Feature Article: DOI. No. 10.1109/MAES.2016.

150285

Relevance of Electromagnetics in Wireless Systems

Design
Tapan K. Sarkar, Syracuse University, Syracuse, NY, USA
Eric L. Mokole, Private Consultant, Burke, VA, USA
Magdalena S. Palma, Universidad Carlos III de Madrid, Madrid, Spain

In his seminal article [3], Gabor observed

INTRODUCTION
Communication theory has up to now developed mainly on
Current research on wireless radio frequency (RF) systems of-
mathematical lines, taking for granted the physical signifi-
ten makes unphysical simplifying assumptions and treats the as-
cance of the quantities which figure in its formalism. But
sociated signal processing and electromagnetic (EM) analyses
communication is the transmission of physical effects from
independently, resulting in improper inclusion of the underlying
one system to another, hence communication theory should
physics (EM theory) of such systems and consequently leading
be considered as a branch of physics. Thus it is necessary to
to inaccurate and erroneous performance predictions. To achieve
embody in its foundations such fundamental physical data as
accurate modeling and predictions, EM theory and signal pro-
the quantum of action, and the discreteness of electric charges
cessing must be intelligently merged. Examples of such errors
…. We observe first that all electric signals are conveyed by
commonly appearing in both fields are the following: not includ-
radiation. Even if lines or cables are used in the transmission,
ing the platform on which an antenna is mounted, which leads
by the Maxwell-Poynting theory the energy can be located
to completely erroneous predictions of antenna coverage [1], a
in empty space. Hence we can apply to our problem the well
serious issue for designing relevant antennas for RF systems,
known results of the theory of radiation.
and optimizing transmitted waveforms without including proper
differentiability requirements imposed by Maxwellian electro- The point here is that in the design of the physical layer one has
magnetics yielding inaccurate and physically unrealizable wave- to apply other concepts besides the principles of communication
forms [2]. Because discussions of these topics appear in [1] and theory.
[2], this article concentrates on the different concepts of chan- More recently, this point was further investigated by Ivrlac and
nel capacity and their implications and on the antenna and its Nossek [4], where they state
relationship to the maximum power transfer theorem. Although
Electromagnetic field theory provides the physics of radio
the discussion addresses communication systems, much of it is
communications, while information theory approaches the
applicable to noncommunication systems. A future of improved
problem from a purely mathematical point of view. While there
performance and better spectral harmony requires coding at RF,
is a law of conservation of energy in physics, there is no such
initially envisioned by Shannon and subsequently investigated
law in information theory. Consequently, when, in informa-
by Viterbi, which is similar to the coding in global positioning
tion theory, reference is made (as it frequently is) to terms like
satellite (GPS) and satellite communication. Such a methodol-
energy, power, noise, or antennas, it is by no means guaranteed
ogy can also be carried out in radar in which, for example, a
that their use is consistent with the physics of the communica-
radar can transmit coded waveforms, such as a Barker code, to
tion system. Information theory serves well as the mathematical
increase detection capabilities.
theory of communication. However, it contains no provision
that makes sure its theorems are consistent with the physical
laws that govern any existing realization of a communication
Authors’ current addresses: T. K. Sarkar, Syracuse University,
system. Therefore, it may not be surprising that applications of
Department of Electrical Engineering and Computer Science,
Syracuse, NY 13235-5060, USA, E-mail: tksarkar@syr.edu. E. information theory or signal processing, as currently practiced,
L. Mokole, Private Consultant, 5816 Fitzhugh Street, Burke, easily turn out to be inconsistent with fundamental principles of
VA 22015-3625, USA. M. S. Palma, Departmento de Teoría de physics, such as the law of conservation of energy.
la Senal y Comunicaciones, Universidad Carlos III de Madrid,
28911 Leganes, Madrid, Spain. Ivrlac and Nossek further elaborate [5]
Manuscript received December 28, 2015, revised April 15,
2016, and ready for publication June 8, 2016. There exist a number of fundamental principles in physics
Review handled by M. Greco. which can be stated as conservation laws, meaning that there
0885/8985/$26.00 © 2016 IEEE are quantities which can be calculated for a physical system at

