere.

281
Figure 5.8. Appearance of the “silence zone” for frequency f > f c for the sky wave

propagation mechanism (flat-earth case).

This area becomes larger, if greater the difference between f and fc is. Reflection will take place
if the angle of incidence is greater than ϕ min and reflection condition may be written as

cos ϕ ≤ cos ϕ min = f c / f , (5.36)

or f ≤ f c sec ϕ , (5.37)

which is known as a secant rule. It allows finding the range of frequencies of the radio waves
that reflect if it is obliquely incident with the angle of ϕ at the lower border of ionosphere.
The frequency range for the vertically incident wave may be estimated based on table 5.1.
Considering a vertical incidence, ϕ = 0 and taking N e , max ≈ 10 6 electrons/cm3 for ionospheric

layer F2, one may define from (5.37)

f kHz ≤ f c , kHz = 80.8 N e , max ≈ 9 MHz. (5.38)

As one can see from (5.37), much wider range of frequencies may be involved in reflection if the
radio wave is obliquely incident ( ϕ ≠ 0, sec ϕ > 1 ). It may seem like the radio wave of any
frequency may satisfy (5.37) if angle of incidence is properly chosen. Larger is the frequency,
closer the value of ϕ may be taken to 90 degrees to let (5.37) be satisfied, i.e. the radio waves
of unlimited frequencies are able to reflect from ionosphere. However some limitation occurs
due to the convexity of earth’s surface, which is considered below.

282
For the practical applications it's more appropriate to express ϕ in right hand side of (5.37) by

horizontal curvilinear distance R between corresponding points shown in Figure 5.9.

Figure 5.9. Geometrical sketch of single-hop sky wave propagation due to reflection from
ionosphere. hmax is the elevation of the maximum plasma concentration (see N e (h)

profile on Figure 5.7)

The following derivations may be performed based on geometrical sketch shown on Figure 5.9.
R
θ= , (5.39)
a
 θ
∆ h = ED = a 1 − cos  (5.40)
 2

 θ
DC = hmax + ∆h = hmax + a1 − cos  , (5.41)
 2
θ
AD = a sin , (5.42)
2
 R
a sin  
ϕ = tan −1
AD
= tan −1  2a  . (5.43)
DC  R
hmax + a 1 − cos 
 2a 
Maximum angle of incidence ϕ = ϕ max may be achieved for the rays AC and CB that are tangent

to the Earth's surface. In that case the triangle AOC becomes a right triangle, thus apparently

283
θ R π
= = − ϕ max . (5.44)
2 2a 2
Thus, taking into account (5.44) the following expression may be derived from (5.43):

 a 
ϕ max = sin −1   , (5.45)
 a + hmax 
which also may easily be found from geometrical sketch in Figure 5.9.
Maximum propagation distance for single-hop path may be found based on (5.44) as

π 
Rmax = 2 a  − ϕ max  . (5.46)
2 
Now we can evaluate the upper limit of the frequencies for radio waves able to reflect from
ionospheric layer. That frequency, called maximum usable frequency (MUF), may be found by
substitution of (5.45) into (5.37):

  a  a + hmax
f max = MUF = sec sin −1   f c = fc . (5.47)
  a + hmax  2ahmax + hmax
2

Along with f c , this is one of the important parameters of the ionospheric reflecting layer that

depends on its altitude hmax and the maximum concentration of the ionospheric plasma N e , max .

The approximate values of the ionospheric parameters calculated by (5.35), (5.45), (5.46) and
(5.47) are given in Table 5.2 1. In HF-communications practice the suffixes are used after MUF
as given in notation examples below:
o MUF-F2-0 - is to identify f c for the normal incidence on F2-layer,

o MUF-F2-4000 - is to identify f max for the maximum incident angle on F2-layer (4000 km)

o MUF-F1-1500 - is to identify f c for the oblique incidence on F1-layer, for single-hop

propagation distance of 1500 km.

1
Because the magnitude of the dielectric constant is fairly close to unity for the frequencies used in
radars and communications, ionospheric D-layer, which exists in day-time only, may not cause a bent of
the ray path. Therefore it is non-reflecting layer, and, therefore, is not included in Table 5.2.

284
Table 5.2.

Reflecting N e , max f max


fc hmax ϕ max Rmax
ionospheric electrons (MUF)
MHz km degrees km
layer per cm3 MHz

E (4 to 5) ⋅10 3 0.57 to 0.64 80 to 100 ≈ 80.8 ≈ 2000 3.3 to 3.7

F1 (2 to 5) ⋅10 5 5 to 12.7 ≈180 ≈ 76.8 ≈ 3000 17 to 28

F2 3⋅10 5 to 2 ⋅10 6 5 to 12.7 ≈ 300 to 350 ≈ 72.2 ≈ 4000 16 to 41

The diagram shown in Figure 5.10 [7, p.292] is designed to simplify obtainment of the MUF for
radio wave reflected from F2 layer, for any arbitrary single-hop distance from 0 to 4000 km, if
those for two extreme distances, MUF-F2-0 and MUF-F2-4000 are initially provided.

Figure 5.10. The diagram for predicting the MUF-F2 for the single-hop arbitrary distances.
Example. Initial data: (1) single-hop distance R = 1500 km, (2) MUF-F2-0 = 6 MHz, (3) MUF-
F2-4000 = 20 MHz. The dashed lines prove the obtainment of the result: MUF-F2-1500 =
12.4 MHz.

Practical calculations as well as numerous observation show, that a variety of different types of
propagation modes may exist between ionospheric layers and earth's surface, i.e. the radio
wave may arrive the destination point by multiple reflections from different layers and earth's

285
surface. Some examples of propagation modes are shown in Figure 5.11, where the reader can
easily realize the principles of their notations borrowed from [7, p.189]:
• The reflections are specified in order of their appearance with respect to the transmitting
point
• A dash is used to indicate the reflection from the earth’s surface; the absence of dash
means the reflection from the top side of the layer, after the previous reflection took
place from another layer; one example of those cases is shown in Figures 5.11b, where
reflection from F2-layer is followed by reflection from F1-layer; another example is shown
in Figure 5.11c, where reflection from E-layer is followed by reflection from F2-layer.

Figure 5.11. Examples of the ray patterns of different sky-wave modes propagating due
to reflections from ionospheric layers and earth's surface subsequently, and their
notations

5.5. THE IMPACT OF EARTH'S MAGNETC FIELD ON


PROPAGATION OF THE RADIO WAVES IN IONOSPHERE

To assess the impact of the earth's magnetic field on the ionospheric plasma consider
the behavior of free electron in uniform, permanent magnetic field. As known from the physics
course, the Lorentz force applies to electron that moves with the velocity of v in a magnetic

field of the intensity H 0 . Direction of the force is defined from the following vector relation

Fe = e [ v × B0 ] = − e µ 0 [ v × H 0 ] , (5.48)

286
i.e. this force is perpendicular to both, v and H 0 vectors, and is bounded to them by the right

hand rule (see Figure 5.12).

Figure 5.12. Trajectory of free electron in permanent H 0 magnetic field

The electron movement is due to collisions with the randomly moving plasma particles, such as
protons, neutral molecules, etc. Because of the low density of ionospheric plasma those
collisions appear quite rarely, so thus electron keeps his kinetic energy unchanged for relatively
long time periods, hence the magnitude of the velocity remains unchanged for those time
intervals. Despite the constant magnitude, the direction of the vector of velocity changes
continuously due to the impact of the permanent magnetic field. The Lorentz force makes
electron move along the circular trajectory. In order to define the period of circulations
(rotations) we have to take a magnitude of Lorentz force equal to centripetal force, which keeps
electron on circular orbit, i.e.
2
me v e
= e µ 0 ve H 0 , (5.49)
RH

where RH is a radius of the trajectory, e = 1.6 ⋅10 −19 C is the charge of electron, µ 0 = 4π ⋅10 − 7

F/m is the absolute magnetic permeability of vacuum, and me = 9.1 ⋅10 −31 kg is the mass of

electron. The radian frequency of electron's movement may be found from (5.49) as
ve e µ0 H 0
ωH = = . (5.50)
RH me

287
ω H is called gyro-magnetic radian frequency 1 , which, as follows from (5.50), doesn't depend
on the initial speed of electron, but only on the magnetic field strength. The radius of circular
trajectory of electron depends on the initial speed, as it may be seen from (5.49) as
me v e
RH = . (5.51)
e µ0 H 0
Due to collisions between electrons and plasma’s composite particles abrupt changes in
electrons’ kinetic energy may occur (losing or gaining the energy). Therefore in real conditions
the linear velocity of electrons may jump abruptly. Regarding (5.51) it will result in jumps of radii
of electrons' trajectories, i.e. the orbit doesn’t remain circular all the time, but rather be
expanding or decaying spirals. However, the rotation frequency will not be disturbed, and is
defined by (5.50), being independent on linear velocity, and hence on kinetic energy. If the
numerical values are substituted into (5.50), then taking into account the earth’s magnetic field
average H 0 ≈ 40 A/m, as well as other constants given above, the Larmor frequency may be

estimated as
ωH e µ0 H 0
fH = = ≈ 1.4 MHz. (5.52)
2π 2 π me
Propagation of the radio waves in such magneto-active ionospheric plasma is considered
below.

5.5.1. LONGITUDINAL PROPAGATION OF THE RADIO WAVE


Assume the direction of propagation of linearly-polarized radio wave is the same as the direction
of the constant earth’s magnetic field H 0 in ionospheric magneto-active plasma, namely H 0 is

directed along Z-axes. As it was shown in section 2.3 any linearly-polarized radio wave may be
represented as a superposition of two circularly-polarized waves with the opposite rotations of
E -vector components in X0Y plane for the viewer facing along positive Z-axes as shown in
Figure 5.13: CW, which is RHCP, and CCW is LHCP by definition.

1
Also called cyclotron or Larmor frequency

288
Figure 5.13. Linearly polarized radio wave in magneto-active plasma: longitudinal
propagation. Π - Pointing vector; E = E + + E − , decomposition of linearly-polarized radio
wave into RHCP (+) and LHCP (-) circularly polarized waves

It's easy to realize, that for H 0 directed along Z-axes the gyro-magnetic rotations of electrons

are in the same plane where vector E is located. Those rotations are of same direction as one
of circular components of the radio wave has, but are opposite to the rotations of the other
circular component of propagating wave (see Figure 5.13).
It's reasonable to assume, that passing through the electrons cloud, different circular
components of the propagating radio wave will experience different impacts from the medium:
the CW rotations of electrons will have one impact on CW-polarized component of the radio
wave, and another impact on CCW-polarized component. In other words the medium behaves
as anisotropic, i.e. it exhibits different parameters for radio waves of different polarizations. The
analytical evaluations for the dielectric constant of lossless magneto-active ionospheric plasma
for the radio waves of different circular polarizations are outside the scope of this text and may
be found in [4 – 7, 9]. Here we just provide the results as follows:
2
f  1
ε + = 1 −  c  , (5.53)
 f  1− fH
f
2
f  1
ε − = 1 −  c  . (5.54)
 f  1+ fH
f

289
where f c and f H are defined by (5.10a) and (5.50) respectively. Expressions (5.53) and (5.54)

are graphically represented in Figure 5.14.

Figure 5.14. Dielectric permittivity of ionospheric magneto-active plasma for longitudinal


propagation of RHCP (+), and LHCP (-) radio waves. a). propagation within F1- and F2-
layers ( f c > f H ), b). propagation within E-layer ( f c < f H )

In presence of the earth's magnetic field dielectric constant becomes equal to zero not for the
frequency f = f c , like for the case of non-magneto-active plasma, H 0 = 0 , but for the

frequencies

f0 + = f c + ( f H / 2) + f H / 2
2 2
(5.55)

f0 − = f c + ( f H / 2) − f H / 2 .
2 2
and (5.56)

The followings may be concluded from the above considerations:


1. Due to difference in dielectric constants for right-hand and left-hand circularly polarized
components of linearly-polarized radio wave ( ε + ≠ ε − ) different propagation velocities appear

for those components within magneto-active medium: v + = v− . After passing any distance r, a
phase shift between components will appear at the end of distance,

290
∆Φ =
ω
v+
r−
ω
v−
r=
ω
c
r ( ε+ − )
ε− . (5.57)

This phase shift ∆Φ will result in rotation of the polarization plane of the initial wave by
∆Ψ angle, relative to the original position (see Figure 5.15).

Figure 5.15. Structure of the linearly polarized radio wave at the input (a) and the output
(b) of magneto-active plasma (dashed vector represents the original position of E—
component).

The angle of rotation of the polarization plane for the linearly-polarized wave may be defined
from Figure 5.15 as

∆Ψ =
∆Φ ω r
2
=
2c
( ε+ − ε− .) (5.58)

This phenomenon that occurs for the linearly polarized radio wave propagating longitudinally in
magneto-active plasma is called Faraday rotation. Closer the radio wave’s frequency is to gyro-
magnetic f H , stronger this effect occurs.

2. As one may notice from figure 5.14 the LHCP radio wave is unable to propagate at the
frequency f 0, − less 1, whereas the RHCP radio wave may propagate not just at the frequencies

greater than f 0, + but also at the frequencies less than f H , due to positive values of dielectric

constant. In other words at the frequencies much lower than the critical a LHCP component of

1
The same statement is thru for the regular plasma (H 0 = 0) at the frequencies less than f 0 (see

sections 5.1 and 5.4)

291
the radio wave will be cut off, whereas the RHCP component will easily propagate along
magnetic field-lines for fairly long distances without considerable attenuations. In plasma
physics this specific type of wave of the relatively low frequencies are known as spiral waves. In
radio communication routine the term whistling atmospherics (or whistlers) is in use for many
decades. The RHCP radio waves of VLF-band may arise from the atmospheric discharges
(lightning) in the polar areas: they are able to penetrate into the ionosphere and propagate
along the magnetic field lines from one earth's pole to the other, causing whistling noises in
radio receivers. The whistling character of those noises may be explained as follows: being
originally generated by the atmospheric lightning /discharges, those noise-type electromagnetic
waves have considerably large frequency spectrum, which is concentrated mostly in the area
from several hertz to hundreds of kilohertz; different harmonics of that spectrum have highly
diverse propagation velocities because of existing highly expressed dispersion of the ionized
medium in low-frequency domain (see Figure 5.14); the lower is the frequency, the smaller
propagation velocity is; thus the higher frequencies are arriving to the receiving point earlier
than the lower ones, causing an audio image of the whistle with the decreasing tone.
3. When the frequency of radio wave becomes close, or equal to gyro-magnetic frequency, f H
i.e. when the radio wave oscillations become synchronous to electrons rotations, then the
interaction between radio wave and electron cloud increases and a considerable part of the
radio wave’s energy transforms into the kinetic energy of electrons rotations, resulting in abrupt
increase of the linear velocity of electrons. Regarding (5.50) it increases dramatically the radii of
electrons rotations. This specific (excited) condition of the electron cloud is called gyro-magnetic
resonance, which leads to significant increase of the number of collisions with other particles in
plasma, and hence increases irreversible (thermal) energy losses. Thus nearby, or at the gyro-
magnetic resonance there're two phenomenon that may take place: (1) the decrease of the
velocity of E + -component of the radio wave, caused by increase of ε + , and (2) intensive

absorptions of the energy of E + component. As a result when the linearly polarized wave of the

frequency f H or close to it inputs that medium, then only LHCP-component may pass through
the medium. When propagating through the ionosphere, it may occur to the radio wave when
f c < f H only (specific for E-layer) and only if f 0 − < f < f H shown in Figure 5.14b.

292
5.5.2. TRANSVERSE PROPAGATION OF THE RADIO WAVE
Now consider the case, when the direction 0Z of linearly-polarized radio wave is perpendicular
to H 0 magnetic field lines (Figure 5.16).

Figure 5.16. Linearly polarized radio wave in magneto-active plasma: transverse


propagation Π - Pointing vector. E = E X + EY decomposition of linearly-polarized radio
wave into two cross orthogonal linearly polarized waves

In this case E vector of the linearly polarized radio wave located in X0Y-plane is decomposed
into of two linearly polarized waves (unlike previous case, where the initial wave has been
decomposed into two circularly polarized waves): one of them, E x is directed along magnetic

field lines (along X-axes), another one, E y is perpendicular to magnetic field lines (along Y-

axes). E x -component forces electron to move along H 0 field, therefore regarding (5.48) no

Lorentz force will be induced and applied to the electron regardless of the existence of the H 0

field. In other words the behavior of free electrons will remain the same as in absence of H 0

field, and the propagation conditions of E x -component in magneto-active plasma will remain the

same as in regular non magneto-active plasma, where H 0 = 0 . Hence this component is called

ordinary ray and dielectric constant for this ray is defined as for non magneto-active plasma for
in absence of the collisions, ξ = 0 (see (5.11), and may be rewritten here:

293
2
f 
ε ord = 1 −  c  . (5.59)
 f 
The force e ⋅ E y , applied to free electron, caused by the other component of the propagating

radio wave, is directed perpendicular to direction of H 0 , thus, regarding (5.48), the Lorentz

force will affect the cyclotron movements of the electrons, resulting in more complicated
interaction between that component and the electron cloud, than that for the ordinary
component. Therefore E y -component is called extraordinary component. Detailed analysis of

the interaction between E y -component and the electron cloud is given in [4 – 7, 9] with the

following result for the dielectric constant:


2
f  1
ε ext = 1 −  c  . (5.60)
 f  ( f H / f )2
1−
1 − ( fc / f )
2

Here the same notations are used as in (5.53) and (5.54). Expressions (5.59) and (5.60) are
shown graphically in Figure 5.17. Note form that figure, that the extraordinary ray may cause a
resonance in electrons cloud, similarly to that in previous case of the longitudinal propagation of
RHCP radio wave. Unlike the previous case the resonance appears not on the frequency f H ,
but on

f∞ = fc + f H .
2 2
(5.61)

Figure 5.17. Frequency dependence of dielectric constant for ordinary and extra-ordinary
rays, for transverse-propagation in magneto-active ionospheric plasma

294
From figure 5.17 one may realize that propagation of the extraordinary ray is possible in two
frequency ranges: (1) f 01 < f < f ∞ , and (2) f > f 02 , where f 01 and f 02 are derived from

(5.60) by taking it equal to zero:


2
 fH  f
f 01 =   + fc − H ,
2
(5.62)
 2  2
2
 fH  f
f 02 =   + fc + H
2
(5.63)
 2  2
Those expressions are the same as (5.55) and (5.56).
Some properties of transverse-propagation in magneto-active plasma are as follows:
1. While propagation in magneto-active plasma, an extraordinary ray generates an additional
electric field component E z along the propagation direction, which, regarding some estimates

provided in [2-4]), is 900 phase shifted relative to E y , and is much smaller by magnitude, than

the initial component E y . If the extraordinary ray is the only one that propagates in considered

ionospheric plasma-medium, then the superposition of those two electric field components, E y

and E z will result in elliptic polarization of the wave, with polarization ellipse lying in Y0Z-plane.
2. Similar to the previous case of longitudinal propagation, in this case of transverse
propagation magneto-active plasma exposes different properties ( ε ord and ε ext ) for ordinary and

extraordinary rays. Therefore each of those two components of linearly-polarized radio waves
will propagate with different velocities, so thus at the end of propagation distance an extra
phase shift between them will occur:

∆Φ =
ω
c
r ( )
ε ord − ε ext . (5.64)

The total field strength, as a superposition of those two components


E = x0 E x + y 0 E y (5.65)

may have different polarization for different values of ∆Φ (see section 2.4). While moving
along the medium one polarization type may smoothly be transformed into another, depending
on distance covered. It must also be noted that the sense of polarization may become even
more complicated if E z component is taken into account in (5.65).

295
3. When the frequency of the radio wave becomes close to resonant, f ≈ f ∞ , then similarly to
the case of longitudinal propagation one of the components of the arbitrary polarized wave,
namely extraordinary component, will experience an intensive attenuation and may disappear
almost completely, being absorbed by the medium. In other words the magneto-active plasma-
medium may be considered as a polarization filter at that particular resonant frequency.

5.5.3. PROPAGATION OF THE RADIO WAVE ARBITRARY ORIENTED


RELATIVE TO THE EARTH’S MAGNETIC FIELD
Consider the general case when the Pointing vector of the radio wave has an arbitrary γ angle

with earth’s magnetic field H 0 . Let’s decompose the magnetic field vector into longitudinal

H L = H 0 cos γ (5.66)

and transversal
H T = H 0 sin γ (5.67)

components as it's shown in Figure 5.18.

Figure 5.18. Disposition of earth’s magnetic field into transversal and longitudinal
components for an arbitrary direction of propagation in magneto-active ionospheric
plasma

For the previous case of transverse propagation the radio wave splits into two linearly-polarized
waves: ordinary ( E x ) and extraordinary ( E y ). Without restrictions to the generality we may

assume, that the magnetic field H 0 lies in XOZ-plane, thus for the linearly polarized plane wave

296
propagating along Z-axes the E x -component is concurrent with the direction of H T and E y is

perpendicular to it. Thus E x occurs as an ordinary component of the radio wave, and E y , as an

extraordinary component. Regarding to longitudinal H L -field the primary wave may be

decomposed into RHCP ( E+ ) and LHCP ( E − ) circularly polarized components. Superposition of

one of the linear components ( E x , for instance) with one of the circular components ( E − , for

instance) will result in left-hand-elliptically-polarized wave E1 , propagating along 0Z-axes 1 and

with the main axes of the polarization ellipse coincide with H T . Similarly, the superposition of

E y with the other circular component E + will result in right-hand-elliptically-polarized wave, E 2

with the main axes of the polarization ellipse orthogonal to H T , and with same propagation
direction (see figure 5.19).

Figure 5.19. The pattern of two elliptically-polarized waves, generated in magneto-active


plasma by primary wave. In this general case the earth’s magnetic field is arbitrarily
oriented relative to the direction of propagation of the primary radio wave

Electric field vector of each elliptically-polarized wave may be represented as

E1 = E ord + E − 
. (5.68)
E 2 = E ext + E + 

1
It's assumed that in general the magnitudes of components are not equal to each-other.

297
For this general case the dielectric constant of magneto-active plasma is defined by the
following formula [4 - 7]
2
2 fc
ε 1, 2 = 1 − 1/ 2
. (5.69)
fT
2  fT
4 
2f2− ± +4 f fL 
2 2

1 − fc / f
2 2
(
 1 − f c 2 / f )
2 2


Here ε 1 and "+" sign relate to propagation of E1 - wave, while ε 2 and "-" sign relate to

propagation of E 2 - wave. Two other notations in (5.69) are

e µ0
fT = HT , (5.70)
2 π me
e µ0
and fL = HL . (5.71)
2 π me
Expression (5.69) may be reduced considerably in two extreme cases:

4
fT
>> 4 f 2 f L
2
Case-1: 2
(5.72)
 f
2

1 − c 
 f2 
 
which is satisfied under the condition f ≈ f c , which usually takes place in HF- or MF-bands.

Then (5.69) may be simplified down to (5.59) and (5.60), so thus the propagation conditions
become the same as for transverse propagation, regardless to the value of γ angle. In other
words the ray will split into two cross-orthogonal linearly-polarized rays, which propagate
independently (ordinary and extraordinary, instead of two elliptically polarized components with
cross-orthogonal polarization ellipses). Therefore this case is often called quasi-transverse
propagation.
4
fT
<< 4 f 2 f L
2
Case-2: 2
(5.73)
 f
2

1 − c 
 f2 
 
which is satisfied under the condition f >> f c , which is typical for the frequencies UHF and

higher. Then (5.69) may simplified down to (5.53) and (5.54), so thus the propagation conditions
become the same as for longitudinal propagation, regardless to the value of γ angle; the ray

298
will split into two circularly-polarized components, which propagate independently. This case is
called quasi-longitudinal propagation.
Note finally, that after transformations of (5.69) the values f L or f T must be kept the same as

from (5.70) and (5.71) respectively, instead of f H .

5.5.4. REFLECTION AND REFRACTION OF THE RADIO WAVES IN


MAGNETO-ACTIVE IONOSPHERE
The presence of earth's magnetic field modifies the patterns of reflection and refraction of the
radio waves in ionosphere, because of dependence of parameters of the medium on
propagation direction and polarization of the radio wave. This is a typical case of propagation in
an anisotropic medium. Consider the oblique incidence of the radio wave on the boundary of
magneto-active ionosphere. The incident wave splits into two independently propagating waves,
as it was discussed above (see Figure 5.20).

Figure 5.20. Split of the radio wave in magneto-active ionosphere: a). reflection pattern
for the extraordinary (1), and ordinary (2) rays, b). refraction pattern for LHCP (1), and
RHCP (2) rays.

The reflections from the ionosphere are generally used for HF long-distance propagation (HF
long-range radio links), so thus the radio wave frequency is usually close to f c . Based on (5.72)

for those links the reflection may be considered for quasi-transverse propagation condition. At
the input of the ionospheric layer the incident wave, radiated from point A, splits into two
independent linearly polarized cross-orthogonal waves: ordinary and extraordinary. The
reflection condition, (5.37) for the ordinary ray may be rewritten here as

299
f ≤ f max = sec ϕ 80.8 ⋅ N e , ordinary , (5.74)

where N e , ordinary is the ionospheric plasma concentration needed to satisfy reflection condition

for the ordinary ray. Now rewrite the reflection condition (5.37) for the extraordinary ray, which
may appear hypothetically in two cases:
f ≤ f 0,1 sec ϕ (5.75)

and f ≤ f 0, 2 sec ϕ , (5.76)

where f 0,1 and f 0, 2 are defined by (5.62) and (5.63) respectively.

Due to the relation f 0,1 < f c < f 0, 2 , the expression (5.76) is easier to satisfy, so we may ignore

condition (5.75). Then the MUF for the extraordinary ray may be found from (5.76) and (5.63) as

f max = sec ϕ  80.8 N e , extraord + ( f H / 2 ) + f H / 2 .


2
(5.77)
 
The same frequency and the same angle of incidence are considered for both, ordinary and
extraordinary rays. Hence taking (5.74) and (5.77) equal to each-other, the following expression
may be written here:

80.8 N e , ordinary = 80.8 N e, extraord + ( f H / 2 ) + f H / 2 .


2
(5.78)

The expression (5.78) shows clearly the following relation between ionospheric plasma
concentration needed to satisfy the reflection conditions for the ordinary and extraordinary rays:
N e , extraord < N e, ordinary . (5.79)

The reflection heights of both rays may be found based on vertical profile of the ionospheric
plasma concentration N e (h ) : either on experimental that is shown in Figure 1.2, or on widely

used parabolic model shown in Figure 5.21.

300
Figure 5.21. Reflection heights definition for the ordinary and extraordinary rays, based
on vertical profile of the parabolic model of ionospheric plasma concentration

It may be seen from the Figure 5.21 that the reflection point for ordinary ray is higher, than for
the extraordinary, i.e. h 2 > h 1 . It must be noted, that the absorption rate, as well as dispersion

rate for the extraordinary ray is much higher than that for ordinary ray. That's the reason of
using for HF communication lines the radio waves of polarization, which produces the ordinary
ray rather than extraordinary. The structure of geomagnetic field is such, that for the low and
medium latitudes the earth's magnetic field-lines are close to horizontal, and directed from north
to south. That's the reason the ordinary ray has a horizontal polarization, therefore the
horizontally polarized waves are at the most preference for the HF communication lines design.
At the higher latitudes, namely for the earth’s polar zones the HF-communications are rarely
used due to the high absorption rate (abnormal absorptions) in those areas.
Considering the refraction of the ray path in ionosphere (e.g. on earth-to-space, or space-to-
earth links) it must be emphasized, that for the frequencies higher than VHF, the radio wave
penetrates through the ionosphere, being shifted from his initial propagation direction due to
refraction as shown in Figure 5.20b. From the general expression (5.69) for dielectric constant
of magneto-active ionospheric plasma it is clear, that the impact of medium will be as low, as
greater is the difference between the currier frequency of radio wave and the cutoff frequency
(plasma frequency), i.e.
f >> f c . (5.80)

Under this condition:


1. The dielectric constant become close to one, so thus the ray path will become close to
the straight line

301
2. Ionospheric absorptions will decrease because regarding (5.9b) the conductivity of the
medium will decrease
3. The polarization disturbances and signal dispersions will also decrease
The condition (5.80) is to be met while designing the earth-to-space radio links as much as
possible. As mentioned above the frequencies VHF and higher are preferably to be used for
earth-to-space or space-to-earth propagation paths, which relates to the quasi-longitudinal
propagation (see condition (5.73)). If the linearly polarized radio wave propagates vertically
trough the ionosphere then the Faraday's rotation may be expected. The random rotations in
polarization plane of LP signal may results in so called polarization losses at the reception point
due to polarization random mismatches between Tx and Rx antennas.
If the propagation track is sloped, then most likely just one of the circularly polarized (CP)
components of the linearly polarized (LP) radio wave will arrive the destination due to different
paths for the CP components, as shown in Figure 5.20b. Hence the usage of LP radio wave
may result in loss of 50% energy (3 dB). To avoid those losses a CP radio wave is considered
as more preferable for the satellite radio links such as communication, navigation, broadcast,
etc.

