This article has been accepted for publication in IEEE Transactions on Antennas and Propagation.

This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 1

Space-Time Modulation of a Multimode Electrically

Small Antenna for Increased Matching and
Efficiency Bandwidths
Zachary Fritts, Student Member, IEEE, Amirhossein Babaee, Student Member, IEEE, Steve M. Young, Senior
Member, IEEE, and Anthony Grbic, Fellow, IEEE

Abstract—In this work, a method for increasing the efficiency-

bandwidth product of an electrically small antenna (ESA) by
using spatially-discrete traveling-wave modulation (SDTWM) is
introduced. This method applies generally to electrically-small 𝟔𝟎 𝒎𝒎
antennas possessing N-fold rotational symmetry. To illustrate the 𝟐𝟎 𝒎𝒎
method, it is applied to a top-hat loaded monopole antenna that
supports a fundamental radiating mode as well as non-radiating 𝑪𝟑 𝒕
modes that are analogous to discrete orbital angular momentum
modes. Time-modulated capacitors, placed in each sector, couple 𝑪𝟒 𝒕
these modes using SDTWM. The modulation scheme is similar 𝟓𝟎 𝛀
to that of a classic negative-resistance parametric amplifier, but 𝑪𝟐 𝒕
instead of using the modulation to obtain gain from the system,
the modulation parameters are chosen to enhance the matching
𝑪𝟏 𝒕
bandwidth of the antenna. We show how this can be done by
choosing an appropriate combination of the modulation depth
and modulation frequency. After extending the definition of Fig. 1. The four-sector top-hat loaded monopole considered in this work.
radiation efficiency and mismatch loss to the case of a paramet- Neighboring sectors are capacitively coupled via interdigitaged capacitors on
rically time-varying antenna, the extended definitions are used the top hats, and the helical elements inductively load the antenna to reduce
to characterize the simulated performance of the time-varying its electrical size to ka = 0.253 at its operating frequency of 335 MHz. A
antenna under two different SDTWM modulation schemes. The time-varying capacitor is placed in series with each sector, and these are then
first modulation scheme couples the radiative mode to a single connected in parallel to an L matching network.
non-radiative mode, while the second modulation scheme couples
the radiative mode to two different non-radiative modes. Both
schemes yield efficiency-bandwidth products that are 6.4-7.6 dB electrically small antennas are commonly defined as those
larger than that of the time-invariant version of the antenna. antennas for which the radius a of the Chu sphere satisfies
ka < 0.5, where k is the wavenumber in the radiating medium
(here free space). Since then, the analytic work yielding the
I. I NTRODUCTION minimum quality factor for ESAs has been refined [4]–[10].
Modern methods, based on the method of moments and convex
F OR many communications and sensing applications, it is
desirable to have an antenna which is small compared
to the operating wavelength. However, there are well-known
optimization, have extended bounds on the quality factor (Q)
of antennas to arbitrary geometries [11], [12]. These bounds
limits on the quality factors, bandwidths, and efficiencies on the Q are significant since Q can function as a proxy for
of electrically small antennas (ESAs) [1]–[3]. Chu derived the bandwidth of a resonant antenna. For example, in [13],
bounds on the radiation quality factor of passive, linear, time- a simple equation for the fractional bandwidth of a single-
shift invariant (LTI) small antennas by using equivalent circuit resonant antenna is derived as a function of its Q and the
models which approximate the energy stored within the lowest allowable impedance mismatch between the antenna and its
order spherical modes [2]. Because of his work, the smallest source or load.
sphere enclosing an antenna is called the Chu sphere, and Once these fundamental limits on passive, LTI electrically
small antennas became known, engineers started working to
This research is supported by the National Science Foundation Graduate overcome them in earnest. A history of some of the resulting
Research Fellowship under Grant No. DGE-2241144 and by the Office of the antennas is recounted in [14]. There are two fundamental
Director of National Intelligence (ODNI), Intelligence Advanced Research
Projects Activity (IARPA), via [2021-2106240007]. The views and conclu- categories of approaches for increasing the bandwidth of
sions contained herein are those of the authors and should not be interpreted ESAs: LTI approaches and non-LTI approaches. It is worth
as necessarily representing the official policies, either expressed or implied, of summarizing several of these as background for the methods
ODNI, IARPA, or the U.S. Government. The U.S. Government is authorized
to reproduce and distribute reprints for government purposes notwithstanding used in this work.
any copy-right annotation therein. (Corresponding author: Anthony Grbic) Arguably the simplest LTI method for increasing the band-
The authors are with the Department of Electrical and Computer width of a resonant antenna is to reduce the quality factor of
Engineering, University of Michigan, Ann Arbor, MI, 48109 USA;
e-mail: [email protected]; [email protected]; [email protected]; agr- the antenna by increasing its loss, whether mismatch loss or
[email protected]. ohmic loss. This trade-off may or may not be acceptable for

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 2

a given application. Improvements in bandwidth can also be 𝒎=𝟎

𝟏
obtained by incorporating a tuned matching network. Some
successful small antennas incorporate a matching network
within the antenna structure itself, thereby eliminating the 𝒎=𝟑 𝒎=𝟏
need for an external matching network which would otherwise
increase the effective size of the antenna [15]–[18].
Closely related to antennas with internal matching networks 𝒎=𝟐
are multimode or multi-resonant ESAs. Early in the develop-
ment of theoretical limits on antenna Q, it was discovered that
an antenna designed to generate a combination of an electric
−𝟏
dipole mode and a magnetic dipole mode could attain a radi-
(a)
ation quality factor smaller than that of an electric dipole [3].
The spherical helix is an example of a practical antenna which
can be configured to radiate both electric and magnetic dipole
modes in this way [19]. It is also possible for the antenna
structure to couple spherical multipole modes of different
orders, e.g., an electric dipole with an electric quadrupole. This
coupling of multiple resonances can act in a way analogous to
a higher-order filter or matching network, thereby providing a
broadband match [20], [21]. Coupling multiple resonances to
create a broadband match is an approach that we build upon
in this work.
All of the above LTI approaches to improving the bandwidth
of ESAs are still constrained by the Bode-Fano bound on
impedance matching [22]–[25]. An additional performance
(b)
constraint is given by a recent extension of Shannon’s theorem
Fig. 2. (a) A visual depiction of the antenna modes. For a given mode,
to band-limited channels, such as those including narrowband the current, voltage, and fields in each sector possess a progressive phase
resonant antennas [26]. A significant result of this work is shift of ϕm = m · 90◦ . (b) The magnitude of the modal admittances as a
to quantify the increases in channel capacity that result from function of frequency. Where the modal admittance reaches its maximum, the
mode is resonant. Note the close agreement between the modal admittances
optimal choices of antenna losses and/or matching networks. computed from full-wave simulations of the antenna (solid curves) and from
The most relevant non-LTI approaches to enhancing small the equivalent circuit model of the antenna (dashed curves). The gain and
antenna bandwidth for this work are those involving para- radiation pattern for each mode are depicted along with the sectoral excitation
pattern.
metric modulation of the antenna or its matching network
[27]–[30]. These approaches use time-modulation to couple
energy between different frequencies. This coupling can be
used to boost the radiation bandwidth of an antenna [28],
inductively load the antenna and reduce its electrical size to
to compensate for intentionally-introduced mismatch losses
ka = 0.253 at the fundamental self-resonant frequency of
of a receiving antenna in a low-noise manner [29], or to
335 MHz. The exact dimensions of the antenna are shown
provide an impedance match to an antenna [30]. However,
in Fig. 1. A time-varying capacitor is placed in series with
these approaches can suffer from the radiation of spurious
the vertical helix in each sector, and the capacitors are then
harmonics and large reflections back to the input of the device
joined at a single feed point. An L-network at this feed point
[28], and may require the inclusion of bulky external idler
matches the input impedance of the time-invariant version of
resonators [29]. In this work, we build on an approach first
the antenna to 50Ω (the time-invariant version of the antenna
introduced in [31]–[33], which uses parametric space- and
is obtained by simply replacing the time-varying capacitors
time-variation to couple energy between radiative and non-
with LTI capacitors having the same average capacitance).
radiative modes of a multimode ESA. Such coupling enables a
Since the four sectors are in such close proximity, they are
broadband match while suppressing the radiation or reflection
strongly coupled to each other. This coupling is predominantly
of spurious harmonics and eliminating the need for external
capacitive coupling between the top hats of each sector. It is
idler resonators.
tuned by the interdigitations between the top hats.