8 IEEE A&E SYSTEMS MAGAZINE OCTOBER 2016

Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
 one time, and when recalculated at a later time come out the
same [6]. An example is the law of conservation of energy ….
The movement of the planets around the sun can be obtained
solely by following the implications of the laws of conserva-
tion of energy and angular momentum [7]. Moreover, the
conservation laws are also deep principles for they relate to
symmetry in physics [8]. For instance, conservation of energy
implies that the laws of Nature are time-invariant, and vice
versa. Also in signal processing and information theory, the
concept of energy is a prominent one. It appears as the energy
required to transfer one bit of information, or one symbol of
the signal alphabet, or sometimes in form of transmit power,
i.e., the rate at which energy must be supplied per unit of time
to keep the communication going. Yet, it is interesting that
the fact that energy is conserved, this very fact that is of such
fundamental importance in physics, apparently plays no role in
standard textbooks on information theory [9], signal process-
ing [10], communication theory [11] or signal theory [12]. The
authors are also not aware of any research work in these areas
where the remarkable fact that energy is conserved is explored
or discussed. The reason for this strange absence of conserva-
tion laws in signal processing, information theory and related
disciplines seems to be the fact that inputs and outputs are
described by single variables each, instead of by a pair of con-
jugated variables, like position and momentum in Hamiltonian
mechanics [8], or voltage and current in circuit theory [13].
Therefore, to address the development of the physical layer ad-
equately and to ensure that systems perform according to design
criteria, it is necessary to merge the principles of electromagnet-
ics, which are primarily related to antennas and maximum power
transfer, to the issues of channel capacity and how it can be quanti-
fied using the principles of physics and the radiation efficiency of
antennas rather than use of the maximum power transfer theorem.
PROPERTIES OF ANTENNAS THAT ARE PARAMOUNT IN
SYSTEM DESIGN
Antennas have unique properties that must be understood and cor-
rectly applied. First, note that an antenna is a spatial filter. Unfortu-
nately, the EM community only considers this topic when developing
methodologies such as beamforming. An antenna is also a temporal
OCTOBER 2016 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
filter, as it can be tuned to a desired bandwidth, which the signal
processing community exploits to develop innovative algorithms
primarily based on the temporal properties of the signals. However,
an antenna is simultaneously both a temporal filter and a spatial filter
that are connected. This space-time relationship and its associated
properties can be characterized in an exact fashion with Maxwell’s
equations. This set of equations is one of the few in physics that has
withstood the erosion and corrosion of progress; even the inclusion
of relativity has had no effect on Maxwell’s equations, because it is
built into them. Recognition of this important space-time relation-
ship can make it possible to address problems that may not be solved
by using temporal or spatial properties alone or independently.
Second, an antenna does not radiate power; it radiates pow-
er density, which is related to the electric (E(w)) and magnetic
(H(w)) fields—the exact relationship is given by the Poynting vec-
tor (E(w)H*(w)/2), where * denotes complex conjugation. When
a power density radiated by a transmitting antenna is incident on
a receiving antenna, the receiving antenna integrates the incident
electric field intensity (which has units of volt per meter) to gener-
ate a voltage at its terminals. In other words, an antenna is excited
by a voltage source or a current source and not by a power source.
Once the voltage applied to the antenna and the environment into
which it subsequently radiates (which dictates the input impedance
of the antenna) are known, the input power applied to the antenna
can be obtained. When comparing performances of systems con-
taining radiating antennas, it is imperative that the input power to
the radiating antenna be kept fixed; otherwise, the comparison may
not be scientific. Hence, it is relevant to deal with antenna radiation
efficiency rather than use of the maximum power transfer theorem
that has no direct connection with the total amount of applied input
power. Note that in nature, the term power amplifier is a misnomer,
as no such device exists. The only physically realizable amplifier is
either a voltage amplifier or a current amplifier that converts the in-
put direct current (DC) power into alternating current (AC) power;
hence, the terminology power amplifier leads to misunderstand-
ings. Consequently, an antenna excitation needs to be implemented
and analytically modeled using one of these two sources—either
a voltage source or a current source, then focusing on the design
of the antenna, based on the radiation efficiency, which relates the
fraction of the input power that is radiated from it.
The last point addresses an important issue in electrical en-
gineering in which two principles are of paramount importance:
9
Relevance of Electromagnetics in Wireless Systems Design
convolution and correlation. Convolution is related to reciprocity
and allows one to find the total response of a system in the trans-
formed domain as the product of the input with the transfer func-
tion; whereas in the original domain, it is worked out as a convolu-
tion. It is well-known that the principle of superposition applies
to convolution. However, the principle of correlation is related to
power as it is equal to VI* (voltage V times the conjugate of the
current I in the frequency domain). Note that superposition does
not apply to correlation; therefore, power cannot be superposed,
except under special conditions! In short, convolution is additive,
and correlation is not additive, in general.
To illustrate these points, consider two different transmitters
that radiate ideal plane waves with power densities of 100 and 1
W/m2 that are allowed to interact with each other as it propagates
through space. Even though one of the waves has only 1% of the
power density of the other, if the two waves interfere construc-
tively or destructively, as they propagate through free space, the
resulting variation in the resultant power density is neither 101 (=
100 + 1) nor 99 (= 100 − 1)W/m2 but rather is
(1)
both of which are derived from E2(2h) = |E(w)H*(w)2| = |E|2(2h),
where E is the electric field intensity in V/m and η is the impedance of
free space (120p = 377 Ω). Thus, the variation of the received power
density is 121 or 81 W/m2, because electric fields only add or subtract
and the power or power densities do not. Consequently, the differ-
ence between the maximum and minimum received power densities
is 40%, instead of the 2% change calculated when inappropriately
adding and subtracting the power densities, because it is the electric
field related to the voltages in an equivalent receiving circuit or the
magnetic field related to the currents in the circuit that can be added
in the electrical engineering context—that is, the principle of super-
position applies to voltages/currents and not to the powers or power
densities. Therefore, if one uses two antennas and each of them radi-
ates 1 W/m2, the total variation of the received field will be between
4 W/m2 and 0 W/m2. Thus, does the result of 4 W/m2 imply there is
some sort of power gain generated by using a multiantenna system?
Of course, the answer is no, because power densities cannot be added
because superposition only applies to the voltages (electric fields)
and currents (magnetic fields) in a system. Also, if one talks about the
increase in the resultant power density then one must also mention
about the power loss to 0! This discussion shows the relevant phys-
ics of signal reception, and nonacceptance of this basic fact can lead
to designs of unrealistic systems and inaccurate predictions of their
performance. The other issue is that these principles require vector
properties in computing the total electric field; therefore, the phases
of the incident fields are relevant and scalar addition is not correct.
WIRELESS TRANSMISSION OF ULTRAWIDEBAND (UWB) SIGNALS
Another relevant issue is that often in the published literature, an
antenna is characterized by a point source generating an isotropic
omnidirectional radiation pattern. In reality, an isotropic omnidi-
rectional point radiator does not exist in electromagnetics. The
10 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
smallest physically realizable radiating current element is a short
(Hertzian) dipole, which can produce omnidirectional patterns only
along certain directions. The practical significance of this truth is
that a propagation path loss is independent of frequency, and by
proper choice of transmitting and receiving antennas, it is possible
to transmit a wideband signal in free space, for example, without
any distortion. This point will be illustrated by using measured data
and illustrate how a 40 GHz bandwidth waveform can be transmit-
ted and received through free space without any distortion.
For illustration purposes, consider the two forms of the Friis
transmission formula in (2),
(2)
where Pr is the power obtained at the receiving antenna, Pt is the
power radiated by the transmitting antenna, Gr and Gt are the gains
of the receiving and transmitting antennas with respect to an iso-
tropic radiator, λ is the wavelength of transmission, and R is the
separation distance between the transmitting and receiving anten-
nas. The first form (middle) is conventionally used, and the second
one (right) is a special form that illustrates the principle of UWB
transmission of waveforms when the transmitting and the receiv-
ing antennas are factored into the channel characterization. In the
second part, AR is the effective area of the receiving antenna and is
related to the gain of the receiving antenna by GR = 4p AR/l2. Now,
if we focus our attention on the second form of Friis transmission
formula, we observe that a UWB transmission without any distor-
tion can be achieved simply by transmitting a waveform with an
antenna that has a nearly constant gain with frequency, instead of
with a nonexisting isotropic omnidirectional point radiator. In ad-
dition, if the receive antenna has an effective aperture whose size
does not vary with frequency, we will achieve our goal of distor-
tionless transmission and reception of UWB signals.
To verify experimentally the various theoretical simulations that
were presented earlier, we include here the experimental results ob-
tained by Andrews [14], [15] and the theoretical analysis that confirms
those experimental results [16]. A miniature UWB antenna range
(Figure 1) was designed to demonstrate the principles of UWB trans-
mission, reception, and propagation. The UWB metrology antennas
demonstrated on this range include the conical, transverse electro-
magnetic (TEM) horn, and waveguide horn antennas in addition to
others. The transmitting and the receiving antennas were placed on a
36-cm × 44-cm aluminum plate. The UWB test input signal for this
antenna range was a 4-V, 9-ps rise time step of the Picosecond Pulse
Laboratory signal generator (Figure 2). All of the waveforms shown
in this section were measured by using an HP-54752B, 50 GHz, 9-ps
rise time oscilloscope. A 35-cm, 9-ps rise time, Gore, subminiature
version A coaxial cable was used to connect the antennas to the oscil-
loscope. Summing all the distortions, the composite rise time of the
pulse generator, coaxial cable, and oscilloscope was 16 ps.
A conical antenna was used for transmission, because it has a
gain that is essentially constant over a wide bandwidth primarily
along the broadside direction. Furthermore, the conical antenna sus-
pended over a large metal ground plane is the preferred antenna for
transmitting known transient EM waves. This type of antenna, used
OCTOBER 2016