5.6. OVER-THE-HORIZON PROPAGATION OF THE RADIO


WAVES BY TROPOSPHERIC SCATTERINGS MECHANISM.
SECONDARY TROPOSPHERIC RADIOLINKS

Although the asymptotic diffraction theory (ADT) examined in section 4.3 appears to be a
powerful tool for radio wave propagation predictions in shadow region behind the horizon line,
however the only way of counting the impact of the atmosphere within ADT concepts is the
introduction of the earth’s equivalent radius that takes into account only a smooth standard
atmospheric refraction. ADT prediction results are in fairly good agreement with measurements
in relatively low frequency bands up to HF. For the VHF frequencies and up, ADT predicts the
abrupt decrease of the field intensity for over-the-horizon distances, i.e. in diffraction zone,
whereas numerous observations performed after WW-II has shown much higher field intensities
than those predicted by ADT. Figure 5.22 demonstrates qualitatively the dependence of the field

302
intensity on wide range of distances that are predicted by ADT, as well as observed in real
conditions at the frequencies VHF and higher: in fact, higher is the frequency, shaper is
predicted field decay in shadow zone.

Figure 5.22. Distance-dependence of field strength for the VHF/Microwave propagation (a


qualitative pattern). R0 – horizon distance, 1 – theoretical prediction from ADT 2 – real

measurements (averaged curve), 3 – reference path (free-space)

In the mid 30-s several pioneering measurements have been carried by G. Marconi and other
investigators, which have shown a considerable difference (up to several hundreds dB-s)
compared to diffraction theory predictions. One of the features of the received field is very deep
fast fading of the signal level at the reception point.
Form mid 40-s to mid 50-s the formal explanation as well as fairly good quantitative description
of this phenomenon has been given by H. Booker and W. Gordon [11] with the later updates by
V. Tatarskii and S. Rytov [8]. The basic explanation of the nature of fairly intensive field at the
receiving point for over-the-horizon distances is as follows:
• In presence of real atmosphere the existence of the field in shadow area is a result of
scattering of the primary radio wave from random irregularities of the tropospheric air
known as turbulences
• The facts such as twinkling stars and the wavering appearance of the objects observed
over the earth’s surface heated by sun testify that the atmospheric air is in random,
erratic movements that exists permanently, regardless on weather condition, geographic
region, and season

303
• Statistically those irregularities appear as spherical volumes (globules, eddies) with the
refraction index slightly different from the surroundings; those globules exist permanently
in troposphere, even in “clear air” condition
• Typical dimensions of those globules that are most intensively involved in scattering of
the UHF and higher band frequencies vary from several millimeters to hundreds of
centimeters, hence maximum effectiveness of those re-radiations must be expected in a
range of the wavelengths comparable to the range of those dimensions, which typically
belong to microwave frequency bands
• Although a scattering from one single globe may never be “sensed”, however the
superposition of the huge amount of those scattered fields, called random ensemble or
Rayleigh ensemble of waves, causes in fairly “sensible” field strength at the observation
point
• Number of those scattering globules is unpredictably large, and is defined by the
common volume between transmitting and receiving antenna beams; that common
volume shown in Figure 5.33 is located at the heights from about 1 to 8 km above the
surface
• Deep fading of the received signal, which is specific for the troposcatter propagation, is a
result of random interference of the scattered (elementary) fields at the reception point,
i.e. is a result of constructive and destructive interference of those elementary fields due
to random phase shifts between them.

5.6.1. ANALYTICAL APPROACHES IN DESCRIPTION OF THE RANDOM


TROPOSPHERIC SCATTERINGS
Assume that the dielectric constant ε of the troposphere is a variable that is randomly
fluctuations in time and space. The time variations of ε are much slower, than the rate of
oscillations of propagating radio wave. Thus it may be assumed, that within several periods of
oscillations of the radio wave the random spatial distribution ε remains unchanged, i.e. the
"randomness" occurs only as a function of coordinates. In other words only the random spatial
fluctuations around the spatial average, ε may be accounted to simplify the analysis, i.e.

304
ε (r ) = ε + ∆ε (r ) , (5.81)

where ∆ε is a randomly fluctuated part of dielectric constant, and r = xˆ 0 x + yˆ 0 y + zˆ 0 z is the

radial vector-coordinate of the observation point. This time-independent model of so called


frozen turbulences is quite convenient for analytical evaluations instead of time-dynamic
approach, and as follows from the numerous of experiments, the final results are close enough
to those observed. For further simplifications we may assume ε = 1 , which is acceptable for

the tropospheric scatterings analysis.


Recall (2.5), which constitutes relation between vectors of the electric field strength ( E ) and its
induction ( D ):
D = ε 0ε E = ε 0 (1 + ∆ε ) E = ε 0 E + P , (5.82)

where P , by its classical definitions, is microscopic vector of polarization of the medium at


particular point of space within a unit volume. From (5.82)
P = ε 0 ∆ε E . (5.82a)

If referred to Figure 5.23, one can realize that time-harmonic polarization P of the medium at
point C is forced by electric field E incident from transmission point A. In other words the field,
scattered from point C (secondary radiation) may hypothetically be presented as an primary
field radiated by the dipole with the moment of polarization
d p = ε 0 ∆ε EC dVsc . (5.82b)

Here d p is a resultant dipole moment from all polarized particles (atoms, molecules) induced

by the incident electric field EC within elementary volume dVsc .

The following vector quantities are introduced to support further analysis (see Figure 5.23):
• k 0 the wave-vector of the incident wave, which is directed orthogonal to incident wave-

front; its magnitude is equal 2π / λ 0

• r1 is the radius-vector of the elementary volume dVsc . It is directed from the transmitting

point A to the point where the elementary volume dVsc is located (see Figure 5.23).

305
Figure 5.23. Positioning of vectors for tropospheric scattering analysis. (Origin of the
coordinate system is located at A point)

Taking into account the above introduced quantities, the expression (4.1) for the amplitude of
the incident field at point C may be rewritten in modified form as

E C =
60 PTx GTx
r1
[ ]
exp i (ω t − k 0 ⋅ r1 ) . (5.83)

For the elementary dipole with the dipole moment d p , placed in point C, the field strength in

reception point B at the distance r2 from point C in complex scalar form is defined by (A3.31)
from Appendix-3, which may here be rewritten as

[ ]
2
k0
dE B = d p sin θ exp − i k sc ⋅ r2 , (5.84)
4πε 0ε r2

where k sc is the wave-vector of the scattered wave, shown in Figure 5.23. Note that the

multiplier sin θ in (5.84) is the normalized radiation pattern of the elementary dipole as
expected from its physical meaning. In our case the dipole axis is collinear to the exciting field

E C , thus θ must be counted form that axes (see Figure 5.23). After combining (5.82b), (5.83)
and (5.84) the following result may be obtained for the scalar complex field strength at the
observation point

k 0 ∆ε (r1 )
[ ]
sin θ exp i (ω t − k 0 ⋅ r1 − k sc ⋅ r2 ) dVsc .
2
60 PT GT
dE B = (5.85)
4π ε r1 r2
Here is assumed, that the vector of the scattered field strength will keep the same direction as

the original vector E C . As one can see from (5.85) the field dE B is proportional to ξ ε = ∆ε / ε .

306
Hence both, the amplitude and phase of dE B will randomly fluctuate as the position of the

scattering point C varies within the scattering volume Vsc .

In order to define the power flow density of the scattered radio wave at the receiving point (the
average magnitude of Pointing vector), consider scattered field strengths dE B′ and dE B′′ arriving

′ ″
at the observation point B from elementary scattering volumes dVsc and dVsc positioned at

point C’ and C” respectively as shown in Figure 5.24.

Figure 5.24. Superposition of two scattered rays at the reception point from different
volumetric elements of the scattering volume

Taking into account (2.25) the total average power flow intensity at the receiving point B may by
expressed in the following form of the volumetric integral:

1  1 ′ ~ ″ 
Π B , ave = Re  ∫ (dE B dE B ) , (5.86)
2 Vsc W0 
~ ″
where W0 = 120π Ohm is the wave impedance of the free space, dE B is a complex conjugate

″ ~ ″
of dE B . Using (5.85) we may substitute both, dE B′ and dE B into (5.86):

1  ′ sin θ ″  ∆ε ′ (r1 ) ⋅.......


4
k0 60 PT GT
Π B , ave = Re  ∫∫ sin θ  ε 
2  Vsc (4π )2 W0 r1 r2 r1 + ρ r2 − ρ  

 ∆ε ′′(r1 + ρ ) ′ ′ ″ ″ ′ ″
.....  exp − i k 0 ⋅ r1 − i k sc ⋅ r2 + i k 0 ⋅ (r1 + ρ ) + i k sc ⋅ (r2 − ρ ) dVsc dVsc  , (5.87)
 ε  
 
 
where the vector-coordinates, shown in Figure 5.24 are related to each-other as follows:
″ ′ ″ ″ ′ ″
r1 = r1 , r1 = r1 + ρ = r1 + ρ , r2 = r2 , r2 = r2 − ρ = r2 − ρ . (5.88)

307
The heavy expression (5.87) may considerably be simplified for the real conditions, when both
antenna beams, Tx and Rx, are extremely narrow, so thus the distances r1 and r2 are much

greater, than the linear dimensions of the common scattering volume Vsc . This volume has an

elevation above the earth's surface much less, than the horizontal distances. Hence
′ ″ ′ ″
k0 ≈ k0 , k sc ≈ k sc , θ′ ≈θ″, r1 + ρ ≈ r1 , r2 − ρ ≈ r2 . (5.89)

For conditions (5.89), as well as W0 = 120π Ohms, the expression (5.87) may be simplified

down to (see also the footnote on the bottom of this page)

k 0 PT GT sin 2 θ  ∆ε ′ (r1 ) ∆ε ′′ (r1 + ρ ) 


∫ exp [ − i (k ]
− k 0 )⋅ ρ  ∫
4

Π B , ave ≈ dVsc  dVsc (5.90)


(4π )3 2
r1 r2
2
Vsc
sc
Vsc ε ε 
Evidently the integral in braces in (5.90) is a the spatial autocorrelation function of the random
scalar field, ξ ε (r ) = ∆ε (r ) / ε , which we denote here as

ψ ε (r1 , ρ ) = ξ ε ′ (r1 + ρ ) ξ ε ″ (r1 ) dVsc .


1

Vsc Vsc
(5.91)

In general, ψ ε depends on r1 , as well as on the distance ρ of two point within the common

volume Vsc . Fortunately the fluctuations of tropospheric refraction index are statistically

homogeneous, i.e. they are independent on position r1 , and are only dependent on distance ρ
between two points. Thus (5.91) may be written as
ψ ε = ψ ε (ρ ) , (5.92)

and (5.90) is represented as

k 0 PT GT Vsc sin 2 θ
4

Π B , ave ≈
(4π )3 r12 r2 2 ∫ ψ ε (ρ ) ⋅ exp [− i q ⋅ ρ ]dV
Vsc
sc , (5.93)

where q = k sc − k 0 (5.94)

is called a scattering vector or vector of spatial wave number. ψ ε (ρ ) is considered as one of

deterministic characteristics of the random field ∆ε (r ) . Then the integral in (5.93) may be

expressed in terms of volumetric Fourier-spectrum of ψ ε (ρ ) field 1:

As seen later the volumetric function ψ s ( ρ ) is symmetric, hence (5.95) becomes a real function.
1

Therefore when transforming (5.87) into (5.90) the sign Re was omitted.

308
S ε (q ) = ψ ε (ρ ) exp (− i q ⋅ ρ ) dVsc ,
1
(2π )3 Vsc∫
(5.95)

with the inverse transform:

ψ ε (ρ ) = ∫ S ε (q ) exp (i q ⋅ ρ ) dVsc . (5.96)


Vsc

The physical meaning of the (5.96) transform is that the autocorrelation function ψ ε (ρ ) is

represented as superposition of the continuum of deterministic plane waves S ε (q ) exp ( i q ⋅ ρ )

each of which has an amplitude of S ε ( q ) as a function of the spatial frequency q (scattering

vector). Note that in the theory of random processes (theory of time-domain random functions)
the transform (5.95) is known as Wiener-Khinchin transform, whereas in theory of random fields
(theory of space-domain random functions) it's known as Shannon-Whittaker transform. If (5.95)
is substituted into (5.93), then the result is

k 0 PT GT sin 2 θ S ε ( q )
4

Π B , ave ≈ 2 2
Vsc . (5.97)
8 r1 r2
Finally we use the replacement θ = π / 2 + θ S that is illustrated in Figure 5.23. Then (5.97) will

be rewritten as

k 0 PT GT cos 2 θ sc S ε ( q )
4

Π B , ave ≈ 2 2
Vsc . (5.97a)
8 r1 r2
Here θ sc is a scattering angle (see Figure 5.23).

5.6.2. PHYSICAL INTERPRETATION OF TROPOSPHERIC


SCATTERINGS

As shown above, spatial distribution of the random field ξ ε (r ) may be expanded into the

superposition of continuum of elementary plane waves. To be more specific, note that the
expansion is applied not to the random field ξ ε (r ) itself, but to one of its deterministic

characteristic such as autocorrelation function ψ ε (ρ ) . It contains the main statistical

parameters, such as mean square deviation, ξ ε2 = (∆ ε / ε ) and the average dimension of


2

random irregularities Lε, i.e. the average sizes of the globules of turbulences (see Figure 5.25).

309
Figure 5.25. a). Flat projections of the autocorrelation function of the random scalar field,
b). Spectrum of the spatial harmonics of autocorrelation function of the random field ∆ε .

One of the properties of Shannon-Whittaker transform is similar to property of the Fourier


transform 1, namely the narrower is the initial function ψ ε (ρ ) , the wider the spectrum becomes,

and vise-versa. There's a following approximate relation between the width of the main lobe of

the correlation function (the average dimension of the irregularities), Lε and the width of spatial

spectrum of irregularities: q max ~ 1 / Lε .

To understand the mechanism of how the random medium impacts the scattering process, let’s
consider a single S ε (q ) component of 3D spectrum of irregularities. If, for instance, q is

directed vertically down, then 3D shape of that component of the spectrum of autocorrelation
function may be represented as a stratified medium with the harmonically changing intensity of
ε in vertical direction (Figure 5.26) with the amplitude Sε and the spatial period of Λ = 2π / q .
Loci of the maximums of spatial densities are the parallel planes, separated by Λ , as shown
with dashed lines on Figure 5.26.

1
As it was mentioned above the Shannon-Whittaker transform is the same as a Fourier transform, which

is applied to spatial correlation function ψ ε (ρ ) , i.e. allows finding the Fourier spectrum of the spatial

harmonics of the given function ψ ε (ρ ) .

310
Figure 5.26. Scattering of the radio wave from single spatial harmonic of the turbulent
fluctuation of dielectric constant in troposphere

As seen from Figure 5.26 all rays reflected towards the receiver will superimpose at the
receiving point with the same phase (constructive interference) if the phase shift between two
adjacent rays is equal to 2π (or is multiple of 2π ). If we take into account that the difference in
distances between two adjacent rays is equal 2 BC , then from the Figure 5.26 it’s easy to derive
geometrically
2π  θ 
k 0 (2 BC ) =  2Λ sin sc  = 2π . (5.98)
λ0  2 
Then the spatial period of the considering harmonic may be found from (5.98) as
λ
Λ= . (5.99)
2 sin (θ sc / 2)
The same expression may be found from the same Figure 5.26 based upon disposition of three
vectors, k 0 , k sc and q . Indeed, taking into account the equal magnitudes


k 0 = k sc = (5.100)
λ

and q = (5.101)
Λ

311
the following relation may be developed form triangle ABC:
θ
k sc − k 0 = 2 k 0 sin = q . (5.102)
2
After substitution (5.100) and (5.101) into (5.102) the expression (5.99) may be verified, which is
known in optics as a Bragg's diffraction condition with the vector form of (5.94). It allows finding
relations between λ , Λ and θ sc , i.e. the relations between directions of incident and diffracted

waves, if the diffraction takes place on one of the spatial harmonics of 3D correlation function of
random field of the dielectric constant.
Now if the expressions (5.97) or (5.97a) are recalled, then one may notice that the average
power flow density at the receiving point is proportional to the amplitude of spatial harmonic.
Hence the intensity of scattered field, Π B , ave will vanish for scattering angles greater than θ max

when S ε becomes very small (see Figure 5.25). The relation between θ max and q max is found

from (5.102) as

 λ0 
θ max = 2 sin −1  q max  . (5.103)
 4π 
As mentioned above, the upper limit of spatial spectrum of ξ ε fluctuations, namely q max is in

inverse proportion to the average dimension of irregularities ( q max ∝ 1 / Lε ), therefore greater the

average dimensions of irregularities, narrower is the spatial spectrum, thus narrower is the
angular aperture of the scattered rays as demonstrated in Figure 5.27.

312
Figure 5.27. Patterns of scatterings of the radio waves on random irregularities of the
tropospheric turbulences and their vector diagrams ( 2θ max -- angular aperture of

scatterings). a). scatterings on small-scale irregularities, b). scatterings on large-scale


irregularities

5.6.3. EFFECTIVE SCATTERING CROSS-SECTION OF THE


TURBULENT TROPOSPHERE
For the engineering applications it's useful to introduce the effective scattering cross-section
(ESCS) of the troposphere. Consider a unit volume (e.g. 1 m3) that surrounds point C located
within the scattering volume Vsc . Now we replace hypothetically the unit volume of scatterers

that surround point C by a flat PEC surface σ sc , called ESCS, which produces at the receiving

point B the same amount of the received power as it is produced by the primary irregularities
located within that unit volume. Similarly to (3.35) we may obtain the incident power flow density
at the scattering point C produced by the transmitter. That is
PT GT
ΠC = , (5.104)
4π r1
2

313
where r1 is the direct distance from transmitting point A to the scattering point C. Then,
regarding the definition for ESCS, the total power flow of the wave scattered by the unit volume
that surrounds point C may be found as
PT GT
PC = Π C σ sc = σ sc . (5.105)
4π r1
2

The power flow density at the receiving point B is


′ PC PT GT σ sc
ΠB = = , (5.106)
4π r2
2
(4π ) 2 r12 r2 2

where P C is substituted from (5.105), and r2 is the direct distance from scattering point C to

receiving point B. Finally the total power flow density at B point, induced by the entire scattering
1
volume may be defined if we multiply (5.106) by Vsc , i.e.

′ PG σ V
Π B = Π B Vsc = T T 2 sc2 2sc . (5.107)
(4π ) r1 r2
If (5.107) is compared with (5.97a), then SSCS may be found in the following form:

σ sc = 2π 2 k 0 4 cos 2 θ sc S ε (q ) . (5.108)

Expression (5.108) demonstrates σ sc (θ sc ) dependence not just due to the presence of the

cos 2 θ sc factor, but mostly because of the presence of the spectral component S ε (q ) , which is
highly dependent on q vector, namely on both, on its magnitude and orientation θ sc . It's easy

to realize, that the maximum value of ESCS will appear at θ sc = 0 , i.e. σ sc , max = σ sc (0 ) . The

graph of the normalized function


σ sc (θ sc )
Σ sc (θ ) = (5.109)
σ sc , max
is called a scattering diagram, and is considered below for particular troposcatter models.

σ scVsc ∫σ
1
For higher accuracy the product must be replaced by the volume integral sc dVsc . However
VSC

for the practical engineering applicationsσ sc is assumed to be uniformly distributed within scattering
volume Vsc , which results in simple product σ scVsc instead of the volume integral.

314
5.6.4. STATISTICAL MODELS OF TROPOSPHERIC TURBULENCES

5.6.4.1. GAUSSIAN MODEL

Most commonly the autocorrelation function ψ ε (ρ ) in problems that relate to statistical radio-

physics is presented in form of Gaussian function. It is assumed, that this function doesn't
depend on direction of the spatial coordinate ρ but only on its magnitude. This kind of
statistically homogeneous random medium is called statistically isotropic random field.

 ρ2 
ψ ε (ρ ) = ξ ε exp  −
2
,
2 
(5.110)
 2 Lε 

where Lε is the average radius of the irregularities of the tropospheric air that are spherically

shaped by statistical means, and

 ∆ε 
2

ξ ε2 =   (5.111)
 ε 
is standard deviation of fluctuations of the relative dielectric permittivity of the clear tropospheric
−7 −6
air that usually belongs to the range from 1.5 ⋅10 to 3.3 ⋅10 . The lower limit corresponds to
the winter season, and the upper limit, to the summer season. It is independent on the
elevations above the earth’s surface (at least for the elevations of up to 5 km).
Spatial spectrum (5.95) of statistically isotropic field presented particularly by (5.110) may be
introduced in the following form after transformations provided in Appendix-7.

S ε (q ) = ∫ ψ ε (ρ ) sin (qρ ) ρ dρ .
1
(5.112)
2π 2 q 0

If (5.110) is substituted into (5.112), then the expression for the spatial spectrum may be derived
as [10, p.495]

ξ ε2 Lε 3  q 2 Lε 2 
S ε (q ) = exp  −  . (5.113)
(2π ) 3 / 2  2 
Now substitute (5.113) into (5.108) and (5.109) and take into account (5.102) to determine final
expressions for SSCS and for the scattering diagram respectively:

π  θ 
σ sc (θ sc ) = k 0 ξ ε2 Lε cos 2 θ sc exp − 2k 0 Lε sin 2 sc  ,
4 3 2 2
(5.114)
2  2 

315
 θ 
Σ sc (θ sc ) = cos 2 θ sc exp  − 2k 0 Lε sin 2 sc  .
2 2
(5.115)
 2 
The 2D-scattering diagrams that are expressed in (5.115) analytic form in polar coordinates are

illustrated in Figure 5.28 for different Lε / λ 0 ratios obtained by using MATLAB routine.

Figure 5.28. Scattering diagrams of the turbulent troposphere plotted in polar


coordinates, for the Gaussian model of turbulences (verifies the concept demonstrated
in Figure 5.27)

5.6.4.2. KOLMOGOROV-OBUKHOV MODEL

Although the previously considered Gaussian model (5.110) has various practical applications,
however, in case of tropospheric turbulences it is applied notionally, and is not based on real
aerodynamic processes related to generation of the tropospheric clear-air inhomogenieties of
the dielectric constant (eddies). More accurate quantitative results may be obtained, if scattering

316
phenomenon analysis is linked to the model, which is developed based on the real aero-
hydrodynamic processes of generation of those inhomogenieties.
Regarding to the principles of hydrodynamics, there're two types of movement of any gaseous
or liquid masses:
• Laminar – a particular type of streamline flow where the gas (or fluid) in thin,
parallel layers tends to maintain uniform velocity with the constant magnitudes
and directions (Figure 5.29a);
• Turbulent – when, at certain condition, a laminar flow of gas (liquid) turns into
stochastically distributed eddies, i.e. when the magnitudes and directions of the
composite atoms and molecules turn to be randomly distributed in time and
space (see Figure 5.29b).

Figure 5.29. Movements of tropospheric air masses; a) Laminar, b). Turbulent

The type of movement of gas or liquid is usually specified by so called Reynolds number:
ϑ v l flow
Re = , (5.116)
ν~
Here ϑ is a density, v is the magnitude of velocity vector v , l flow is the cross-sectional

dimension of the flow, and ν~ is the viscosity of the medium. Each gas or liquid has the specific
critical value of Reynolds number, Re cr . So if for particular medium Re > Re cr then the

movement of gas or liquid becomes turbulent, otherwise it is laminar. The values of variables in
(5.116) for the real troposphere are such that the movement of the air masses is almost always
turbulent, which results in permanent presence of the fluctuations of its dielectric constant.

317
Figure 5.30 demonstrates the spectrum of spatial harmonics of the tropospheric dielectric
constant fluctuations as a function of the “mechanical” spatial wave numbers. As seen from the
figure, the spectrum is divided into three ranges:
1. Range of formation of the eddies (globules) of largest sizes, L0 called the outer scale of

irregularities, which may vary in the range from 100 to 1000 m, and even larger
2. Inertial range. Kinetic energy of the movements remains unchanged. Eddies become
smaller in size gradually, hence with gradually increasing velocity of spinning of the air
masses inside each eddy. The range of sizes is l 0 < l < L0 ( 2π / L0 < q < 2π / l 0 ), where

l 0 is its lower limit called the inner scale of irregularities, which varies in range of
dimensions approximately from 1 centimeters to 1 millimeter
3. Dissipation range, where the lower limit l 0 of the eddy’s size is reached, thus the

maximum spinning velocity results in destruction of the eddies followed by the


dissipation of their kinetic energy, i.e. irreversible transformation of that energy into heat
while disappearance of the globules

Figure 5.30. Spatial spectrum of random fluctuations of dielectric constant of the


troposphere. 1 – range of eddies formation 2 – energy conservation range 3 – energy
dissipation range

Regarding to theoretical investigations by Kolmogorov and Obukhov the spatial spectrum of


correlation function for random fluctuations of dielectric constant of the troposphere in the area
of inertial transforms is expressed in the following analytical form [8]

318
11

Sε (q ) = 0.033 Cε q
2 3
, (5.117)
2
In (5.117) Cε here is in m-2/3 is called a structural constant that is dependent on season of the
2
year and on elevation of the scattering point. Sometimes the values of Cε are given in m-2/3
2
units. To transform Cε from one unit to the other recall that (1cm)-2/3 = (0.01m)-2/3 =
2
= 21.54(1m)-2/3, which means the value of Cε in cm-2/3 must be multiplied by 21.54 to get it in

m-2/3 units. Seasonal variations of the structural constant near the earth's surface are in the
range from 10 −14 cm-2/3 (summer) to 10 −16 cm-2/3 (winter).
2
The average vertical profile of Cε may be expressed as
−α
h
Cε ( h ) = Cε ( h0 )  
2 2
, (5.118)
 h0 
where Cε
2
( h ) is defined for the elevation of h
0 0 = 30 m. α is a constant, which has the

following values, defined empirically: α = 2 / 3 for the winter season, and α = 4 / 3 for the
summer season (see Figure 5.31).

2
Figure 5.31. Vertical profiles of the structural constant Cε for the turbulent troposphere

Note, that expressions (5.108) and (5.109) become meaningless if θ sc → 0 , because (5.107) is

valid only in inertial range shown in Figure 5.30, i.e. is valid for the spatial harmonics q starting
with their minimum value of

319
θ sc 2π
q = 2 k 0 sin ≥ qmin = (5.119)
2 L0
up to the maximum q max . Indeed, for the scattering angle θ sc = 0 the Kolmogorov-Obukhov

model violates the physical picture: it results in infinite scattering intensity, while limited value is
expected. This is the shortcoming of this model. On the other hand if that is compared with the
Gaussian model then one may realize that the Kolmogorov-Obukhov's model considers an
existence of multi-scale irregularities, which is typical for the troposphere, while the Gaussian
model considers only a mono-scale irregularities, with the dimensions, dispersed (spread)
around its mean value, Lε given in (5.110). Despite this fact, the use of Gaussian model is still

applicable if the angles θ sc are close to zero.