II. T HE A NTENNA If we consider four ports located where the helical feed of
each sector reaches the ground plane (indicated by black disks
A. The Top-Hat Loaded Monopole and Its Modes in Fig. 1), the voltage and current at each port are related
The antenna consists of a four-sector, top-hat loaded by a 4 × 4 admittance matrix, Y(ω). Rotational and mirror
monopole above a ground plane (see Fig. 1). The design symmetry imply that the port parameters of the structure
is inspired by prior ESAs known to have large efficiencies are represented by a circulant matrix (or a permutation of
and resonant bandwidths [15], [17], [20]. In each sector, a circulant matrix), containing at most three independent
the vertical monopolar element is twisted into a helix to parameters:

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 3

  for a vertically polarized cylindrical antenna [35]. Therefore,

Y11 Y12 Y13 Y12 the LTI antenna is a good ESA even apart from time-varying
Y12 Y11 Y12 Y13 
Y= . (1) effects.
Y13 Y12 Y11 Y12  Note that even though we have chosen a top-hat loaded
Y12 Y13 Y12 Y11 monopole design for the antenna structure, the method of
Since Y is circulant, its set of eigenvectors using SDTWM to couple radiative and non-radiative modes
{V0 , V1 , V2 , V3 } is given by the complex discrete Fourier could also apply to other antenna geometries, e.g., coupled
transform basis: loop antennas or planar sectorized antennas like that reported
in [36].
 2πm 4πm 6πm T
Vm = 1 ej 4 ej 4 ej 4 , (2)
B. The Circuit Model
where m ∈ {0, 1, 2, 3} and YVm = ym Vm , such that ym Since the antenna is electrically small, it is well approxi-
is the eigen-admittance for mode m. These admittance eigen- mated by a circuit model [17]. A circuit model captures the
modes are discretized versions of orbital angular momentum frequency-dependent behavior of the antenna parameters while
(OAM) modes, where each eigenmode is characterized by an simplifying analysis of the effects due to time-modulation.
azimuthal index m. For the mth eigenmode, the voltage and Specifically, creating a circuit model for the antenna allows the
current in adjacent sectors have an identical magnitude and modulated device to be analyzed in the same way as a classic
a progressive phase shift of m · 90◦ . These eigenmodes are parametric amplifier, thereby providing a convenient mathe-
depicted in Fig. 2(a), and the modal admittances are shown as matical tool for predicting and explaining its performance [37].
functions of frequency in Fig. 2(b). Note that each eigenmode The four-sector top-hat loaded monopole is well represented
has a modal resonant frequency ωm at which the modal by the circuit model shown in Fig. 3. Each element of the
admittance is maximized and becomes purely real. Modes 1 circuit model corresponds to a physical feature of the antenna:
and 3 are degenerate, since they represent OAM modes with inductor L represents the inductance of the vertical helix,
the same wavenumber but opposite rotational senses. capacitor Ce the capacitance of the fringing fields between
Throughout this work, any reference to a mode of the the top hat and the ground plane, capacitor Cc the inter-
antenna should be taken to mean an eigenmode of the time- sector coupling capacitance, and capacitance Ca the parasitic
invariant structure. Even though these modes were defined capacitance between the feeds and the ground plane.
as eigenvectors of the admittance matrix (1), the symmetry The four-path nature of the circuit in Fig. 3(b) illustrates
inherent in the structure causes these same eigenvectors to why solely the m = 0 mode is radiative [38]. The current
also be eigenvectors of any port-based representation of the through the radiation resistance is the sum of the currents in
network parameters (e.g., S parameters or Z parameters). The the four paths of the circuit. Suppose that the current in path
symmetry of the feed network also means that in the time- n is related to the current in path n − 1 by a phase shift of
invariant structure, only the m = 0 mode can be excited m·90◦ . Then, the total current through the radiation resistance
from the RF source. It is the spatial variation introduced by is zero except when m = 0. Consequently, only the m = 0
SDTWM which allows the m = 0 mode to couple to other mode can contribute to radiated power.
modes. An identical argument about the cancellation of currents for
Only the fundamental m = 0 mode couples efficiently to modes m ̸= 0 applies to currents entering or exiting the node
the far field; all other modes are predominantly confined to to the right of the inductor Lmatch in Fig. 3(b). As a result, the
the near field of the antenna. As a result, the 83% radiation non-radiative modes are isolated from the matching network
efficiency of the m = 0 mode is relatively good for a small and (more importantly) from the 50 Ω source impedance.
antenna. Conversely, the radiation efficiencies of the other This isolation is significant because it prevents the source
modes are vanishingly small: 0.41% for the (m = 1, 3) modes impedance from damping the non-radiative modes. This allows
and 0.003% for the (m = 2) mode, according to full-wave them to have the high quality factor needed to serve as low-
simulations in HFSS. Therefore, we term the m = 0 mode loss idler tanks within parametric processes [37].
the “radiative” or “bright” mode and the other modes “non- Equivalent circuit parameters for the circuit model were
radiative” or “dark” modes. chosen to yield the best fit to the port parameters of the
The poor radiation efficiencies of the non-radiative modes antenna, computed via full-wave simulation in ANSYS HFSS.
have an intuitive explanation. The sectors of the antenna are Fig. 2(b) shows good agreement between the modal admit-
nearly spatially collocated, in the sense that their separation tances for the circuit model and the full-wave simulation
is deeply sub-wavelength. Furthermore, the modal surface results. In addition, the loss resistances RL and Re,loss were
currents on each sector are identical except for the per-sector fitted to yield the appropriate modal Q factors and radiation
phase delay of m · 90◦ . Consequently, the contributions of efficiency of the radiative mode. Table I shows the parameters
each sector to far-field radiation destructively interfere [34]. for the equivalent circuit model.
However, the currents in each sector are in-phase for the
m = 0 mode, causing their contributions to the far-field III. S PATIALLY-D ISCRETE T RAVELING -WAVE
radiation to sum constructively. The resulting radiation pattern M ODULATION AND THE I NTERPATH R ELATION
is that of a single top-hat loaded monopole. The radiative Spatially-discrete traveling-wave modulation (SDTWM) is
mode also has a bandwidth that is 86% of the Gustafsson limit a class of space-time modulation in which one or more

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 4

𝑪𝟏 𝒕 𝑳 𝑹𝑳 𝑪𝒆

𝑪𝒂
𝑪𝒆 𝑪𝒄 𝑪𝒄
𝑳 𝑪𝟐 𝒕
𝑪𝒂
𝑹 𝑳𝒎𝒂𝒕𝒄𝒉
𝟓𝟎 𝛀 𝑹𝒆,𝒍𝒐𝒔𝒔
𝑪𝟑 𝒕

𝑪𝟒 𝒕
𝑪𝒎𝒂𝒕𝒄𝒉 𝑪𝟒 𝒕 𝑹𝒓𝒂𝒅
𝟓𝟎 𝛀
𝑪𝟐 𝒕

𝑪𝟏 𝒕 𝑪𝟑 𝒕

(a) (b)
Fig. 3. An equivalent circuit model for the antenna. (a) Here the equivalent circuit elements are drawn on top of the antenna to emphasize their correspondence
to the physical structure of the antenna. For example, Cc is the inter-sector coupling capacitance, L the inductance of the helical element, and Ce the capacitance
between the top-hat and the ground plane. (b) The full circuit model drawn as 4-path circuit. Current in the four paths can be decomposed into in-phase
and out-of-phase components. At the node where the paths rejoin to pass through the radiation resistance Rrad , only the in-phase components will sum
constructively. Thus the m = 0 mode is the only radiative mode for the circuit model (as for the antenna). The shaded gray region indicates the location of
the ground plane.

TABLE I For the nth sector, the simplest modulated capacitance varies
E QUIVALENT C IRCUIT PARAMETERS sinusoidally around an average capacitance value C0 :

Lmatch Cmatch L Ca Ce
Cn (t) = C0 (1 + 2M cos (ωmod (t − ntp )))
7 nH 48 pF 291 nH 0.45 pF 0.81 pF   
(4)
2πp
= C0 1 + 2M cos ωmod t − n ,
Cc RL Re,loss Rrad C0 4

