2870 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO.

6, JUNE 2018

Holographic-Inspired Multibeam Reflectarray

With Linear Polarization
Majid Karimipour and Nader Komjani

Abstract— In this paper, a novel systematic method based on in their configurations. This drawback will be more remarkable
the holographic theory to design a reflectarray (RA) antenna for designing RAs with specific features on their radiation
is presented. According to the holographic technique, one can beams, where almost it is needed to use iterative optimization
record features of electromagnetic waves impinging on a specific
surface and then, whenever needed, reconstruct each of them. algorithms. In most cases, the optimization algorithms utilize
We exploit this property to design a single-feed RA with a one of each radiation pattern computing techniques in their
multibeam linearly polarized pattern without using any iterative subroutines. Therefore, the need for implementing a fast and
method such as optimization algorithms in the design process. accurate analysis technique seems to be vital for everyone that
At first, various kinds of interferograms along with related wants to use an RA in its equipment.
input admittance distributions are introduced according to the
polarization and number of radiated beams. Afterward, a simple To date, numerous efforts have been devoted for describing
subwavelength metallic patch as a scalar admittance unit cell the radiation behavior of an RA. In general, it can divide
is employed to implement the input admittance distribution of all of the RA analysis methods into two branches. First,
the interferograms. Moreover, a fast analytical method, which is the methods focused on calculating the phase response of
fundamentally inspired by the method of moments, has been the radiating elements numerically and analytically [3]–[5],
employed to calculate the input admittance of the utilized
element. Finally, as a typical case, an 18 cm × 18 cm linearly and the second one is the methods focused on estimating
polarized RA with a two-beam radiation pattern has been the radiation characteristics of the far-field pattern, such as
manufactured and tested at 15 GHz. The sidelobe level, cross- the main beam direction, gain level, sidelobe level (SLL),
pol, and efficiency of the antenna are found −26 dB, lower than co-pol and cross-pol of the pattern, and finally shape of the
−10 dB, and 47.2%, respectively, which meet the expectations. pattern [1]. Doing so, three main classical analysis approaches
Index Terms— Artificial impedance surface, holographic for designing RA have been introduced so far:
technique, method of moments (MoM), reflectarray (RA). 1) the array theory [6]–[8];
2) the aperture field [9];
I. I NTRODUCTION 3) the equivalent current approach based on a continuous

R EFLECTARRAY (RA) antennas are very good candidate

for a wide variety of applications from microwave to
terahertz regimes such as military and surveillance radars,
plane wave spectrum [10]–[12].
To our best knowledge, in most cases the synthesis proce-
dures of RA configurations which are required for beam
point-to-point telecommunications, broadcasting systems, shaping, multibeam pattern, and contour beam applications
radiometric remote sensing, and imaging [1]. Low cost, simple are based on direct optimization methods such as evolutionary
configuration, low profile, low mass, and simple manufac- intelligent algorithms with computational cores the same
turing process (thanks to lithography techniques) are several as those aforementioned methods described for the analysis
advantages of RAs in comparison of solid parabolic reflectors of RAs [13]–[22]. The common aspect between all of the
and phased array antennas [2]. The most important challenge proposed methods for analyzing and synthesizing of RAs is the
relevant to RA antennas is to provide a fast and accurate adjustment of the phase distribution of the reflective surface
method to describe and predict the radiation characteristics. of RA concerning their specific applications. In other words,
The estimation of the far-field radiation pattern of these the radiation elements play a role of spatial phase shifter at
antennas is a crucial task, which should be accomplished a given frequency. Therefore, these methods are also called
with high precision. This fact becomes more attractive when phase-only tuning methods. This kind of attitude toward the
everybody finds out that it is a very time-consuming process to RA elements will follow its own some limitations which we
simulate RAs with full-wave simulators, because, in the vast aim to introduce them briefly in the following.
majority of cases, several hundred radiation elements are used In most RA configurations, the reflective surface is consid-
ered as a regular lattice of discrete phase distribution because
Manuscript received July 29, 2017; revised March 8, 2018; accepted
March 29, 2018. Date of publication April 6, 2018; date of current version of phase swing limitations. It means that the dimensions of
May 31, 2018. This work was supported by the Department of Electrical the periodic unit cells are comparable with a wavelength that
Engineering, Iran University of Science and Technology. (Corresponding leads to several restrictions. The limitation of providing at least
author: Nader Komjani.)
The authors are with the Department of Electrical Engineering, Iran 360° of phase delay on the radiating elements, while keeping
University of Science and Technology, Tehran 16844, Iran (e-mail: a gentle behavior on the phase characteristic curve at every
[email protected]; [email protected]). frequency band of interest, is the main prohibitive reason for
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. diminishing the unit cell dimensions smaller than λ/3. This
Digital Object Identifier 10.1109/TAP.2018.2823776 limits the bandwidth of these antennas [1]. Meanwhile, if the
0018-926X © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
KARIMIPOUR AND KOMJANI: HOLOGRAPHIC-INSPIRED MULTIBEAM RA 2871