In MMW, and optical bands, when the values of scattering vector become large enough, i.e.,
when the interaction between incident waves with the turbulent medium takes place in
dissipation range of the spectrum, the following relation
θ sc 4π θ sc 2π
q = 2 k 0 sin = sin ≥ = qmax , (5.120)
2 λ 2 l0
dictates the usage of an additional exponential multiplier to be inserted into (5.117), which
allows to take into account an abrupt decrease of S ε (q ) seen from Figure 5.28:

 q 
S ε (q ) = 0.033 Cε q −11 / 3 exp  −  .
2
(5.121)
 q max 
For the small values of the magnitude of scattering vector, i.e. in range of formation of the
tropospheric turbulences, where
θ sc 2π
q = 2 k 0 sin ≤ = qmin , (5.122)
2 L0
the data obtained from the Kolmogorov-Obukhov model considerably distinguishes from real
observed values of S ε (q ) . Carman [8] introduced later update of the model by transforming

(5.121) into his final form, applicable for the practical applications:

 q 
S ε (q ) = 0.033 Cε (q 2 + q min ) −11 / 6 exp  −  .
2 2
(5.123)
 q max 

320
It was shown theoretically by Kolmogorov and Obukhov that the outer scale of atmospheric
turbulences, L0 is linked to the mean-square value of fluctuations dielectric constant of the

troposphere ξ ε2 and to the structural constant Cε by the following expression [8]


2

2 ξ ε2
Cε =
2
2/3
. (5.124)
L0

If the decaying tendency of the vertical profile Cε (h ) is taken into account from (5.118) or from
2

Figure 5.31, then it may be realized, that the higher is the elevation, the greater the outer scale
of the turbulences is, i.e. greater the dimensions of the new-born eddies are.
Now we substitute (5.123) into (5.108) and (5.109) to define ESCS and scattering diagram in
analytical form for this model:
−11 / 6
4
π 0.033 k 0 Cε L 0
2 11 / 3   L θ 
2
  l0 θ 
σ sc (θ ) = 3 cos θ sc 1 +  2 0 sin sc
2

  exp − 2 sin sc  , (5.125)
4 (2π ) 2   λ0 2    λ0 2 
    
−11 / 6
  L θ 
2
  l0 θ 
Σ sc (θ sc ) ≈ cos 2 θ sc 1 +  2 sin sc   exp − 2 sin sc .
0
(5.126)
  λ 0 2 
  
 λ0 2 

 
Examples of scattering diagrams calculated based on (5.126) for the Kolmogorov-Obukhov
model of the tropospheric turbulences is shown in Figure 5.32.

Figure 5.32. Scattering diagrams of the turbulent troposphere plotted in Cartesian


coordinates, based on (5.128) model for the fixed values of the inner ( l 0 ) and outer ( L 0 )

scales, and for different wavelengths (verifies the concept demonstrated in Figure 5.27)

321
5.6.5. PROPAGATION FACTOR ON SECONDARY TROPOSPHERIC
RADIO LINKS

In order to obtain the total power received at the point B the power flow density Π B form (5.107)
must be multiplied by antenna's effective aperture form (3.30), i.e.

PTx GTx G Rx λ 0 σ sc (θ sc )Vsc


2

PB = . (5.127)
(4π r1 r2 )2 4π
If this expression is compared with (3.38), then the propagation factor may be found as

r σ sc (θ sc )Vsc
F= , (5.128)
r1 r2 4π
where r1 and r2 are the distances from scattering point C and transmitting, A and receiving, B
points respectively, and r is the shortest distance between transmitting and receiving points;
Expression (5.128) may be rewritten for the total horizontal distance R between transmission
and reception points if assumed r1 ≈ r 2 ≈ R / 2 :

2
F= σ scVsc . (5.128a)
R π

The scattering volume Vsc is defined as intersection of main beams of the transmitting and

receiving antennas, as shown in Figure 5.33.


For simplicity we'll assume both diagrams symmetric relative to the main radiation direction, i.e.
half power beam widths (HPBW) are equal in both, E- and H-planes for each antenna:

γ 1E = γ 1H = γ 1 , (5.129)

γ 2E = γ 2H = γ 2 . (5.130)
Then by definition (3.13) given in chapter 3 antenna gains may be expressed in terms of beam
widths as
4π 4π
GTx ≈ and G Rx ≈ . (5.131)
γ1 2
γ 22
For maximum efficiency of the secondary tropospheric radio links the angles of elevation of both
antennas are chosen close to zero, i.e. the main beams of both antennas are directed tangential
to the earth’s surface, i.e. directed horizontally. That allows having a common scattering volume
Vsc to be as close to earth’s ground as possible: the reason is because the lower the elevation

322
of the scattering volume is, then smaller is the scattering angle, thus higher the intensity of
scatterings are. From Figure 5.33b
VS = Area MNLP ⋅ MM ′ , (5.132)

Figure 5.33. The definition of the scattering volume of troposphere.


a). The intercept of the radiation diagrams of Tx and Rx antennas, b). The detailed sketch
of the scattering volume. (note that the vertical scale is highly exaggerated for clarity)

b
where Area MNLP = d , and d = MM ′ . (5.133)
sin θ
If C is the center point of the scattering volume, then from Figure 5.33a the following
approximate expressions may be written down:
d ≈ AC ⋅ γ 1 = r1 γ 1 
 (5.134)
b ≈ BC ⋅ γ 2 = r2 γ 2 
It’s easy to realize that if the elevations of antennas main beams are close to horizontal then the
scattering angle θ sc is the same as the geo-central angle. For the real conditions the following

approximation may be applied:


R
sin θ sc ≈ θ sc ≈ , (5.135)
ae
where ae = 8500 km is the earth's equivalent radius, and R is the horizontal distance between

corresponding points. Another assumption is r1 ≈ r 2 ≈ R / 2 . If γ 1 and γ 2 are found from

(5.131), then, by combining (5.132) - (5.135), we find

323
R 2 ae (4π )3 / 2
Vsc ≈ . (5.136)
8 GT G R

Yet was assumed GT > G R , because the width d of the volume VS was defined based on the

beam width of the antenna with higher gain (transmitting antenna in this case). It's easy to

show, that for GT < G R denominator in (5.126) will become G R GT . In real conditions the

mean-geometric of those two values may be used instead:

GTx G Rx G Rx GTx = GTx


3/ 4 3/ 4
G Rx . (5.137)

Thus (5.136) may be rewritten as

R 2 ae (4π )3 / 2
Vsc ≈ . (5.138)
8 GTx 3 / 4 G Rx 3 / 4
This expression may easily be transformed to

( R γ ) 3 R 3 (4π ) 3 / 2
Vsc ≈ ≈ , (5.138a)
8θ sc 8θ sc G 3 / 2
where (5.135) is taken into account as well as the similarity between corresponding antennas,
i.e. G = GTx ≈ G Rx ≈ 4π / γ 2 .

If (5.138) is substituted into (5.128a), then the following formula may be found for the power
propagation factor
σ sc (θ sc )
F 2 ≈ 4 π ae , (5.139)
∆G
∆G = GTx
3/ 4 3/ 4
where G Rx . (5.140)

Now before we refer to σ sc (θ sc ) to make expression (5.139) applicable for practical estimates

we have to take into account the following statements:


• The path length of the troposcatter links is practically limited from 200 km (for relatively
wideband systems with the bandwidth up to several MHz) to 1000 km (for narrowband
systems with the bandwidth no more tens of kHz)
• The carrying frequencies are limited from 200 MHz (due to antenna size limitations) to 5
GHz (due to increasing of the atmospheric absorptions)
It’s easy to estimate that for the given distances the geo-central angle θ sc = R / ae along the

great circle is limited to 0.0235 < θ sc < 0.1177 radians. Then the magnitude of scattering vector

from expression (5.102), q = q = 2 k 0 sin θ sc / 2 will be ranged within 0.0986 m-1 < q < 12.3 m-1

324
limits. If those limits are compared with the limits ( q min ≤ 0.063 m-1, q max ≥ 628 m-1) of the energy

conservation range (range-2 on Figure 5.30), then easy to realize that only those irregularities
are responsible for the scattering, which are located in range-2 only. Therefore for further
evaluation of (5.139) for the power propagation factor the original version (5.117) of the
spectrum of turbulences may be employed, without counting on Carman’s corrections made to
adjust both ends of the spectrum. After combining (5.108), (5.117), and (5.139) we obtain

4 π ae
F2 ≈ 2π 2 k 04 cos 2 θ sc 0.033 Cε2 q −11 / 3 = ...
∆G
−11 / 3
39.26 ⋅10 6 4  θ 
... = k 0 cos 2 θ sc Cε2  2k 0 sin sc  . (5.141)
∆G  2 
The following notations are used in (5.141):
• ∆G is defined by (5.142) as unitless
• k 0 is a free space wave number in 1/m

• ae = 8.5 ⋅10 6 m is equivalent earth’s radius

• θ sc is a scattering angle defined by (5.135)

• Cε2 is a structural constant for atmospheric turbulences in m-2/3.


-------------------------------------------------------------------------------------------------
Example 5.2

Estimate the propagation factor on the tropospheric radio link for the frequency 1 GHz (free
space wavelength, λ 0 = 30 cm, wave number, k 0 = 2π / λ = 20.944 , 1/m), and horizontal

distance of R = 300 km between transmitting and receiving antennas with the gains
GTx = G Rx = 50 dB ( 10 5 unitless). The value of the structural constant in vicinity of the earth

surface, at the height h 0 = 30 m, may approximately be taken: Cε = 10 −14 cm-2/3 for summer
2

season and Cε = 10 −16 cm-2/3 for winter season (see Figure 5.31). Find the power received, if
2

the radiated power is PTx = 2 kW.

Solution
• Scattering angle is calculated taking it equal to geo-central angle,
θ = r / aeq = 300 / 8500 = 0.0353 rad ,
• The height of scattering volume is found from triangle AOC on Figure E5.1:

325
Figure E5.1. Definition of the scattering volume elevation above the ground

 1   1 
hC = ae  − 1 = 8500  − 1 ≈ 5.3 km ,
 cos(θ / 2)   cos (0.0353 / 2 ) 
• We define the structural constant at the elevation hC from (5.118) (or roughly estimated

from Figure 5.31):


2
 Cε (5.3 km) = 10-16(5300/30)-2/3 = 3.18x10-18 cm-2/3 = 6.85x10-17 m-2/3 (for winter

season)
2
 Cε (5.3 km) = 10-14(5300/30)-4/3 = 10-17 cm-2/3 = 2.15x10-16 m-2/3 (for summer

season)
 Note: here we ignore seasonal variations of the smooth tropospheric refraction
that may result in variations of the common volume elevation above the ground;
standard tropospheric refraction is assumed here
• Power propagation factor is calculated from (5.143):
 for winter seasons
−11 / 3
39.26 ⋅10 6  0.0353 
F2 = 20.944 4 cos 2 0.0353 ⋅ 6.85 ⋅10 −17  2 ⋅ 20.944 ⋅ sin  = ...
5 3/ 2
(10 )  2 
... = 4.95 ⋅10 −11 (-103 dB) (Answer)
 for summer seasons
−11 / 3
39.26 ⋅10 6  0.0353 
F =
2
20.944 4 cos 2 0.0353 ⋅ 2.15 ⋅10 −16  2 ⋅ 20.944 ⋅ sin  = ...
5 3/ 2
(10 )  2 

... = 1.55 ⋅10 −10 (-98.1 dB) (Answer)


• Power received on reference propagation path (free space) is found from (3.36):

326
PTx GTx G Rx λ 20 2000 ⋅10 5 ⋅10 5 ⋅ 0.32
PRx , 0 = = = 0.1267 W (126.7 mW)
(4π R )2 (4π ⋅ 300000) 2

• Power received on real propagation path is found as PR = PR , 0 ⋅ F 2 :

 for winter seasons, PRx = 6.27 ⋅10 −12 W (-112 dBW) (Answer)

 for summer seasons, PRx = 1.96 ⋅10 −11 W (-107 dBW) (Answer)

--------------------------------------------------------------------------------------------

The above considerations of the propagation factor and power received on troposcatter radio
links allow make only the rough assessments of their median values. Unfortunately more
accurate calculations may not be performed due to lack of complete and precise set of data
about the parameters of the atmospheric turbulences: we do mean the data of structural
constant global distributions, and its spatial and time variations.
For more accurate results the reader may be referred to ITU-R document [12], which provides
an empirical calculation method based on numerous of observation on existing troposcatter
radio links, but it is not based on analysis of physical mechanisms of propagation.

5.6.6. THE SPECIFICS OF THE SECONDARY TROPOSPHERIC RADIO


LINKS PERFORMANCE

5.6.6.1. ANTENNAS GAIN EFFECT ON LINK PERFORMANCE

As seen form (5.141) the propagation factor on secondary tropospheric radio links is highly
dependent on antenna gains GTx and G Rx . The propagation factor is getting lower by ∆G when

either GTx or G Rx (or both) increase unlike other types of radio links, where propagation factor is

independent on antennas. In other words the higher antennas performance, the greater the
propagation factor is for all other conditions remaining.
The cause of this phenomenon is quite clear: the increase of the antenna gains will result in
decrease of the beam widths γ 1 and γ 2 , thus, regarding (5.138) and (5.138a) will result in

decrease of the scattering volume VS as demonstrated in Figure 5.33.

If (5.140) is expressed in dB-s

327
∆GdB =
3
4
(GTx, dB + GRx, dB ) . (5.142)

then, as follows from the plot of (5.142) shown in Figure 5.34, the loss of the antennas' gains (in
dB-s) as a function of total antenna gain is linearly increasing with the slope of 3/4.

Figure 5.34. Antennas' total gain losses on secondary tropospheric radio-link

In reality ∆G becomes considerable when the total gain exceeds 65 -- 70 dB, as may be seen
from Figure 5.34.
Another reasonable explanation of this phenomenon is as follows: due to turbulences, scattered
field is a subject of intensive amplitude and phase fluctuations at the reception point along the
antenna’s aperture; correlation distance of those fluctuations across the propagation path is
comparable to the antennas dimensions; if the antenna’s dimension is less than that correlation
distance, then within the antenna’s aperture received field is correlated, i.e. its structure
becomes close to the structure of plane wave; if the antenna’s dimension is greater than
correlation distance, then uncorrelated fluctuation along the receiving aperture (especially phase
fluctuations) result in destructive integration of the field values, hence resulting in destruction of
the antenna gain; it is clear that when transmitting and receiving antennas are not identical, then
antenna with larger dimension suffers more than that of smaller dimension. Total antennas gain
of GTx + G Rx = 90 --100 dB seems to be the limit, which is meaningless to exceed; further

investments are not paid off. Hence the value for each antenna gain is usually taken not larger
than 45--50 dB.

328
5.6.6.2. SIGNAL LEVEL FLUCTUATIONS AT THE RECEIVING POINT (FADING)

One of the specifics of troposcatter radio link is in presence of intensive random fluctuations of
the signal level at the receiving point, which may be classified into three independent
categories:
• Fast fading,
• Slow fading,
• Seasonal variations.
As pointed above, the cause of fast fading is a random interference of the multitude of
secondary waves, interfering at the receiving point 1. On this type of radio-links the multitude of
secondary, interfering partial waves may be considered as almost ideal Rayleigh ensemble,
thus the fast fading is considerably deep, and have the Rayleigh-type probability distribution of
the field strength (see chapter 6 for details).
Slow fading on troposcatter radio-links is described by lognormal probability distribution with fair
accuracy (see subsection 6.1.3.b for more details). Standard deviation of the median signal
level, σ y depends on propagation path length and season of the year as shown in Figure 5.35.

Figure 5.35. Dependence of the standard deviation of slow fading of the received field
level on distance for troposcatter radio link

To understand the reason for the decease in intensity in slow fades distance-dependence recall
that greater the distance is, higher is the elevation of the scattering volume VS , hence more

1
Detailed analysis of the random interference (multipath interference) is given in Chapter 6, along with
the references for terms and definitions.

329
stable the atmospheric conditions are. Those conditions are less dependent on seasonal
variations at high elevations, rather than at low elevations.
The seasonal dependence of those variations may become clear if taken into account the fact,
that during winter season the earth's surface is heated less, therefore less is the vertical
gradient of the atmospheric refractivity, dN / dh ; regarding (5.25) the curvature of ray trace due
to regular smooth tropospheric refraction is less, hence higher is position of the scattering
volume VS . Higher is the elevation of scattering volume, less is the impact of the seasonal

variations. In spite the intensity of slow fades is less during the winter season, however the
short-term median signal at the reception point is always less than in summer. Therefore it is
common practice to design such radio-links for the winter season that guaranties the
performance all year long. On the other hand in terms of stability of the signal level at the
receiving point, the worst conditions arise during the summer season, because of increasing of
deepness of fading (both, fast and slow). Therefore the estimates for the stability of those radio
links is carried out preferably for the summer seasons, based on approaches given in chapter 6,
and the values of σ y obtained from Figure 5.35.

One of the specific features of troposcatter communication systems is the use of diversity
receiving principles that allows struggling against very deep fading. Note that the above
approaches allow estimating the long-term median received power only. If the sensitivity of the
receiver is taken equal to that median power level, then apparently the designer may not expect
the overall performance quality (steadiness) of the system greater than 50% by definition. To
reach the required performance quality, the increase of the radiated power, i.e. the imbedding of
power margin is not the solution for such systems because of the limitations in power budgets.
The only means of achieving the required performance steadiness is a combination of the
power margin with diversity reception techniques that are widely used in troposcatter
communication lines. Those techniques are considered in subsection 6.3.3 in details.

5.6.6.3. LIMITATIONS TO SIGNAL TRANSMISSION BANDWIDTH

Consider an amplitude-modulated signal with the carrier frequency f 0 and simple double-

sideband sinusoidal modulation. The waveform has a periodical shape with the amplitude
envelope of the sinusoidal form. If Fmod is modulating frequency, then the frequency bandwidth

is ∆ f = 2Fmod , hence the spectrum is located in frequency range from f min = f 0 − Fmod to

f max = f 0 + Fmod .

330
To analyze transmission of this signal through the troposcatter propagation path we refer to
Figure 5.33 in order to define the phase shifts between spectral components f min and f max .

Two extreme rays scattered from the highest and lowest points of the scattering volume VS ,

ANB and APB have the difference in distances, which may roughly be estimated as

R2 γ
∆ r = APB − ANB ≈ R γ θ = , (5.143)
ae
for γ 1 = γ 2 = γ and equal distances r1 ≈ r2 = R from transmission/reception point to the center

of the common volume Vsc along the great circle. Expression (5.143) may be rewritten, taking

into account (5.131):

R2
∆r ≈ 4π / G , (5.144)
ae
where ae = 8500 km is the earth’s equivalent radius.

The phase shift between two rays passed through different extreme paths for the lowest
spectral harmonic is
2π f min 2π ( f 0 − Fmod )
∆ϕ min = ⋅∆r = ⋅∆r , (5.145)
c c
whereas for the highest spectral harmonic it is equal to
2π f min 2π ( f 0 + Fmod )
∆ϕ max = ⋅∆r = ⋅∆r , (5.146)
c c
where c = 3⋅10 8 m/s is a speed of light in free space.
As noted in subsection 4.1.3 the phase distortions of the signal may be neglected if the
maximum difference in phase shifts between above extreme rays and extreme spectral
harmonics is less than π / 2 , i.e.
2 π Fmod π
δϕ = ∆ϕ max − ∆ϕ min = 2 ∆r ≤ . (5.147)
c 2
Then the limit for signal transmission bandwidth may be defined by substitution of (5.144) into
(5.147)

c c ae G
∆ f = 2 Fmod ≤ ≈ . (5.148)
4∆ r 4 R2 4π
An example below demonstrates, in particular, how the signal transmission frequency band may
depend on troposcatter propagation path length.

331
----------------------------------------------------------------------------------------------
Example 5.3

Calculate and plot the distance dependence of the maximum signal transmission bandwidth for
the range of distances from 200 to 1000 km. Assume identical antennas of the gain G = 47 dB
(G = 50000) for both transmitting and receiving stations.
Solution
Based on (5.142) the calculation results are presented in Table E5.3 below and plotted in Figure
E5.3.
Table E5.3
R, km 100 200 300 400 500 600 700 800 900 1000
∆f, kHz 4021.2 1005.3 446.8 251.3 160.8 111.7 82.1 62.8 49.6 40.2

Figure E5.3. Example of the distance dependence of maximum bandwidth


for troposcatter signal transmission

---------------------------------------------------------------------------------------------

From the considered example it may be seen that the bandwidth of the secondary-tropospheric
transmitting system is strictly limited and cannot exceed several megahertz. From (5.148) it may
be concluded, that some improvement of the bandwidth can be achieved by increasing of the
antennas gains. However the gain limitations are to be applies. Unfortunately the restrictions do
not allow transmission of the video signals or high speed digital data.

332
5.7. ATTENUATION OF THE RADIO WAVES IN THE
ATMOSPHERE

5.7.1. ATTENUATIONS IN TROPOSPHERE


Propagation of the radio wave through the atmospheric air is accompanied by the losses due to
transfer of the energy of the wave into the thermal movements of atoms and molecules of the
composites: gases and hydrometeors. Those composites are considered to be uniformly
distributed along the propagation path, so thus the attenuation coefficient in (2.90) is counted as
a sum
α = αg +αh , (5.149)

where α g and α h are the attenuations per unit distance (attenuation coefficients 1) in

atmospheric gases, and hydrometeors respectively. The most significant contributions into
gaseous component of the attenuation coefficient, namely α g occurs in molecular oxygen (O2),

called “dry air specific attenuation”, and in water vapors (H2O), i.e. the total attenuation
coefficient due to losses in atmospheric gases in dB/km is
α g ≈ αO + α H O .
2 2
(5.150)

Frequency dependencies of α O2 and α H 2O are shown on graphs in Figure 5.36; it’s seen that

both terms in (5.150) are highly frequency-sensitive, called therefore frequency-selective


absorptions [13].

1
Cited in [13] as specific attenuation.

333
Figure 5.36. Attenuation coefficient of sea-level atmosphere due to molecular oxygen and
water vapour for standard atmosphere: air pressure, 1013 mbar, temperature, 15 0C,
water vapour density, 7.5 g/m3 (for α H 2O curve only).

From the given figure one may realize, that the attenuation of energy of radio waves becomes
considerable for the frequencies higher than 5 GHz. For the frequencies less than 5 GHz those
attenuations can be neglected for any type of radio links including terrestrial, as well as earth-to-
space and space-to-earth. Considering the frequency range of f ≥ 5 GHz, the value of α g

increases with the significant jumps of attenuations caused by mechanical resonances in


molecular structures of O2 and H2O as seen from Figure 5.36. Apparently the use of those
frequencies is meaningless. From the same figure it may be seen that there're also so called
transparency windows with relatively low attenuations; only those windows are acceptable for
the radio links design. Particularly one of the windows is located in a frequency range between
22 GHz and 60 GHz with the minimum value of specific attenuation factor of approximately
α g ≈ 0.15 dB/km at f = 35 GHz.
Total attenuation (losses in dB) on terrestrial radio links due to absorptions in atmospheric
gases may easily be found as
Lg = α g R , dB (5.151)

334
where R is a horizontal distance between corresponding antennas in kilometers. Note that for
the slant propagation paths, such as those for earth-to-space, space-to-earth, as well as for the
VHF, UHF, and microwave communication links between ground-based and airborne stations
expression (5.151) is not valid any more. The reason is uneven vertical distribution of the
atmospheric air temperature, pressure, and density of the water vapours. Last two factors have
most significant impact on the values of α O2 and α H 2O . For the standard atmospheric conditions

they both decrease exponentially. To obtain the total attenuation on slant propagation path the
product (5.151) is to be replaced by the (2.94) integral. If ∆ h is the difference in altitudes of the

communicating antennas, the equivalent height ∆ heq < ∆ h is introduced [13] for both, dry air

attenuations, ∆ heq , O2 and water vapours attenuations, ∆ heq , H 2O . It must be emphasized that the

air pressure decreases much slower than water vapour density, therefore ∆ heq , H 2O < ∆ heq , O2 .

Then for the zenith angle θ of the slant propagation path the value of the total attenuation may
be calculated based on secant law, i.e.
( )
Lg = α O2 ∆ heq , O2 + α H 2O ∆ heq , H 2O secθ . (5.152)

The values of α O2 and α H 2O are taken at the height of the lowest antenna. Figure 5.36 may be

used only for the rough graphical estimates of α g . For precise (line-by-line calculation method),

and approximate analytical calculations of α g , as well as calculation of ∆ heq , O2 and ∆ heq , H 2O the

ITU document [13] may be used.


The second term in (5.149), namely the attenuation coefficient due to hydrometeors, has two
composite parts:
α h = α prec + α fog , (5.153)

where α prec is a component, caused by precipitations (rain, snow, hail), and α fog is a

component, caused by clouds and fog. Note, that each hydrometeor may be considered as a
small particle (raindrops, ice-balls, or snowflakes) with the semi-conducting properties. The
propagating radio wave excites a displacement and/or conducting currents in that particle.
Regarding (2.44) the intensity of displacement currents is in direct proportion to dielectric
constant of propagation medium. The density of these currents within each raindrop is
significant because the dielectric constant of water is 80 times greater than in air. On the other
hand they are proportional to the frequency; hence they become heavier at the higher
frequency. From the same expression (2.44) one may realize that the displacement currents do

335
not result in dissipation of energy of the radio wave into heat within the particle, because they’re
shifted by 900 relative to electric field. However they result in decay of energy due to scatterings
similar to scatterings on tropospheric turbulences: the incident radio wave turns each particle
(raindrop, snowflake) into the elementary radiator with wide radiation pattern, thus the energy of
directed radiation transforms into the radiation widely spread into surrounding space. This
scattering mechanism may have a significant impact on loss of energy especially if the
wavelength is comparable to the size of particles. Those attenuations may be neglected for the
frequencies less than 1 GHz, whereas they become considerable in microwave bands.
An additional attenuation in hydrometeors is caused by conducting currents, which turn
irreversibly part of the energy of radio wave into heat that dissipates within the particle.
The frequency-dependencies of composite parts of α h are shown in Figure 5.37 [3]

Figure 5.37. a). Frequency dependence of the attenuation coefficient, α h due to

hydrometeors (rain and fog): (1) Drizzling rain (0.25 mm/hr); (2) Light rain (1mm/hr); (3)
Moderate rain (4 mm/hr); (4) Heavy rain (15 mm/hr); (5) Light fog (0.03 g/m3 – 600 m
visibility); (6) Medium fog (0.3 g/m3 – 120 m visibility); (7) Dense fog (2.3 g/m3 – 30 m
visibility).
b). Propagation path range within precipitation area for earth-to-space communication
link

Graphs on Figure 5.37 may be used for rough, approximate references only. For precise
calculations the reader may be referred to [14,15]. Note that for the dry snow and hail α prec

336
becomes almost 25 times less than that for the same intensity of the rain at the same frequency.
This statement may be taken into account when similar calculations are performed for snow and
hail based on the data from Figure 5.37a. However, the wet snow results in almost the same
α prec as a rain of the same intensity. Total loss due to precipitations on earth-to-space

communication path of length ∆ prec depends on the zenith angle θ (see Figure 5.37b) as

follows:
hclouds
L prec = α prec ∆ prec =α prec . (5.154)
cosθ
More complete data including geographical and seasonal statistical distributions may be found
in [14].