Figure 1.
Miniature UWB antenna range. Antennas are a conical antenna, a TEM
horn, in addition to a monopole, and D dot.
Figure 2.
A 4-V, 9-ps rise time, step pulse used for UWB antenna testing. Mea-
sured by 50 GHz, 9-ps rise time, HP sampling oscilloscope (Keysight
Technology-current company of Agilent which evolved from Hewlett
Packard [HP]). Vertical and horizontal divisions respectively are 750
mV/div and 10 ps/division.
by National Institute of Standards and Technology as their reference
standard for a transient transmitting antenna, radiates an EM field
that is a perfect replica of the driving-point voltage waveform. How-
ever, when the pulse from the feed point reaches the top of the cone,
it is reflected back, so the perfect replica property of the transmit-
ted pulse no longer holds, unless resistive loading on the far end of
the antenna is used to help suppress multiple reflections. The upper
bandwidth of a conical antenna is mainly determined by the fidelity
of the coaxial connector to the transition region of the conical anten-
na. Its gain along the broadside direction is essentially independent
of frequency. If a conical antenna is used as a receiving antenna, its
output is the integral of the incident E field, because from the fun-
damental EM principles, it is clear that the impulse response of an
antenna on transmit is essentially the time derivative of the impulse
response of the same antenna on receive [14]! For these measure-
ments, the conical antenna was constructed from a brass sheet and
had a half solid angle of 12° and a height of 7 cm (Figure 1).
A TEM horn is the most preferred metrology for a receiving an-
tenna to make direct measurements of transient EM fields. A TEM
horn antenna is basically an open-ended parallel-plate transmission
OCTOBER 2016 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
Sarkar, Mokole, and Palma
line. Again, the rational for choosing it as a receiving antenna for this
example is that it has a fixed aperture over a wideband. A TEM horn
typically tapers from a large aperture at the receiving end down to a
small aperture at the coaxial connector output. The height-to-width
ratio of the parallel plate is constant along the length of the antenna
to maintain uniform characteristic impedance. However, to optimize
sensitivity, most TEM antennas are designed with a 100-Ω antenna
impedance. Practical UWB TEM horn antennas are usually designed
with resistive loading near the mouth of the antenna to help sup-
press multiple reflections. The upper bandwidth of a TEM antenna is
mainly determined by the size of its aperture and secondarily by the
transition from the parallel plate to the coaxial connector. To demon-
strate faithful UWB radiation and propagation, the conical antenna
and TEM horn in Figure 1 were used on the test range. The TEM horn
was also fabricated from brass with a length of 15 cm. The height at
the mouth was 2.6 cm, and its width was 3.8 cm. At the mouth, it was
supported by a nylon spacer. The approximate incline angle was 9°.
The 4-V 9-ps rise time pulse (Figure 2) was inputted directly to the
conical antenna. The output of the 15-cm TEM horn antenna was
connected through the 35-cm cable to the 50 GHz oscilloscope and
had an impedance of 106 Ω. The conical antenna was 7 cm high and
had an input impedance of 132 Ω. For the experimental data to be
presented, the separation between the transmitting conical antenna
with the receiving TEM horn antenna was 25 cm. The received out-
put at the load of the TEM horn antenna has a 23-ps rise time step
waveform that is in good agreement with the input waveform (Figure
2). The received signal when the two antennas were separated by 5 m
(Figure 4) is exactly the same in wave shape to the 25-cm path signal
(Figure 3), except that it is weaker in amplitude. It is clear that very
little dispersion/distortion is introduced by the transmitting-receiving
system, which was also independently verified through simulations
[16]. The path loss then is independent of frequency!
Figure 3.
Transmission from a conical antenna to a TEM horn antenna over a
25-cm path.
Figure 4.
Transmission from a conical antenna to a TEM horn antenna over a 5-m path.
11
Relevance of Electromagnetics in Wireless Systems Design
Note that an antenna does not radiate DC. Because these are
waveforms observed in a time domain reflectometer, note that one
is observing only the positive cycle on the oscilloscope. However,
by simply observing Figures 3 and 4, one would get the impres-
sion that DC is being radiated. This observation is not true. In the
figures, we focus only on the rise time of a single pulse to illus-
trate the fidelity with which the wave shapes have been radiated
and received by the different antennas. Also, Figures 3 and 4 illus-
trate that path loss is independent of frequency if the proper anten-
nas are chosen for transmission and reception of waveforms. The
key point here is that the useful popularized properties of antennas
in the frequency domain do not apply to the analysis of antennas
in the time domain. For example, an antenna operating in the time
domain has no antenna pattern; hence; the popular belief that the
radiation pattern of the antenna is equivalent to the reception pat-
tern of the same antenna does not hold in the time domain. More-
over, the impulse response of an antenna on transmit is the time
derivative of the impulse response of the same antenna on receive.
In addition, a single antenna is a nonreciprocal system; whereas
the transmission and reception using two antennas are reciprocal.
The previously mentioned figures have tremendous implications
when dealing with broadband wireless systems. Many researchers are
used to performing channel modeling of free space for a broadband
wireless system routinely. From the perspective of the Figures 1–4, it
appears that modeling of a broadband wireless system in free space is
equivalent to channel modeling of free space, which is dispersionless!
Many researchers only look at the channel and are not considering the
effect of the transmitting and receiving antennas, which are integral
parts of a wireless system. Alternately, antennas are often modeled
as point sources, thereby excluding the fundamental physics from
the analysis. Now, if one looks at the Friis transmission formula and
selects the appropriate transmitting and receiving antennas, then it is
possible to transmit and receive a UWB signal traveling through free
space without distortion! On the point, vector EM problems differ
from scalar acoustic problems. Specifically, observe that an acous-
tic signal is not distorted when it is converted to electrical energy
through any microphone or when the electrical energy is converted
to acoustic energy through any loudspeaker. However, the problem
is quite different for the vector EM problem. In the time domain, the
impulse response of an antenna when it operates in the transmit mode
is the time derivative of the impulse response when the same antenna
operates in the receive mode. Hence, even a point source, which is
often used for modeling the antennas, differentiates the input wave-
form on transmit. Therefore, the modeling of any wireless system and
moreover that of a broadband system must include the antenna effects
to obtain a physically meaningful solution [14], [15], [16]. We next
discuss another important issue related to antennas.
MAXIMUM POWER TRANSFER
Another popular property that invariably arises in antenna system
design is the principle of maximum power transfer.
Maximum Power Transfer Theorem
For a source network, the maximum amount of power will be dis-
sipated by a load resistance when that load resistance is equal to
12 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
the Thevenin or Norton resistance of the network supplying the
power. If the load resistance is not equal to the Thevenin-Norton
resistance of the source network, the network’s dissipated power
will be less than maximal.
Unfortunately, this highly misused theorem is irrelevant to
antenna systems, because an antenna is a radiator of fields that
are excited by a voltage source. In an antenna system, the goal
should be to improve the radiation efficiency of an antenna and
not to focus on the maximum power transfer, just like what is
done in simple household appliances, when increased efficiency
is the objective and not maximum power transfer. If the appli-
ance industry had sought to maximize power transfer, it would
have achieved a mere 50% efficiency, with half of the power lost
(wasted) because it is not absorbed by the desired load and is
equivalent to the amount radiated by the antenna. Therefore, it
is necessary to pay attention to the efficiency of power or energy
transfer and not to maximum power transfer. This approach is
illustrated next.
Consider the electrical circuit of a voltage divider in Figure 5,
where the antenna is characterized by the load RL. Recall that an
antenna is excited by a voltage source and not by a power source.
The total input power to the antenna is calculated by multiplying
the excitation voltage VS and the input current I. Thus, what is re-
quired here is efficiency in converting most of the power from the
source to the load. The efficiency η of the circuit is the proportion
of the power dissipated in the load to all the power dissipated in the
circuit (i.e., supplied by the source)
(3)
Maximum power is transferred to the load when the internal re-
sistance of the source RS equals the resistance of the load RL, that
is, RL = RS, which yields an efficiency of h = 0.5. When the exter-
nal load resistance RL can be varied and the internal resistance of
the excitation source RS is kept constant, then when RL = ∞, the
radiation efficiency is h = 1.0. Alternately, if the load resistance
RL is kept fixed, then maximum power transfer from the source to
the load occurs for zero internal resistance; that is, RS = 0 which
implies h = 1.0. However, if the source resistance RS is kept fixed
and the external load resistance RL is varied, then h = 0, when
RL = 0. The behaviors of efficiency and power transfer versus
RL/RS are illustrated in Figure 6. Thus, in household appliances,
where the goal is to obtain the maximum energy transferred to
the load, one is aiming for an efficiency exceeding 90%, which
is quite different from maximum power transfer where the ef-
ficiency of conversion is only 50%, and therefore is not a good
practical solution! This principle becomes apparent when a com-
parison of two systems is made by keeping the input power to
each system exactly the same. To calculate the input power for
a general AC circuit, it is also apparent that one needs to have
values of both the voltage and the current. Just considering the
voltage and relating its correlation to power spectral density only
applies to stationary systems, where there are no inductors and
capacitors in the system because they destroy the stationarity as-
sumption—thus, stationarity is only valid for resistive circuits.
Therefore, to compute the radiated power density in antenna
OCTOBER 2016