1.35 pF 0.69 Ω 0.133 Ω 1.487 Ω 10 pF where M is the modulation depth, tp = pTmod 4 and Tmod is the
temporal modulation period. Note that the modulation index p
plays a similar role in the modulation waveform as the modal
parameters of each unit cell in a periodic structure are varied index m in the set of eigenmodes {Vm } of the antenna - both
in time, but with a time-delay in the modulation waveform indices can be thought of as azimuthal orders. In fact, we will
between adjacent unit cells [39]. This form of space-time show through the interpath relation that p determines which
modulation allows us to selectively couple the radiating mode eigenmodes are coupled by the modulation.
of the antenna to specific non-radiating modes. This will To simplify the proof, we will assume the system is linear
be shown mathematically by proving a form of the inter- with respect to its input. Because the interpath relation is
path relation [40]. It makes intuitive sense that we need general, the proof will also avoid reference to the specific way
both time- and space-variation in order to couple different that SDTWM is applied to the antenna in Fig. 1. We will rely
modal resonances of the antenna, as these occur at different only on the symmetries fundamental to the design. The matrix
(temporal) frequencies, as well as different angular (spatial) operators and vectors will be written with the appropriate size
wavenumbers. for our four-sector system. However, the analysis holds for a
system with any number of unit cells.
A key feature of the interpath relation for our antenna is
that harmonics of the operating frequency of the antenna can It is helpful to begin by introducing a few simple mathemat-
be confined to non-radiating modes, suppressing the radiation ical definitions and observations. First, the eigenvectors {Vm }
of undesired spurious harmonics. In [41], this property of of the antenna are related to each other by the diagonal matrix
SDTWM is used to develop an antenna that performs effi- D:
cient subharmonic mixing by confining the first and second
modulation harmonics to non-radiating modes of the antenna, D = diag{1, ej2π/4 , ej4π/4 , ej6π/4 }, (5)
while allowing the third harmonic to fall into a radiating mode.
In this work, however, the desired radiating frequency is the such that Vm may be given in terms of V0 according to the
fundamental harmonic, ω. relation:
The time-modulated elements of the top-hat loaded Vm = Dm V0 . (6)
monopole are the capacitances Cn (t), as depicted in Fig.
In fact, the index m, previously restricted to m ∈ {0, 1, 2, 3},
1. These time-varying capacitors obey the voltage-current
can be extended to all integers as long as the eigenvalue index
relationship given by
is understood to mean mod4 (m).
d Next, note that for a SDTWM structure, any parameter P
i(t) = [Cn (t)v(t)] . (3) of unit cell (sector) n is related to the same parameter on the
dt
Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 5

prior unit cell by a time shift. This can be written in equation which simply means that at any given time, voltages in each
form as, unit cell are related to voltages in a prior unit cell by time shift

pTmod
 and a phase shift. As observed in [42], this is highly similar
Pn (t) = Pn−1 (t − tp ) = Pn−1 t − . (7) to the Floquet theorem.
4
The parametric time-variation at frequency ωmod has the
By creating a vector for this parameter in each sector, P(t) = effect of mixing the fundamental frequency ω to various
[P1 (t), P2 (t), P3 (t), P4 (t)]T , and defining the circular shift modulation harmonics, making it reasonable to assume a
matrix R, where solution of the form:
  ∞
0 0 0 1 X
1 0 0 0 v(t) = V(ℓ) ej(ω+ℓωmod )t , (18)
R= 0 1 0 0 ,
 (8) ℓ=−∞

0 0 1 0 where ℓ is an index indicating the harmonic frequency. Each

coefficient vector V(ℓ) in the assumed solution contains the
then (7) can be written simply as:
voltage phasors of the four sectors for the ℓth harmonic.
P(t) = RP(t − tp ). (9) By substituting the assumed solution (18) into (17), it is
straightforward to show that
Note that the eigenvectors of the LTI antenna {Vm } from (2) 2πpℓ
are also eigenvectors of R, with associated eigenvalues given V(ℓ) = e−j 4 RV(ℓ) , (19)
by e−j2πm/4 :
2πm which means that for the ℓth harmonic, the voltages in each
Vm = ej 4 RVm . (10) sector are related to those on the prior sector by a phase
The orthogonality of the eigenvectors {Vm } implies that if shift that depends both on the harmonic index ℓ and on the
any vector V satisfies (10), then it must be proportional to azimuthal order of the modulation p.
Vm : Applying (11) to (19) implies the following significant
2πm
V = ej 4 RV ⇔ V ∝ Vm . (11) result:
V(ℓ) ∝ Vmod4 (−pℓ) = D−pℓ V0 . (20)
Finally, we define an operator L that maps the voltage at
corresponding locations within each sector (denoted v(t)) to This means that the vector containing the voltages for the ℓth
an input signal vin (t), given the parameters P(t) of the system. harmonic in each sector must be proportional to the mth
We denote the action of this operator as: eigenvector of the time-invariant structure, Vm , where the
modal index m obeys the selection rule m = mod4 (−pℓ).
v(t) = L{P(t); vin (t)}. (12) Therefore, SDTWM can be used to couple specific harmonics
to specific modes of the antenna, and the the azimuthal order
If the vector of parameters is rotated, it is the same as
of the modulation p determines which modes are coupled.
rotating the structure itself (along with its voltages). Therefore,
Matrix D−pℓ is a type of “delay matrix,” and plays an
Rv(t) = L{RP(t); vin (t)}. (13) important part in the analysis of SDTWM structures [39].

With these preliminaries settled, we are ready to prove the

IV. E FFICIENCY-BANDWIDTH M ETRICS FOR L INEAR
interpath relation. If we shift the time origin by tp , then (13)
T IME -H ARMONIC S YSTEMS
becomes:
For LTI small antennas, the efficiency-bandwidth product
Rv(t − tp ) = L{RP(t − tp ); vin (t − tp )}. (14) is a common performance metric, as it is well-known that
the radiating bandwidth can be increased if dissipation or
Applying (9), we can conclude that
mismatch losses are permitted. In this section, it is shown that
Rv(t − tp ) = L{P(t); vin (t − tp )}, (15) as long as one carefully defines the efficiency of the antenna,
it is also possible to define a meaningful efficiency-bandwidth
which is a time-domain version of the interpath relation. To get product for linear, time-varying ESAs.
it into a more useful form, let the input be a single harmonic, In an LTI antenna, power is only radiated, dissipated, or
vin (t) = ejωt , and apply the assumption that the operator is reflected over the bandwidth of the signal being transmitted
linear with respect to its input, yielding: through the antenna. However, in a system with time-varying
Rv(t − tp ) = L{P(t); ejω(t−tp ) } parameters, mixing products are (intentionally) produced, and
these contribute to dissipated and reflected powers. In addition,
= e−jωtp L{P(t), vin (t)} (16)
the mechanism by which parameters of the system are made
−jωtp
Rv(t − tp ) = e v(t) to vary in time (hereafter referred to as the modulation source)
Rearranging, it is clear that for time-harmonic input voltages, can transfer power into the system. Both the presence of
the interpath relation implies that voltages within the structure mixing products and the power introduced by the modula-
obey the following relation: tion source must be taken into account when extending the
usual definitions of efficiency to parametrically time-varying
v(t) = ejωtp Rv(t − tp ), (17) systems.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 6

A general definition of the radiation efficiency of an antenna Negative-R Negative-R

𝒎=𝟎
is the radiated power normalized by the total accepted power. Coupling with Detuning
𝝎𝟎
In an LTI transmitting antenna, the accepted power, Pacc (ω), 𝒎=𝟑 𝒎=𝟏
is the sum of the radiated power, Prad (ω), and the power 𝝎𝟏
dissipated in the antenna due to dielectric and conductor 𝒎=𝟐

𝝎𝒎𝒐𝒅

𝝎𝒎𝒐𝒅
losses, Pdis (ω). In a time-varying antenna, the accepted power 𝝎𝟐
𝝎=𝟎
must be re-defined to include contributions from all modu-
−𝝎𝟐
lation harmonics and from the power supplied by the time
modulation. This constrains the efficiency to be less than unity. −𝝎𝟏
Accordingly, we define:
−𝝎𝟎
∞
X
′
Pacc (ω) = [Prad (ωℓ ) + Pdis (ωℓ ) + Pmod (ωℓ )] , (21)
Fig. 4. An illustration of the relationship between the modal resonant
ℓ=−∞ frequencies and the modulation frequency ωmod . The resonant frequency of
where Pdis is the power dissipated due to losses, and Pmod mode m is indicated by ωm . The double-sided arrow indicates which modal
resonances are being coupled for a particular parametric modulation scheme.
is the power supplied by the modulation source. The index ℓ “Negative-R coupling” here means choosing the modulation frequency to
indicates the ℓth modulation harmonic of ω. For a parametric exactly couple the resonance of the radiative mode, ω0 , with the negative-
system with a fundamental modulation frequency of ωmod , the frequency image of the resonance of a non-radiative mode, in this case −ω1 .
However, ωmod can also be detuned from exactly coupling these resonances,
modulation harmonics are of the form ωℓ = ω + ℓωmod . This which is shown in the rightmost part of the figure.
allows us to define the radiation efficiency of a time-varying
antenna as:
If ωL and ωU represent lower and upper frequency limits of the
Prad (ω) useful bandwidth of the antenna, then the efficiency-bandwidth
ηrad = ′ . (22)
Pacc (ω) product is the integral of the efficiency over this bandwidth:
Note that in the above equation, Prad (ω) specifically refers Z ωU
only to the radiated power at the desired operating frequency η · BW = η(ω)dω, (25)
ωL
ω; otherwise radiation at spurious harmonics would increase
the radiation efficiency. where η can be either the radiation efficiency as defined in
Since the radiation efficiency does not account for mismatch (22) or the realized radiation efficiency as defined in (24).
losses, we can define an additional figure of merit that is
the product of the radiation efficiency and the mismatch loss V. N EGATIVE -R ESISTANCE M ODULATION FOR
factor. This is the “realized radiation efficiency,” as defined in E FFICIENCY-BANDWIDTH I NCREASES
[43] – in analogy with the definition of antenna realized gain
A. Modulation Effects
(which also accounts for the mismatch loss factor). For an LTI
antenna, the mismatch loss factor is P acc (ω) As discussed in Section III, the interpath relation implies
Pinc (ω) , where Pinc (ω)
is the power incident from the RF source. For an LTI antenna, that time-varying capacitances of the form of (4) can be used
the realized radiation efficiency would then be defined as: to couple an RF signal centered around fundamental frequency
ω to one of the non-radiative modes of the antenna. In this
Prad (ω) section, it is shown that the radiation and matching bandwidths
η̃rlz (ω) = . (23)
Pinc (ω) of the antenna can be increased through this modal coupling,
if appropriate values are chosen for the modulation frequency
Extending this definition to a time-varying antenna, we
ωmod and modulation depth M .
include the power supplied by the modulation source in the
Let us start by choosing the modulation frequency to couple
denominator of the realized radiation efficiency. If this change
the resonance of the radiative mode (ω0 ) with the negative-
were not made, the apparent efficiency of the antenna could
frequency image of the resonance of a non-radiative mode
exceed unity, since the radiated power can exceed the incident
(e.g., ω1 ). Practically, this means that one of the harmonics of
RF power. This is possible because parametric modulation
ω0 must be equal to −ω1 : −ω1 = ω0 + ℓωmod . If ℓ = −1,
yields a (small) RF power gain at certain frequencies (see
then the modulation frequency becomes:
Fig. 7, for example). Therefore, for the time-varying antenna:

Prad (ω) ωmod = ω0 + ω1 . (26)

ηrlz (ω) = P . (24)
Pinc (ω) + ℓ Pmod (ωℓ ) Since this modulation frequency can be used to generate
Note that the power from the modulation source is summed negative-resistance parametric amplification, we refer to it as
over all harmonics present in the system, while the incident RF “negative-R modulation” [37]. If the modulation frequency is
power is assumed to be present only at the desired operating chosen to deviate from this value by some frequency ∆, then
frequency ω. this is termed “detuned negative-R modulation”, for which:
With definitions (22) and (24), efficiency-bandwidth prod-
ucts can be defined for a parametrically time-varying antenna. ωmod = ω0 + ω1 + ∆, (27)

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 7

𝑀 = 0.05

Prad /Pinc dB
Radiated Power/Incident Power (dB)

𝑀 = 0.10

Pref /Pinc dB
𝑀 = 0.20

Frequency MHz

Fig. 6. Radiated power (top) and total reflected power (bottom) vs. RF
Frequency MHz signal frequency for a range of different modulation depths. The radiated
and reflected powers are both normalized to the available power from the
RF source. For each modulation depth, the modulation frequency is chosen
Fig. 5. Simulation results showing the normalized radiated power through such that the total reflected power remains 9.54 dB below the incident RF
the antenna vs. the frequency of the RF signal when the modulation scheme power. As the modulation depth increases, the antenna is able to remain
from Fig. 4 is applied. Each curve represents a different combination of matched over a wider range of RF frequencies. For frequencies at which
modulation depth M and modulation frequency ωmod . Three of the curves the antenna is matched, the radiated power is approximately constant. The
which have the same modulation frequency but different modulation depths quantity Pref /Pinc is analogous to |S11 |2 for an LTI antenna, but also
have been highlighted in red. Top: for small modulation depths there is a accounts for reflected power at all harmonics, as discussed in Section IV.
single modulation frequency for which the gain (radiated power normalized to
incident RF power) is the largest. Middle: as the modulation depth increases,
there are two possible choices of modulation frequency that yield maximum
gain: one detuned to a smaller value and one detuned to a larger value. Bottom: computed by multiplying the power accepted into the ports
as the modulation depth continues to increase, the modulation frequencies
yielding maximum gain spread further apart. of the radiator at the fundamental harmonic by the radiation
efficiency at that frequency (as computed in HFSS). By energy
conservation, the power supplied by the modulation source
where ∆ is the detuning parameter. The choices of modulation can be computed as the difference between the total power
frequency given by (26) and (27) are illustrated in Fig. 4. In dissipated/radiated in the antenna and the total power accepted
general, (20) implies that if we would like to couple the reso- at the RF port.
nance of the radiative mode at frequency ω0 to the resonance The effect of the detuned negative-R modulation on the
of mode m at frequency ωm by negative-R coupling, then the antenna’s radiated power spectrum is shown in Fig. 5. The
modulation frequency should be chosen as ωmod = ω0 + ωm , modulation index and modulation frequencies are chosen to
and the modulation index should be chosen using the selection couple the m = 0 mode to the m = 1 mode (as in Fig. 4).
rule p = mod4 {m}. In this example, the appropriate choice There are three families of curves in Fig. 5, where each is char-
of modulation index is p = 1. acterized by a specific modulation depth M . The individual
To simulate the time-varying antenna shown in Fig. 1, a hy- curves correspond to particular modulation frequencies. For
brid approach was used that combines full-wave simulations of low modulation depths, there is a single modulation frequency
the antenna structure with harmonic balance simulations. All for which the gain (radiated power normalized by incident
full-wave simulations of the antenna structure were performed power) attains its maximum value. This modulation frequency
in ANSYS HFSS. Due to the symmetry of the structure, the corresponds to the case when the m = 0 and m = 1 modal
complete 4-port S-parameters of the device could be obtained resonances are exactly coupled, without detuning. This effect
by simulating a single sector (with the appropriate boundary can be visually seen in the top plot of Fig. 5, where the peak
conditions applied) as described in [40]. These S-parameters with the largest gain is located exactly above the center of the
were then imported into harmonic balance simulations of the LTI m = 0 resonance. For larger modulation depths, however,
full, time-varying antenna. The harmonic balance simulations there are two modulation frequencies at which maximum
were performed in Keysight ADS. gain is achieved. One of these modulation frequencies has
Results from the harmonic balance simulations were used a positive detuning parameter and the other has a negative
to determine the powers that appear in (22) and (24). For detuning parameter. We call the modulation depth at which
example, the reflected power from the matching network is this transition occurs the critical modulation depth. For the
the difference of the total accepted power into the matching example in Fig. 5, this critical modulation depth is Mc = 0.06.
network and the incident (available) RF power from the source. At modulation depths that are larger than the critical mod-
The radiated power at the fundamental harmonic can be ulation depth, it is possible to choose a detuned modulation

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 8

frequency such that the antenna exhibits low gain, but a large

𝑃/𝑃inc dB
radiating bandwidth. This is exemplified by the red curves in
the middle and bottom plots of Fig. 5, which have the same
modulation frequency but different modulation depths. In the
middle sub-plot (M = 0.10), the radiated power spectrum
highlighted in red has a large gain over a narrow bandwidth.
However, in the bottom sub-plot (M = 0.20), the same choice
of modulation frequency yields a radiated power spectrum with

𝑃/𝑃inc dB
gain near unity but over a broader bandwidth.
In addition to determining the radiated power spectrum, the
choice of the modulation frequency at a particular modulation
depth also determines the reflected power spectrum. In fact, it
is possible to select a detuned modulation frequency for which
the total power reflected back to the RF source remains less