--------------------------------------------------------------------------------------------
Example 5.4

Calculate approximate diameter of parabolic dish antenna onboard geostationary satellite for
the space-to-earth communication link for the initial data outlined below. Assume receiving
station positioned in under-the-satellite point ( θ = 0 ).
• The wavelength λ 0 = 7.5 cm ( f = 4 GHz),
• Power radiated PT = 30 W (14.8 dB/W),

• Minimum power received PR = 1.6 ⋅ 10 −11 W = 16 pW (-108 dB/W),


• Receiving antenna gain 50 dB ,
• The distance r = 36000 km , (distance to geostationary
orbit)
• Intensity of the rain 15 mm/hr ,
• Visibility in foggy condition 120 m,
• Height of the precipitation ceiling 1 km
• Losses in antenna feeding lines, and total losses in ionosphere ignored
• Consider for simplicity the equivalent path lengths:
∆ heq , O2 = 7 km, and ∆ heq , H 2O = 1.5 km
Solution
1. Specific attenuation (attenuation coefficient) is found from Figure 5.36
approximately as: α O2 = 0.007 dB/km, α H 2O = 0.001 dB/km

337
2. Total attenuation loss, due to atmospheric gases found from (5.152):
Lg = (0.007 ⋅ 7 + 0.001 ⋅1.5) sec 0 = 0.0505 dB
3. attenuation coefficient due to rain and fog defined from Figure 5.37a:
for rain α prec ≈ 0.015 dB/km

for fog α fog ≈ 0.02 dB/km


4. As a worst-case scenario we consider the same path length for both, rain and
fog. Then the total loss due to attenuations in hydrometeors:
Lh ≈ ( 0.015 + 0.02) ⋅1 / cos 0 = 0.035 dB
5. Total losses due to attenuations in troposphere (dry air and hydrometeors):
LTrop = L g + L h = 0.0505+0.035 ≈ 0.1 dB
6. From the expression (3.38a) the satellite-based transmitting antennas gain:
GT = PR , dB − PT , dB − G R + L0 + Ladd =

= −108 − 14.8 − 50 + 0.1 + 195.6 = 22.9 dB ( ≈ 195 unitless) ,


2
 4π r 
 = 10 log  4π ⋅ 36,000,000  = 195.6 dB is a free-space
2

where L0 = 10 log 
 λ0   0.075 
 
loss
7. Now we refer to (3.30) and (3.31) and take ν ≈ 0.6 . Then the geometrical
aperture of the satellite transmitting antenna may be calculated as:

GT λ 0
2
195.6 ⋅ 0.075 2 π D2
S= = ≈ 0.146 m2 = ,
4π ν 4 ⋅ 3.14 ⋅ 0.6 4
where D is the diameter of the satellite transmitting antenna
8. Finally, form the previous expression

4S 4 ⋅ 0.146
D= = = 0.43 m (Answer)
π 3.14
-----------------------------------------------------------------------------------------

5.7.2. ATTENUATIONS IN IONOSPHERE


Ionosphere is considered as a low-loss dielectric with the attenuation coefficient defined by
(2.85) for tan δ << 1 condition:

338
π ε tan δ
α≈ ε tan δ = ω ε 0 µ 0 , (5.155)
λ0 2

where 2π / λ 0 = k 0 = ω ε 0 µ 0 is a free space wave number and ε is a real part of the

10 −9
dielectric constant. If tan δ is substituted from (2.7), along with ε 0 = F/m, and
36 π

µ 0 = 4π ⋅ 10 −7 H/m, then (5.155) may be rewritten as


60 π σ
α= . (5.156)
ε
Now we estimate the value of α for one of the ionospheric layers (e.g. for E-layer) and for any
fixed frequency (e.g. f = 1 MHz). Using (5.8) and (5.9), as well as the data from the Table 5.1
first it may be seen, that the following assumptions may be undertaken without considerable
error: ε ≈ 1 , and for ω 2 >> ξ 2 . Then (5.9) may be rewritten as

2.8 ξ N elec / cm3 ξ N elec / cm


σ≈ ⋅10 −2 ≈ 7.1 ⋅10 −16 , S/m,
3
(5.157)
(2π ) 2
f Hz
2
f MHz
2

where N elec / cm3 is a number of free electrons per cubic centimeter, and ξ is a number of free

electrons collisions per second. The substitution of (5.157) into (5.156) will result in
ξ N elec / cm
α ≈ 1.34 10 −13 , Np/m.
3

2
(5.158)
f MHz
For the simple estimates we may assume the values ξ and N e remaining constant within the

absorbing layer. Particularly for the E-layer the number of collisions is taken form Table 5.1 as
ξ ≈ 10 5 1/sec, then (5.149) can be expressed as
2
f c E , MHz
α E ≈ 1.66 2
⋅10 −4 , Np/m , (5.159)
f MHz

where f c E , MHz = 80.0 N elec / cm3 ⋅10 −3 (5.160)

is the plasma frequency (otherwise called critical or Langmuir frequency) of the ionospheric E-
layer in MHz. Because the value of α E is taken approximately constant within the absorbing
layer, then the total losses in that particular layer can be estimated based on (2.90) (instead of
(2.94)). The distance z, covered by the radio wave within the absorbing layer may be estimated
based on geometric sketch given in figure 5.38a.

339
Figure 5.38. The reflection pattern of the radio wave from F2 Ionospheric layer in
presence of the absorbing layers D, E and F1 layers

z = ∆ hE sec ϕ E , (5.161)

where ∆hE is a thickness of the E-layer, and ϕ E is the angle of radio wave incidence on E-
layer. By substitutions of (5.150) and (5.152) as well as the average thickness of the E-layer
∆ hE = 30 km into (2.90), a formula for total attenuation of the radio wave in E-layer may be
presented as

AE
AE = 2
, (5.162)
f MHz

AE ≈ 5 f c , E , MHz sec ϕ E
2
where (5.163)

is the attenuation for the carrier frequency of 1 MHz (unit is: Np·MHz2).
The expressions (5.162) – (5.163) may be generalized for the other absorbing layers such as D-
and F1-layer. To be more accurate it must be recalled, that those expressions have been
developed without taking into account the following factors for simplicity:
• Within each layer the value of α doesn't remain constant due to variations of ξ (h ) and

N e (h ) (see section 1.3). It results in changes of α from point-to-point along the ray
path, thus (2.90) may not provide enough accuracy; (2.94) must be used instead
• The earth's magnetic field has a proper impact on the value of attenuation, therefore it
must be taken into account as well
The exact evaluations provide the following expression for the total attenuation, consolidating all
three absorbing layers

340


AΣ = , (5.164)
(f + fL )
2

where
′ ′ ′ ′
AΣ = AD + AE + AF 1 = f c , E , MHz (3 sec ϕ D + 2.5 sec ϕ E + 0.4 sec ϕ F 1 )
2
(5.165)

is for the day-time propagation, and f L is defined from (5.52), (5.66) and (5.71) as

f L = f H cos γ , (5.166)

where f H ≈ 1.4 MHz, and γ is the angle between the direction of propagation and the
geomagnetic meridian (see Figure 5.18), which is almost the same as the direction of the earth's
′ ′ ′
magnetic field, especially for the low and middle latitudes.. The fact that AD , AE and AF 1 are

defined by only one parameter, f c , E , MHz is because the critical frequencies of all three

ionospheric layers, f c , D , f c , E and f c , F 1 are dependent on just one single factor, namely on

the solar activity. In order to simplify the engineering approaches it's reasonable to use just one
of those three parameters, e.g. f c , E , keeping in mind that all three attenuations are bounded


originally with just one source. The family of curves for the total attenuation AΣ f c , E as a ( )
function of the length of a single-hop propagation distance (horizontal distance along the earth’s
surface), R h is presented in Figure 5.39.

Figure 5.39. Total nonreflecting attenuation of the radio wave of the frequency 1 MHz in
ionosphere as a function of the E-layer’s plasma frequency f c E , MHz

341

As seen form (5.165) the increase of single-hop distance results in increase of AΣ due to

increase of the angles of incident ϕ D , ϕ E and ϕ F 1 .

Besides the nonreflecting attenuation, some attenuation takes place while reflection form F2 -
layer, which may be calculated by using the following formula

AF 2 = BF 2 f MHz ,
2
(5.167)

where BF 2 may be defined from Figure 5.40 for the given distance and effective height of F2

layer, heff shown in Figure 5.38b.

Figure 5.40. Attenuation of the radio wave in reflecting, F2 ionospheric layer for the

frequency of 1 MHz as a function of single-hop distance, R h and the effective reflection

height, heff

The decaying shape of the family of attenuation curves on Figure 5.40 is due to the increase of
the angle of incidence, ϕ F 2 when the horizontal distance Rh is increased, and as a result

lowering the altitude of the reflection point C ′ , shown in Figure 5.38b. Lower is the height of the
reflection point, smaller the distance MC ′N will be covered by the propagation path, less
attenuation will occur. The height of the vertex C ′ of the track heff depends on the ratio of the

342
carrier frequency and the critical frequency of the reflecting layer, namely f / f c , F 2 , of

ionospheric MUF-predictions; otherwise it may be taken roughly heff ≈ 300 to 350 km. Those

advance monthly predictions have been provided in the US for several decades by CRPL
(Central Radio Propagation Laboratory) of the National Bureau of Standards (currently NIST,
National Institute of Standards and Technology). Similar governmental services exist in other
countries. Ionospheric predictions are in strong correlation with the index of solar activity
measured in so called Wolf number-s (also known as Zürich number-s), which is highly
predictable for many years in advance, taking into account the 11-year solar activity cycle.
The total attenuations from all layers, nonreflecting and reflecting, is defined for the day-time as
A = AΣ + AF 2 . (5.168)

Note finally that sometimes the reflections from the layers lower than F2 , such as E or F1 , are
possible (see Figure 5.11). The specific approaches must be applied in each particular case.
For instance if reflection takes place from the E-layer, then the total attenuation may be found
as follows

AD f
A≈ + BE f (5.169)
(f + fL )
2
f + fL

4
where BE ≈ cos 2 ϕ E (5.170)
f c , E , MHz
is a semi-empiric expression, which represents the reflecting attenuation at the frequency of 1
MHz. The first term represents the non-reflective absorption from D-layer only, whereas the
second term represents the reflective absorption within E-layer.
The median value of the field strength at the receiving point for the multi-hop propagation path
may be found from the following empirical expression:
~
1 + Γ ~ n −1
E B = E0 Γ exp (− A) , (5.171)
2
where
• E 0 is the free space field strength (3.34) for the radio wave, which covers a distance

counted along the earth's surface between radiation, A and reception, B points
• n is a number of hopes
• A is total attenuation in ionospheric layers, in nepers (Np), including both, reflective and
non-reflective absorptions. They’re defined either by (5.168) or (5.169)

343
( ~
)
• The multiplier 1 + Γ / 2 takes into account the impact of the earth surface at the
~
transmitting and receiving points for Γ as the average reflection coefficient 1 at those
points. For the practical calculations it may be taken approximately equal to 0.8.
Finally it is to be mentioned that the approaches considered in this subsection are mostly
applicable to HF communication lines assessments/designs. Those lines are still in use by
military and commercial users despite high competitiveness of the satellite communication
(satcom) systems; however HF communication systems are still used as substitutes or back-ups
for those highly developed satcom systems.

-------------------------------------------------------------------------------------------------------
Example 5.5

Determine which ionospheric layer reflects the radio wave on a single-hop HF communication
line ( n = 1 ) at the frequency f = 28 MHz, and estimate the RMS field strength at the reception

point if the horizontal propagation distance is R = 1800 km. The angle between the direction of
propagation and geomagnetic meridian is γ = 45 degrees. Power radiated by transmitter is

PT = 200 W with the antenna gain of GT = 20 dB (100 unitless). For ionospheric layers
parameters assume the following average values taken from Table 5.1 for a day-time:
• D-layer - N e = 5 ⋅10 3 electrons/cm3, hmax = 75 km (non-reflecting layer)

• E-layer - N e = 2.8 ⋅10 5 electrons/cm3, hmax = 120 km

• F1-layer - N e = 3.25 ⋅10 5 electrons/cm3, hmax = 210 km

• F2-layer - N e = 2 ⋅10 6 electrons/cm3, hmax = 315 km

Solution
1. Check the reflection conditions for each, E F1 and F2 layer: (a) find the reflection angles
from (5.43), (b) calculate plasma frequencies from (5.10a), (c) calculate MUF-s for the
given distance from (5.37), (d) check if the condition f < MUF is satisfied

• For E-layer – (a) ϕ = 1.369 rad (78.44 degrees), (b) f c = 4.76 MHz,

(c) MUF-E-1800=23.7 MHz, (d) f > MUF (this is non-reflecting layer)

• For F1-layer – (a) ϕ = 1.275 rad (73.04degrees), (b) f c = 5.12 MHz,

1 ~
R is not to be confused with the horizontal distance R .
344
(c) MUF-E-1800=17.56 MHz, (d) f > MUF (this is non-reflecting layer)

• For F2-layer – (a) ϕ = 1.1715 rad (67.12 degrees), (b) f c = 12.71 MHz,

(c) MUF-E-1800=32.7 MHz, (d) f < MUF (this is reflecting layer)


2. Find the angle of incidence on absorbing D- E- and F1-layer, considering F2 as a
reflecting layer. For the geo-central angle between communicating points along the
great circle θ = R / a = 1800 / 6370 = 0.2826 rad (16.19 degrees),
ψ = π − (ϕ + θ / 2) = 1.8288 rad (104.78 degrees, see Figure E5.5).

Figure E5.5. Reflecting F2 and absorbing layers of ionosphere: habs is the height of the

absorbing layer, ϕ abs is the angle of incidence of the radio wave onto absorbing layer

In triangle ADO, DO= a + h Abs . Then based on the law of sines

sinψ sin ϕ Abs  a 


= , hence ϕ Abs = sin −1  sinψ  (E5.5.1)
a + hAbs a  a + h Abs 
Angles of incidence for proper layers are calculated from (E5.5.1) as follows:
• D-layer - ϕ D = 1.27186 rad (72.872 degrees)

• E-layer - ϕ E = 1.25 rad (71.63 degrees)

• F1-layer - ϕ F 1 = 1.21 rad (69.4 degrees)

3. Calculating f L from expression (5.166): f L = f H cos γ = 1.4 cos 45 0 = 0.99 MHz


4. AΣ for non-reflecting layers is defined from(5.165)


AΣ = 4.76 2 (3 / cos1.272 + 2.5 / cos1.25 + 0.4 / cos1.21) = 436.22

345
5. Total non-reflecting absorption in D- E- and F1-layers is defined from (5.164):

AΣ 436.22
AΣ = = = 0.52 Np (4.51 dB)
(f + fL )
2
(28 + 0.99) 2

6. The reflective absorption in F2 layer is defined by (5.167) for BF 2 = 7 ⋅10 − 4 obtained from

Figure 5.40 approximately is: AF 2 = BF 2 f MHz = 7 ⋅10 − 4 ⋅ 28 2 = 0.55 Np (4.76 dB)


2

7. The total ionospheric absorptions from (5.168) are A = AΣ + AF 2 = 0.52 + 0.55 = 1.07 Np
(9.29 dB)
8. RMS value of the electric field strength in free space (reference conditions) is defined by

30 PT GT 30 ⋅ 200 ⋅100
expression (3.34) as: E0 = = = 4.4 ⋅10 − 4 V/m = 0.44 mV/m
R 1800 ⋅10 3

9. The final RMS value of the electric field strength is defined by (5.171) as:
~
1 + Γ ~ n −1 1 + 0.8
E = E0 Γ exp (− A) = 0.44 ⋅ ⋅ exp (−1.07) = 0.136 mV/m (Answer)
2 2
~
for Γ = 0.8 , and n = 1 .

---------------------------------------------------------------------------------------------------

REFERENCES

[1] Levis, C. A., Johnson, J.T., and Teixeira, F.L., Radiowave propagation: physics and
applications, John Wiley & Sons, Inc., 2010
[2] Staelin, D.H., Morgenthaler, A.W., and Kong, J.A., Electromagnetic waves, NJ: Prentice-Hall,
Inc., 1994.
[3] Dolukhanov, M. P., Propagation of Radio Waves, Moscow, USSR: Mir Publishers, 1971
[4] Rawer, K., Wave Propagation in the Ionosphere. Theory and Applications, Vol. 5, Springer
Verlag,1993
[5] Al'pert, Ya. L., Radio Wave Propagation and Ionosphere, Vol. I, The Ionosphere,
NY: Consultants Bureau, 1974
[6] Budden, K.G., The Propagation of Radio Waves. The Theory of Radio Waves of Low Power
in the Ionosphere and Magnetosphere, NY: Cambridge University Press, 1988.
[7] Davies, K., Ionospheric Radio Propagation, NBS, Washington, DC, 1965.

346
[8] Rytov, S.M., Kravtsov, Yu.A., and Tatarskii, V.I., Principles of Statistical Radiophysics,
Springer Verlag, 1989
[9] Collin, R.E., Antennas and Radio Wave Propagation, NY: McGraw-Hill Book Co.,1985
[10] Gradshtein, I.S., Ryzhik, I.M., Tables of Integrals, Series, and Products, Seventh Ed.,
Elsevier, Inc., 2007
[11] Booker, H.G., and Gordon, W.E., “Radio Scattering in the Troposphere,” Proc. IRE, Vol. 38,
April 1950, pp. 401-421
[12] ITU-R Recommendation P.617-1, “Propagation Prediction Techniques and Data Required
for the Design of Trans-Horizon Radio-Relay Systems,” International Telecommunication Union,
1992
[13] ITU-R Recommendation P.676-7, “Attenuation by atmospheric gases,” International
Telecommunication Union, 2007
[14] Crane, R.K., Electromagnetic Wave Propagation through Rain, Wiley, 1996
[15] ITU-R Recommendation P.838-3, “Specific attenuation model for rain for use in prediction
methods,” International Telecommunication Union, 2003

PROBLEMS

P5.1. LOS communication radio link has a transmitting antenna of horizontal polarization with
the gain of 15 dB. It is placed at the height of h1 = 40 m. Transmitter’s output power is PT = 5 W
at the frequency f = 3 GHz. The receiving antenna of the height h2 = 30 m is placed at the
distance of 20 km. Calculate the values of field strength of the received signal (in dB/ per mV/m)
and plot its dependence on gradient of refractivity of the troposphere within the following range:
− 0.157 < dN / dh < 0 . Assume the magnitude of reflection coefficient equal unity for the flat-
Earth approximation. Explain the behavior of the plot.
Note: Use (4.69) to calculate propagation factor. Replace H (0) by H (0) + ∆H to take into

account the tropospheric refraction, with ∆H from (5.29).

P5.2. A local TV station broadcasts on UHF channel-77 at the frequency of 850 MHz.
Transmitted signal has a horizontal polarization with omnidirectional EIRP = 35 dBW. Calculate

347
and compare field strengths at the receiving point without (E1) and with (E2) impact of the
standard (normal) tropospheric refraction for the distance of 20 km and antenna heights of h1 =
50 m (transmitting) and receiving, h2 = 10 m. Consider the ideal reflection from the earth’s
surface for simplicity.
Note: Use the approaches given in subsection 4.1.2. To take the atmospheric refraction into
account replace the real earth’s radius by equivalent.
Answer: E1 = 7.06 mV/m, E2 = 8.39 mV/m

P5.3. Confirm the values of equivalent earth’s radius ae and radius of ray curvature A provided

in Table 5.2 for the given values of the gradient of refractivity index dN / dh .

P5.4. Reconnaissance radar detected a target at the range of r 0 = 40 km and the elevation of

h = 5 km. Radar is calibrated for the standard atmosphere. However, the current condition of the
troposphere is such that ae = 2.5 a (relation between equivalent and real earth’s radii).

Calculate the error in elevation measurement ∆ h by assuming the measured range remaining
unchanged.
Answer: ∆ h = 43 m

P5.5. Estimate the values of scattering angles 2 θ max for the radio wave propagating in the

troposphere with the average size of the turbulent inhomogenieties Lε ≈ 15 cm for the following

frequencies: 0.5 GHz, 5 GHz, 50 GHz.


1
Hint: Assume q max ∝ for the rough estimate.

Answer: 74.40, 7.30, 0.720

P5.6. Repeat calculations from the Example 5.2 for the frequency 3 GHz and compare the
results.

P5.7. Calculate the frequency bandwidth of the tropo-scatter radio-link of the length R = 300 km
if transmitting and receiving antennas are identical with the beam width of the radiation pattern
γ = 2.5O. Does the regular tropospheric refraction impact the bandwidth? Explain.
Answer: 162.4 kHz

348
P5.8. Estimate the decrease of the level of the signal received from the TV broadcasting
satellite. The drop of the signal level (in dB) is due to rain of intensity 50 mm/hr, and the fog with
visibility of 100 m compared to “clear weather” condition. Frequency – 12 GHz. Assume the rain
and fog ceiling equal 0.8 km. The zenith angle of arrival of the radio wave from satellite to
reception point is 45O.
Answer: 2.34 dB (decreased 1.714 times)

P5.9. Calculate the angle of Faraday rotation for the radio wave propagating in magneto-active
ionosphere along the geomagnetic field line for the distance of 20 km if the carrier frequency is
30 MHz and the concentration of free electrons (ionospheric plasma concentration) is 106
el/cm3.
Answer: 145.4 degrees

P5.10. Calculate MUF and angle of incidence onto F2-layer for HF broadcast link of the length
6000 km, N e , max = 10 6 electrons/cm3, and hmax = 300 km. You may use the diagram on Figure

5.10 to verify your analytical estimate.


Note: 1. Assume that the propagation mode is based on the least number of ionospheric
reflections.
2. Ignore the impact of the earth’s magnetic field
Answer: MUF=29.5 MHz ϕ = 72.270

P5.11. The HF communication link of the path length 2600 km uses reflections from F2 layer of
the ionosphere. The daily variations of electrons concentration, obtained from ionospheric
observations, are shown in Figure P5.11 provided below. Design the schedule of changes of the
carrier frequency, f carrier = 0.8 ⋅ MUF − F2 − 2600 , i.e. fill out the Table P5.11. Show your

calculations.

349
Figure P5.11

Table P5.11
Observation time MUF - MUF - fcarrier
Nel per cm3 MUF0 = fc
From To 4000 2600

1:00 7:00
7:00 17:00
17:00 1:00

Notes: 1. Assume the height of F2 layer hmax = 300 km.


2. Ignore the impact of earth’s magnetic field.

350
APPENDIX-7
VOLUMETRIC SPECTRUM FOR AUTOCORRELATION FUNCTION OF
STATISTICALLY HOMOGENEOUS AND ISOTROPIC RANDOM FIELD

For the statistically isotropic random field ∆ε (r ) the volumetric autocorrelation function

depends only on the magnitude of the distance between two correlating points, ρ = r2 − r1 i.e.

ψ ε ( r1 , r2 ) = ψ ε (ρ ) = ψ ε (ρ ) . (A7.1)

Assume the function ψ ε (ρ ) is given in spherical coordinates ( ρ , ζ , ϕ ) . First we define the

elementary volume dV in spherical coordinates as (see Figure A7.1b below)

dV = ρ 2 sin ζ ⋅ dρ ⋅ dϕ ⋅ dζ . (A7.2)

Figure A7.1. a). Positions of vectors q and ρ ; b). The pattern of elementary volume dV
in spherical coordinates.

Vector q of the spatial frequency is conventionally taken directed along negative Z-axes as
shown in Figure A7.1a. Then the scalar product q ⋅ ρ is needed for (5.95):

q ⋅ ρ = q ⋅ ρ ⋅ cos (π − ζ ) = − q ⋅ ρ ⋅ cos ζ . (A7.3)


Now substitute (A7.2) and (A7.3) into (5.95), taking into account statistical isotropy (A7.1).

351
S ε (q ) = S ε (q ) = ⋅ ∫ψ ε (ρ ) ⋅ exp (i q ρ cos ζ ) ⋅ ρ 2 sin ζ ⋅ dρ ⋅ dϕ ⋅ dζ .
1
(A7.4)
(2π ) 3
V

Variables in (A7.4) may be separated and volumetric integral may be written as


2π ∞
π 
⋅ ∫ dϕ ⋅∫ψ ε (ρ ) ⋅  ∫ exp ( i q ρ ⋅ cos ζ ) ⋅ sin ζ ⋅ dζ  ⋅ ρ 2 dρ .
1
S ε (q) = (A7.5)
(2π )3 0 0 0 
Now consider the brackets in last expression:
π π

∫ exp ( i qρ ⋅ cos ζ ) ⋅ sin ζ ⋅ dζ = −∫ exp (i qρ ⋅ cos ζ ) ⋅ d (cos ζ ) =


0 0

2 exp (i q ρ ) − exp (− i qρ )
⋅ sin (q ρ ) .
1 π 2
=− ⋅ exp(iqρ ⋅ cos ζ ) = ⋅ = (A7.6)
i qρ 0
qρ 2i qρ


If dϕ = 2π is substituted into (A7.5) along with (A7.6), then final expression for the spatial
0

spectrum of the isotropic, homogeneous field ψ ε (ρ ) may be obtained as



S ε (q ) = ⋅ ∫ψ ε (ρ ) ⋅ sin (qρ ) ⋅ ρ dρ ,
1
(A7.7)
2π 2 q 0

which is ready to be used with different ψ ε (ρ ) models.

352
Chapter 6. RECEIVING OF THE RADIO WAVES:
BASIC OUTLINES

6.1. MULTIPLICATIVE INTERFERENCES (SIGNAL FADES)

6.1.1. FLUCTUATION PROCESSES AND STABILITY OF RADIO LINKS


The filed strength of the radio wave at the reception point in the real conditions is counted based
on propagation factor F in (3.37), which is introduced in section 3.2 to accommodate the
reference path (ideal propagation track) to the real conditions. In other words the propagation
factor allows taking into account the impact of the earth's surface and the atmospheric effects.
The electro-magnetic parameters of the earth's surface and the atmospheric propagation
medium are not constant, but randomly fluctuating in time and space, therefore generally the
magnitude and initial phase of the complex propagation factor F , must be considered as
random variables in time domain at any particular receiving point. Those random time-
variations/fluctuations may be expressed as
F = F ( t )exp [ i Φ F ( t )] , (6.1)

where F ( t ) is a random magnitude of the propagation factor, and Φ F (t ) is its random phase.
Based on (3.37) for the RMS (effective) value of the electric field strength, the expression for
complex time-harmonic field strength, may be written in the following form
E (t ) = E m, 0 F exp(iω t ) = E m, 0 exp {i [ω t + Φ E ( t ) ] }F , (6.2)

where
E m , 0 = E m , 0 exp[i Φ E (t )] (6.3)

is the amplitude phasor of the signal in free space with the magnitude of

60 PT GT
E m, 0 = E0 2= , (6.4)
r
and E 0 is the effective (RMS) value of the reference field strength of the radio wave (compare

with (3.37)). The randomly fluctuating amplitude of the field strength may be defined from (6.1)
and (6.2) as a product
Em ( t ) = Em, 0 F ( t ) . (6.5)

353
From (6.5) one may notice that in real condition the field amplitude at the reception point
randomly fluctuates because of random variations of F (t ) . The value of amplitude E m , 0 for the

ideal propagation conditions is multiplied by a random factor that represents the interferences
caused by fluctuations on real propagation path. Therefore this type of interference is called
multiplicative interference or fading. In other words fading is considered as fluctuation of the
wanted signal caused by random changes of the signal propagation conditions 1.
Methods of the probability theory and statistics are employed for the quantitative description of
fading statistics. For numerical estimates the probability of any value E of the continuous
random variable E ′ is formally defined as a ratio of (1) the sum of the time periods, ∆ Tn , when

E ′ faded below E ( E ′ ≤ E ) to (2) the total observation time tTot (see Figure 6.1), i.e.

∑ ∆T n
∆T1 + ∆T2 + ∆T3 + ∆T4
P( E ) = n
= . (6.6)
tTot tTot

Figure 6.1. Random variations of effective field strength of the signal at the receiving
point in presence of multiplicative interference (fading).

1
Taking into account (6.1) and (6.2), the random total phase of the radio wave at the reception point is

Φ E ( t ) + Φ F (t ) . For the AM systems the initial phase Φ E ( t ) = Φ 0 is constant, i.e. doesn’t carry
information . In FM and PM systems Φ E (t ) carries the information, thus at the output of demodulator the
component Φ F (t ) appears as an additive interference (noise) (see section 6.2). Fortunately the rate of
fluctuations of Φ F (t ) is usually much slower, than the rate of time-variations of Φ E (t ) due to the
informative signal. Hence the power spectrum of this additive noise is limited to less than 10 Hz and may
be easily cleared off with the high-pass filter.

354
As one may see form Figure 6.1, the greater the value of E, the larger the probability P (E ) is.
This functional dependence is called the cumulative distribution function (CDF).
Another important statistical measure of the random variable is probability density function
(PDF). Suppose ∆P(E ) is the probability of the field strength E ranged between E and

E + ∆E , i.e. in expression (6.6) the quantity ∑ ∆T


n
n is defined not as the cumulative time

interval, when the random field strength remains below the value E , but the cumulative time
interval, when it remains within E and E + ∆E limits. Then PDF is defined as:
∆P dP
w ( E ) = lim = = P ′(E ) , (6.7)
∆E →0 ∆E dE
i.e. it is the probability of the random variable E ′ (t ) remained within the unit interval that

surrounds the particular value E . From (6.7) it's easy to see that the relation between CDF and
PDF is
E
P( E ) = ∫ w ( E ) dE . (6.8)
0

Per definition, the cumulative probability, CDF may not exceed the unity, therefore w ( E ) must
satisfy the normalization condition, i.e.