Figure 5.
A voltage divider.
theory, one needs to have the quantifications of both the electric
and the magnetic fields. Unless one is operating in the far field of
an antenna, where only one of the field components provides the
power density, as the electric and the magnetic fields are related
by the impedance of free space (377 Ω), both the electric and the
magnetic field components are required to evaluate the radiated
power density. This issue will be addressed later.
In radio circuits, the internal resistance of a source of power
cannot usually be reduced to a small value, so the load resistance
is made equal to it for maximum power transfer. The powers in-
volved are very small, and losses are unimportant—only the
amount of power that comes through is important. The situation
is quite different for power transmission, when losses are impor-
tant—or equivalently for the antenna radiation problem where ef-
ficiency is paramount! Clearly, maximum power transfer does not
coincide with maximum efficiency. Application of the maximum
power transfer theorem to AC power distribution will result in nei-
ther a maximum nor high efficiency. The goal of high efficiency is
more important for AC power distribution, which dictates a rela-
tively low generator impedance compared to load impedance and
consequently should be the approach when deploying antennas.
For example, similar to AC power distribution, high-fidelity au-
dio amplifiers are designed for relatively low output impedances
and relatively high load impedances of the speakers. The ratio of
“output impedance” to “load impedance” is known as the damping
factor and is typically in the range of 100 to 1,000. Also, maximum
power transfer does not coincide with the goal of lowest noise.
For example, the low-level RF amplifier between an antenna and a
radio receiver is often designed to have the lowest possible noise.
This design often requires a mismatch of the amplifier’s input im-
pedance to the antenna instead of that dictated by the maximum
power transfer theorem.
In summary, the maximum power transfer theorem states that
the maximum amount of power will be dissipated by a load re-
sistance if it is equal to the Thevenin or Norton resistance of the
network supplying the power resulting in a 50% efficiency, as
only half of the input power is dissipated in the load. This theorem
does not satisfy the goal of maximum efficiency (a transformer
or a heater in a power system needs to be >90% efficient), where
the supplied power remains constant. Note that under conditions
of maximum power transfer, as much power is dissipated in the
source as in the load. The maximum power transfer condition is
used in communication systems (usually at RF) when the source
resistance cannot be made low, the power levels are relatively low,
OCTOBER 2016 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
Sarkar, Mokole, and Palma
Figure 6.
Efficiency versus power transfer.
and it is of paramount importance to supply as much signal power
as possible to the receiving end of the system (the load).
Use/Misuse of the S-Parameters
The previous discussion leads to a related topic, the use of the S-
parameters in analyzing antenna problems. Unfortunately, the S-
parameters are not suitable for characterizing antenna problems
when the impedance changes as a function of frequency. Treating
the forward and the backward waves in the formulation of the S-
parameters as a sum of weighted voltages and currents makes it
possible to apply the principle of superposition of power in the S-
parameter formulation, because the waves representing power are
related to the voltages and currents to which superposition applies!
In antenna problems, one wants to increase the radiation efficiency
just as in household electrical appliances and to avoid involve-
ment in power transfer, as maximum power transfer implies an ef-
ficiency of only 50% with half of the generated power wasted! The
crucial takeaway is that an antenna does not radiate power; rather
it radiates power density, which is related to the radiated electric
field. Therefore in antenna design, the goal should be to radiate the
maximum electric field intensity as the input power to the system
is fixed and not to dwell on power where the principle of superpo-
sition does not apply!
As an example, consider a center fed dipole antenna [17] of
length 0.475 λ with a half-length of 0.2375 λ. This structure is
resonant as for an input exciting voltage at the center fed delta
gap of 1.0 V, the current at the feed point is 13.84 − j 0.0227
mA. The broadside radiated field is 0.063 − j 0.82 V/m for this
excitation. Therefore, for an input V-mA (volt-milliamperes) of
1.0 the maximum absolute value for the radiated field is 0.0594
mVA (milli-Volt-amperes). Next, consider a dipole of length 0.52
λ that is nonresonant, as its half-length is 0.26 λ. For a center
fed applied voltage of 1.0 V, the complex value of the input cur-
rent in mA is 5.72 − j 4.9. The value of the radiated field in the
broadside direction is 0.39 − j 0.37 V/m. Thus, for an input V-mA
(volt-milliamperes) of 1.0, which is the same as that of the previ-
ous case, the maximum absolute value for the radiated field is
0.0714. Hence this nonresonant antenna radiates more energy in
the broadside direction for the same input power in V-A (Volt-
amperes). Hence, a resonant structure does not imply that it is
13
Relevance of Electromagnetics in Wireless Systems Design
going to radiate more power—a fallacious assumption often used
in the matching of antennas, where it is assumed that if the cur-
rent is resonant the structure will radiate more power! Hence, in
antenna design, the emphasis should be on radiation efficiency
and not on the maximum power transfer principles.
The final topic addressed in this section are the near field and
far field of antennas and the computation of the power density to
characterize an efficient radiating antenna. An antenna has no pat-
tern nulls in the near field, where the radiating wave fronts are not
planar but rather spherical. The antenna pattern is only defined in
the far field, where the wave fronts radiated by the antenna are ap-
proximately planar. In the near field, the radiated power is a com-
plex number with the real and imaginary parts associated with radi-
ation and the input reactance of the antenna, respectively. In the far
field, the radiated power becomes real, and the wave fronts become
nearly planar, generating a transverse EM propagating wave; then it
is possible to calculate the power only in terms of the electric field
or the magnetic field, as they are related in the far field by the char-
acteristic impedance of the medium (377 Ω). To the point, an an-
tenna pattern null can only be produced in the far field. In the near
field an antenna has no pattern nulls and so space division multiple
access (SDMA) has no relevance in the near field of an antenna, as
in that region, an antenna radiation pattern cannot be defined.
As a rule of thumb [17], if one is within a radius of 2D2/λ from
the broadside direction of the antenna, where D is the largest di-
mension of the antenna and λ is the operating wavelength, then one
will be in the near field of the antenna, beyond which the far field
of the antenna starts. For example, if one is operating a half-wave
dipole at a frequency of 1 GHz, then the far field will start at a dis-
tance of 2D2/λ = λ/2 = 0.5 m. However, if the antenna is deployed
on top of a 20-m tower over the Earth, representing an imperfect-
ly conducting ground, the equivalent dimension of the radiating
source will not be λ/2 = 0.5 m; rather, it will be 40 m, because the
equivalent aperture of the radiating system will not only involve
the antenna, but also the mounting structure above the ground plus
its image below the ground. In that case, the far field will start at
2D2/λ = 2 × 40 × 40/0.3 = 10.6 km, so the antenna pattern will
begin at a distance that is outside the cell (typically a cell size in
wireless communication will have a maximum of 5 km in radius).
This simple, yet misunderstood, concept about where a far field
begins leads to erroneous conclusions about base station antenna pat-
terns and the performances of communications systems. Essentially,
antenna patterns cannot be defined for typical base station place-
ments because the antennas will have no pattern nulls in the near
field that cover the entire cells of the base stations. In particular, this
concept has had tremendous relevance for the implementation failure
of SDMA in designing cellular systems, because they all typically op-
erate in the near field, where an antenna pattern cannot be character-
ized, as is clearly seen from the basic physical principles of antennas.
DISCUSSIONS ON CHANNEL CAPACITY FROM A SYSTEM
VIEWPOINT
In this section, the three notions of communication channel capac-
ity by Shannon, Gabor, and Hartley-Nyquist-Tuller are discussed.
14 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
In particular, the physical basis, or lack thereof, issues, and appro-
priate usage for each formulation are explored.
SHANNON CHANNEL CAPACITY
The most popular form of the channel capacity theorem was devel-
oped by Claude Shannon [18]. As he pointed out
The ordinary capacity C of a noisy channel may be thought of
as follows. There exists a sequence of codes for the channel of
increasing block length such that the input rate of transmission
approaches C and the probability of error in decoding at the
receiving point approaches zero. Furthermore this is not true for
any value higher than C. In some situations it may be of interest
to consider, rather than codes with probability of error approach-
ing zero, codes for which the probability is zero and to inves-
tigate the highest possible rate of transmission for these codes.
This rate C0 is the main object of investigation of the paper.
Consequently, if there is no coding, then the channel capacity car-
ries little information.
This point was very well illustrated by Viterbi [19]:
From the beginning, the additive white Gaussian noise
(AWGN) channel was a favorite vehicle of information theo-
rists. But the launch of the first artificial satellite generated
a need for conserving transmitted power over a real channel
that is closely approximated by this model, and thus served
as a powerful stimulus to further theoretical work, particu-
larly toward the goal of minimum-complexity decoding. In
the sixties, the demands of satellite networks carrying large
amount of data traffic increased the need for coding. In fact,
by the end of sixties the state of the art of digital technology,
together with the evolution of the theory, rendered feasible
coding for the AWGN channel with an error probabilities of
10─5, could conserve 5-6 dB of power relative to the uncoded
operation, even at data rates in megabits per second. To gauge
the economic advantage, consider that an alternative method
of gaining 5 dB power would be to more than triple the area
of the receiving antenna.
Hence, the contribution of Shannon made satellite communi-
cation possible, where the received signal can be well below the
level of thermal noise and the information can still be extracted in
such an environment if adequate redundancy is introduced in the
form of a code at the RF stage of transmission. Note that accord-
ing to Shannon the transmitter is the coder [20]. Shannon further
noted, “An encoding or transmitting element. Mathematically this
amounts to a transformation applied to a message to produce the
signal, i.e., the encoded message.” He also talks about pulse code
modulation and pulse position modulation in which the coding is
done at the transmitter stage at the RF frequency. Currently, there
are two popular systems that code at RF, namely, satellite com-
munication and GPS. In wireless communication, coding is done
at baseband and not at the final RF stage, unless one uses totally
digital transmission as advocated by Viterbi [21]. Some radars use
the Barker code during transmission, and this coding can be con-
sidered to be part of this group.
OCTOBER 2016
 Sarkar, Mokole, and Palma