Efficiency
than a threshold. In Fig. 6, total reflected power spectra are
shown for a range of modulation depths, each with a different
modulation frequency. The modulation frequency for each
curve is specifically chosen to keep the total reflected power,
including all mixing products, 9.54 dB below the incident
RF power (corresponding to a maximum VSWR of 2). As Frequency MHz
the modulation depth increases beyond the critical modulation
depth, so does the achievable matching bandwidth and radiat- Fig. 7. Comparison of the results of the LTI and parametrically time-
ing bandwidth. Designing the critical modulation depth to be varying antennas for single-tone negative-R modulation. Solid curves indicate
results from the hybrid full-wave/harmonic balance simulations, while dashed
small therefore aids in obtaining larger bandwidths from the curves indicate results from harmonic balance simulations of the equivalent
time-varying antenna. One way the critical modulation depth circuit model. Top: The radiated power for both the LTI and time-varying
can be decreased is by reducing the losses in the idler mode(s), antennas, normalized to the incident RF power, as a function of the RF
signal frequency. The time-varying antenna (red) is able to radiate power
which highlights in the importance of isolating the idler modes over a broader bandwidth than the LTI antenna (blue). Power supplied by
from the impedance of the RF source (see Section II-B). the modulation source (green), also normalized to the incident RF power, is
Therefore, by using SDTWM to couple the radiative mode only significant for frequencies near the lower edge of the passband. Power
radiated in spurious harmonics is also shown (gray). Power radiated from
to a non-radiative mode, it is possible to improve the matching the idler harmonic remains 30 dB below the transmitted signal, and most
and radiating bandwidths of the time-varying antenna over of the spurious power comes from a single harmonic (black). Middle: As
those of the time-invariant antenna. a result of applying the parametric modulation, the power reflected from
the time-varying antenna (red) remains below the -9.54 dB threshold over
a much broader bandwidth than is achieved by the LTI antenna (blue). The
threshold corresponds to a VSWR of 2, and the vertical dashed lines indicate
B. Single-Tone Modulation Scheme the edges of the passband for each antenna. Bottom: The radiation efficiency
for the time-varying antenna (dark red) is lower than the radiation efficiency
Having shown how the detuned negative-R modulation for the LTI antenna (dark blue), as expected. However, the realized radiation
could be used to obtain increased matching and radiation efficiency of the time-varying antenna (light red) exceeds the realized radiation
bandwidths, let us now use the insights from the prior section efficiency of the LTI antenna (light blue) outside the bandwidth of the LTI
antenna.
to choose the modulation parameters and report the resulting
efficiency-bandwidth products. As before, the modulation in-
dex in (4) is chosen as p = 1 in order to couple the radiating
mode to the m = 1 mode. Since simulations revealed that flat – there is no significant passband ripple. The radiated
larger modulation depths yield larger bandwidths, we choose and reflected powers computed using the circuit model closely
a relatively large modulation depth of M = 0.45. By setting agree with the results of the hybrid full-wave/harmonic balance
the modulation frequency to 515.8 MHz, we can keep the total approach, indicating the utility of the circuit model.
reflected power in the passband of the antenna at least 9.54 Next, it is worth noting that the power supplied by the
dB below the incident RF power. This value of the modulation modulation source does not make up a significant fraction of
frequency is detuned by ∆ = −4.0 MHz. All simulation the radiated power, except near the lowest edge of the pass-
results were obtained using the hybrid full-wave/harmonic band. This implies that the modulation source is not simply
balance method described in Section V-A. amplifying RF signal power that happens to be accepted into
The radiation, matching, and efficiency characteristics of the antenna. Rather, the match provided by the parametric
the time-varying antenna with these modulation parameters modulation allows the incident RF power to radiate through
are shown in Fig. 7 as a function of the RF signal frequency. the antenna, instead of being reflected back to the RF source.
There are several interesting features of these results to point The upper and lower band limits of the antennas can be
out. First, the reflected power from the time-varying antenna taken as the frequencies where the reflected power is exactly
remains below the threshold over a bandwidth that is much 9.54 dB below the RF incident power (corresponding to
larger than that obtained by the LTI antenna. Over this VSWR = 2). Note that the center frequency of the time-
passband, the radiated power from the antenna is relatively varying antenna is smaller than that of the LTI antenna. This

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 9

is because the detuning parameter (27) was negative for this Two-Tone Negative-R
design. It is also possible to use a positive detuning parameter, 𝒎=𝟎
with Detuning
in which case the center frequency of the time-varying antenna 𝝎𝟎
𝒎=𝟑 𝒎=𝟏

𝝎𝒎𝒐𝒅,𝟏

𝝎𝒎𝒐𝒅,𝟐
moves to a higher frequency than that of the LTI antenna.
𝝎𝟏
Using the upper and lower band limits of the antennas, the 𝒎=𝟐
efficiency-bandwidth metrics can be computed as in (25). The 𝝎𝟐
efficiency-bandwidth products for the LTI and parametrically 𝝎=𝟎
time-varying versions of the antenna are compared in Table
II.

TABLE II
E FFICIENCY-BANDWIDTH R ESULTS FOR THE S INGLE -T ONE M ODULATION
S CHEME

Fig. 8. A graphical representation of the two-tone negative-R modulation

Time-Varying Factor of scheme, showing the relationship between the modal resonant frequencies and
Metric LTI Antenna
Antenna Improvement the modulation frequencies. The time-varying capacitance in each sector of
the antenna is modulated with two different frequencies, ωmod,1 and ωmod,2 .
Matching BW 2.51 MHz 12.34 MHz 4.92 (6.92 dB) The modal resonances {ω0 , ω1 , ω2 } coupled by these modulation frequencies
are linked by double-ended arrows. The first modulation frequency couples the
ηrad · BW 2.08 MHz 9.28 MHz 4.45 (6.49 dB) radiative m = 0 mode to negative-frequency resonance of the non-radiative
m = 1 mode and is detuned positively. The second modulation frequency
ηrlz · BW 2.00 MHz 8.86 MHz 4.42 (6.45 dB) couples to the negative-frequency resonance of the m = 2 mode and is
detuned negatively.

Because the radiation efficiency of the time-varying antenna

(22) also has to account for power lost in mixing products with a positive detuning parameter, and the other couples
and power supplied by the modulation source, it will always the radiative mode to the m = 2 mode with a negative
be lower than that of the LTI antenna. However, the realized detuning parameter. This two-tone modulation scheme extends
radiation efficiency of the time-varying antenna (24) can be the matching and radiation bandwidths of the time-varying
larger than the realized radiation efficiency of the LTI antenna. antenna beyond those of the antenna with the single-tone
This is because the realized radiation efficiency penalizes modulation scheme.
mismatch. Both radiation efficiency-bandwidth and realized The time-varying capacitance on sector n for coupling the
radiation efficiency-bandwidth products are increased by more radiative mode to two different non-radiative modes takes the
than 6.4 dB for the time-varying antenna – compared to form:
the LTI antenna. Including parasitic losses typical of varactor
diodes leaves the matching bandwidth essentially unaltered,  
but reduces the improvement in efficiency-bandwidth product 2πp1
Cn (t) = C0 + 2M1 C0 cos ωmod,1 t − n
by about 0.5 dB. 4
  (28)
Fig. 7 also shows that the power radiated in the idler 2πp2
harmonic (at frequency f − fmod ) remains more than 30 + 2M2 C0 cos ωmod,2 t − n .
4
dB below the radiated RF signal. This is expected, since
this harmonic is confined to the m = 1 dark mode of the Since ωmod,1 couples the radiative mode to the m = 1 mode,
antenna. However, there are higher-order harmonics generated it is required that the modulation index on tone 1 be p1 =
in the system due to the time-variation, which can contribute 1. Similarly, since ωmod,2 will couple the radiative mode to
to the radiation of spurious harmonics. In this case, most the m = 2 mode, the modulation index on tone 2 must be
of the spurious radiated power comes from the harmonic at p2 = 2. The modulation frequencies ωmod,1 and ωmod,2 are
frequency −f + 4fmod . This harmonic radiates more strongly detuned positively and negatively, respectively. This two-tone
than the idler harmonic, but still remains nearly 20 dB below modulation scheme is depicted in Figure 8.
the radiated RF signal throughout the passband of the antenna. Simulations of the time-varying antenna with two modula-
The presence of this spurious harmonic is explained in Section tion tones reveal that multiple choices of modulation depths
VI, where strategies for further suppression of this harmonic M1 , M2 and modulation frequencies fmod,1 , fmod,2 can yield
are also suggested. the desired behavior. Heuristically, increasing the modulation
depth of one of the modulation frequencies also alters the
effective modulation depth of the other modulation frequency.
C. Two-Tone Modulation Scheme Therefore, choosing appropriate combinations of modulation
Coupling the radiative mode to two different non-radiative parameters is an iterative process.
modes simultaneously can yield a broader bandwidth match Representative performance characteristics of the time-
that is symmetric about the center of the passband. This effect varying antenna under the the two-tone modulation scheme
can be achieved by modulating the capacitance in each sector given by (28) are shown in Fig. 9. For these results, the
with two different modulation frequencies, fmod,1 and fmod,2 . modulation parameters were M1 = 0.37, M2 = 0.08,
One frequency couples the radiative mode to the m = 1 mode fmod,1 = 555.16 MHz, and fmod,2 = 502.16 MHz.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 10

TABLE III
E FFICIENCY-BANDWIDTH R ESULTS FOR THE T WO -T ONE M ODULATION
𝑃/𝑃inc dB
S CHEME

Time-Varying Factor of
Metric LTI Antenna
Antenna Improvement

Matching BW 2.51 MHz 16.47 MHz 6.56 (8.17 dB)

𝑃/𝑃inc dB

ηrad · BW 2.08 MHz 12.10 MHz 5.81 (7.64 dB)

ηrlz · BW 2.00 MHz 11.58 MHz 5.78 (7.62 dB)