∫ w ( E ) dE = 1 .
0
(6.9)

One of the useful statistical measures of the random E ′ (t ) process that is widely used in the

radio communications and radar engineering applications is the median value, E med defined

from the following equation


P( E med ) = 0.5 . (6.10)

As follows from (6.10) E med has a meaning of deterministic parameter of the random variable

E ′ (t ) that represents the value of E being exceeded with the probability of 0.5, i.e. being
exceeded within 50% of the total observation time. Another important parameter of the random
variable is the mean value E , called average value or statistical expectation defined from the
following averaging integral
tTot
1
E =
tTot ∫ E ′ (t ) dt
0
(6.10a)

355
The definitions (6.10) and (6.10a) are considerably different, so thus the quantities E and

E med are different in general. However, for the statistics those describe fading of the radio

waves, the values E and E med are fairly close to each other, so the difference may often be

ignored.
It’s easy to realize that neither E , nor E med allow specify the depth of fades. It is customary to

obtain the fade depth based on deciles E0.1 and E0.9 , which show the levels of observing signal

(usually in dB) that are exceeded within 10% and 90% of total observation time respectively 1. In
other words from the practical point as a measure of the fades depth the use of the difference
E0.1 − E0.9 (in dB) in RF propagation is more convenient than standard deviation that is widely
used in other probability and statistics applications.
In engineering practice the term channel reliability is introduced as a measure of quality of the
service: it is defined as percentage of the total serving time while the performance of the
communication system is not less than its minimum (threshold) 2; this threshold is usually
specified by the industry standards. Therefore, in practical applications another statistical
measure, namely complementary cumulative distribution function (CCDF) is introduced similar
to that given in subsection 4.2.3 for the complementary error function. CCDF is more commonly
used in communications and radars, rather than cumulative distribution function: CCDF
indicates the percentage of total observation time while random variable E ′ is greater or equal
than considered fixed value E . Hence, taking into account (6.8) and (6.9), the expression for
CCDF may be written as

CCDF = W ( E ) = 1 − P( E ) = ∫ w( E )dE . (6.11)
E

Within time intervals ∆t1 , ∆t 2 , and ∆t 3 of the total observation time, t Tot , shown in Figure 6.2

the fluctuating field E may fade down the minimum allowable value E min , which causes

communication outage (break). E min is the lowest electric field, which receiver is able to “sense”
in presence of additive noises in terms of its ability to extract the modulating signal from the

1
Note that E med = E0.5 thus E med may be considered as a decile.
2
In some references [16], the term channel reliability = 100 ⋅ (1 − t 0 / t tot ) = t a / t tot is used: t 0 is the total
outage (break) time, t tot is total observation time, and t a is total availability time.

356
carrier. In order to estimate channel reliability one may refer to Figure 6.2, where the short-
term 1 pattern of the field strength at the receiving point is presented.

Figure 6.2. Sketch of the complementary cumulative distribution function W(E) of the
fading field of the radio wave

From the above definition the quantitative value of channel reliability may be found based upon
CCDF as (see Figure 6.2)
W0 = W ( E min ) . (6.12)

It is equal to probability of the fluctuating field at the reception point being above its minimum
value, i.e. it is equal W (E ) for the value of E = E min . It may be realized, that higher is the

margin between E med and E min , greater is the channel reliability W0 . In other words for any

particular radio link the greater is ratio E med / E min , the higher is channel reliability. This ratio is

called fade margin that in most cases is expressed in dB-s, as follows:


fade margin = E med , dB − E min, dB .

On the other hand the probability of communication break is apparently equal to


T0 = 1 − W0 . (6.13)

1
Short-term is introduced to emphasize that during the total observation time-period tTot the median
value of electric field remains constant, i.e. this random process remains stationary. Therefore we call it
interval of stationarity, t Stat . For different types of radio links t Stat has different values. For instance for
HF-links t Stat is equal approximately several to tens of minutes.

357
Generally all above statements are applicable not just to randomly fluctuating E (t ) , but also to
Signal-to-Noise Ratio (SNR), which actually behaves as a random variable with the same
statistical properties as for E (t ) , if assumed the noise intensity unchanged within the
observation time interval. Channel reliability standards for most of the service types are defined
in terms of SNR rather than in terms of E (t ) , i.e. the quantities such as SNRmed , SNRmin , fade

margin for SNR are commonly used in radar and communications engineering.
If, instead of short-term observation, the long-term observations (hours, days, months, etc.) are
conducted, then in predominant propagation cases the median field strength, as well as other
statistical parameters, do not remain constant, but also are changing randomly as shown in
Figure 6.3. This means that generally the fading is a complex, non-stationary random process,
i.e. the process that have non-stable statistical parameters, namely average, median, standard
deviation, etc., which are also randomly fluctuating. It was noticed from numerous observations,
that in real conditions any long-time interval may be divided into a several short-time intervals so
that the statistical parameters within each of those short-time intervals may be considered as
being almost constant, i.e. each of those short-time intervals may practically be considered as
intervals of stationarity, t stat (see Figure 6.3).

Figure 6.3. Random variations of the effective field strength at the receiving point for
long term observation interval

From the above consideration one may conclude, that two types of fading persist in radio links:
fast and slow. Fast fading appears as random variations within the stationarity intervals,
whereas slow fading is considered as long-term, slow variations of the median value of the field
strength, which usually appears together with variations of other statistical parameters, such as
standard deviation and mean value.

358
The natures of fast and slow fades are totally different and independent. The cause of fast fades
relates to multi-path propagation. A typical example is tropospheric scattered propagation
considered in section 5.6. For multi-path propagation different independent rays approach the
reception point by passing different paths with random propagation distances (∆ r ) , causing so

called random interference. Those fluctuations of ∆ r are the result of fluctuating positions of
scattering irregularities of the tropospheric "clear air" on tropo-scatter radio links (Figure 6.4a),
or by random changes of the reflection points and reflection heights on HF-ionospheric radio-
links (Figure 6.4b). Hence the phase shifts between different rays ∆Φ = (2π / λ ) ∆ r becomes a

random variable. It’s "sensitive" to both: fluctuations of paths differences ∆ r , and the

wavelength. Smaller is the wavelength λ , the larger the fluctuations of phase shifts are, and
therefore higher is the fast fading rate.

Figure 6.4. The multi-path propagation mechanisms on tropospheric (a) and


ionospheric (b) radio links

When arriving the destination point B with random phase shifts within wide range [0, 2π]
different rays are superimposing differently. The extreme cases are: Equal-phase superposition
causing constructive interference, and opposite-phase superposition causing destructive
interference. The random interference is typical for most of RF links. The greater the number of
superimposing rays, the deeper the fading is, and more considerable is its impact on
communication stability.
Slow fading of the RW is caused by random changes of attenuations along the radio
propagation path, which are considered as a result of slow changes in the electromagnetic
parameters of earth's surface and atmosphere, due to the random variations of temperature,
humidity, air pressure, ionization, etc.
From the above one may realize, that fast and slow fading are two independent random
processes acting simultaneously within antenna-medium-antenna propagation path. Therefore

359
statistical distributions are expected to be composite functions, combining fast and slow fade
statistics. This issue will be considered in section 6.1.4.

6.1.2. FAST FADING STATISTICAL DISTRIBUTIONS


Here we will start with simple case of two-ray random interference, continue to the case of multi-
ray superposition of the independent wavelets (Rayleigh ensemble), and then further consider
more complex cases in subsection 6.1.2d to provide generalized approaches.

6.1.2.a. Two-Ray Random Interference

Consider a random interference of two rays, i.e. a superposition of two sine waves with the
constant, equal amplitudes E max , same spatial orientation of electric field and random phase

shift, Φ (t ) with evenly distributed probability within the range [ 0 ÷ π ]. The sum of electric field
collinear vectors of those two waves may be replaced by a sum of two complex scalars, i.e.
E (t ) = E max exp (iω t ) + E max exp [i (ω t − Φ )] =

= E max exp ( iω t ) ⋅ [1 + cos Φ − i sin Φ ] . (6.14)

From (6.14) the effective (RMS) field strength value may be found as

E = E1 (1 + cos Φ )2 + sin 2 Φ = 2 E1 cos Φ , (6.15)


2
where E1 = E max / 2 is an effective field strength of one of the rays.
Now we differentiate both sides of (6.15) 1 as follows:
Φ
dE = E1 ⋅ sin dΦ , (6.16)
2
From (6.15) we define sin (Φ / 2) , substitute into (6.16) and divide both sides by π :

1 dE
dΦ = . (6.17)
π E2
π E1 −
2

4
For evenly distributed probability density of the random phase Φ shown in Figure 6.5, the left
hand side of is a probability of random variable Φ being located in the range [Φ, Φ + dΦ ] : in

1
Minus sign is ignored, because it doesn’t effect the final result, as it may be seen later.

360
general form it may be expressed as w (Φ ) d Φ , where w (Φ ) = 1 / π is a probability density for
this particular case 1.

Figure 6.5. Probability density distribution of the resultant phase for randomly
superimposing two independent rays

Hence the right hand side of (6.17) may be considered as a probability of the resultant randomly
fluctuating E (t ) field being within the range [ E , E + dE ] , and may be written in the form

of w (E ) d E . Therefore the following expression may be found for the PDDF, w (E ) from (6.17):

w (E ) =
1
. (6.18)
π E1 − E 2 / 4
2

Based on (6.18) the CPDF may be found, taking into account the fact that the upper limit in (6.8)
is equal 2 E1 , i.e. may not exceed the double value of the effective field E1 of the single wave.

 E 
2 E1

W (E ) =
1 2
∫π
E E1 − E / 4
2 2
dE = 1 −
π
sin −1   .
 2 E1 
(6.19)

The definition (6.10) for the median value of random variable E may now be applied to (6.19) as

E 
W (E med ) = 1 −
2
sin −1  med  = 0.5 , (6.20)
π  2 E1 
then from (6.20) the value of E med may be expressed in terms of E1 :

E med = 2 E1 . (6.21)

1
Note, that w (Φ ) distribution is considered only in {0, π }, because the values of E are repeated in rest
of the range, i.e. within {π , 2π }.
361
The reason for that replacement is that the median value of the electric field at the reception
point is measurable, whereas the effective value of the electric field of only one of the
components of the total field E1 is really hard to separate in practical reception procedure.

Using (6.21), the value of E1 in (6.19) may be replaced by E med as follows:

 
W (E ) = 1 −
2 E
sin −1   . (6.22)
π  2 E med 
 
The final expression (6.22) for the CPDF of the random variable E (t ) is shown graphically in

Figure 6.6 as a function of the ratio E / E med . Note that due to the condition 0 ≤ E ≤ 2 E1 the

argument remains in the range 0 ≤ E / E med ≤ 2.

Figure 6.6. Cumulative Probability Distribution Functions for different scenarios of


multi-path random interference

6.1.2.b. Random interference of the large number of independent wavelets


In most applications a large number of partial, independent waves of equal amplitudes and
random phases, evenly distributed in [0, 2π ] range, are interfering at the reception point. The
intensity of each wavelet is vanishing. However the number of those partial wavelets tends to
infinity resulting in a finite overall power. This group of partial rays, called Rayleigh ensemble, is
a theoretical model that closely presents the real conditions that are specific for scatterings from

362
tropospheric turbulences or scatterings from the ionospheric irregularities due to micro-meteor
ionized trails 1 (see Figure 6.4a).

Figure 6.7. a). Partial waves ( E n and E k ) presentation in complex plane b). Probability

density distribution of the components of the resultant E - magnitude of Rayleigh


ensemble

Each wavelet may be represented in the complex plane as a phasor with the constant amplitude
and random phase as shown in Figure 6.7a. As pointed above, the number of independent
partial waves is extremely large, i.e. N → ∞ , with vanishing intensities of each wave. Then
each of them may be decomposed into real and imaginary parts, either positive or negative,
depending on random initial phase shift Φ of the single partial wavelet. Note, we silently
assume that all those wavelets propagate parallel to each-other, resulting in collinear electric
fields. Hence one may replace a vector sum by algebraic summation in complex plane, as it’s
been done in previous cases of random interference, i.e.
N N N
E = ∑ E n = ∑ Re E n + i ⋅ ∑ Im E n = X + i Y . (6.23)
n =1 n =1 n =1

1
Rayleigh model of the random interference is equally applicable to diffuse scatterings (such as
scatterings from the tropospheric turbulences) and to discrete scatterings (such as scatterings from the
micro-meteor trail-paths in ionosphere)

363
Regarding to the central limit theorem, if N is large enough, then both terms in right hand side of
(6.23) have a Gaussian probability distribution, regardless to what the distribution of each
N
component of the sum ∑ E
n =1
n is, i.e.

 X2 
w (X ) =
1
exp  −  , (6.24)
2π σ  2σ
2

 Y2 
w (Y ) =
1
exp  −  , (6.25)
2π σ  2σ
2

where σ is a standard deviation of X and Y random variables that are the resultant components
of the total vector E 1 in complex plane.

In order to evaluate the probability distribution of the E = E magnitude of the electric field,

let’s find first the probability for the tip of vector E to be located in the area d S that surrounds

point A with the coordinates ( X , Y ) as shown in Figure 6.7b. The joint probability density

function, w ( X , Y ) of the events X ∈ [ X , X + dX ] , and Y ∈ [Y , Y + dY ] , which appear


simultaneously, may be counted as a product of the proper distribution functions, due to
independency of X and Y random variables, i.e.

 X 2 +Y 2 
w ( X , Y ) = w ( X ) ⋅ w (Y ) =
1
exp  −  . (6.26)
2π σ 2  2σ 2 
2
It may be seen from Figure 6.7b that E 2 = E = X 2 + Y 2 , therefore (6.26) may be considered

also as the joint probability density of the magnitude and phase of E = E exp ( iΦ ) phasor:

 E2 
w ( X , Y ) = w (E , Φ ) =
1
exp  −  . (6.27)
2π σ 2  2σ
2

Now, based on probability density distribution function w (E , Φ ) we may find the probability for

the tip of vector E to be located in the area d S = E dE dΦ (see Figure 6.7b).

d W [E ∈ (E , E + dE ), Φ ∈ (Φ, Φ + dΦ )] =

1
Note that the standard deviation σ is taken the same for both, X and Y because it’s invariant to the
rotations of complex coordinates.

364
 E2 
= w (E , Φ ) d S =
1
exp  −  ⋅ E dE dΦ . (6.28)
2π σ  2σ
2

In order to find the probability distribution of the magnitude E , the phase Φ to be excluded
from (6.28). In other words (6.28) is to be integrated for Φ form 0 to 2π , which allows finding
the probability for the tip of E to be located within the ring E ∈ (E , E + ∆E ) shown in Figure
6.7b.

dW [E ∈ (E , E + dE )] = ∫ dW [E ∈ (E , E + dE ), Φ ∈ (Φ, Φ + dΦ )]dΦ =
0


1  E2  E  E2 
= ∫
0 2π σ 2
exp  −
 2σ
2
 E dE dΦ = 2 exp  −
 σ  2σ
2
 dE .

(6.29)

From (6.29) the probability density distribution function for the Rayleigh ensemble, or so called
differential Rayleigh distribution function, may be found as

 E2 
w (E ) =
dW E
= 2 exp  −  . (6.30)
dE σ  2σ
2

The graph of (6.30) is shown in Figure 6.8a.
Cumulative Rayleigh distribution function may be obtained by substitution of (6.30) into (6.11).
∞ ∞
 y2   E2 
W (E ) = ∫ w ( y ) dy = ∫
1
exp  −  y dy = exp  −  . (6.31)
Eσ  2σ  2σ
2 2 2
E  
Similarly to the previous case the parameter σ is likely be replaced by the median value, E med .

By definition

 E med 2 
W (E med ) = 0.5 = exp  − 2 
. (6.32)
 2σ 
Then σ may be found by solving (6.32):
2 2
E med E
2σ 2 = = med . (6.33)
ln 2 0.693

365
Figure 6.8. Rayleigh distribution: a).Probability density function.
b). Cumulative probability function

Final expression for CPDF as a function of E / E med may be found if (6.33) is substituted into

(6.30):
  E 
2

W (E ) = exp − 0.693    . (6.34)
  E med  

The graph of probability density distribution function (6.34) is presented in figure 6.6a in linear
scale and cumulative probability distribution function is shown in Figure 6.8b in so called
Rayleighian scale, which ends up with the straight line, as shown.

6.1.2.c. Further Generalization of the Fast Fading Statistics


In his pure form the Rayleigh distribution is not quite common for fast fading statistical
description in practical applications in radars and communications practices. Its main restriction
is the lack of flexibility that is in needed to support a variety of propagation mechanisms, hence
the variety of statistical properties, particularly the fading depths and shapes of probability
distribution functions. Variety of types of radio links and propagation mechanisms leads to the
need of more complex analytical expressions for fading statistics that are able to:
1. Specifics of the physical processes in radio waves propagation
2. Adjust to statistical properties of the particular propagation path of interest
Further updates of the fast fading statistics that involve more detailed propagation mechanisms
are outlined in this subsection without going inside the mathematical evaluations procedures.

366
One of the advantages of those statistics is the ability to adjust the probability distribution
functions to

1). Rice-distribution (Generalized Rayleigh-distribution) appears as a result of superposition


of Rayleigh ensemble with the strong monochromatic (deterministic) component of E 0 field

strength (see Figure 6.9). This approach has been originally presented in [4], where the
probability density distribution function was evaluated in the following form (see graph in Figure
6.10a)

E  E 2 + E0 2   E E0 
w ( E ) = 2 exp  −  ⋅ I0  2  . (6.35)
σ 2σ 2   σ 
 
Here I 0 is the modified Bessel function of the first kind, zero order. If E 0 = 0 , then the

distribution reduces to a regular Rayleigh distribution. The adjustment parameter β = E 0 / σ is

to adopt (6.35) to the real conditions for the given ratio between deterministic component E 0

and total intensity of Rayleigh ensemble. This distribution is specific for the terrestrial LOS or
Earth-to-Space / Space-to-Earth communication links at UHF and higher frequencies. CPDF is
calculated numerically, and is shown in Figure 6.10b in Rayleighian scale.

Figure 6.9. Superposition of a strong, stand-alone monochromatic component


with the continuum of single-scattered 1 uncorrelated wavelets (Rayleigh ensemble).

1
Superposition of the continuum of multi-scattered waves with intensive monochromatic component is
mostly observable on atmospheric-optical communication links rather than on RF links. Multi scattering of
the partial waves results in log-normal probability distribution (see [7] for references).

367
Figure 6.10. Rice-distribution: a). Probability density distribution, b). CPDF

2). m-distribution of Nakagami has been originally presented in [5]. It is one of the most
generalized descriptions of the random interference for the superimposing wavelets of various
amplitude and phases distributions. The probability density distribution function is presented by

2m m E 2 m−1  mE 2 
wm ( E ) = exp − 2  ,
Γ ( m) σ 2m
(6.36)
 σ 
where m and σ are the adjustment parameters. This expression is applicable to any type of
radio links as a fast fading statistic (see graph on Figure 6.11).

Figure 6.11. Probability density of m-distribution Nakagami

Based on (6.8), and (6.36) the CPDF for m-distribution may be defined as

368
∞ ∞
2 mm  m 
Wm ( E ) = t m −1 exp (− t ) dt ,
1
2m ∫ ∫
z 2 m −1 exp  − 2 z 2  dz = (6.37)
Γ(m )σ E  σ  Γ (m ) B

mE2
where B= . (6.38)
σ2
The integral in (6.37) may be expressed in closed from through the incomplete gamma-function
[1] as follows:

 m 
γ  m, E2 
 σ 2
.
Wm ( E ) = 1 − (6.39)
Γ (m )
It may be shown, that the second order mathematical moment of the random variable E (t ) is
defined from (6.39) as

E2 = σ 2 , (6.40)
and it is constant and independent on m.
Median values may be found from Wm (E med ) = 0.5 , by using (6.39) in the following form

 E
2

γ  m, m med2 
 σ 
= 0.5 . (6.41)
Γ (m )

The ratio E med / σ 2 is m-dependent. If we denote


2

p (m ) = E med / σ 2 ,
2
(6.42)

then values of p (m ) may be calculated: the results are given in Table 6.1.
Table 6.1
m 0.5 0.75 1.0 1.5 2.0 3.0 4.0 5.0 6.0
p ( m ) 0.456 0.606 0.693 0.789 0.839 0.891 0.918 0.922 0.945

p (m ) values may also be found from the graph shown in Figure 6.12. Taking into account
(6.42) the expression (6.39) may be rewritten as
γ (m, m ⋅ q (m ) ⋅ x 2 )
Wm ( x ) = 1 − , (6.43)
Γ (m )
E
where x= (6.44)
E med

369
is E-level, relative to median value. The sketch of CPDF for m-distribution is shown in Figure
6.12b in Rayleighian scale, so thus for m = 1 (for Rayleigh distribution) the graph becomes a
straight line.

Figure 6.12. a). Graph of the p (m) parameter, b). CPDF of m-distribution

3). n-distribution (Quasi-Rayleigh distribution) is a statistical model that was verified based
on numerous observations of fast fading on HF radio links, and has been originally presented in
[6]. The adjusting parameter n is introduced to modify Rayleigh distribution and make it
adjustable to the observed fade statistics on real propagation paths. The idea relates to diversity
reception in auto-selection mode that results in (6.150). Similar expression is adopted here for
CPDF of Quasi-Rayleigh distribution as follows:
n
   E 
2
 
Wn ( E ) = 1 −  1 − exp − 0.693 ⋅ q (n) ⋅    , (6.45)
   E med   

 
2
E med  
q (n ) =
1
where = ln   (6.46)
2σ 2  1 
1 − exp  n ln 0.5  
 
is n-dependent parameter, so thus Wn (E ) becomes adjustable. The calculated values of q (n )

are given in Table 6.2 and are presented graphically in Figure 6.13. CPDF of the n-distribution
(quazi-Rayleigh distribution) is shown in Figure 6.14 with Rayleigh-scale for horizontal axis.

370
Table 6.2
n 0.25 0.5 1 2 3
q(n) 0.0645 0.2877 0.6931 1.2279 1.5784

Figure 6.13. Quasi-Rayleigh distribution: a) Graph of the q (n) parameter, b). CPDF

6.1.3. SLOW FADING STATISTICAL DISTRIBUTION


As noted above the cause of slow fading (slow random variations of the short-term median field
at the reception point) are the random changes in attenuations along the propagation path.
Those attenuations depend on frequencies and propagation path configurations'. They may
arise in different atmospheric layers and along Earth's surface, being impacted by random
changes in temperature, humidity, pressure, ionization intensity or other physical conditions.
Numerous observations on various radio links exhibited the logarithmic-normal (lognormal)
probability distribution of slow fades.

6.1.3.a. Normal (Gaussian) distribution of the random variable.

First consider the normally distributed (Gaussian distribution) random variable z. This type of
random variable is described by the following probability density distribution function:

 ( z − z )2 
w (z ) =
1 ,
exp  −  (6.47)
2π σ z  2σ z
2

371
The graph is shown in figure 6.14, where z is a mean value of random variable z (statistical

expectation), and σ z = ( z − z )2 is its standard deviation.

Figure 6.14. Normal (Gaussian) probability density distribution function for z = 1

Based on (6.10) for this case the value of zmed may be found by solving the equation below.

1  ( z − z )2 
∫ exp − 2 
dz = 0.5 (6.48)
z med 2π σ z  2σ z 
The result is
z med = z (6.49)

because of the symmetry of the function (6.47) about z = z line 1.

6.1.3.b. Lognormal distribution of the random variable.

Term "lognormal" means the normal (Gaussian) distribution of the logarithm of random variable
E, i.e.
z = ln E . (6.50)

1
Otherwise, for non-symmetric distributions, z med ≠ z .

372
For the practical applications it's more convenient to rewrite (6.47) as a function of E-variable,
taking into account (6.50):

dW ′ [z ] = w (z ) d z = w [z (E )]⋅ ⋅ d E = d W (E ) .
dz
(6.51)
dE
Probability density distribution function may be found from (6.51) as (Figure 6.15)

w (E ) = w [z (E )] ⋅
dz
=
1
exp −
(
 ln E − ln E )  .
2

(6.52)
2π σ z 2σ z
2
dE E  

Figure 6.15. Lognormal probability density distribution (simplified version of (6.56))

Based on (6.49) and (6.50) the expressions for statistical expectation and standard deviation
may be rewritten as

ln E = ln Emed , (6.53a)

 
σz = (z − z )2 ≈  ln
E
 = y2 = σ y , (6.53b)
 Emed 
E
where y = ln , [Np] (6.54)
Emed
is a relative level of field strength expressed in nepers.
Cumulative (integral) distribution function for this case may be evaluated if (6.52) is substituted
into (6.11) as follows:

W (E ) = ∫ w (E ) dE = ∫

1
exp −
(
 ln E − ln E )  dE .
2

(6.55)
2π σ z 2σ z
2
E E   E

If replacements (6.53) and (6.54) are applied to (6.55), then the result is

373
∞  y′2 
W (y) =
1
∫  − 2 σ 2
2π σ y y
exp  d y′ ,

(6.56)
 y 
i.e. the slow fading of the relative level (in dB or Np) of short-term medians of the electric field
strength is normally distributed. In order to “standardize” this expression the replacement
y
x= (6.57)
2σ y
1
is used, which results in

W ( y ) = W (x ) = exp(− x ′ ) dx ′ = [1 − erf ( x )] ,
1 1

2
(6.58)
π x
2

exp (− x′ ) dx′ = −erf (− x )


x
erf ( x ) =
2
π ∫
2
where (6.59)
0

is so called error function, given in numerous of math references, e.g. [1].


Examples of lognormal cumulative distributions are shown in Figure 6.16, where nonlinear scale
was adopted for the horizontal axes (so called "Gaussian" scale), which allows to sketch the
graphs as straight lines.

Figure 6.16. Lognormal cumulative probability distribution function

The graphs on Figure 6.16 are drawn based on the following procedure:

∫ exp (− x′ )dx′ = 1 is used to evaluate (6.62).


1 1 2
The normalization
π −∞

374
• if x = 1 / 2 , i.e. y = σ y , then from (6.58) W = 0.16 (16 %) , and

• if x = −1 / 2 , i.e. y = − σ y , then from (6.58) W = 0.84 (84 %) .

• note also the fact, that for any σ y graph will cross the point (0, 50%), which

comes from the definition of the median value. Hence in Gaussian coordinate
system the any graph of CPDF is a straight line that goes through the points:
(− σ y , 84) , (0, 50) , and (σ y , 84) .
Log-normal distribution is appropriate as a fast fading statistics for atmospheric-optical
communication lines only. That’s because of multi-scatter propagation of each partial wave
coming to the reception point along with the strong “deterministic” component [7], as it’s shown
in Figure 6.17.

Figure 6.17. Mechanism of the superposition of strong “deterministic” component with


an ensemble of partial, multi-scattered waves

In (6.53) and (6.56) σ y is an independent parameter, which specifies the deepness of fading

and may be used to adopt the graph to real data observed. Note again, the analytical
expressions, given in subsection 6.1.2d, describing fast fading also include that kind of

"adopting" parameter, e.g. parameter E 0 /( 2 σ ) in Rice-distribution, parameter m in

Nakagami-distribution, or parameter n in quasi-Rayleigh distribution.

375
6.1.4. COMBINED DISTRIBUTION OF FAST AND SLOW FADES.
SIGNAL STABILITY IN LONG-TERM OBSERVATIONS

Fast and slow fades always act simultaneously, i.e. within the short-term stationarity intervals of
observations the median value of the field strength remains almost constant, whereas during
long-term observation periods the random variations of the median E med ( t ) become observable

and are considered as slow fading (see Figure 6.3).


The evaluation provided below is based on assumption, that the fast fading may be counted by
introduction of the normalized random variable κ fast (t ) with the median value equal to one.

Thus within the interval of stationarity (short term observation) the field strength at the reception
point may be written in following form:
E (t ) = κ fast (t ) ⋅ E med . (6.60)

If observation period is prolonged, becoming greater than the interval of stationarity, then in
order to count slow random variations of the median field E med ( t ) we may introduce another

variable κ slow (t ) , similar to that for the fast fading, so the fast and slow fades may be taken into

account simultaneously by the following expression:


E (t ) = κ fast (t ) ⋅ κ slow (t ) ⋅ E med , slow , (6.61)

where E med , slow is an overall median value of the field strength for a long-term observation

period, i.e. the median of the slow fading.


Now we introduce the level of signal as a unitless quantity referenced to the long-term median
E med , slow :

E (t )
Z (t ) = ln . (6.62)
E med , slow
Then taking the logarithm from both sides, the expression (6.61) may be rewritten as
Z (t ) = X (t ) + Y (t ) , (6.63)
where
X (t ) = ln κ slow (t ) 
. (6.64)
Y (t ) = ln κ fast (t ) 
In (6.64) X (t ) and Y (t ) represent slow and fast fading components in Np, respectively. They
easily may be converted into dB by using the relation 1 Np = 8.69 dB.

376
From (6.63) one may notice, that the randomly fluctuating signal level Z ( t ) is displayed in form

of the sum of two random variables, X ( t ) and Y ( t ) that are representing slow and fast fades

respectively. It's known from the probability theory, that if w ( X , Y ) is the joint probability density

distribution function of X and Y random variables, then in order to figure out what is the joint
probability of the combination ( X , Y ) . Based on (6.63) one may realize that in XOY-plane only

S-region, shown in Figure 6.18 is to be considered as the area of existence for X and Y
variables.