For those who ignored and continue to ignore this subtle but modulation, this bandwidth will be twice that of the baseband sig-
important point made by Shannon, namely, coding at the transmit- nal. Therefore, just because the carrier frequency is 26 GHz, it has
ter, he offered a cautionary note in [22]: no effect on (4) because it is related to the bandwidth of the compo-
nents used in the system and this bandwidth of system components
Information theory has, in the last few years, become
seldom exceeds 500 MHz for any individual component, as the
something of a scientific bandwagon. Starting as a techni-
group delay and hence the rates of distortion of the components
cal tool for the communication engineer, it has received an
increase beyond such a bandwidth. The RF carrier frequency really
extraordinary amount of publicity in the popular as well as the
is not relevant when defining the capacity.
scientific press. In part, this has been due to connections with
For a wireless system where a transmitting antenna is involved,
such fashionable fields as computing machines, cybernet-
it is necessary to consider the radiation efficiency of the antenna, in
ics, and automation; and in part, to the novelty of its subject
addition to coding, because to compare the channel capacity of the
matter. As a consequence, it has perhaps been ballooned to an
various systems, the input power to the transmitting antenna must
importance beyond its actual accomplishments. Our fellow
be fixed and the received power is the transmitted power density
scientists in many different fields, attracted by the fanfare
incident on a receiving antenna. As noted earlier, an antenna does
and by the new avenues opened to scientific analysis, are
not radiate power, it radiates a power density through the radiating
using these ideas in their own problems. Applications are
fields, and when the received power density is integrated by the re-
being made to biology, psychology, linguistics, fundamental
ceiving antenna, it provides the received power. Therefore, without
physics, economics, the theory of organization, and many
specifying the transmitter and the receiving antennas and without
others. In short, information theory is currently partaking of
keeping the input power to the transmitting antenna fixed, it is dif-
a somewhat heady draught of general popularity …. While
ficult to characterize the received power! It must also be pointed
we feel that information theory is indeed a valuable tool in
out that the noise here is purely random; thus multipath and other
providing fundamental insights into the nature of commu-
interference sources are not incorporated in Shannon’s formula.
nication problems and will continue to grow in importance,
Thus, the first question that arises in connection with Shannon’s
it is certainly no panacea for the communication engineer
definition of capacity is how to apply his formula in a near-field
or, a fortiori, for anyone else. Seldom do more than a few of
environment, which is pertinent for operations in microcell and
nature’s secrets give way at one time. What can be done to
pico-cell environments. In the near field of an antenna, the wave
inject a note of moderation in this situation? In the first place,
fronts are nearly planar, and the power has both real and reactive
workers in other fields should realize that the basic results of
forms as discussed in the last section. From a Maxwellian point
the subject are aimed in a very specific direction, a direction
of view, the Poynting vector in the near field is a complex-valued
that is not necessarily relevant to such fields as psychology,
quantity. Thus, what value of the Poynting vector should one put in
economics, and other social sciences. Indeed, the hard core
the capacity formula for the signal power S? Is it the real part or the
of information theory is, essentially, a branch of mathemat-
magnitude of the Poynting vector? This question poses an interest-
ics, a strictly deductive system. A thorough understanding of
ing dilemma. Another point is that the capacity also should vary as
the mathematical foundation and its communication applica-
a function of the square of the distance between the transmitter and
tion is surely a prerequisite to other applications. I personally
the receiver, particularly if one uses an array of antenna elements.
believe that many of the concepts of information theory will
Therefore, to perceive a total electromagnetically accurate picture,
prove useful in these other fields-and, indeed, some results
one must relate the S/N in Shannon’s formula, the Poynting vector
are already quite promising-but the establishing of such ap-
associated with the transmitting antenna, and the Friis transmission
plications is not a trivial matter of translating words to a new
formula. This interconnectivity implies that the channel capacity is
domain, but rather the slow tedious process of hypothesis and
a function of the separation distance between the transmitter and
experimental verification.
the receiver; therefore, Shannon’s formula is applicable only at the
If applied in the proper context, the Shannon channel capacity CS receiver terminals. If the transmitting and the receiving systems
is defined by are deployed over the Earth, which is an imperfectly conducting
ground, then the images of the transmitting-receiving structures
are generated by the presence of the Earth. Hence, the effect of the
(4)
ground reflected through the images generated by the transmitting-
receiving structures should also be incorporated in the analysis.
where B is the one-sided 3 dB bandwidth of the channel, S is the Shannon’s formula does not say anything about the transmit power
magnitude of the received signal power from the transmitter, and of the system; only the received power is defined with respect to
S/N is the signal-to-noise ratio of the received signal. Shannon con- the background thermal noise.
sidered this formula for point-to-point communication between a In most modeling of communication systems, one typically
transmitter and a receiver. Hence, it is imperative to describe the deals with the voltages induced on the antennas, which are treated
transmitting and the receiving antennas, as without them in place as point sources, and then takes the square of the absolute value to
S cannot be clearly defined. It is quite important to emphasize that estimate the power by relating the autocorrelation function of the
B here is the bandwidth of the RF channel and is not related to the voltage to the power spectral density through the Fourier trans-
carrier frequency of the RF channel; that is, for a low-pass signal form. The squared absolute value of any quantity, particularly the