polarization considerations add to the design requirements, this

limitation will be more prominent. The usual assessed way to
enhance the RA bandwidth is the use of radiating elements
consisting of two or more stacked printed patches [23]–[26].
As such, conventional broadband RAs are almost implemented
on multilayer and complex structures with suboptimal perfor-
mance. Recently, however, several RA configurations with
subwavelength unit cell have been represented, but design
simple-shape and single-layer radiation elements with accept-
able performances are already major challenge [3], [27]–[29].
Furthermore, some attempts have been done in the last years
to compensate the overall phase error accumulated in different
element-feed paths for all frequency bands of interest [14].
According to the above-mentioned descriptions, the aim of
this paper is to provide a novel and accurate methodology Fig. 1. Primitive schematic of an HRA. Inset: applied scalar unit cell
based on the holographic theory to design an RA and then composed of a single-layer metal patch structure.
exploit this technique to develop a multibeam RA with linear
polarization for the first time. The proposed technique not only
gives a new physical insight from radiation behavior of RA but describe the aperture EM characteristics of an antenna with
also facilitates the RA antenna design process and improves respect to the object and reference waves. In this fashion,
its overall performance, such as the operational bandwidth, the desired radiation pattern is equivalent to the object wave
efficiency, SLL, and cross-pol level. It will be shown that and the radiation pattern of antenna feed is considered as the
the proposed procedure has several advantages in comparison reference wave. To sum up, several potential advantages of
with classical methods such as simplicity and nonsensitive holographic theory in the antenna design are as follows:
to the radiation and impinging wave characteristics such as 1) frequency independent nature of the holographic
polarization, shape, and number of the beams. For this reason, method;
the holographic theory has a potential to be used for designing 2) nonsensitive to the overall shape of both object and
antennas with a wide variety of applications such as power reference waves;
combining, beam shaping, and multibeam applications. This 3) capable of implementing on conformal structures.
technique has been employed to design various kinds of one- Considering the holographic theory, the radiating aperture
and/or two-dimensionally leaky-wave antennas [30]–[36]. of RA [which we called it hereinafter holographic RA (HRA)]
In order to support the validity of the design procedure, is considered as a hologram surface which records phase and
several examples, including single- and multiple-beam holo- amplitude properties of reference (feed pattern of horn in RA
graphic RAs, are proposed and full-wave simulation results case) and object waves (radiation pattern). As a result, two
compared with those of other works, specifically the band- principal steps should be implemented to design an HRA:
width, SLL, efficiency, and cross-pol level quantities. Finally, 1) preparing the hologram surface by recording two sets of
a prototype of two-beam linearly polarized holographic RA reference(s) and object wave(s) on it and 2) reconstructing
fabricated and tested with close agreements was obtained with the object wave(s) by impinging the reference wave(s) on
simulation results. the hologram. For practical implementation of the hologram
This paper is organized as follows. In Section II, surface, which is characterized by their electromagnetic (EM)
the basic concepts of the holographic technique are intro- features such as impedance distributions, one can use several
duced and preliminary tools necessary to design a typical structures such as dielectric-only surfaces with modulating at
holographic RA, such as impedance distribution of holo- their thicknesses or employing artificial impedance surfaces
grams associated with various object waves, will be provided. consisting of an array of metal patches [36].
Section III is dedicated to implement practically an artificial
impedance surface by extracting the input impedance values B. Holographic Reflectarray
of a metal-patch unit cell embedded in a periodic environment. A schematic of the design process of an HRA is depicted
Some examples are represented and discussed in Section IV. in Fig. 1. As can be seen in Fig. 1, the desired far field
Finally, Section V includes the conclusion. and reference waves have been assumed as two impinging
plane waves originated from the far zone with respect to
II. H OLOGRAPHIC T HEORY the hologram surface. In planar structures such as leaky-
wave antennas, where the surface wave is considered as the
A. Basic Concepts reference wave, the interferogram can be characterized by
The holographic technique, in the original implementation, modulating the surface impedance quantity, which is the ratio
has been recently used to record and reconstruct the object of tangential electric field to tangential magnetic field of
image in real 3-D space based on the principle of interference surface wave [34], [35]. Since in RA configurations, we deal
defined at the optical regime [31]. This idea can exactly with space wave instead of surface wave; the interferogram can
be utilized for antenna design in the microwave regime to properly be described by the input impedance or admittance