Figure 6.18. Area S of integration in (6.69)

Then having w ( X , Y ) as a joint differential probability distribution function, the integral PDF may
be written

W = ∫∫ w ( X , Y ) dX dY . (6.65)
S

As known from the probability theory course, for the independent X and Y random variables
the joint differential probability density distribution function may be simplified to the product
w ( X , Y ) = wslow ( X ) ⋅ w fast (Y ) . (6.66)

Hence, taking into account (6.63), the integration limits in (6.65) may be set as follows:
+∞ +∞

W (Z ) = ∫ wslow ( X ) ∫ w (Y ) dY dX =
fast
−∞ Z−X

377
+∞ +∞

= ∫ w fast (Y ) ∫ w ( X ) dX dY .
slow (6.67)
−∞ Z −Y

The convolution integral (6.67) in probability theory is called composition of probabilities, and a
special symbol, " ⊗ " is used to express (6.67) in compact presentation form.
Taking into account (6.11), we may rewrite the convolutions (6.67) as
W (Z ) = Wslow ( X ) ⊗ W fast (Y ) = ...
+∞ +∞

... = ∫ wslow ( X ) ⋅W fast (Z − X )dX = ∫ w fast (Y ) ⋅Wslow (Z − Y )dY .


−∞ −∞
(6.68)

The expression (6.67) for the cumulative probability distribution function in form of double
integral may be transformed to the single-integral form, written for the probability density
1
distribution functions of both, fast and slow fades. I.e. (6.67) may be rewritten as follows
∞ ∞
w (Z ) = ∫ w ( X ) ⋅ w (Z − X ) dX = ∫ w (Y ) ⋅ w (Z − Y ) dY .
slow fast fast slow (6.68a)
−∞ −∞

It’s may be shown, that if one of the functions, w fast (Y ) or wslow ( X ) is even (symmetric about

vertical axis), then (6.68a) may be written as


∞ ∞
w (Z ) = ∫ w ( X ) ⋅ w (Z + X ) dX = ∫ w (Y ) ⋅ w (Z + Y ) dY ,
slow fast fast slow (6.68b)
−∞ −∞

which means, that in this care of symmetry both expressions, (6.68a) and (6.68b) may be
applied to either sum of two random variables, Z = X + Y , or to their difference, Z = X − Y .
Note: (6.68a) type convolutions may easier be calculated based upon the fact that the Fourier

transforms of both functions, w fast (Y ) and wslow ( X ) are just to be multiplied.

---------------------------------------------------------------------------------------------------------
Example 6.1

Consider a combined distribution of fast and slow fades if both are log-normally distributed

with σ 1 and σ 2 , standard deviations respectively. Note that this is typical case for the
atmospheric optical communication links (not for RF links). Find the joint probability distribution
function as well as its standard deviation.

∞ ∞

∫ w ( X ) dX = W (Z − X ) , or ∫ w (Y ) dY = W (Z − X ) are employed here.


1
Expressions slow slow fast fast
Z −Y Z−X

378
Solution

The expressions probability distribution differential functions for the logarithmic variables of fast
and slow fades are presented the in the first column of Table E6.1, with proper Fourier
transforms in the second column. The expression for the Fourier transform of the joint
distribution density is a product:

F (ξ ) = F1 (ξ ) ⋅ F2 (ξ ) =
1

 ξ2
exp − ( 
σ 12 + σ 2 2  , ) (E6.1)
 2 
As seen from (E6.1) the inverse transform is also expected to be a log-normal, and may be
easily obtained from [1] with same form as w1 ( X ) , or w 2 (Y ) , and with the standard deviation

σ 2 = σ 1 2 + σ 2 2 , dB (E6.2)

Table E6.1.

Probability density distributions Fourier cosine transforms [1, p.1127]

 X2   ξ 2σ 1 2 
w1 ( X ) = F1 (ξ ) =
1  1 
exp −  exp − 
2π σ 1  2σ 1 2π
2
  2 
 X2   ξ 2σ 2 2 
w 2 (X ) = F2 (ξ ) =
1 1
exp −  exp − 

2π σ 2  2σ 2  2π 2
 2   

----------------------------------------------------------------------------------------------------------------

The expressions (6.68) and (6.68a) allow determining the long-term probability distribution by
combining lognormal distribution for slow fades, and one of the distribution models given in
section 6.1.2 for fast fades. Unfortunately for RF links any combinations between fast and slow
fading models may not be evaluated in closed form and, therefore, only numerical evaluations
are applicable. The graphs of several combined distributions are shown in Figures 6.19 – 6.21.

379
Figure 6.19. Combination of lognormal and m-distributions

Figure 6.20. Combination of lognormal and n-distributions

380
Figure 6.21. Combination of lognormal and Rayleigh distributions

To generalize the above statements one may consider three, four and more independently
acting fading mechanisms instead of two: the problem is to obtain the resultant distribution
function if the cumulative distribution functions (W1, W 2, … W N) due to different independent
mechanisms of fades are known. The following approach may be proposed, based on (6.68):
first W 1 must be composed with W 2, then the result with W 3 and so on , i.e.
W = { [ (W1 ⊗W2 ) ⊗ W3 ]........ ⊗W N } . (6.69)

The approaches presented in this subsection may be considered as a background for the
assessments of the RF link stability.

---------------------------------------------------------------------------------------
Example 6.2
Assume, E ( t ) = Eth = 4.3 µV/m is a minimum value of the electric field the receiver is able to

“sense” (threshold value) for the conditions described in Example 5.5. Fast fading of the signal
has Nakagami statistics with parameter m = 1.5, and slow fading is log-normally distributed with
parameter σ = 8 dB. Estimate the stability of communication link.
Solution
If the output power of the transmitter is chosen to induce the long-term median electric field at
the reception point equal to threshold, E med , slow = Eth then apparently from (6.66) follows Z = 0 ,

which, regarding to Figure 6.19 results in 50% of stability. For the considered case

381
E med , slow = 136 µV/m, therefore Z = 20 log( E / E med , slow ) = − 30 dB. From Figure 6.19 for the given
parameter m = 1.5 we may determine approximately: W = 99.9 % (Answer)
As seen, the power margin of 30 dB results in fairly high communication link stability

----------------------------------------------------------------------------------------

Now consider the case, when the power margin is such that received signal stability is high
enough, like it is demonstrated in Example 6.2. It means only a very deep fades may result in
signal outage, when it drops down the threshold level. The reason is a significant difference
between threshold and long-term median fields.
Now rewrite (6.68) for the combined probability distribution function:
+∞

W(Z)= ∫ w ( X ) ⋅W (Z − X )dX .
slow fast (6.70)
−∞

As seen, in case of deep fades, i.e. for large negative values of Z the cumulative distribution
W fast (Z − X ) is fairly close to unity (100%) in wide range of variations of the argument X ,

where the contribution of wslow ( X ) is significant, the value of W fast may approximately be

considered as a constant and taken out of the integral. Then based on (6.67), i.e. Z − X = Y ,
the expression (6.70) may be rewritten as
W (Z ) ≈ Wslow ( X ) ⋅ W fast (Y ) (6.70a)

in other words the convolution (6.70) may be replaced by the product (6.70a) of CPDF-s of slow
and fast fades, with acceptable accuracy. The same approach may be applied to the case when
more than two independent fading mechanisms remain in force.
From (6.11) that shows the relation between IPDF and CPDF we may obtain:
P = 1−W . (6.71)
P represents the probability of dropping down the fixed value of the signal (e.g. down the
threshold value). Based on (6.70a) CPDF for two independent fades may be rewritten as
Wresult ≈ W1 ⋅ W2 = (1 − P1 ) ⋅ (1 − P2 ) = 1 − (P1 + P2 ) + P1 ⋅ P2 = 1 − Presult , (6.72)

hence Presult = P1 + P2 − P1 ⋅ P2 . (6.73)

The last term is the probability of simultaneous appearance of two drop-downs, caused by two
different independent mechanisms. Under the assumption P1 << 1 and P2 << 1 the last term in
(6.73) becomes a second order infinitesimal quantity, and may be ignored. Therefore

382
Presult ≈ P1 + P2 . (6.74)

Expression (6.74) may be generalized to:


N
Presult ≈ ∑ Pk . (6.75)
k

The physical meaning of (6.74) may be clarified if referred to Figure 6.22.

Figure 6.22. Fading patterns of two independent fading mechanisms.


a). Fading due to the first mechanism, b). Fading due to the second mechanism,
c). Resultant pattern due to simultaneous action of both mechanisms

As seen from the figure


∆t11 + ∆t 21 + ∆t 22 + ∆t12 ∆t11 + ∆t12 ∆t 21 + ∆t 22
Presult = = + = P1 + P2 , (6.76)
t total t total t total
where t total is a total period of observation, ∆ t11 , ∆ t12 are the breaks caused by the first fading

mechanism, ∆ t 21 , ∆ t 22 are the breaks caused by the second fading mechanism.


Note again, that expressions (6.74) and (6.76) are valid if:
1. Both probabilities of outages, P1 and P2 are small
2. Fading mechanisms are independent
3. Overlaps of the outages are disregarded

383
6.2. ADDITVE INTERFERENCES (NOISES)

6.2.1. INTERNAL NOISES OF ONE- AND TWO-PORT NETWORKS.


NOISE FIGURE
In late 1920-s J.B. Johnson at Bell Labs discovered experimentally, that any temperature other
than absolute zero always results in a small randomly fluctuating voltage on terminals of the
resistive element. Shortly after the American physicist, H. Nyquist has shown theoretically, that
this voltage is generated by the thermal agitation, namely chaotic movements of the charge
carriers (usually free electrons) within an electrical conductor at equilibrium condition. Those
fluctuating voltages are called thermal noise 1.
As mentioned in section 1.4 this noise voltage, u N (t ) is a stationary, random Gaussian process,

i.e. a process with normal probability distribution. It is specified by zero mean and constant RMS
value in time domain.
Consider a one-port network (OPN) that has an impedance Z. As shown by Nyquist, the power
spectral density 2 of the noise voltage is defined as

N T (ω ) = k B T0 Re { Z ( jω )} ,
2
(6.77)
π
W
where, k B = 1.38 ⋅10 −23 is Boltzman’s constant, and T0 is the absolute temperature of
K ⋅ Hz
the network under consideration, in kelvins. For purely resistive OPN, when Z = R = const ,
expression (6.77) assumes that the value of N T (ω ) remains constant (frequency independent)
over the entire RF range. Therefore it’s called “white noise”.
One of the basic features of the white noise is the fact that it is the uncorrelated random
process. Based on (6.77) the mean square value of the noise voltage at the OPN terminals
within a frequency band ∆f = f max − f min = (1 / 2π ) (ω max − ω min ) may be defined as
ω max
uN = ∫ N (ω ) dω = 4k T Re (Z ) ∆f ,
2
T B 0 (6.78)
ω min

and the proper effective, RMS value of the voltage becomes

1
Sometimes it’s called Johnson-Nyqiust noise.
2
( )
N T ω is defined as a noise power per unit frequency band

384
u N , eff = uN = 4k B T0 Re (Z ) ∆f .
2
(6.79)

As seen from (6.78) and (6.79) for the purely resistive OPN the intensity of generated noise
depends only on frequency band of the measuring instrument, but not on absolute values of fmin
and fmax.
For the analytical evaluations it’s more applicable to represent a “noisy” OPM as either a voltage
or a current source, shown in Fig.6.23.a and Fig.6.23.b.

Figure 6.23. Equivalent Thevenin (a) and Norton (b) replacements of the resistive noise
source

An ideal noise voltage source is considered in series with noiseless (“silent”) resistance R,
whereas an ideal noise current source is in parallel with noiseless conductance G = 1 / R . As
known from circuit theory course, maximum power of
2
u N , eff
PN = = k BT0 ∆f (6.80)
4R
that is transferred from the source of internal impedance Z to the load ZL may be achieved if
∗ ∗
Z = Z L , where Z L is a complex conjugate of the load impedance. Note that expression (6.79)
is written for the particular case of both, Z and Z L purely resistive impedances. One may
realized that for the ideal case of any standalone passive two-port network (TPN), which is
composed only on passive elements (including resistive elements), thermal noise voltage may
appear at the output even if the input is shorted, and if no any active elements (such as
transistors, microchips, etc.) are included. In addition the extra noises such as (1) environmental
(external) noise that penetrates from the input as well as (2) the receiver’s (internal) noise that is
induced by the active components of the considered TPN may result in even worst case, i.e. in
much higher intensity of the resultant noise at the output of TPN.

385
The overall RF system performance (either communication or radar) is highly dependent not just
on the signal at the output, but on the relation between intensities of the wanted signal and
unwanted noise. Therefore the noise parameters are defined in terms of ratio of the wanted
signal power to the power of unwanted noise. For that assessment we consider a linear TPN
with the power transform coefficient K, when a mix of wanted signal and unwanted noise 1 is
applied to the input. This is a typical case when antenna is connected to the first stage of the
receiving system, or for the interface between different stages of the multi-stage receiver. In
case of multi-stage receiver the output impedance of the previous stage may be interpreted as
an equivalent, “noisy” one-port-network connected to the input in series with the source of
wanted signal as demonstrated in Figure 6.24. The TPN is assumed to be matched from both,
input and output terminals.

Figure 6.24. To the definition of noise figure of two-port-network

The following two noise parameters are widely used for the assessments in any particular TPN,
or in the receiving systems overall:
• NF , noise figure, and
• T , noise temperature.
To introduce those two parameters we’ll define the input power of the wanted signal as
2
u S , in
PS , in = , (6.81)
4R
and PS , out the output power of the wanted signal; PN , in and PN , out are the input and output

powers of the unwanted noise respectively. Then the noise figure of the real, “noisy” TPN is
conventionally defined as

1
This noise is considered as an environmental (external) noise, i.e. the noise that is originated outside
the TPN.

386
(PS / PN )in
NF = . (6.82)
(PS / PN )out
It’s easy to realize from (6.82) the physical meaning of noise figure: it demonstrates the amount
of degradation in the Signal-to-Noise Ratio (SNR) while passing through the network.
If the expression for power transmission coefficient, K = PS , out / PS , in is taken into account, then

(6.82) may be rewritten in the following form


PN , out PN , out
NF = = . (6.83)
PS , out K ⋅ PN , in
PN , in
PS , in
The power of the output noise comprises two components:
• First component, K PN , in is the power of input noise that is transferred to the output

• Second component, PN , own is the own, internal noise of the network, which is generated

by the elements within the TPN.


Now we’ll rewrite (6.83) as
K ⋅ PN , in + PN , own PN , own
NF = = 1+ . (6.84)
K ⋅ PN , in K ⋅ PN , in
One may conclude from (6.84) that N is always greater than one. Only in ideal case of absence
of the own (internal) noises in TPN ( PN , own = 0 ) the value of N becomes equal to one.

Now we define the term noise temperature, T based on the following:


• Assume the resistor under the real, normal temperature, 1 T0 is connected to the input of

ideal (noiseless) TPN and matched to it; hence the input noise power is
PN , in = k B T0 ∆f (6.85)

• Now we increase hypothetically the temperature of the resistor until the noise power,
that passes from input to output, becomes equal to that of generated inside the real
(“noisy”) TPN, i.e.
PN , own = K ⋅ k B T ∆f . (6.86)

If (6.85) and (6.86) are substituted into (6.84), then the following expression may be found for
the noise temperature.
T = T0 (NF − 1) . (6.87)

1
T0 = 290 K is commonly accepted value for the normal atmospheric temperature.

387
The physical meaning of the noise temperature, T is as follows: a hypothetical temperature that
is assigned to the matched resistance at the input of a TPN such that a resistance at T
produces a TPN noise output component equal to the contribution of noise sources internal to
the TPN.

6.2.2. NOISE FIGURE AND NOISE TEMPERATURE OF THE CASCADED


TWO-PORT NETWORKS
Consider two cascaded TPNs with noise figures, NF1 and NF2, power transform coefficients, K1
and K2 and noise temperatures T1 and T2 respectively (Figure 6.25).

Fig. 6.25. Cascaded connection of two-port networks

Regarding (6.83) the resultant noise figure for the TPN between “A” and “C” terminals may be
introduced as
PN , out , 3 PN , in ,1 K1 K 2 + PN , own ,1 K 2 + PN , own , 2 PN , own ,1 1
NF = = = 1+ + ...
K1 K 2 ⋅ PN , in ,1 K1 K 2 ⋅ PN , in ,1 PN , in ,1 K1

PN , own , 2 1
... + , (6.88)
PN , in ,1 K1 K 2
where subscripts 1 and 2 relate to proper measures for the first and second stages of TPN-s
respectively. The absence of the subscript 1 or 2 relates to the measures for the resultant (total)
TPN, i.e. to the measures between A an C terminals.
In order to analyze the last term in (6.88) one may take into account that in (6.84) PN , in is a

noise power, which comes from an independent source, regardless to what TPN it is applied to.

388
Thus (6.88) be modified, by using (6.84) if assumed that the same noise source PN , in ,1 = PN , in , 2

applied to each, TPN1 and TPN2 independently 1, i.e.


PN , own , 2 1 PN , own , 2 1 NF2 − 1
= = . (6.89)
PN , in ,1 K1 K 2 PN , in , 2 K1 K 2 K1
Then (6.88) may be rewritten as
NF2 − 1
NF = NF1 + . (6.90)
K1
Expression (6.90) may be generalized for cascaded connection of n independent TPN-s:
NF2 − 1 NF3 − 1 NFn − 1
NF = NF1 + + + ......... + . (6.91)
K1 K1 K 2 K1 K 2 ⋅ ⋅ ⋅ K n −1
The resultant noise temperature of the cascaded TPN-s may be derived from (6.98) by using
(6.87). Finally
T2 T Tn
T = T1 + + 3 + ...... + , (6.92)
K1 K1 K 2 K 1 K 2 ⋅ ⋅ ⋅ K n −1
where T1 , T2 , …… Tn are the noise temperatures of each separate TPN, and K 1 , K 2 ,

……… K n −1 are the power transfer coefficients.

Expressions (6.91) and (6.92) show that the resultant noise figure N and resultant noise
temperature T of the receiver are highly dependent on the first stage, which immediately
follows the antenna, especially when each stage has a large value of power transform
coefficient. For large enough values of K 1 , K 2 , .... K n −1 (especially first two stages) one may

come up with NF ≈ NF1 and T ≈ T1 . In other words to reduce the resultant noise parameters,

namely N and T , i.e. to improve the system performance one has to decrease noise figure of
the first stage and increase its power transform coefficient.

6.2.3. NOISE FIGURE OF THE PASSIVE TWO-PORT NETWORK


Noise parameters of passive TPN-s such as passive filters, transmission lines, matching
networks, etc. have some specifics that allow making simplifications. In order to estimate the

1
It may seem, that in denominator of (6.89) instead of PN , in , 2 must be the value of PN , in ,1 K1 + PN , own ,1 . It
is not, because the noise figure of each TPN is defined independently, as a standalone TPN, regardless
to what is before or after. An independent noise source may be used at the input.

389
values of PN , in and PN , out in (6.83), assume that the TPN is matched from input and output with

preceding and following stages, i.e. the equivalent matched impedances, Z and ZL are
connected to the input and output of the passive TPN respectively as shown in Figure 6.26.

Figure 6.26. Equivalent replacement of the passive TPN for noise parameters definition
(Zin and Zout are the input and output impedances of the TPN, K is the power transform
coefficient).
Note: Virtual elements are shown with dashed lines

It’s easy to realize, that in matched conditions both noise powers, namely at the input and at the
output of TPN, have the same values,
PN , in = PN , out = k BT0 ∆f , (6.93)

which follows also from the principle of conservation of the power. Therefore in this case (6.82)
may be rewritten as
1
NF = (6.94)
K
For the passive TPN-s we have K < 1 , i.e. they not just push down the level of the signal, but
also decrease the SNR from input to output by contributing the internal noise of that passive
TPN to the total noise power. Particularly if a piece of transmission line of length l f is

considered as an antenna feed line with the attenuation coefficient α f , then

K f = exp (− α f l f ), and therefore NF feed = exp ( α f l f ) . (6.95)

From (6.95) one may conclude that longer is feed line, larger the contribution of the internal
noises is, and larger the value of noise figure NF feed is.

390
---------------------------------------------------------------------------------------------
Example 6.3
Calculate noise parameters (noise figure and noise temperature) for the RF unit of the ground-
based receiving station of the satellite communication system: the RF unit includes cascaded
LNA (low noise RF amplifier with NFLNA = 1.3 dB, K LNA = 15 dB), downconverter (mixer)

( NFmix = 8 dB, K mix = −5 dB), and IF pre-amplifier ( NFIF = 10 dB, K IF = 10 dB). Consider

following two possible scenarios:


• RF unit is placed right in the focal point of the parabolic dish antenna (no feed line
between antenna and RF-unit); the IF signal is transmitted from the output of the RF unit
to the main ground station through the coaxial cable (Figure E6.3.1a)
• RF unit is placed within the main ground station, l = 20 m away from the receiving
antenna (Figure E6.3.1b). Signal from the antenna output travels to the RF unit through
the waveguide (attenuation coefficient is α f = 0.2 dB/m)

Figure E6.3.1. a). Case-1scenario, b). Case-2 scenario, c). Block-diagram of the RF-unit

Compare those scenarios to achieve better noise parameters.


Solution
• First we convert initial data to their unitless (linear) values:

391
o For LNA : NFLNA = 1.35 , K LNA = 31.6 ,

o For Downconverter: NFmix = 6.3 , K mix = 0.316

o For IF pre-amplifier: NFIF = 10 , K IF = 10


• First-case scenario:
o The noise figure is defined by (6.98):
NFmix − 1 NFIF − 1 6.3 − 1 10 − 1
NF = NFLNA + + = 1.35 + + = 2.42 (3.83 dB)
K LNA K LNA K mix 31.6 31.6 ⋅ 0.316
o The equivalent noise temperature is defined by (6.94):
T = T0 (NF − 1) = 290 ( 2.42 − 1) = 412 K
• Second-case scenario:
o For the waveguide transmission line preceding the LNA we have:
K feed = − l ⋅ α f = −20 ⋅ 0.2 = −4 dB (0.398 unitless); NF feed = 1 / K feed = 2.51 (4 dB)
o The noise figure is defined by (6.98):
NFLNA − 1 NFmix − 1 NFIF − 1
NF = NF feed + + + = ...
K feed K feed K LNA K feed K LNA K mix

1.35 − 1 6.3 − 1 10 − 1
... = 2.51 + + + = 6.08 (7.83 dB)
0.398 0.398 ⋅ 31.6 0.398 ⋅ 31.6 ⋅ 0.316
o The equivalent noise temperature is defined by (6.94):
T = T0 ( NF − 1) = 290 ( 6.08 − 1) = 1473 K
• Significant difference between two scenarios is due to the feed line in front of the LNA in
second-case scenario: the feed line contributes a significant amount of the thermal noise
to LNA’s input resulting in degradation of the noise parameters compared to the first
case, where the wanted signal is amplified first, before it proceeds further, so it becomes
stronger compared to the noises attributed by the following stages.

----------------------------------------------------------------------------------------

6.2.4. ANTENNA NOISE TEMPERATURE


From the antenna’s equivalent schematic shown on Figure 3.5 it’s easy to realize, that both
resistive elements, Rant , 0 and Rant , Σ may be considered as thermal noise sources that are

applied to the receiver’s input and thus impact significantly the final SNR of the entire receiving

392
system. Those two types of antenna noises have different origins: first comes from the metallic
body of the antenna that is under the ambient temperature TA, 0 ; the second is due to radiation

resistance Rant , Σ , which, in case of receiving antenna, introduces a power received from the

surrounding space, including both, the power of wanted signal, and power of unwanted noise.
The unwanted noise that is hypothetically assigned to Rant , Σ is actually the external noise

generated by atmospheric, cosmic, industrial, etc. processes. Based on previous definition,


(6.78) the mean squares of those two noise voltages may be expressed as

u N , 0 = 4 k BT0 Rant , 0 ∆f ,
2
(6.96)

u N , Σ = 4 k BTΣ Rant , Σ ∆f .
2
(6.97)

In (6.97) TΣ is a temperature, which is conventionally assigned to resistance Rant , Σ so that the

2
noise voltage u N , Σ is considered as a hypothetic thermal noise of resistance Rant , Σ . Hence the

total mean square noise voltage, induced by the resultant resistance Rant = Rant , 0 + Rant , Σ at the

antenna’s output, is equal to

u N , ant = 4 k BTant Rant ∆f = u N , 0 + u N , Σ = 4 k B ∆f (T0 Rant , 0 + TΣ Rant , Σ ).


2 2 2
(6.98)

The final antenna noise temperature may be found form (6.98) as


Rant , 0 Rant , Σ
Tant = T0 + TΣ . (6.99)
Rant Rant
If (3.16) is taken into account, then (6.99) may be presented as
Tant = ( 1 − η ant )T0 + η ant TΣ , (6.100)

where η ant is antenna efficiency. From (6.100) it may be seen that in some cases for highly

efficient antennas (η ≈ 1 ) an approximation T A ≈ TΣ may practically be accepted.

The value of TΣ is a part of the antenna noise temperature due to noise power received from
the external sources such as 1:
• atmospheric noise radiations ( T Atm )

• cosmic noises ( TCosm )

• thermal radiations of the Earth’s surface ( TEarth )

1
In some references TΣ is called brightness temperature

393
This statement may be presented in following form:
TΣ = TAtm + TCosm + TEarth . (6.101)

Each term in (6.101) will certainly depend on the direction of the antenna’s main beam as well
as on its beam width (see below). Consider, as an example, the receiving antenna onboard the
satellite, of the satellite uplink, which is directed towards the Earth: in this case TCosm may be

ignored. For the satellite downlink, when the receiving antenna of the ground station is directed
towards the sky, another component of (6.101), namely TEarth may be ignored: for the downlinks

main contribution into TΣ comes from T Atm and TCosm .

6.2.5. RECEIVER SENSITIVITY AND SIGNAL THRESHOLD DEFINITION


Part of the receiving system between antenna and detector includes transmission (feed) line
and receiver’s linear (pre-detector) section, as shown in Figure 6.27.

Fig.6.27. Simplified block-diagram of the receiving system

The term “linear” indicates the fact that any stage-to-stage transforms of the signal within linear
section such as amplification, mixing, etc. do not result in changes in shape of the spectrum of
signal but only result in shift from one frequency domain to another, despite some of those
transforms are basically nonlinear processes. In other words the shape of the spectrum of
desired signal remains the same within the linear section of receiver, or simply receiver.

An important function of the receiver is to support desired signal-to-noise ratio (SNR) at


detector’s input in order to satisfy performance quality. Here we denote the minimum acceptable

value of this SNR 1 as

1
Note that in some references pre-detector signal-to-noise ratio (SNR) is called carrier-to-noise ratio
(CNR) to emphasize that we’re dealing with signal, which is carried by the sinusoidal wave. However,

here and below we’ll use SNRDet , in and SNRDet , out for the pre- and post-detector signal-to-noise ratios.

394
P 
SNRDet , in =  S  . (6.102)
 PN  Det , in
It highly depends on type of the modulation/detection used, as well as on either standard
requirements, or on particular application needs. For example it highly depends on desired
customer-defined quality of sound (image) if analog audio (video) signal is processed or on
desired/allowed value of the BER (bit error rate), if digital signal is processed.
If the receiver noise parameters are known, then receiver sensitivity may be defined as a
threshold (minimal) desired signal power at the receiver’s input to meet given SNRDet , in .

First we find the internal noise power, generated within the receiver that occurs at the output
terminals: based on (6.86) and (6.87) for the given value of the receiver’s noise figure N Rx and

its power transmission coefficient, K Rx the receiver’s internal noise power may be written as

PN , Rx = k B T0 ∆f (NFRx − 1) K Rx . (6.103)

Here PN , Rx = PN , Det , in , i.e. it is detector’s input noise that does not count on environmental

(external) noises. If (6.103) is substituted into (6.102) as a noise power at detector’s input, then
the threshold power of the receiver input signal (receiver sensitivity) may be obtained as
PS , Det , min SNRDet , in PN , Det , in
PRx , min = = = SNRDet , in k BT0 ∆f ( NFRx − 1) . (6.104)
K Rx K Rx
In engineering applications it is appropriate to setup a required value of the post-detector signal-
to-noise ratio SNRDet , out that is linked to its pre-detector value SNRDet , in through the detection

SNR-gain χ = SNRDet , out / SNRDet , in (see Table 6.6 for examples). Usually SNRDet , out is outlined

either by industry standards, or defined to adapt to customer needs. Then (6.104) more
conveniently may be rewritten as
PRx , min = ( SNRDet , out / χ ) k BT0 ∆f (NFRx − 1) . (6.104a)

The following statements can be made based on (6.104):


• The definition of receiver sensitivity PRx , min doesn’t include noises external to the

receiver. It is considered as one of “internal” parameters of the receiver, regardless to


the external conditions. Note that sometimes noises from feed line may also be included
into NFRx similarly to that demonstrated in Example 6.3.