Due to the low radiation efficiencies of the dark modes,

the radiated power in the idler harmonics (at frequencies
Efficiency

f − fmod,1 and f − fmod,2 ) remains more than 30 dB below

the radiated power in the RF signal. As was seen for the
single-tone modulation scheme, higher-order harmonics also
contribute to spurious radiation. In this case, there are two
higher-order harmonics that are responsible for the majority
Frequency MHz of the spurious radiation: one at frequency −f + 4fmod,1 and
another at frequency f + 4fmod,1 − 2fmod,2 . Despite the pres-
Fig. 9. Comparison of the results of the LTI and parametrically time-varying ence of these harmonics, the spurious radiated power remains
antennas for two-tone negative-R modulation. Solid curves indicate results nearly 20 dB below the signal power over the passband of
from the hybrid full-wave/harmonic balance simulations, while dashed curves the antenna. Additional attention is given to explaining the
indicate results from harmonic balance simulations of the equivalent circuit
model. Top: The radiated power for both the LTI (blue) and time-varying (red) presence of these spurious harmonics in the following section.
antennas, normalized to the incident RF power, as a function of the RF signal
frequency. The power supplied by the modulation source, normalized to the VI. D ISCUSSION
incident RF power, is also shown (green). The power radiated in spurious
harmonics remains nearly 20 dB below the radiated RF signal throughout the Both of the negative-R modulation schemes discussed in
passband (gray), and the idler harmonics remain more than 30 dB below the the prior section allow the antenna to radiate an RF signal
radiated RF signal. Most of the spurious radiation comes from two higher-
order harmonics (black). Middle: the total power reflected back to the RF efficiently over a bandwidth that is much wider than that of the
source for the LTI (blue) and time-varying (red) antennas, with the limits of LTI antenna. Since the radiation and realized efficiencies were
the reflection bandwidth shown in vertical dashed lines. Bottom: The radiation defined in Section IV to account for power supplied by the
efficiency for the time-varying antenna (dark red) and for the LTI antenna
(dark blue), as well as the realized efficiencies of both antennas. modulation source, the enhanced efficiency-bandwidth metrics
do not come at the expense of a hidden power draw. Nor is the
power supplied by the modulation simply amplifying the RF
As was seen for single-tone modulation, the time-varying input signal. Rather, the parametric modulation provides an
antenna can radiate power over a broader bandwidth than the effective input impedance match to the antenna. This allows
LTI antenna while maintaining an effective impedance match the time-varying antenna to radiate power that would have
at the RF input of the antenna. In this case, the response otherwise been reflected back to the RF source.
of the time-varying antenna is roughly symmetric about the Not only does the time-varying antenna improve upon
center of the passband. The results computed from the cir- the base LTI antenna, it also achieves efficiency-bandwidth
cuit model align well with the radiated and reflected power products that are only realizable by larger LTI electrically-
responses computed using the hybrid full-wave/harmonic bal- small antennas. This can be seen in Fig. 10, which compares
ance method. the realized efficiency-bandwidth products from simulations of
Unlike the single-tone modulation case, the center frequency the time-varying antenna with LTI antennas of different elec-
of the time-varying antenna is larger than that of the LTI trical sizes [44]. The performance points of the antennas are
antenna. However, if the modulation depths were changed also compared with upper limits on the efficiency-bandwidth
so that M2 > M1 , then maintaining an impedance match product of small antennas as a function of electrical size –
would require reducing both modulation frequencies. This assuming a single-tuned matching network with a maximum
would have the effect of reducing the center frequency of allowable VSWR of two.
the time-varying antenna below that of the LTI antenna. The Several of the antennas reported in Fig. 10 are multi-mode
particular choice of modulation parameters in the present antennas that essentially use the antenna structure to form a
design was determined so as to suppress the radiation of higher-order matching network [15], [18], [20]. This allows
spurious harmonics (see Section VI). them to approach or surpass the efficiency-bandwidth limit
The efficiency-bandwidth products of the time-varying an- for small antennas with a single-tuned matching network. It
tenna are both improved over those of the LTI antenna by 7.6 is worth noting that these designs, which are among the best-
dB, as shown in Table III. This is caused by the increased performing LTI antennas, each have an electrical size larger
matching bandwidth of the parametrically modulated antenna. than ka > 0.5. The space-time varying antenna concept of

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 11

This work This work

(single-tone (two-tone [15] [18]
Idler
modulation) modulation)
−𝒇 + 𝟒𝒇𝒎𝒐𝒅
𝐵𝑊 ∙ 𝜂

[20]
Limit: 𝑀 = 1
Limit: 𝑀 = 2
Gustafsson limit [35]
This work
Space-filling antenna
Idler −𝒇 + 𝟒𝒇𝒎𝒐𝒅,𝟐
(LTI)
Planar antenna 𝒇 − 𝟐𝒇𝒎𝒐𝒅,𝟏 + 𝟒𝒇𝒎𝒐𝒅,𝟐

𝑘𝑎

Fig. 10. A comparison of the efficiency-bandwidth products of antennas of

different electrical sizes, ka. The performance of the time-varying antennas
from this work are shown for both one and two-tone modulation schemes
(green and red dots, respectively), as well as the corresponding four-sector
LTI antenna upon which these time-varying antennas are based (blue dot). The
other antennas on the plot are drawn from the review in [44], except where
explicit citations are given. The limits shown in the black curves represent the
achievable efficiency-bandwidth product of LTI small antennas given the Chu Fig. 11. Radiated power spectra for a the one-tone modulation scheme
limit, assuming a single-tuned matching network with a maximum allowable (top) and for the two-tone modulation scheme (bottom). The fundamental
VSWR of 2; the M = 1 curve applies when a single electric or magnetic tone (starred) is chosen near the center of the passband of the time-varying
dipole mode is radiated, while the M = 2 curve applies with both electric antennas, and all other spectral components represent spurious higher-order
and magnetic dipole modes radiate. The LTI antennas that are explicitly cited harmonics. The specific harmonics singled out in Figs. 7 and 9 are labelled,
are examples of multi-mode antennas. These essentially incorporate a double and the mode that each harmonic falls into is indicated by the style of marker.
tuned matching network into the antenna structure, explaining why they seem Note that the most significant spurious harmonics tend to be in the radiating
capable of surpassing the M = 1 limit. Gustafsson has proposed a tighter m = 0 mode, which is especially clear for the two-tone modulation scheme.
performance bound based on the physical envelope of a particular LTI antenna
[11]. This Gustafsson limit is plotted as a blue curve for a cylindrical envelope
that fits the specific aspect ratio of the four-sector monopole in this work by at least 30 dB from the signal power level). However,
[35], showing that the LTI antenna has an efficiency-bandwidth product that
approaches this tighter bound.
both modulation schemes resulted in the radiation of one
or more spurious harmonics containing more power than the
idler harmonics. The presence of these spurious harmonics
this work also couples multiple modes of the antenna to form can be explained by the interpath relation. For the single-
an impedance match, but at a reduced electrical size around tone modulation scheme, the modulation index is p = 1, so
ka = 0.25. according to (20), every harmonic f + ℓfmod obeying the
The main difference between the single-tone and two- selection rule
tone modulation schemes is that the latter results in larger mod4 {ℓ} = 0 (29)
efficiency-bandwidth products (and yields symmetric passband falls within the m = 0 mode. Similarly, for the two-tone
responses). However, a comparison of the results in Tables II modulation scheme shown in Fig. 8, the modulation indices are
and III shows that coupling to two idler modes instead of p1 = 1 and p2 = 2, so every harmonic f +ℓ1 fmod,1 +ℓ2 fmod,2
one only increases the efficiency-bandwidth metrics by about for which the selection rule
1 dB. Therefore, there are diminishing returns to increased
modulation scheme complexity, just as there are diminishing mod4 {ℓ1 + 2ℓ2 } = 0 (30)
returns in increasing the order of an LTI matching network. holds will fall within the radiating mode. The most significant
This is a consequence of constraining the capacitance in each spurious harmonics shown in Figs. 7 and 9 satisfy these
sector, as given by (28), to be a positive, real-valued function. conditions. Since they fall into the radiating mode of the
Applying this constraint limits the combined magnitudes of antenna (unlike the idler harmonics), it makes sense that they
modulation depths such that M1 + M2 < 0.5. Since larger account for most of the spurious radiated power.
modulation depths yield larger bandwidths, the constraint on This intuition is verified in Fig. 11, which shows the
the magnitude of the modulation depths limits the achievable spurious harmonic spectra for both one-tone and two-tone
bandwidth. This constraint on the sum of the modulation modulation schemes, assuming a single input frequency near
is not necessarily fundamental. For example, using separate the center of the matching bandwidth. The harmonics that
time-varying capacitors for each modulation tone would relax fall into the m = 0 mode tend to be larger in amplitude
the modulation depth constraint to two separate constraints: than harmonics falling into the non-radiating modes. It is also
M1,2 < 0.5. clear from this figure that there are more spurious harmonics
As expected, the power radiated from the idler harmonics generated by the two-tone modulation scheme than by the one-
in both modulation schemes was strongly suppressed (down tone modulation scheme.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 12