OCTOBER 2016 IEEE A&E SYSTEMS MAGAZINE 15

Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
Relevance of Electromagnetics in Wireless Systems Design

voltage or the current, provides an estimate of power only for a (phase and waveform); only the bandwidth, carrier frequency, and
strictly resistive circuit. For circuits containing complex-valued noise power are known. These beats resulted in increased energy
impedances that can be either inductive or capacitive, as is always fluctuations. To illustrate this phenomenon, superpose a signal of
the case for any realistic antennas, using only the voltage or the amplitude A1 and an interfering signal of variable amplitude A2(t)
current is not sufficient to give the value for the power. However, with a variable phase ψ(t) to obtain
the real part of the power can be evaluated by knowing either the
voltage or the current on the resistors in the circuit. One needs the
(5)
current, the voltage, and the phase angle between them to compute
the total power. The salient points of this discussion are that elec-
tromagnetics are the basis of electrical engineering and that one is which implies
treating a vector problem and not a scalar problem when dealing
with wireless systems. Hence, for a general system in electrical
(6)
engineering, the total complex power can never be computed ex-
clusively from the power spectral density of either only the volt-
age or the current! This point was made by Ivrlac and Nossek [4], Now, averaging over a few periods of the oscillation, represented
[5], who observe that both the voltage and the current are needed by an overbar, and assuming slow variations of A2 and ψ yield
to compute the complex power for a general circuit. Only for the
special case of a purely resistive circuit can power be obtained
(7)
from either the voltage or the current only.
The fundamental principle that one needs to apply in wireless
applications is use both the voltage and the current to compute The averaged power over such a relatively short interval of time is
the power for real antennas, a crucially essential fact that is of-
ten overlooked by communication theory practitioners, leading
to erroneous conclusions. For example, it is widely believed in (8)
communication theory and the signal processing literature that
deployment of multiple antennas will provide a better signal-to-
where P1 and P2, respectively, are the powers associated with sig-
noise ratio at the receiver. This belief cannot be far from the truth
nals 1 and 2. When the average is taken over a long interval of
in a vector problem, as the total field induced on a receiving
time, only the first two terms in (8) remain,
antenna is the vector sum of all the different field components
responsible for inducing currents on the system. Thus, the prob-
lem is how one connects the Shannon channel capacity formula (9)
to the Poynting vector, which actually provides the radiation ef-
ficiency and the gain of an antenna. Note here is that the source
because cosy is zero on the average [24]. The third term of (8)
of noise is outside the antenna and it acts independently, and
represents the beats between the signal and the undesired noise. It
furthermore, there is no assumed loss in the receiving antenna
plays an important role in the power fluctuations as
system.

(10)
GABOR CHANNEL CAPACITY
In many situations, thermal noise may not be the limiting factor
Consequently, the average of the fluctuations is
in preventing a communication channel from performing its job.
The environment may be interference limited. In this situation,
perhaps multipath may limit the detection capability of the signal,
(11)
even though the signal may be several times stronger than the ther-
mal noise. Gabor developed a channel capacity for this scenario,
when the channel is not limited by thermal noise, but is limited by
Because for the noise we have and , it is seen
interference.
from (10) that the fluctuations increase with the signal power P1
Gabor very thoroughly investigated the energy levels that can
[24].
be used in each cell of his logon concept to represent distinguish-
These results can be used for computing the number m of dis-
able signals [3], [23]. Because the concept of the Gabor channel
tinguishable power levels to evaluate the channel capacity. Choose
capacity that provides a signal processing explanation of the in-
depth and formal work performed by Gabor is not well-known, the step of size to be consistent with Shannon’s formula
it is included for completeness. In his analysis, Gabor introduced [24]. This step size is twice the one originally proposed by Gabor.
the phenomenon of beats between the signal and the noise (which If PM is the maximum power used in the signal transmission, the
in this case are deterministic unwanted signals). His development number of steps mG (including the one at P = 0) available for sig-
deals with a receiver without any knowledge of the desired signal naling is

16 IEEE A&E SYSTEMS MAGAZINE OCTOBER 2016

Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.