• Receiver sensitivity depends on intensity of the internal noises of receiver, as well as on


predefined minimum signal-to-noise ratio acceptable by detector

395
If we convert NFRx into receiver noise temperature by employing (6.87),

TRx = T0 ( NFRx − 1) (6.105)

then a system noise temperature may be introduced as


Tsys = Tant + TRx . (6.106)

Note that in this expression we assume a feed line as a part of the antenna, which means two
issues are to be taken into account: (1) antenna total gain includes a feed line loss and (2)
antenna noise temperature counts for the feed line internal noise as an extension to Tant . Then

based on (6.106) a system noise power is equal


PN , sys = k B Tsys ∆f . (6.107)

Physical meaning of PN , sys is a total noise power generated along the entire propagation path

from transmitting antenna up to the detector’s input, which is hypothetically assigned to the
receiving antenna’s aperture. In other words Tsys is a hypothetical temperature that is assigned

to the receiving antenna’s input to replace all noises that occur along the entire path, including
receiver it selves, and results the same noise power at detector’s input as it is in real conditions.
This statement allows introducing a system signal-to-noise ratio at the receiving antenna input
(on its aperture) based on (6.107), (3.29), and (3.30) as follows:

PRx Π S eff  Πλ2  G Rx 


SNRsys = = =  ,

(6.108)
PN , sys k BTsys ∆f  4π k B ∆f  sys 
T

where Π is a signal power flow density at the reception point, and G Rx is the receiving antenna

gain, ∆f is receiver’s RF frequency bandwidth. As seen from (6.108) the expression in brackets

in right hand side do not depend on receiver at all, whereas the last ratio, G Rx / Tsys ( in dB/K)

may be used to specify the overall SNRsys , i.e. to be used as the performance figure of merit of

the receiving system in whole (including antenna). G Rx / Tsys is widely used in communications

and radars practices. Note that unlike sensitivity PRx , min that depends exceptionally on receiver’s

configuration, the ratio G Rx / Tsys depends, in addition, on receiving antenna orientation, on

ambient conditions at the receiving point, such as weather, ambient temperature, etc. due to
Tant as a part of Tsys .
In HF and lower frequency bands more commonly the receiver sensitivity is expressed in terms
of minimum acceptable electric field strength. Here we take into account that for the matched

396
antenna-receiver circuit the electro-motive force emf = E Rx , min leff induced in antenna of the

effective length leff will result in receiver’s input power

emf 2
PRx , min = . (6.109)
4 Rant

Here PRx , min is the power sensitivity of receiver. Using (3.25), (6.104), and (6.87) the value of

minimum field strength that receiver is able to “sense” may be derived as

2π 120 TRx
E Rx , min = SNRDet , min k B ∆f , (6.110)
λ G Rx

where TRx is defined from (6.105), and λ is in meters. Note that a conclusion similar to that

from (6.108) can be made here, namely higher is the ratio G Rx / TRx smaller is the required

minimum electric field E Rx , min the receiver is able to “sense”, hence better is the quality of the

receiver.

6.2.6. ENVIRONMENTAL (EXTERNAL) NOISE

As one may notice from (6.100) and (6.101) when highly efficient antennas ( η ≈ 1 ) are used a
significant contribution to the antenna’s total noise power comes from the external sources. The
estimate of those contributions, which is considered below, is a delicate task and in general is
dependent on carrier frequency, geographic location, season of the year, etc.
6.2.6.a. Atmospheric noise
This type of is generated by several independent mechanisms that are as follows:
• Noise radiation due to chaotic, thermal movements in atmospheric gases, mainly by O2
and H2O molecules, therefore called atmospheric thermal noise. Proper component of so
called brightness temperature T Atm in (6.105) is denoted as TGases 1. The graphs of TGases

in frequency domain are shown in Figure 6.28 for different elevation angles, and different
density of atmospheric gases. It must be emphasized, that regarding the principle of the

1
As shown in introductory chapter a predominant percentage of the atmospheric gases is contributed by
molecular nitrogen, N2. However, it was discovered from numerous observations that for the entire RF
range the noises caused by chaotic movements of N2 molecules are much smaller than those, caused by
O2 and H2O molecules, i.e. the thermal noise generated by atmospheric molecular nitrogen may be
ignored.

397
power conservation, the resonant and non-resonant absorptions of thermal energy are
being returned as a noise radiations with the same shape of the spectral distribution as
the frequency dependence of the atmospheric absorption graphs. Therefore the shape
of curves on Figure 6.28 is similar to those on figure 5.36. It may be seen from Figure
6.28 that for the frequencies lower than 100 MHz those noises may be ignored. Detailed
family of graphs of TGases for the standard atmospheric conditions may be found in [10].

Fig. 6.28. Frequency dependence of TGases component of the atmospheric noise

temperature, TAtm for different elevation angles of the receiving antenna beam ∆:

1. Dry atmosphere, 2. Water vapour density is 10 g/m3, 3. Water vapour density is 20 g/m3

• Atmospheric precipitations noise is mostly contributed by the rain. The proper


component, TRain of the atmospheric noise temperature TAtm may approximately be

calculated based on the following empiric formula [17]


[ ]
TRain = 280 1 − exp (− L prec / 4.34 ) , K (6.111)

398
where 280 K is so called effective rain temperature, which is different from the actual
antenna temperature because scatterings and antenna effects. L prec is a total

attenuation due to precipitations that is discussed in subsection 5.7.1.

• Atmospheric non-thermal noise is mostly due the lightning discharges generated by the
thunderstorms that occur worldwide. According to [12] approximately 2000
thunderstorms are in progress globally at one time, resulting in 30 to 100 cloud-to-
ground discharges per second. Those discharges result in electromagnetic short pulses
with the duration of 0.1 to 3 ms, and extremely large magnitudes. The power spectral
density of those pulses is predominantly located in the area of audio frequencies, i.e.
from 300 Hz up to 10 kHz, with the extension to RF range, where it drops down nearly
inverse proportional to frequency. Ones those spectral components are generated
they’re able to propagate for long distances in form of sky-waves that are supported by
earth-atmosphere waveguide. At any geographic point on the earth the overlay of those
pulses from the large number of sources worldwide may result in permanent noise
background (non-stationary noise hum) with the power spectral density spread up to 30
MHz. Frequency dependence of the power spectral density of this atmospheric noise is
quite complicated and is not only due to the spectrum of the discharges, but also due to
the frequency dependence of the propagation losses in earth-atmosphere channel for
LF, MF and HF bands. For the frequencies higher than HF the above distribution
mechanism is no longer in effect 1, therefore for VHF and higher frequency bands this
type of noise decreases rapidly and becomes ignorable. If the source of electric
discharge is located in vicinity of the receiving point (up to about hundred miles), then it
results in intensive short pulse which jumps up above the permanent “hum”. The
propagation of those short pulses is supported by ground waves mechanism. Statistical
features of those pulses that overlay with the permanent, non-stationary “hum” are
defined by amplitude probability distribution (APD), and is considered in details in [9]. It
mostly affects a bit-error-rate (BER) in digital system and is not a subject of our further
discussions. A quantitative description of the temporal and geographic variations of the
intensity of atmospheric non-thermal noise (“hum”) may certainly be described only by
empirical models, based on statistical data that are obtained from large number of
measurements worldwide over many years. It was first referenced in [8], and later

1
See section 5.5.

399
updated in [9] and [10]. The above mentioned model (procedure) of the atmospheric
non-thermal noise assessment for the LF-, MF-, and HF radio links design is outlined
below.
Unlike previous types of atmospheric noises, where the noise temperature is a basic
parameter commonly used to specify noise intensity, the atmospheric non-thermal noise
power is specified by the noise figure 1. The ITU recommendation P.372-7 [10], which is
accepted as an official guide for the atmospheric non-thermal noise estimates, defines
noise factor as a noise power received by antenna relative to the noise power generated
by the input-matched resistor at the standard atmospheric temperature T0 2:

PAtm, NT TAtm, NT
NFA = = . (6.112)
k BT0 ∆f T0
Here we’re using the notation NFA instead of commonly used FA (including [10]), in

order to prevent confusions with propagation factor. In (6.112) PAtm , NT = k B T Atm , NT ∆f ,

and T Atm , NT is a brightness temperature due to non-thermal noise received by ideal,

noise-less antenna.
References [8-10] provide the worldwide distribution of isolines of the long-term median
values of NFA (medians denoted as NFAm ), for the frequency of 1 MHz. Those values

are given for each season within 4 hour time blocks. The supplemental charts are also
provided, which allow recalculating NFAm from 1 MHz to any other frequency. Additional

statistical parameters of NFA are also provided in [8-10]. One of them is σ Fam , the

standard deviation of NFA . Note that the noise power level PN , dB = 10 log PN is normally

distributed as expressed in (6.114). An example of worldwide distribution of NFAm for 1

MHz is given in Figure 6.29. In addition, Figure 6.30 provides graphs of NFAm and σ Fam

vs. frequency for the spring season, 12:00 – 16:00 time-block. The isolines and graphs
of NFAm that are shown in Figures 6.29 and 6.30 include also other types of noises (e.g.

1
Note that a term noise factor is used in [10] instead of the term noise figure.
2
This definition is fairly consistent to what follows the expression (6.82) if assumed that the TPN has

common input and output terminals with one single resistor between them. Then formally PS , in and

PS , out are equal and may be cancelled out. Thus only a ratio on noise powers is left in (6.82)

400
cosmic noises – see below); those noises spring up at frequencies higher than 10 MHz,
being able to penetrate through the ionosphere from space at those frequencies.

Figure 6.29. Worldwide distribution of the median values of NFAm at 1 MHz


for Spring season; 12:00-16:00 time-block

Figure 6.30. Atmospheric non-thermal noise data for Spring; 12:00-16:00 hrs

a). Variations of NFAm with frequency, b). Variations of σFam with frequency

401
It has been shown by numerous of measurements that statistical distribution of short-
term variations of NFA , i.e. statistical distribution within a given season and 4-hour time

block is described by lognormal distribution with the standard deviation σ Fam .

-------------------------------------------------------------------------------------
Example 6.4
Determine median value of NFA (i.e. NFAm ), as well as the value of NFA that is exceeded

within 90% of the observation time for Spring, 12:00 – 16:00 time block, and for the reception
point located at the geographic coordinates: 43N-Latitude, 90W-Longitude.
Frequency is 10 MHz.
Solution
• From Figure 6.29 for the given geographic coordinates we may find NFAm = 50 dB for

the frequency 1 MHz


• Transform NFAm from 1 MHz to 10 MHz by using Figure 6.30a: NFAm = 35 dB. By

definition, the value NFAm = 35 dB is exceeded within 50% of the observation time.

• To define NFA for the required percentage of the observation time we need to find from

Figure 6.30b σ Fam = 5 dB. Then referring to Figure 6.16 for the lognormal CPDF an

additional margin of minus 7 dB may be found for the given σ Fam = 5 dB. Thus the final

result is NFA = 35 − 7 = 28 dB, which is exceeded within 90% of the observation time.
--------------------------------------------------------------------------------------------------------------

For HF and lower frequency bands both, input signal and input noise powers are
randomly fluctuating independent variables, therefore the variations of SNR become
more complicated. The signal-to-noise ratio in dB-s may be written here as

P 
SNRdB =  S  = 10 ⋅ log PS − 10 ⋅ log PN = PS , dB − PN , dB . (6.113)
 PN  dB
It is considered as a difference of two independent random variables PS , dB and PN , dB

that are given in relative units, e.g. dB per Watt. Fluctuations of the signal level PS , dB as

a result of fast and slow fades were discussed in section 6.1, whereas the fluctuations of

402
noise level PN , dB are normally distributed ( PN is log-normally distributed), as mentioned

above, i.e. the probability density function (PDF) of the noise is presented as

1  PN , dB 2 
wN = exp − . (6.114)
2π σ Fam  2σ Fam 2 
 
The resultant statistical distribution of the random variable SNRdB may be found as a

composition of distributions of signal level, PS , dB and noise level, PN , dB similar to those

given by (6.68) or (6.68a). In other words a statistical distribution of the SNRdB is

counted as a three-fold convolution


WSNR = Wslow ⊗ W fast ⊗ W N . (6.115)

For practical applications one may take into account a commutative property of (6.115),
as well as the fact that slow fading of wanted signal level (logarithmic variable) and
fluctuations of the level of noise both have a Gaussian statistic. Based on the results of
Example 6.1 (p 26), the following approach may be recommended for (6.115)
calculation:
 Consider composition W = Wslow ⊗ W N as a Gaussian distribution with the

standard deviation σ y + σ N
2 2

 Evaluate a composition WSNR = W ⊗ W fast either analytically or numerically

(whichever is applicable)
 Use WSNR for defining a power margin to ensure a specified communication

stability.

• Man-made radio noise is unintentional, broadband noise caused by power transmission


lines, automotive ignition systems, electrical machinery, switching devices, etc.
Experimental graphs of the values of NFa vs. frequency are shown in Figure 6.31 [11].

Narrowband man-made noises caused by coherent sources such as computers, out-of-


band transmitters, etc. are the subjects of special treatments, and are not included in
these graphs.

403
Figure 6.31. Median values of the man-made noise factor NFa vs. frequency for different

areas

6.2.6.b. Thermal noise of the earth’s surface


TEarth component in (6.101) is usually taken into account for the satellite uplinks, when the
receiving antenna onboard the satellite is directed towards the earth’s surface. According to [10]
the earth’s effective thermal radiation temperature (brightness) is in direct proportion with its real
temperature TEarth , real , and linked to it by the emissivity ∈ of the surface:

TEarth =∈ TEarth , real (6.116)

Emissivity ∈ is a physical property of the surface: that is the ratio of radiation intensity of the
real material to that of the black bogy at the same temperature. It’s always less than a unity.
More reflective is the surface less is its emissivity. If ∈= 1 − Γ 2 is employed [18, p106], then the
following expression may be used for estimates:
( )
TEarth = TEarth , real 1 − Γ 2 , (6.116a)

where TEarth , real is the earth’s real (thermodynamic) temperature and Γ 2 is its magnitude of the

power reflection coefficient for the vertical incidence.

6.2.6.c. Cosmic noise


This noise is presented in (6.101) for antenna total noise temperature by TCosm . It consists of

two components:

404
• Common background sky noise, denoted as TSky , CB (Figure 6.32a) that is spread evenly

along the entire sky and doesn’t depend on time and direction of the main beam of
receiving antenna, i.e. doesn’t depend on direction of reception of radio wave
• Noise radiations from the cosmic objects, TSky , Obj such as sun, moon, constellations,

stars, planets, etc. (see examples on figure 6.32b)

Figure 6.32. Cosmic noise temperature: (a) Background noise radiation component,
(b) Cosmic objects’ noise radiation component

Note, that the solid angular size of the cosmic object ∆Ω Obj is much smaller than the receiving

antenna solid beam width ∆Ω A, HP . Therefore the equivalent noise temperature of that particular

object may be found approximately by averaging the noise temperature within the main lobe of
the radiation pattern of the receiving antenna (see Figure 6.33). Analytically it may be expressed
as [14,19]

∆Ω Obj  ∆ϕ Obj
2

′ , Obj = TCosm , Obj
TCosm = TCosm , Obj   , (6.117)
∆Ω A, HP  HPBW 
where TSky , Obj is a real (thermodynamic) temperature of the cosmic object, ∆ϕ Obj is a flat

angular size of the object, and HPBW = 2∆ϕ 0.5 is a half power beam width of the receiving

antenna (see subsection 3.1.1).

405
Figure 6.33. Cosmic noise source within the receiving antenna’s beam width

As an example noise radiation temperatures and angular dimensions of planets are shown in
Table 6.3. More detailed information may be found in [14].
Table 6.3

Name of the Planet ∆ϕ Obj , in angular seconds TSky , Obj , in kelvins

Mercury 11.3” 673


Venus 60.6 (1’ 6”) 367
Earth - 293
Mars 17.4” 303
Jupiter 45.8” 144
Saturn 19.4” 121

It may be noted, that for the frequencies less than 10 MHz the ionosphere is opaque, and
therefore for those frequencies sky noise may be ignored.

6.3. METHODS OF IMPROVEMENT OF THE RADIO WAVES


RECEPTION PERFORMANCE
Major difference between wire-line communications 1 and radio communications 1 is in presence
of multiplicative interferences on radio links (fading), which is a random fluctuation of SNR at the

1
The term wire-line communications means communication through the physical path made of metallic
conductors (wires, cables, waveguides) to support signals transmission between corresponding points.

406
reception point. During several short periods of the total observation time desired signal may
drop down the minimum level, become “hidden” within the noise, i.e. resulting in outages.
As seen in section 6.2.1 the durations of those outages, as well as their number may be
reduced, i.e. the improvement of communication stability may be achieved by increasing the
fade margin, the gap between the median value of signal-to-noise ratio, SNRmed and its
threshold value, SNRmin. Greater is the fade margin lower is the probability of SNR to fall down
the minimum acceptable value SNRmin. In other words desired communication stability may be
achieved by choosing the proper value of fade margin.
A straight forward approach in increasing the fade margin for the given fading profile 2 is either
(1) the increase of the median value of the signal at the reception point, or (2) by suppressing
the noise intensity. The first of those two approaches may be achieved by increasing the EIRP
of the Tx-antenna (see section 3.2). Unfortunately it’s not always feasible because of limitations
on radiated power, set by FCC and international agencies, as well as limitations on physical
dimensions on Tx and Rx antennas that limit gains of both antennas. On the other hand
suppressing the noise level is also limited because of limitations on the receiver sensitivity:
internal and external noises may not always be brought down to desirable level to achieve
proper sensitivity. In other words in some cases getting over those limitations is beyond the
designer’s control.
Except of those straight forward approaches, three basic techniques that are applicable for
further improvement of the radio waves reception quality, i.e. applicable for improvement of the
communication stability / channel reliability are shortly described below.

3
6.3.1. NOISE-SUPPRESSING MODEMS IN ANALOG CW-SYSTEMS
SNR at the output of the receiver is highly dependent on structure of signals used, (namely,
digital or analog, modulation type, etc.) as well as on signal processing: they may support
reduction of both, multiplicative and additive interferences. The analog information transmission
CW systems, which are considered here as examples, may allow the additive noise suppression
by using proper modulation / demodulation scheme. The values for the gain, χ in SNR for

1
In radio communications free propagating electromagnetic waves are used to support the transmission
of the electric signals between corresponding points. Electromagnetic waves are not guided by physical
paths such as wires, waveguides, or optical fibers.
2
I.e.for the given fading conditions on particular propagation track
3
CW – Continuous Wave

407
different types of modulation/demodulation are shown in Table 6.4 without going into details of
mathematical evaluations. The term “gain” shows here the improvement in SNR from detector’s
input to its output. Expression for χ is similar to that introduced by expression (6.82) for noise
figure, which usually shows to reduction in SNR for the stages like amplifiers, filters, etc.
Expressions and values for χ in middle column of the table are shown for the sinusoidal
modulating waveform. In case of Phase Modulation (PM) and analog pulse modulation types
such as pulse amplitude modulation (PAM), or pulse position modulation (PPM), and for digital
modulation types one may refer to [15, 20-21].
Comparison of the modulation types shown in Table 6.4 may emphasize that three linear
modulation types (AM, DSB-SC, and SSB) result in SNR gain at most equals 3 dB (2 unitless).
At the same time the non-linear type of modulation such as FM and PM may result in much
higher value of SNR gain by proper choice of modulation index. Seems like any value of SNR
gain may be obtained by choosing the value of modulation index m FM for FM. In other words
seems like even in case when the input signal is “buried” within the additive noise, the output
signal may spring up from the noise after demodulation, if m FM is chosen large enough. This
statement is not true because of existence of the threshold that is specific for FM systems: the
linear relation χ = SNRDet , out / SNRDet , in between input and output SNR-s is not in effect for the

small SNRDet , in -s; the output SNR goes down abruptly when the input SNR is close or less than

the threshold, as shown in Figure 6.34.


Table 6.6
SNR gain
Modulation type χ = SNRDet , out / SNRDet , in Notes and explanations

at detector’s output

2
0 < m AM < 1 is the
Traditional amplitude 2m AM
modulation coefficient.
m AM + 2
2
modulation (AM)
Note: In this case there’s a loss of SNR
Double side-band AM
with suppressed
amplitude of the 2 3 dB gain
carrier
(DSB-SC)

408
Single side-band AM
1 0 dB gain
(SSB)

m FM = ∆f dev / F >> 1 is a modulation

index; α = 2 ⋅ (1 + m FM + m FM ) = ∆f / F ;
Wideband frequency 3
m FM α
2 ∆f dev is a maximum deviation of carrying
modulation (FM) 4
frequency; F is an initial, modulating
frequency; ∆f is width of FM signal
spectrum

Figure 6.34. Demonstration of SNR threshold for FM signal (both axes are in logarithmic
scale). Note: if the input SNR is less than threshold then FM becomes unacceptable

Significant improvement in SNR makes FM systems widely used for broadcasting, and for VHF
and microwave communication applications. That improvement is achieved for expense of the
signal bandwidth, 2 ⋅ ∆f max = α ⋅ F which becomes much wider than that for linear types of

modulation such as AM, DSB-SC and SSB. If SNRDet , out for detector’s output, as well as the

allowable width of signal spectrum is pre-assigned (either by existing standard, or it is user-


defined) then SNRDet , in for detector’s input becomes linked to those values. That allows defining

receiver’s sensitivity, based on developments given in 6.2.5.

409
The modems such as AM, DSB-SC, SSB, and FM are just examples of how detection of the
signal may affect both, Tx-Rx structure design, and link budget calculations. Complete analysis
of all types of analog and digital modems is behind the scope of this text, and may be subject for
additional reading.

6.3.2. USE OF SPREAD-SPECTRUM DISCRETE SIGNALS


For the special class of discrete signals called spread-spectrum signals a significant
improvement in SNR may be achieved when a matched-filtering reception 1 is used. The idea of
the matched-filtering was first introduced by D.O. North in 1943 [22], and further evaluated in
late 40-s and early 50-s [23]. Below is a brief, heuristic outline of the basics of the matched-
filtering.
Consider a pulse signal s (t ) of duration T and of the arbitrary waveform that is applied to the

input of the linear two-port network resulting in output response s out (t ) of duration Tout . The goal

of the matched filter (MF) 2 is to “squeeze” (compress) the signal from input to output with
maximum possible compression as shown in Figure 6.35. Then the envelope of the output

Figure 6.35. Signal waveforms at input/output of the matched filter

pulse presumably is to be sin(t ) / t -shaped with duration

Tout ≈ 1 / ∆f , (6.118)

where ∆f is a bandwidth of the input signal. For the proper reason ∆f remains the same from
input to output of MF. Considering MF is lossless network, the total energy of the signal remains
unchanged from input to output, i.e.

1
Optimum filtering or correlation detection terms are also in use.
2
Not to be confused with abbreviation for Medium Frequency (see Chapter 1)

410
∞ ∞
E = ∫ s (t )dt = T ⋅ s = ∫s (t )dt = Tout ⋅ s out ,
2 2 2 2
out (6.119)
−∞ −∞

2
where s 2 and s out are the values of mean power 1 of input and output signals respectively.

Then, taking into account (6.118), the following relation between input and output RMS voltages
may be written:
2
sout , rms sout T
= = = T ∆f = B, (6.120)
s rms s 2 Tout

where the time-bandwidth product B = T ∆f is called pulse compression ratio when having a
deal with discrete signal. It shows how many times the pulse duration decreases from input to
output of the matched filter, and as a result, how many times increases the peak power of the
input signal (not the energy of the signal). As shown below in presence of white Gaussian
uncorrelated noise, which equally affects both, input and output signals the ratio (6.120) is
nothing but the ratio of SNR from input to output, i.e. it is a SNR gain for this type of transform.
The values of that gain may become significant, especially for the radar applications, where the
value of B may achieve 1000, and even more.
In order to develop a structure of the matched (optimal) receiver that maximizes SNR consider a
signal s (t ) of the complex spectrum

S ( jω ) = S (ω ) ⋅ exp[ jΦ S (ω )] (6.121)

that passes through MF along with the Gaussian white noise n(t ) of the spectral density N 0 . In

(6.121) S (ω ) is the amplitude and Φ S (ω ) is a phase spectrums of the input signal respectively.

Let MF has a transmission function in complex from:


H ( jω ) = H (ω ) ⋅ exp[ jΦ H (ω )] , (6.122)

where H (ω ) and Φ H (ω ) are amplitude and phase transmission coefficients of MF. If one
needs to superimpose all spectral components of signal at the output with the same phase
(constructive superposition) at any time moment t 0 , then the signal’s phase variations vs.

2
1
Actually s 2 and s out are the mean squares of the voltages within the pulse durations. They turn to
become mean powers if the voltage drops are considered on unit resistances of 1 Ohms. This
assumption does not put any restrictions on further considerations

411
frequency Φ S (ω ) must be compensated by MF in order to obtain the phase spectrum of output

signal in the form of (−ω t 0 ) . Hence Φ S , out = Φ S (ω ) + Φ H (ω ) = −ω t 0 , therefore

Φ H (ω ) = −Φ S (ω ) − ω t 0 . (6.123)

Indeed, the output response of MF will appear at the time moment of dΦ S , out / dω = − t 0 , i.e. it is

shifted from start-point of the input signal by t 0 . As pointed, the condition (6.123) allows

maximizing the output signal by superimposing all spectral components at the time moment t 0 .

To minimize the noise intensity we may assume that MF is “matched” to amplitude spectrum of
signal, i.e.
H (ω ) = a ⋅ S (ω ) , (6.124)
where a is a real constant. By combining (6.123) and (6.124) the transmission function of MF

may be expressed as a complex conjugate Sˆ ( jω ) of the spectrum of primary signal, i.e.

H ( jω ) = a ⋅ S (− jω ) ⋅ exp(− jω t 0 ) = a ⋅ Sˆ ( jω ) ⋅ exp(− jω t 0 ) . (6.125)

For further evaluations the following statements from the theory of Fourier transforms must be
recalled:
• The Fourier transform of convolution of two functions is a product of the Fourier
transforms of proper functions, i.e.

∫ s(t ′) ⋅ h(t − t ′) ⋅ dt ′ ⇔ S ( jω ) ⋅ H ( jω )
−∞
(6.126)

• If in expression (6.126) the replacement H ( jω ) = Sˆ ( jω ) is applied, then



S ( jω ) ⋅ Sˆ ( jω ) ⇔ ∫ s(t ′) ⋅ s(t ′ − t ) ⋅ dt ′
−∞
= ψ (t ) (6.127)

it becomes an autocorrelation function for s (t ) .

• Time shift property:


exp(− jω t 0 ) ⋅ S ( jω ) ⇔ s (t − t 0 ) . (6.128)

Based on (6.126) - (6.128) the expression for output signal may be written as
S out ( jω ) = S ( jω ) ⋅ H ( jω ) ⇔

⇔ s out (t ) = a ⋅ ∫ s (t ′) ⋅ s[t ′ − (t − t 0 )] ⋅ dt ′ = a ⋅ψ (t − t 0 ) , (6.129)
−∞

412
where ψ (t − t 0 ) is the autocorrelation function of the input signal 1. It may be seen, that for the

time moment t = t 0 the expression (6.129) becomes



s out (t 0 ) = aψ (0) = a ∫ s 2 (t ′) dt ′ = a E , (6.130)
−∞

i.e. the maximum value of the output signal at the time moment t = t 0 is directly proportional to

the total energy of input signal (see Figure 6.36).

Figure 6.36. Optimum reception of spread-spectrum signal: a). Block-diagram of the


signal correlation processing in matched (optimal) receiver, b). Input signal, c). Output
signal

Finally let’s derive the SNR gain based on above analyses. For the input SNR, the signal power

may be defined from (6.119) as PS , in = E / T = s 2 . The input white Gaussian noise spectral

density N 0 is evenly distributed in frequency range ∆f = f max − f min along the entire spectrum

of the signal, thus the input noise power is PN , in = N 0 ∆f . Then taking into account B = T ∆f we

may write

s2
SNRin = = E / ( N 0 B) . (6.131)
N 0 ∆f

1
(6.129) is known as Wiener-Khinchin theorem, which states, that Fourier transform of the signal power
spectrum, S 2 (ω ) = S ( jω ) ⋅ Sˆ ( jω ) is equal to autocorrelation function of the signal.