The presence of spurious harmonics is a challenge in those that would be used to yield maximum parametric gain.
transmit mode because of FCC requirements on out-of-band The non-radiative modes of the antenna play the role of low-
emissions. However, coupling to higher-order harmonics can loss idler resonators and inherently suppress the radiation of
also be problematic in receive mode, where noise or interfering the idler harmonics. Making use of traveling-wave modulation
signals coinciding with spurious harmonics of the system significantly reduced the likelihood of spurious coupling to
can be mixed into the received signal. This motivated the higher-order resonances of the radiating mode. The radiated
definition of an “equivalent external noise temperature” in power from spurious harmonics remained nearly 20 dB be-
[45]. Therefore, unlike in the subharmonic mixing antenna low the RF signal level. It was also found that the time-
reported in [41], avoiding the radiation or reception of these modulated antenna could be designed to show more than 7.6
higher order harmonics is desirable. dB improvement in its efficiency-bandwidth product compared
There are a number of possible design approaches for to the LTI version of the antenna. This indicates that the
suppressing these spurious harmonics. The first is suggested bandwidth broadening was not simply obtained at the expense
by the selection rules (29) and (30): increase the number of of losses due to dissipation, mismatching, or power required
sectors from N = 4 so that fewer harmonics fall into the to modulate the antenna.
radiative mode. It is worth noting that these selection rules are
a direct consequence of using SDTWM. If a different form of R EFERENCES
modulation were used, these selection rules would not apply,
and every harmonic could potentially couple to the radiative [1] H. A. Wheeler, “Fundamental limitations of small antennas,” Proc. IRE,
vol. 35, no. 12, pp. 1479–1484, 1947.
mode. [2] L. J. Chu, “Physical limitations of omni-directional antennas,” J. Appl.
Spurious harmonics may also be suppressed by ensuring Phys., vol. 19, no. 12, pp. 1163–1175, 1948.
that they do not fall near a resonance of the antenna. This [3] R. F. Harrington, “Effect of antenna size on gain, bandwidth, and
efficiency,” J. Res. Nat. Bur. Stand, vol. 64, no. 1, pp. 1–12, 1960.
can be done by changing the detuning parameter(s), which [4] R. Collin and S. Rothschild, “Evaluation of antenna Q,” IEEE Trans.
alters the frequencies of the spurious harmonics. In the case Antennas Propag., vol. 12, no. 1, pp. 23–27, 1964.
of one-tone modulation, the detuning parameter can either [5] R. Fante, “Quality factor of general ideal antennas,” IEEE Trans.
Antennas Propag., vol. 17, no. 2, pp. 151–155, 1969.
be chosen to be positive or negative to yield an impedance [6] J. S. McLean, “A re-examination of the fundamental limits on the
match. As explained in Section V-C, there is a continuum fadiation Q of electrically small antennas,” IEEE Trans. Antennas
of permissible choices of the two detuning parameters for Propag., vol. 44, no. 5, p. 672, 1996.
two-tone modulation. Therefore, one has the design freedom [7] A. D. Yaghjian and H. R. Stuart, “Lower bounds on the Q of electrically
small dipole antennas,” IEEE Trans. Antennas Propag., vol. 58, no. 10,
to change the modulation frequencies, thereby translating pp. 3114–3121, 2010.
problematic harmonics away from antenna resonances. [8] H. L. Thal, “Q bounds for arbitrary small antennas: A circuit approach,”
IEEE Trans. Antennas Propag., vol. 60, no. 7, pp. 3120–3128, 2012.
A final approach to suppressing spurious harmonics is to
[9] A. Yaghjian, M. Gustafsson, and L. Jonsson, “Minimum Q for lossy and
reduce the radiation efficiency of the antenna at the harmonic lossless electrically small dipole antennas,” Prog. Electromagn. Res., vol.
frequencies by altering the structure of the antenna. 143, pp. 641–673, 2013.
The radiated and reflected power responses of the time- [10] C. Pfeiffer and A. Grbic, “A circuit model for electrically small
antennas,” IEEE Trans. Antennas Propag., vol. 60, no. 4, pp. 1671–
varying antenna designs were well-predicted by the circuit 1683, 2012.
model. Given the above discussion of spurious harmonics, it is [11] M. Gustafsson, C. Sohl, and G. Kristensson, “Physical limitations on
worth noting that the circuit model cannot predict the coupling antennas of arbitrary shape,” Proc. Math. Phys., vol. 463, no. 2086, pp.
2589–2607, 2007.
of higher-order harmonics to higher-order antenna resonances. [12] M. Gustafsson, M. Capek, and K. Schab, “Tradeoff between antenna
This is because the circuit model only accurately captures the efficiency and Q-factor,” IEEE Trans. Antennas Propag., vol. 67, no. 4,
fundamental resonances of the antenna. The circuit model can, pp. 2482–2493, 2019.
[13] A. D. Yaghjian and S. R. Best, “Impedance, bandwidth, and Q of
in principle, predict the coupling of higher-order harmonics to antennas,” IEEE Trans. Antennas Propag., vol. 53, no. 4, pp. 1298–
a fundamental antenna resonance. That said, the strength of the 1324, 2005.
circuit model is in providing a tool for understanding how the [14] M. Manteghi, “Fundamental limits, bandwidth, and information rate of
electrically small antennas: Increasing the throughput of an antenna
detuned negative-R modulation schemes provide an enhanced without violating the thermodynamic Q-factor,” IEEE Antennas Propag.
matching bandwidth, which we plan to show in future work. Mag., vol. 61, no. 3, pp. 14–26, 2019.
Other future work includes an experimental realization of the [15] G. Goubau, “Multi-element monopole antennas,” in Proc. ECOM-ARO
Workshop on Electrically Small Antennas. Army Electronics Command,
space-time varying antenna concept developed here. Such a 1976, pp. 63–67.
realization could implement the time-varying capacitors with [16] N. Puri, G. Goubau, S. Stavridis, and S. Fich, “Mathematical modeling
pumped varactor diodes, as in [37], [46], [47]. of multi-element monopole antennas,” Rutgers, The State University,
Tech. Rep., 1981.
[17] C. Friedman, “Wide-band matching of a small disk-loaded monopole,”
IEEE Trans. Antennas Propag., vol. 33, no. 10, pp. 1142–1148, 1985.
VII. C ONCLUSION
[18] T.-H. Yang and D. F. Sievenpiper, “Intuitive broadband matching
By coupling radiative and non-radiative modes of a rota- technique for top-loaded monopole antennas,” IEEE Trans. Antennas
Propag., vol. 67, no. 12, pp. 7611–7616, 2019.
tionally symmetric ESA using SDTWM, the radiation and [19] S. R. Best, “Low Q electrically small linear and elliptical polarized
matching bandwidths of the antenna were increased by more spherical dipole antennas,” IEEE Trans. Antennas Propag., vol. 53, no. 3,
than 6.9 dB. In this approach, the antenna behaves in a manner pp. 1047–1053, 2005.
[20] H. R. Stuart, S. R. Best, and A. D. Yaghjian, “Limitations in relating
analogous to a negative-resistance parametric amplifier, but quality factor to bandwidth in a double resonance small antenna,” IEEE
with modulation frequencies detuned to frequencies away from Antennas Wirel. Propag. Lett., vol. 6, pp. 460–463, 2007.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 13