(12)
As illustrated in [24], for small values of the ratio PM /P2, mG re-
duces to [1 + 0.5 (PM /P2)] = [1 + (Pav/P2)], where Pav is the average
value. For large values of S/N, mG increases more slowly with PM.
For actual observation conditions, it will be necessary to inves-
tigate whether the additional fluctuations due to beats between the
signal and noise are important or not. That discussion will deter-
mine whether one should use Shannon’s or Gabor’s channel capac-
ity formula. Gabor’s channel capacity formula is given by
(13)
When a long time of transmission is selected, the term averages
to zero, and the beats between the signal and the noise can be ignored,
a situation which justifies the use of Shannon’s formula in the limit.
TULLER CHANNEL CAPACITY
The third form of the channel capacity was developed by Tuller, who
designed and built the first Private Line Air Traffic Control Com-
munication System. The Tuller channel capacity parallel the works
of Nyquist and Hartley. As Tuller wrote [25], [26], [27], “The general
principles of the application of information theory to system design
is discussed here and an example is given showing its use in study-
ing the Private Line Air Traffic Control Communication System.”
His “private line” communication link connected the aircraft traffic
controllers of his day with the aircraft under their surveillance. To
develop this system, he realized and used the fact that power plays
little role when dealing with a practical communication system but
rather that the sensitivity of the system depends on the value of the
fields. Because the sensitivity of a receiver is determined by the mini-
mum electric field intensity (typically on the order of 1 μV/m) it can
receive, and below that value, the receiver will not recognize any
signal, so Tuller developed a formula for the channel capacity that in-
corporates the fields and not the power. He followed the formulation
of Nyquist [28] for the development of his channel capacity formula,
but unlike Nyquist, Tuller took the effect of noise into account. It is
also interesting to note that Tuller was aware of Shannon's work and
made reference to it in his 1949 paper [25]. It is also interesting to
observe that Shannon was also aware of Tuller's work and made ref-
erence to it in his paper A Mathematical Theory of Communication.
A practical form of the channel capacity, which is more ap-
plicable to real systems, was first developed by Hartley [29], [30],
[31] in 1928. Among his conclusions, he stated
The total amount of information which may be transmitted
over a system whose transmission is limited to frequencies
lying in a restricted range is proportional to the product of the
OCTOBER 2016 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
Sarkar, Mokole, and Palma
frequency range which it transmits by the time during which it
is available for transmission.
Hartley’s treatment represented a first step in the direction of
measuring a message and the message-transmitting capacity of a
system [30]. Both Nyquist [28], [32] and Hartley [29], [30] showed
what must indeed have been common knowledge from the ear-
liest days of telegraphy, namely, the number of different signals
available to a single receiver is limited by the relative magnitude
of interference [30]. Also, as the signals traverse the transmission
medium (or channel, as it is commonly called), they are distorted,
and noise and interfering signals are added. Hence, it is a major
challenge to interpret the signals correctly at a desired location
[33]. However, real physical systems do not allow the practitioner
to increase the rate of signal change indefinitely or to distinguish
an arbitrarily large number of voltage amplitudes or levels [33]:
1. All of our systems have energy-storage devices present, such
as inductances and capacitances, and changing the signal implies
changing the energy content. Limits on making these changes are
determined by the bandwidth of a particular system.
2. Every system provides inherent (even if small) variations or fluc-
tuations in voltage, or whatever parameter is used to measure the sig-
nal amplitude. One cannot subdivide amplitudes indefinitely. These
unwanted fluctuations of a parameter to be varied are called noise.
The preceding two limitations are tied together by a simple ex-
pression developed for system capacity [33]: C = 1/r log2 n, where
r is the minimum time required for the system to respond to signal
changes and n is the number of distinguishable signal levels. The
minimum response time is proportional to the reciprocal of the
system bandwidth. Because system capacity can be written as C
= B log2 n, where B is the system bandwidth in hertz, the informa-
tion about the number of different amplitude levels can be sent
by fewer numbers simply by binary encoding the different levels.
For example, if a signal has 16 different levels, then four bits are
required. If the signal has 32 different levels, then five bits are
required. More generally, log2 n bits are required for characterizing
n levels of the signal. The Hartley capacity can be shown to be
equivalent to Shannon’s definition of capacity, when the transmit-
ting and receiving antenna systems are conjugately matched. The
basic form of Shannon’s theorem states [18] that
to distinguish between M different signal functions
of duration T on a channel, we can say that the
channel can transmit log2 M bits in time T. The rate
of transmission is then . More precisely, the
channel capacity Cʹ may be defined as
(14)
By following Hartley, one can obtain a simplified expression for
the capacity without delving into the concept of power, if one requires
the received signal to be separated into 2Q distinct levels, so the num-
ber of amplitudes that can be reasonably well distinguished is 2Q.
Furthermore, as Shannon points out, “since in time T there are 2TB
independent amplitudes, then the total number of distinct signals is”
17
Relevance of Electromagnetics in Wireless Systems Design
(15)
where B is the one-sided bandwidth of the signal. Hence, the num-
ber of bits that can be sent in time T is
(16)
This expression is more pertinent in characterizing near-field en-
vironments and is similar to the Hartley-Nyquist-Tuller form of
the channel-capacity theorem, which has been elaborated in [26],
[27], [28],
(17)
where As is the received rms amplitude voltage of the signal and
ΔV is the rms level of the quantization noise voltage related to
the discretization of the received signal. Using this formula for the
channel capacity is simpler and may provide a different value for
the channel capacity than that obtained when it is computed with
the power relationship. In treating the voltage instead of the power,
this form is ideally suited for wireless systems, because the capac-
ity is directly related to the source, namely, the fields, and is valid
both for near-field and far-field characterizations of the antennas.
In this formula, the limiting factor is the lowest discretization level
to which the voltages are quantized.
Note that these three formulas for channel capacity (Shannon,
Gabor, and Tuller) do not represent the same transmission perfor-
mance: one is essentially error free (Tuller); the others are error free
only in the statistical sense in that the average rate of errors can be
made arbitrarily small (Shannon and Gabor). Therefore, if one as-
sumes that the background noise characteristics are very similar in
the Shannon and Gabor representations, then the two expressions
might yield similar results for conjugately matched transmitting
and receiving antennas, as then they will be radiating the maximum
power available and the antennas will be resonant, implying that the
input power is real and directly related to the voltage.
In summary, to introduce the concept of channel capacity, one
needs to have coding at the RF stage, because without the coding,
discussions of channel capacity are not very fruitful. Such is the
case for satellite communication and for GPS systems in which
the code can be a gigabit long. Thus, references [21], [34], [35]
are quite relevant as they describe such scenarios. In other words,
the Maxwell-Poynting theory associated with the radiated fields
from antennas needs to be included in channel capacity formulas to
make them realistic and more amenable to system design.
CONCLUSION
In this article, an attempt has been made to illustrate that the prop-
erties of physical systems must be accurately incorporated in sys-
tem analyses. Equally importantly, the relevant physics need to be
paid proper attention and be connected with any abstract math-
ematical system analyses to make real systems perform according
18 IEEE A&E SYSTEMS MAGAZINE
Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.
to the design predictions. Currently, this connection between the
physics and mathematics of physically realizable systems is rather
weak, and much effort needs to be directed at strengthening their
bonds, if real working systems are to be realized.
ACKNOWLEDGMENTS
Grateful acknowledgement is made to the reviewers for suggesting
ways to improve the readability of the manuscript.
REFERENCES
[1] Mokole, E. L., and Wicks, M. C. A short history of waveform diver-
sity. In Principles of Waveform Diversity & Design, M. Wicks, E.
Mokole, S. Blunt, V. Amuso, and R. Schneible, Eds. Raleigh, NC:
SciTech Publishing, Inc., 2010, pp. 1–21.
[2] Samaddar, S. N., and Mokole, E. L. Some basic properties of antennas
associated with ultrawideband radiation. In Ultra-Wideband, Short-
Pulse Electromagnetics 3, C. E. Baum, L. Carin, and A. P. Stone, Eds.
New York: Plenum Press, 1997, pp. 147–164.
[3] Gabor, D. Communication theory and physics. Transactions on the
IRE Professional Group on Information Theory, Vol. 1, 1 (1953),
48–59.
[4] Ivrlac, M. T., and Nossek, J. A. Towards a circuit theory for com-
munication. IEEE Transactions on Circuits and Systems I, Vol. 57, 7
(2010), 1663–1683.
[5] Ivrlac, M. T., and Nossek, J. A. The multiport communication theory.
IEEE Circuits and Systems Magazine, Vol. 14, 3 (2014), 27–44.
[6] Feynman, R. P. The Character of Physical Law. New York: Random
House, 1994.
[7] Sommerfeld, A. Vorlesungen Über Theoretische Physik, Band I:
Mechanik. Thun, Germany: Verlag Harri Deutsch, 1994.
[8] Landau, L., and Lifshitzs, E. Mechanics: Course of Theoretical Phys-
ics, Vol. 1. Burlington, MA: Butterworth-Heinemann, 1979.
[9] Cover, T., and Thomas, J. Elements of Information Theory. New York:
Wiley, 1991.
[10] Scharf, L. Statistical Signal Processing. Reading, MA: Addison-Wes-
ley, 1991.
[11] Proakis, J. Digital Communications. New York: McGraw-Hill, 1995.
[12] Oppenheim, A., Willsky, A., and Nawab, S. Signal and Systems. En-
glewood Cliffs, NJ: Prentice-Hall, 1983.
[13] Belevitch, V. Classical Network Theory. San Francisco, CA: Holden
Day, 1968.
[14] Sarkar, T. K., Salazar-Palma, M., and Mokole, E. L. Physics of Mul-
tiantenna Systems and Broadband Processing. Hoboken, NJ: Wiley,
2008.
[15] Andrews, J. R. UWB signal sources, antennas and propagation. In
Proceedings of the IEEE Tropical Conference on Wireless Communi-
cation Technology, Honolulu, Hawaii, Oct. 2003, 439–440.
[16] Mei, Z., Sarkar, T. K., and Salazar-Palma, M. The design of an ul-
trawideband T-pulse with a linear phase fitting the FCC mask. IEEE
Transactions on Antennas and Propagation, Vol. 59, 4 (2011), 1432–
1436.
[17] Orfanidis, S. J. Electromagnetic Waves and Antennas. Piscataway,
N.J., Rutgers University, 2012. http://www.ece.rutgers.edu/~orfanidi/
ewa/.
OCTOBER 2016
 Sarkar, Mokole, and Palma