413
For the output SNR the noise power spectral density may be defined if N 0 is multiplied by

power transmission function of the MF from (6.124): N out (ω ) = N 0 a 2 ⋅ S 2 (ω ) . Then the total

output noise power may be defined by integrating the output noise power density within the
effective frequency band:
f max f max
1 1
PN , out =
2π ∫ N out (ω ) dω =
f min
2π ∫N
f min
0 a 2 S 2 (ω ) dω =N 0 a 2 E. (6.132)

Taking into account (6.130) and (6.132) the output SNR may be represented as
PS , out
SNRout = = E / N0 . (6.133)
PN , out
Hence, the SNR gain is found as
SNRout
χ= = B. (6.134)
SNRin
An important conclusion from (6.134): matched (optimal) reception of the spread-spectrum
signals allows achieving significant gain in SNR by increasing the base of signal; thus the
proper gain will allow detecting the signals, which are completely buried within the noise.
The following two types of signals with the large pulse compression ratio B are widely used in
modern radars and communication systems:
• “Colored” pulses, with linear frequency modulation (FM) waveform (Figure 6.37a ); those
signals are sometimes called "chirp" signals
• Phase-coded pulses, with several jumps of carrier’s phase between 0 and 1800 (Figure
6.37b) within signal’s duration

Figure 6.37. Examples of spread-spectrum signals:


a). “Colored” pulse of linear FM waveform (chirp pulse), b). Four-element phase-coded
pulse

414
In those two types of signals the energy is “spread” along a wide frequency range. The reader
may obtain the details about practical utilization of the compression for those types of signals in
[23 - 24].

6.3.3. DIVERSITY RECEPTION TECHNIQUE


An alternative for improvement of the reception reliability is diversity reception technique (DRT).
The use of DRT allows decreasing of the fast fade depth by altering its statistics. In other words
the same communication stability may be achieved for less power fade margin for particular
propagation conditions compared to that for non-DRT reception.
The idea of DRT is based on the use of two or more copies of the same signal (Signal-1, Signal-
2, etc.) sent to the reception point through different statistically-independent tracks, followed by
merging those signals in receiver to decrease fading depth. Those independent tracks are
called diversity branches. The independent (uncorrelated) copies of the signal are merged in
receiver by using different techniques to achieve less communication breaks. Statistically
independent copies of the signal have different fluctuation patterns, so thus if one of them falls
down the desired SNR-level, the others most likely will remain above that level, so the proper
combination of independent copies of signal may result in decrease of probability of outages,
and therefore in increase of transmission channel reliability, as illustrated in Figure 6.38.

Figure 6.38. Illustration of the fade depth decrease while combining of two statistically
independent copies of signal by using DRT technique

415
The methods of achieving statistically-independent diversity branches are:
• Space diversity, when different spatially separated receiving antennas are used: the
separation between antennas (that is called diversity base, L – see figure 6.39a) must be
greater than the spatial correlation distance, which is usually taken as L ≥ (5 ÷10 ) λ

• Polarization diversity, when two orthogonally polarized radio waves of the same
frequency and propagation path are used to carry different copies of the same signal; this
type of diversity branches are fairly effective for HF radio links
• Angle diversity, when different angles of arrivals of the radio waves are used to form the
independent copies of the signal; for instance if a dish antenna is used, then multiple
diversity branches may be formed by using multiple feeds, shifted off the focal point
perpendicular to the axial line of dish reflector
• Frequency diversity, when different carrying frequencies are used: the separation
between carrying frequencies must be large enough in order the copies to become
uncorrelated
• Time diversity, when the signal is repeated two or more times with the time shift that is
greater, than fading correlation interval in time domain; the gain in reception quality is
achieved by sufficient decrease in transmission rate
• Rake technique. When the pulse-signal arrives the reception point being carried by the
radio waves propagating along different paths, then results in time-series of the same
single pulse at the reception point. Those series are the result of different time delays
between different propagation paths; shift-summation (discrete convolution) of those
series may allow overlaying those pulses at a specific time moment, and therefore
improve SNR at the receiver’s output.

Last two of listed approaches are used mostly for digital data transmissions links, when the
discrete signals are transmitted; those two approaches are considered in details in [15].
Simplified block-diagrams of DRT-systems that allow realizing those approaches are shown in
Figure 6.39.

416
Figure 6.39. Simplified block-diagrams of basic DRT-s (two-fold DRT cases):
a). Space diversity, b). Polarization diversity, c). Angle diversity, d). Frequency-diversity

An important issue is how to combine the copies of signal at the reception point in order to
achieve maximum efficiency in SNR, or maximum communications stability.
Consider two separate diversity branches with u1 (t ) and u 2 (t ) signal voltages mixed with the

additive noise voltages u N1 (t ) and u N 2 (t ) in each channel, i.e.

1 − st branch u1 (t ) + u N 1 (t ), 
 (6.135)
2 − nd branch u 2 (t ) + u N 2 (t ) . 
Assume, that the mixes, coming from diversity branches, are combined by adding to each-other
with weighting coefficients a1 and a 2 respectively, i.e. the resultant output voltage is

U out (t ) = u out (t ) + u N , out (t ) = a1 [u1 (t ) + u N 1 (t )] + a 2 [u 2 (t ) + u N 2 (t )] . (6.136)

The power carried by the total wanted signal at the output is in direct proportion to the following
mean-square

417
u out ( t ) = [a1 ⋅ u1 (t ) + a 2 ⋅ u 2 (t )]
2 2
(6.137)

whereas the power carried by total unwanted noise at the output is in direct proportion to

u N , out (t ) = a1 ⋅ u N 1 (t ) + a2 2 ⋅ u N 2 2 (t ) .
2 2 2
(6.138)

Here u N 1 (t ) and u N 2 (t ) are the mean-square noise voltages in channel-1 and channel-2
2 2

respectively. The difference in presentations (6.137) and (6.138) comes from the fact, that u1 (t )

and u 2 (t ) are strongly correlated in contrast to u N1 (t ) and u N 2 (t ) that are uncorrelated

processes therefore their mean product becomes equal to zero. Then for the output SNR the
following expression may be written:

SNRout =
u out
2

=
[a1 ⋅ u1 (t ) + a2 ⋅ u 2 (t )] 2 . (6.139)
a1 ⋅ u N 1 (t ) + a 2 ⋅ u N 2 (t )
2 2 2 2 2
u N , out
Now we assign signal-to-noise ratios

u1 (t ) u 2 (t )
2 2
SNR1 = , and SNR 2 = (6.140)
u N 1 (t ) u N 2 (t )
2 2

to each channel before we combine them. Then, taking into account (6.140), the expression
(6.139) may be rewritten as

a1 ⋅ SNR1 ⋅ u N 1 + 2a1 a 2 ⋅ u1 (t ) ⋅ u 2 (t ) + a 2 ⋅ SNR2 ⋅ u N 2


2 2 2 2

SNRout = . (6.141)
a1 ⋅ u N 1 + a 2 ⋅ u N 2
2 2 2 2

If a1 ⋅ SNR2 ⋅ u N 1 − 2a1 a 2 ⋅ u1 (t ) ⋅ u 2 (t ) + a 2 ⋅ SNR1 ⋅ u N 2


2 2 2 2
is added and subtracted in numerator,

then after proper simplifications the expression (6.141) may be rewritten as


2
 2 
 a1 ⋅ SNR2 ⋅ u N 1 − a 2 ⋅ SNR1 ⋅ u N 2 
2

SNRout = SNR1 + SNR2 −   . (6.142)


a1 ⋅ u N 1 + a 2 ⋅ u N 2
2 2 2 2

In (6.142) the replacement

u1 (t ) ⋅ u 2 (t ) = u1 ⋅ u 2 = SNR1 ⋅ u N 1 ⋅ SNR2 ⋅ u N 2 ,
2 2 2 2
(6.143)

is used based on fact that both signals, u1 (t ) and u 2 (t ) are exact scaled copies of each-other.

An important conclusion can be made from (6.142), namely, maximum SNRout after combining

SNRout max
= SNR1 + SNR2 (6.144)

may be achieved if the expression in parenthesis in (6.142) is equal to zero, i.e. if

418
2 2
a1 u N1 a2 uN 2
= = C = const . (6.145)
SNR1 SNR2
If (6.140) is taken into account, then (6.145) may be rewritten as

a1 ⋅ u N 1 a2 ⋅ u N 2
2 2

= =C. (6.146)
2 2
u1 u2
From (6.146) the unknown weighting coefficients a1, a2 … an may be found for the general case
of n-folded diversity as follows
2 2 2
u1 u2 un
a1 = C ⋅ , a2 = C ⋅ , ………. a n = C ⋅ . (6.147)
2 2 2
u N1 uN 2 uN n
From (6.144) it may be concluded that maximum SNR (maximum SNR combining technique 1)
may achieved if the receiver transfer function of each diversity branch is adjusted individually to
keep the ratio of wanted signal (voltage or current) and unwanted noise power the same and
equal to each-other before summation. One of the appropriate block-diagrams based on this
algorithm is shown in Figure 6.40a.

Figure 6.40. Examples of two-folded DRT block-diagrams: (a) for maximum SNR
combining, (b) for equal gain combining, (c) for switching (signal auto-selection).

1
This principle is called also weighted summation technique.

419
Each diversity branch has two AGC (automatic gain control) loops: one of them to control the
signal voltage level (numerator in (6.147)), another one is to control the noise power level
(denominator in (6.147)) so thus the ratio remains constant. In order to make it applicable, a
pilot-tone is mixed up with the wanted signal at the transmission. This block-diagram is used in
most advanced systems, which allows obtaining of the maximum SNR.
The simplified version of DRT called equal gain combining is represented in Figure 6.40b. In this
case all weighting coefficients are taken equal to each-other.
Finally the selection combining is presented in figure.6.40c. In this case the diversity branch that
has greatest value of the SNR is switched (connected) to detector’s input. This principle is
sometimes called auto-selection DRT.
As a particular case we’ll consider an analytical expression for output fast fades statistics of the
auto-selection DRT to confirm the decrease of deepness of fading (see Figure 6.38 for the
qualitative demonstration). Consider the case of n-folded diversity, when signal-1, signal-2, …
signal-n from diversity branches are due to independently fading fields E1, E2, … En; each of
those fields assumed to have the same cumulative probability distribution function W (E ) . Then

integral probability distribution function is P(E ) = 1 − W (E ) . Because all branch signals are
statistically independent, then the probability of combined signal being below the E-value at the
selector’s output is the product of probabilities of individual branch signals, i.e.

Tn (E ) = T n (E ) = [1 − W (E ) ] .
n
(6.148)

Thus the probability of E-value being exceeded by all diversity branch fields simultaneously
(cumulative probability distribution function) becomes

Wn (E ) = 1 − Tn (E ) = 1 − [1 − W (E )] .
n
(6.149)

If Rayleigh distribution is considered for W (E ) as most common CDF, that describes most deep
fades of single-branch signal, then in (6.153) it may be replaced by (6.35), so
n
  E 2 
Wn (E ) = 1 −  1 − exp  −  .
2 
(6.150)
  2 σ 
Expression (6.150) is referred in section 6.1.2.d as quasi-Rayleigh distribution, which provides
statistical description of the non-Rayleigh type fades with variable depth. Expression (6.45) is
nothing but a modified version of (6.150). The difference between those two expressions is that
the parameter n in quasi-Rayleigh distribution may vary within the range ( 0, ∞ ) continuously,
whereas parameter n in (6.150) may only belong to the discrete sequence of natural numbers.

420
Recall the argument of the expression (6.45), and insert the noise power PN into numerator and

denominator:
2
 E  E2 E 2 / PN SNR
  = γ = 2
= 2
= , (6.151)
 E med  E med E med / PN SNRmed
where the ratios are replaced by proper SNR-s taking into account that signal power is in direct
proportion with proper electric field strength. Then expression (6.45) may be rewritten as

Wn (γ ) = 1 − {1 − exp[− 0.693 ⋅ q (n) ⋅ γ ]} ,


n
(6.152)

where γ is SNR relative to its median per expression (6.151). The values of q (n) may be taken
from Table 6.2 or from Figure 6.13.
As mentioned above, the expression (6.152) is only applicable to selection combining along with
the graph in Figure 6.13b. In two other considered cases, namely maximum SNR combining
and equal gain combining, the CDF, Wn (γ ) is expressed in terms of χ 2 -distribution function 1

(chi-square distribution) that is given in [15] in details. Graphs shown in figure 6.21 [17] may be
conveniently used for the estimates of the gain in fade margin (i.e. how much the fade margin
may be decreased in comparison to non-diverse reception) as a function of number of diversity
folds n and for different types of DRT combining schemes.

Figure 6.41. SNR gain for n-folded diversity reception for following DRT combing
schemes:
1 – Maximum SNR combining, 2 – Equal gain combining, 3 – Auto-selection

1
Do not confuse with χ that represents the SNR gain

421
REFERENCES

1. Gradshtein, I.S., Ryzhik, I.M. Tables of Integrals, Series, and Products. Seventh Ed.,
Elsevier, Inc., 2007
2. Vaughan, R., Andersen, J.B. Channels, Propagation and Antennas for Mobile
Communications. IEEE, 2003.
3. Dolukhanov, M.P., Propagation of Radio Waves. Translated from the Russian by Boris
Kuznetsov. Moskow, Mir Publishers, 1971
4. Rice, S. O. Mathematical Analysis of Random Noise. Bell System Technical Journal 24
(1945), pp. 46–156
5. Nakagami, M. "The m-Distribution, a general formula of intensity of rapid fading". In W.
G. Hoffman, editor, Statistical Methods in Radio Wave Propagation: Proceedings of a
Symposium held at the University of California, pp 3-36. Permagon Press, 1960.
6. Долуханов, М.П., Саакян, А.С., Энтина, Н.Н. Квазирелеевские замирания на
коротких волнах. Известия АН Арм.ССР, серия XXVIII, Апрель, 1975.
(Dolukhanov, M.P., Saakian, A.S., Entina, N.N., “Quasi-Rayleighan Faded on HF”,
Proceedings of the Academy of Sciences of the Rep of Armenia, Series XXVII, April,
1975, in Russian).
7. Rytov, S.M., Kravtsov, Ya.A., Tatarskii, V.I. Principles of statistical radio-physics.
Springer-Verlag, 1987.
8. CCIR. Report 322, “World Distribution and Characteristics of Atmospheric Radio Noise”,
Xth Plenary Assembly, Geneva, 1963.
9. Spaulding, A.D., Washburn, J.S. Atmospheric Radio Noise: Worldwide Levels and Other
Characteristics. NTIA Report 85-173, US Dept of Commerce, 1985.
10. ITU-R Recommendation P.372-7, “Radio Noise,” International Telecommunication
Union, 2001
11. Smith, A.A. Radio Frequency Principles and Applications. IEEE Press, 1998
12. Uman, M.A. Understanding Lightning. Bek Technical Publications, Carnegie, PA, 1971
13. Freeman, R.L., Reference Manual for Telecommunications Engineering. V.1, Wiley-
Interscience Publication, 2002.
14. Kraus, J.D., Radio Astronomy. McGraw-Hill Book Co. 1966.
15. Schwartz, M., Bennett, W.R., Stein, S. Communication systems and techniques. IEEE
Press, NY, 1996.

422
16. Weik, M.H. Communications Standard Dictionary. Van Nostrand Reinold Co., N.Y., 1983
17. Pratt, T., Bostian, C.W., Satellite Communications, John Wiley & Sons, 1986
18. Balanis, C.A., Antenna Theory, Analysis and Design, John Wiley & Sons, 2005
19. Stutzman, W.L., Thiele, G.A., Antenna Theory and Design, John Wiley & Sons, 1998
20. Ziemer, R.E., Tranter, W.H. Principles of communications. Houghton Mifflin Co., Boston,
1990.
21. Carlson A.B. Communication systems. McGraw-Hill Publishing Co., 1986
22. North, D.O., An Analysis of the Factors Which Determine Signal/Noise Discrimination in
Pulsed-carrier Systems, RCA Tech.Rept. PTR-6C, June, 1943 (reprinted, Proc IEEE 51,
No. 7, Jul 1963, 1016-1027; reprinted: Detection and estimation, (S. S. Haykin, ed.),
Halstad Press, 1976, 10-21)
23. Cook, C.E, Bernfeld, M., Radar Signals. An Introduction to Theory and Application,
Academic Press, 1967 (reprinted, Artech House, 1993)
24. Skolnik, M.I. Introduction to Radar Systems. 3-rd edition, McGraw-Hill Co., 2002
25. O’Flinn, M., Moriarty, E. Linear Systems. John Wiley & Sons, NY, 1987.

PROBLEMS

P6.1. Use (6.30) to confirm it satisfies normalization condition (6.9). Express median value
E med of the Rayleigh distribution in terms of parameter σ .

Answer: E med = 1.177 σ

P6.2. Find the depth of the Rayleigh fading relative to its median value (E0.1/Emed)|dB –
(E0.9/Emed)|dB and assess the result by comparing with that found from the graph on Figure 6.8a.
Answer: 13.4 dB
P6.3. The stability of HF communication radio-link is affected by log-normally distributed slow
fading with the standard deviation of σ = 8 dB, and by Rayleighian fast fading. The power
margin on transmitting side is to compensate both fades and to achieve communication stability
of 99%. Find the value of the power margin by using two approaches:
1) As a sum of margins assessed for slow and fast fades separately from Figures 6.16 and
6.12b (m = 1) respectively, and
2) From the combined distribution graphs shown in Figure 6.19 (m = 1)
Which one is larger? Why? Explain in your own words.

423
Answer: 1). 20 dB + 18.5 dB = 38.5 dB
2). 25 dB
P6.4. For the probability density distribution function (PDDF) of two-ray random interference
that is given by (6.18) find the mean value of fluctuating electric field E expressed in terms of
E1 and compare with median value given by (6.21).
x2
Hint: For any random variable x mean value is defined as x = ∫ x w ( x) dx ; for the considering
x1

case x1 = 0 , x 2 = 2 E1 .

4E1
Answer: E =
π
P6.5. A three-stage passive filter has a transformation coefficients of the stages, K1 , K 2 , and

K 3 respectively. Based on (6.91) and (6.92) show that the overall noise parameters for the filter

may be represented as N = 1 /( K1 K 2 K 3 ) , and T = T0 ( N − 1) .

P6.6. In Example 6.3 for the case-1 scenario calculate the minimum value of the gain for LNA if
predefined (required) value of the system noise temperature for the RF-unit is T = 800K ?
Answer: 11.47 dB
P6.7. Confirm (6.142) by modifying (6.141).

P6.8. Estimate the increase in maximum detectable radar range if sounding monochrome pulse
signal (case-1) is replaced by linear-FM spread-spectrum signal with time-bandwidth product
equals B = T ∆f = 50 , and optimal filtering is applied (case-2). Transmitter’s EIRP and the
system’s bandwidth are assumed to remain unchanged.
Hint: use expressions (3.48a), (6.104a), and (6.134). Take amplitude modulation depth for
monochrome pulse equal 100% (mAM = 1).
Answer: increased 2.94 times

P6.9. For the sufficient reception quality of the satellite TV broadcast FM-signal a required
signal-to-noise ratio at the detector’s output is SNRDet, out = 26.2 dB (416.9 unitless) 1. Standard
TV video signal spectrum is limited to the maximum frequency of F = 4.5 MHz. System’s RF

424
bandwidth is ∆f = 30 MHz (same as the width of FM signal spectrum). System noise

temperature is Tsys = 145K. Based on this data make the following estimates:

• Modulation index m FM and detector’s SNR gain χ (Table 6.6)

• Minimum input signal power PRx , min required to meet the value of χ (6.104a)

• Compare PRx , min with the power received in real conditions, PRx if signal power density

at the reception point (magnitude of Poynting vector) is -110 dBW/m2 ( Π = 10 pW/m2). A


dish antenna of diameter d = 0.6 m with the aperture efficiency ν = 0.67 is used at the
receiver’s input.
• Find the power margin ∆PRx = PRx / PRx , min in dB

What changes will take place if the modulation index m FM is increased? Explain.
1
Note: In real systems an additional 18.8 dB improvement in SNRDet, out is
achieved due to technical modifications that are not discussed here; so the
resultant SNRDet, out is 26.2+18.8 = 45 dB. For details refer to Pratt,T.,
Bostian,C.W., Jeremy,E.A., Satellite Communications, John Wiley & Sons, 2003
Answer: m FM = 4.44 , χ = 20 dB, PRx , min = -126 dBW, PRx = -117.2 dBW, ∆PRx = 8.8 dBW

425
LIST OF SYMBOLS AND ABBREVIATIONS

A Diffraction loss
Parameter in Okumura-Hata propagation model
′ ′ ′ ′
A , A′ Attenuations in Ionosphere (e.g., AΣ , AΣ , AD , AD , AE , AE , AF 1 , AF 1 , AF 2 )
A Magnetic vector potential, Wb/m
ADT Asymptotic diffraction theory
AM Amplitude modulation
a Major axis of the ellipse
Earth’s average geometric radius (6370 km)
B Time-bandwidth product for spread-spectrum signal
Parameter in Okumura-Hata propagation model
B Vector of the magnetic field induction, or magnetic flux density, T
BF 2 Specific attenuation in ionospheric reflecting layer F2 (at 1 MHz)
b Minor axis of the ellipse
C Parameter in Okumura-Hata propagation model
CF Fresnel cosine integral

Cε Structural constant of the tropospheric turbulences

CPDF Cumulative probability distribution function


CW Continuous wave
c Speed of light in vacuum ( 3 ⋅ 10 8 m/s)
D Antenna directivity
Parameter in Okumura-Hata propagation model
D Vector of the electric field induction (Electric flux density), C/m2
Ddiv Divergence factor

E Parameter in Okumura-Hata propagation model


E med Median electric field strength, V/m

E Average electric field strength, V/m

E Vector of the electric field strength, V/m


EIRP Effective isotropic radiated power, W

426
ELF Extremely Low Frequency (< 3 kHz)
EHF Extremely High Frequency (30 – 300 GHz)
−19
e Electron’s charge, 1.6 ⋅10 C
erf Error function
erfc Complementary error function
emf Electro-motive force
F Complex propagation factor
F Force, N
F Modulating frequency, Hz
FA Atmospheric noise factor (relative brightness temperature)
FNBW Fist null beam width of the antenna radiation pattern, rad (or degrees)
FM Frequency modulation
f Carrier frequency, Hz
∆f RF signal bandwidth, Hz
Receiver’s passband, Hz
G Antenna gain
H (0) LOS path clearance without atmospheric effect, m

H (N ) Total path clearance on LOS link in presence of the atmospheric refraction, m

H Vector of the magnetic field strength, A/m


Hh Natural unit of heights, m

HF High Frequency (3 – 30 MHz)


HPBW Half power beam width of the antenna radiation pattern, rad (or deg)
h Height, elevation, m
Planck's constant = 6.626 ⋅ 10 − 34 J ⋅ s
I Current, A
IΣ Radiation intensity, W/sr
IEEE Institute of Electrical and Electronics Engineers
IF Intermediate Frequency
IPDF Integral probability distribution function
ITU-R International Telecommunication Union – Radio Communication Sector

i= −1 Imaginary unit

427
J Vector of electric current spatial density, A/m2

JS Vector of conducting current surface density, A/m

K Transmission coefficient of TPN


k Relative distance of the reflection point on LOS microwave radio link

k Propagation constant (complex), 1/m

kB Boltzman’s constant [ 1.38 ⋅10 −23 W /( K ⋅ Hz ) ]


L Total propagation path loss
LF Propagation path loss in real conditions

L0 Free space (reference) propagation loss

Outer scale of tropospheric turbulences


Lmsd Multiple screen diffraction loss in urban area, dB

Lrst Natural unit of distance in ADT approach, m

Lr Roof-to-street diffraction loss, dB


LF Low Frequency (30 – 300 kHz)
LHCP Left Hand Circular Polarization
LOS Line-of-Sight
l eff Antenna effective length, m

l0 Inner scale of tropospheric turbulences

M Parameter in Lee propagation model


Large parameter in ADT approach
MF Medium Frequency (0.3 – 3 MHz)
Matched filter
MUF Maximum usable frequency
m Parameter in m-distribution of Nakagami
−31
me Mass of the electron ( 9.1 ⋅10 kg)

N Refractivity
Noise figure
Ne Plasma concentration, 1/m3

NT Noise power spectral density, W/Hz


n Refraction index

428
Parameter in Lee propagation model
Parameter in Quasi-Rayleigh distribution (n-distribution)
n Unit vector normal to surface
P Power, W
PM Phase modulation
PN Noise power, W

PS Signal power, W

P(E ) IPDF for the electric field fades

P Polarization vector of the unit volume, C/m2


PΣ Radiated power, W

PTx Power transmitted (applied to antenna’s input), W

PRx Power received (from the antenna’s output), W

PDDF Probability density distribution function


PEC Perfect Electric Conductor
p (m) Parameter in m-distribution of Nakagami
q Earth’s ground parameter in ADT approach
q Scattering vector

q (n) Parameter in n-distribution


R Horizontal distance between corresponding antennas along the earth surface, m
Radius, m
Rant Real part of the antenna input impedance, Ohm

RΣ Antenna radiation resistance, Ohm


RHCP Right Hand Circular Polarization
RMS Root mean square value of any variable
RF Radio Frequency
r Direct distance between two points, m
S Area of the surface, m2
S eff Area of the antenna effective aperture, m2

SF Fresnel sine integral

Sε Spatial spectrum of the autocorrelation function for dielectric permittivity

429
SHF Super High Frequency (3 – 30 GHz)
SNR Signal-to-noise ratio
s Distance scale, 1/m
T Temperature (including noise temperature), K
Tesla (Unit for the magnetic field induction/ flux density)
T E Electric field transmission coefficient (complex)
T H Magnetic field transmission coefficient (complex)
TP Power transmission coefficient
TEM Transverse Electromagnetic
TPN Two-port network
t Time, s
U (x) Attenuation factor in ADT approach
UHF Ultra High Frequency (0.3 – 3 GHz)
V Volume, m3
V ( y) Height-gain function in ADT approach
VLF Very Low Frequency (3 – 30 kHz)
VHF Very High Frequency (30 – 300 MHz)
v Velocity of the radio wave, m/s
W (E ) CPDF for the electric field fades

W Intrinsic impedance (complex), Ohm


We Ionization energy, J

Spatial density of the energy of electric field, J/m3


Wm Spatial density of the energy of magnetic field, J/m3

WLAN Wireless local area network


w (t ) Airy function

w (E ) PDDF of the electric field fades


X Normalized distance in ADT approach
X ant Imaginary part of the antenna input impedance, Ohm

x Cartesian coordinate

Relative electric field strength (


x = E / E med )

x Numerical distance in W&VdP method for ground wave propagation

430
x0 Unit vector in Cartesian system

y Normalized height in ADT approach


Cartesian coordinate
y0 Unit vector in Cartesian system

Z ant Antenna input impedance, Ohm

z Cartesian coordinate
z0 Unit vector in Cartesian system

α Attenuation coefficient, Np/m (or dB/m)


Angle, rad
β Phase coefficient of the radio wave, rad/m
Angle, rad
Parameter in generalized Rayleigh distribution function (Rice distribution)
Γ E Electric field reflection coefficient
Γ H Magnetic field reflection coefficient
ΓP Power reflection coefficient
γ Slant angle (elevation angle)

∆ The allowable average height of the surface roughness


δε Loss angle of the dielectric medium, rad

ε Complex relative dielectric permittivity

ε0 Absolute dielectric permittivity of free space, (1 / 36π ) ⋅10 − 9 F/m

ξ Number of collisions in ionospheric plasma, 1/s

ξε Relative fluctuation of the dielectric permittivity ( ξ ε = ∆ε / ε )

η ant Antenna efficiency

Θ Geocentric angle, rad


θ Zenith angle in spherical coordinates, rad (or dergree)
κ fast Fast fading factor

κ slow Slow fading factor

λ Wavelength, m
µ Relative magnetic permeability

431
µ0 Absolute magnetic permeability of free space (vacuum), H/m

ν Antenna aperture efficiency

Π Pointing vector (power flow spatial density of the radio wave), W/m2
ρ Electric charge volumetric density, C/m3
σ Conductivity of the medium, S/m
Standard deviation of the normally distributed random variable
σ sc Scattering cross-section of the turbulent troposphere (per unit volume), 1/m

σ RCS Scattering cross-section of the target in radar applications, m2

Φ Phase of the radio wave, rad


Scalar magnetic potential, V
ΦF Phase of propagation factor, rad

ΦΓ Reflection phase, rad


ϕ Azimuth angle in spherical coordinates
Angle of incidence
χ Detector’s SNR gain
Ψ Amplitude radiation pattern
Angle of transmission of the radio wave on the interface between two media
ψε Spatial autocorrelation function of the random dielectric permittivity

Ω Solid angle, sr
ω Angular frequency, rad/s
∇ Nabla vector operator
∇2 Laplacian Operator

432

View publication stats

S-ar putea să vă placă și