[21] H. R. Stuart and S. R. Best, “A small wideband multimode antenna,” [45] K. Schab and K. C. Kerby-Patel, “Equivalent External Noise Tempera-
in Proc. IEEE Antennas Propag. Soc. Int. Symp., San Diego, CA, USA, ture of Time-Varying Receivers,” Jun. 2024, arXiv:2405.20471.
Jul. 2008, pp. 1–4. [46] M. T. Faber, “Microwave Parametric Amplifiers,” in Encyclopedia of
[22] H. W. Bode, Network Analysis and Feedback Amplifier Design. D. Van Electrical and Electronics Engineering. John Wiley & Sons, 2005.
Nostrand Company, Inc., 1945. [47] N. A. Estep, D. L. Sounas, J. Soric, and A. Alù, “Magnetic-free non-
[23] H. A. Wheeler, “Wideband impedance matching,” Wheeler Laboratories, reciprocity and isolation based on parametrically modulated coupled-
Inc., Tech. Rep. 418, 1950. resonator loops,” Nature Physics, vol. 10, no. 12, pp. 923–927, 2014.
[24] R. M. Fano, “Theoretical limitations on the broadband matching of
arbitrary impedances,” J. Frank. Inst., vol. 249, no. 1, pp. 57–83, Jan.
1950.
[25] A. Lopez, “Review of narrowband impedance-matching limitations,”
IEEE Antennas Propag. Mag., vol. 46, no. 4, pp. 88–90, Aug. 2004.
[26] C. Pfeiffer and B.-I. Wu, “On the maximum communication data rate
of resonant antennas,” IEEE Trans. Commun., vol. 71, no. 3, pp. 1377–
Zachary Fritts (Student Member, IEEE) received
1387, Mar. 2023.
the B.S. degree in bioengineering from the Univer-
[27] H. Li, A. Mekawy, and A. Alù, “Beyond Chu’s limit with Floquet sity of Pittsburgh in Pittsburgh, PA, USA in 2020,
impedance matching,” Phys. Rev. Lett., vol. 123, no. 16, p. 164102, graduating Summa Cum Laude.
2019. In 2015, 2016, and 2017 he was a research as-
[28] A. Mekawy, H. Li, Y. Radi, and A. Alù, “Parametric enhancement of sistant at the Ohio Agricultural Research and De-
radiation from electrically small antennas,” Phys. Rev. Appl., vol. 15, velopment Center (OARDC), Wooster, OH, USA.
no. 5, p. 054063, 2021. From 2018-2020 he was an undergraduate researcher
[29] P. Loghmannia and M. Manteghi, “Broadband parametric impedance in the 7T Bioengineering Research Program at the
matching for small antennas using the Bode-Fano limit: Improving on University of Pittsburgh. He is currently a Ph.D.
Chu’s limit for loaded small antennas,” IEEE Antennas Propag. Mag., student in the group of Prof. Anthony Grbic at
vol. 64, no. 5, pp. 55–68, 2021. the University of Michigan, Ann Arbor, MI, USA. His research interests
[30] M. Mostafa, N. Ha-Van, P. Jayathurathnage, X. Wang, G. Ptitcyn, and include time-varying systems, electrically-small antennas, network analysis
S. Tretyakov, “Antenna bandwidth engineering through time-varying and modeling, electromagnetic metamaterials, and electromagnetic theory.
resistance,” Appl. Phys. Lett., vol. 122, no. 17, 2023. Mr. Fritts was awarded a fellowship through the NSF Graduate Research
[31] Z. Fritts, S. M. Young, C. Scarborough, and A. Grbic, “Increasing the Fellowships Program in 2020.
efficiency-bandwidth of small antennas by coupling radiative and non-
radiative modes using time-variation,” in 2023 United States National
Committee of URSI National Radio Science Meeting (USNC-URSI
NRSM), Boulder, CO, USA, 2023, pp. 77–78.
[32] S. M. Young, Z. Fritts, and A. Grbic, “Increasing the efficiency-
bandwidth product of electrically-small antennas by time-dependent
parametric coupling of characteristic modes,” presented at the 2023 IEEE
AP-S International Symposium, Portland, OR, USA, July 24, 2023, Amirhossein Babaee (Student Member, IEEE) re-
2097. ceived the B.Sc. degree in Electrical Engineering
[33] Z. Fritts, A. Babaee, S. M. Young, and A. Grbic, “Increasing the from Iran University of Science and Technology
efficiency-bandwidth product and impedance bandwidth of electrically- (IUST), Tehran, Iran, in 2021. He is currently pursu-
small antennas through parametric space-time variation,” in 2024 18th ing the Ph.D. degree in Electrical Engineering with
European Conference on Antennas and Propagation (EuCAP), Glasgow, the University of Michigan, Ann Arbor, MI, USA.
Scotland, UK, 2024. During his undergraduate studies, he was with
the Applied Electromagnetic Laboratory, where his
[34] C. A. Balanis, “Antenna theory: Analysis and design,” pp. 133–137,
research focused on metamaterials, metasurfaces,
2016.
and optical computing. His current research interests
[35] M. Gustafsson, “AntennaQ,” https://www.mathworks.com/matlabcentral/
include the design and analysis of electromagnetic
fileexchange/26806-antennaq, 2009, [Online; Accessed November,
metasurfaces, spatiotemporally modulated structures, crystal optics, and non-
2023].
linear optics. He is particularly interested in exploring time-varying phenom-
[36] W. Hong and K. Sarabandi, “Low profile miniaturized planar antenna
ena and their applications in wave manipulation and dynamic electromagnetic
with omnidirectional vertically polarized radiation,” IEEE Trans. Anten-
structures.
nas Propag., vol. 56, no. 6, pp. 1533–1540, 2008.
[37] R. E. Collin, Foundations for Microwave Engineering. John Wiley &
Sons, 2007, pp. 814–821.
[38] E. Mensink, E. A. Klumperink, and B. Nauta, “Distortion cancellation by
polyphase multipath circuits,” IEEE Trans. Circuits Syst. I, Reg. Papers,
vol. 52, no. 9, pp. 1785–1794, 2005.
[39] C. Scarborough and A. Grbic, “Generalized eigenvalue problem for spa-
tially discrete traveling-wave-modulated circuit networks,” IEEE Trans. Steve Young (Senior Member, IEEE) received the
Microw. Theory Tech., vol. 71, no. 2, pp. 511–521, 2023. B.S. degree in electrical engineering from Stanford
[40] C. Scarborough, Z. Wu, and A. Grbic, “Efficient computation of spatially University, Stanford, CA, USA, in 2003; the M.S.E.S
discrete traveling-wave modulated structures,” IEEE Trans. Antennas degree in mechanical engineering from the Naval
Propag., vol. 69, no. 12, pp. 8512–8525, 2021. Postgraduate School, Monterey, CA, USA, in 2006;
[41] C. Scarborough and A. Grbic, “Efficient subharmonic frequency conver- and the Ph.D. in applied physics from the University
sion using space-time induced bound states in the continuum,” in 2021 of Michigan, Ann Arbor, MI, USA in 2016.
Fifteenth International Congress on Artificial Materials for Novel Wave From 2003 to 2009, he served in the United States
Phenomena (Metamaterials), 2021, pp. 395–397. Navy and worked as an instrumentation and control
[42] ——, “Modified Floquet boundary condition for open boundary prob- systems engineer at Naval Reactors in Washington,
lems with N-path symmetry,” in 2020 Fourteenth International Congress D.C., USA. Since 2016, he has been the President of
on Artificial Materials for Novel Wave Phenomena (Metamaterials), Xondas, Inc., an electromagnetics research and consultancy company. In 2018,
2020, pp. 027–029. he joined the Department of Electrical Engineering and Computer Science,
[43] “IEEE Standard for Definitions of Terms for Antennas,” IEEE Std. 145- University of Michigan, Ann Arbor, MI as a Postdoctoral Fellow and, later,
2013, p. 76, 2013. Assistant Research Scientist. His current research interests include electromag-
[44] D. F. Sievenpiper, D. C. Dawson, M. M. Jacob, T. Kanar, S. Kim, netic metasurfaces, dielectric structures, electrically small antennas, wireless
J. Long, and R. G. Quarfoth, “Experimental validation of performance power transfer, time-varying structures, and nonlinear electromagnetics.
limits and design guidelines for small antennas,” IEEE Trans. Antennas
Propag., vol. 60, no. 1, pp. 8–19, 2011.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
 This article has been accepted for publication in IEEE Transactions on Antennas and Propagation. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TAP.2024.3508081

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. #, NO. #, MONTH YEAR 14

Anthony Grbic (S’00 - M’06 - SM’14 – F’16)

received the B.A.Sc., M.A.Sc., and Ph.D. degrees
in electrical engineering from the University of
Toronto, Canada, in 1998, 2000, and 2005, re-
spectively. In 2006, he joined the Department of
Electrical Engineering and Computer Science, Uni-
versity of Michigan, Ann Arbor, MI, USA, where
he is currently the John L. Tishman Professor
of Engineering. His research interests include en-
gineered electromagnetic structures (metamaterials,
metasurfaces, frequency selective surfaces, photonic
bandgap structures), antennas, microwave circuits, time varying and space-
time varying electromagnetic systems, cylindrical vector beams, wireless
power transmission, and analytical electromagnetics/optics.
Dr. Grbic served as Technical Program Co-Chair in 2012 and Topic Co-
Chair in 2016 and 2017 for the IEEE International Symposium on Antennas
and Propagation and USNC-URSI National Radio Science Meeting. He was an
Associate Editor for IEEE Antennas and Wireless Propagation Letters from
2010 to 2015. Dr. Grbic was the recipient of AFOSR Young Investigator
Award as well as NSF Faculty Early Career Development Award in 2008,
the Presidential Early Career Award for Scientists and Engineers in January
2010. He also received an Outstanding Young Engineer Award from the IEEE
Microwave Theory and Techniques Society, a Henry Russel Award from the
University of Michigan, and a Booker Fellowship from the United States
National Committee of the International Union of Radio Science in 2011.
He was the inaugural recipient of the Ernest and Bettine Kuh Distinguished
Faculty Scholar Award in the Department of Electrical and Computer Science,
University of Michigan in 2012. In 2018, Prof. Grbic received a University of
Michigan Faculty Recognition Award for outstanding achievement in scholarly
research, excellence as a teacher, advisor and mentor, and distinguished
service to the institution and profession. In 2022, he received the David E.
Liddle Research Excellence Award, College of Engineering, University of
Michigan. He is currently an IEEE Microwave Theory and Techniques Society
Distinguished Microwave Lecturer and a member of the Scientific Advisory
Board of the International Congress on Artificial Materials for Novel Wave
Phenomena – Metamaterials.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 13,2025 at 15:51:08 UTC from IEEE Xplore. Restrictions apply.
© 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.