[18] Shannon, C. E. Communication in the presence of noise. Proceedings [28] Nyquist, H. Certain factors affecting telegraph speed. Bell System
of the IEEE, Vol. 86, 2 (Feb 1998), 447–457. Technical Journal, Vol. 3 (1924), 324.
[19] Viterbi, A. J. Information theory in the sixties. IEEE Transactions on [29] Hartley, R. V. L. Transmission of information. Bell System Technical
Information Theory, Vol. 19, 3 (1973), 257–262. Journal, Vol. 7 (1928), 535–563.
[20] Shannon, C. E. General treatment of the problem of coding. Transac- [30] Black, H. S. Modulation Theory. New York: D. Van Nostrand Com-
tions of the IRE Professional Group on Information Theory, Vol. 1, 1 pany Inc., 1953.
(1953), 102–104. [31] Sarkar, T., Burintramart, S., Yllmazer, N., Zhang, Y., De, A., Sala-
[21] Viterbi, A. J. The evolution of digital wireless technology from space zar-Palma, M., Lagunas, M., Mokole, E., and Wicks, M. A look
exploration to personal communication services. IEEE Transactions at the concept of channel capacity from a Maxwellian viewpoint.
on Vehicular Technology, Vol. 43, 3 (1994), 638–644. IEEE Antennas and Propagation Magazine, Vol. 50, 3 (2008),
[22] Shannon, C. E. The Bandwagon. IRE Transactions on Information 21–50.
Theory, Vol. 2, 1 (1956), 3. [32] Nyquist, H. Certain topics in telegraph transmission theory. Transac-
[23] Gabor, D. Lectures in communication theory. Tech. Rep. 238. Cam- tions of the American Institute of Electrical Engineers, 47 (Apr. 1928),
bridge, MA: Massachusetts Institute of Technology, Research Labora- 617–644.
tory of Electronics, Apr.1992. [33] Schwartz, M. Information, Transmission, Modulation, and Noise.
[24] Brillouin, L. Science and Information Theory, 2nd ed. New York: Aca- New York, McGraw Hill, 1959.
demic Press, 1962. [34] Viterbi, A. J. Wireless digital communication: a view based on three
[25] Tuller, W. G. Theoretical limitations on the rate of transmission of lessons learned. IEEE Communications Magazine, Vol. 29, 9 (1991),
information. Proceedings of the IRE, Vol. 37, 5 (1949), 468–478. 33–36.
[26] Tuller, W. G. Application of information theory to system design. [35] Viterbi, A. A fresh look at the physical layer of terrestrial mobile
Electrical Engineering, Vol. 70, 2 (1949), 124–126. multiple access networks. In Proceedings of the IEEE International
[27] Tuller, W. G. Information theory applied to system design. Transactions of Conference on Personal Wireless Communications, New Delhi, India,
American Institute of Electrical Engineers, Vol. 69, 2 (1950), 1612–1614. 1996.

OCTOBER 2016 IEEE A&E SYSTEMS MAGAZINE 19

Authorized licensed use limited to: Southeast University. Downloaded on March 06,2021 at 13:41:04 UTC from IEEE Xplore. Restrictions apply.