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
2872 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO. 6, JUNE 2018

quantity which is the ratio of tangential electric field to C. Required Hologram of RA to Generate Multibeam
tangential magnetic field of the space wave near the reflecting Linearly Polarized Radiation Pattern
surface. Similarly, the hologram surface can record the features of
In light of above consideration and according to the holo- more than one object wave or equivalently one radiated beam.
graphic theory, the scalar input impedance or admittance It involves using the superposition technique and describing
distribution associated with linearly polarized waves for a the EM characteristics of the hologram in terms of all radiated
hologram located in the xy plane is given as beams. As will be shown later, these holograms are useful to
design multiple-beam HRAs. The input impedance or admit-
∗
Z |Y input (x, y) = j [X + MRe(ψrad ψref )] (1) tance description of the hologram in this case is as follows:
  n  
M 
∗
where Z |Y input is the desired input impedance or admittance Z |Y input (x, y) = j X + Re ψrad
i
ψref (3)
n
of the hologram to radiate the beam ψrad . The parameters X i=1
and M are the arbitrary average impedance or admittance value where ψrad
i quantity is
and the modulation depth, respectively. Apart from leaky-wave
antennas, where the parameters X and M affect directly on i
ψrad = Ae− j k0 (xsinθi cosφi +ysinθi sinφi ) |at z=0 . (4)
the beam direction and the attenuation constant, which the
latter affects on the beamwidth and SLL of the radiation In (4), (θi , φi ) is the outgoing angle of nth radiated electric
pattern [35], in RA antenna similar to optical holography, field in the spherical coordinate system.
where the hologram should only satisfy amplitude and phase The input admittance distribution associated with several
of interferogram, these limitations do not appear. Thus, these holograms to get the single and multiple radiation beams
two parameters should be determined in such a way that has been demonstrated in Fig. 2. In multibeam case,
the entire range of available impedance or admittance of the the main beams are considered with symmetric and asym-
hologram would be implemented by an ordinary quasi-periodic metric outgoing directions to each other from the hologram
impedance surface. Accordingly, these two parameters strictly plane point of view. In all cases, the operating frequency, the
depend on the type of the unit cell element and consequently hologram dimension, the position of the feed as a radiator of
on its range of impedance variation that can be generated. reference wave, the average admittance value, and the modu-
However, the parameter M should not be selected too small, lation depth have been considered as follows: f = 15 GHz,
because the modulation process of the hologram surface will hologram size is 20 cm ×20 cm, (x f , y f , H ) = (0, 0, 20 cm),
not be done very well and undoubtedly the beam direction and M = 0.006, and X = −0.0022. Note that, in this step,
SLL will be deteriorated. This is why the hologram approaches we aim to show how the interferograms are generated based
to a nonmodulated admittance surface. on the radiation and impinging waves and depict the overall
In order to describe the hologram impedance variations for a template of admittance variations. Therefore, the considered
linearly polarized RA in the general form, it can be assumed amount for two parameters M and X are merely selected for
that the radiation pattern is a plane wave with a collimated purposes of demonstration of surface admittance distribution.
beam at the (θ, φ) direction in the spherical coordinate system But in practical implementation, these amounts strictly depend
and the RA aperture is placed on the z = 0 plane. As such, on the unit cell configuration and should be chosen so that they
we can write ψrad = e− j k0 (xsinθcosφ+ysinθsinφ)|atz=0 , where k0 is can be practically realized by any EM structures like artificial
the free space wavenumber. Also, impedance surfaces.
 for reference wave, we have
ψref = e − j k0 r , where r = (x − x f )2 + (y − y f )2 + H 2
and (x f , y f , H ) is the position of the feed horn in the rectan- III. I MPLEMENTATION OF A RTIFICIAL
gular coordinate system. Substituting ψref and ψrad into (1), I MPEDANCE S URFACE
the input impedance or admittance will be obtained as follows: There are several procedures to realize a hologram surface
practically. A dielectric structure with sinusoidally modulated
Z |Y input (x, y) = j [X + Mcosk0 (xsinθ cosφ + ysinθ sinφ −r )]. substrate thickness and artificial impedance surfaces with
modulated dimensions are two well-known solutions that are
(2) extensively applied in holographic structures [36].
In order to hologram implementation by artificial impedance
As can be seen in (2), the input impedance or admittance surface, they must be designed so as to sample locally
has sinusoidal variations with respect to the x and y positions. the impedance distribution of the hologram. According to
It is worthwhile to note that the representation of the hologram the Nyquist theory, it should be sampled at least half of the
surface according to the admittance distribution is useful for period of the impedance variations to track all changes of the
some artificial impedance surfaces that have a resonance point impedance distribution, accurately. A well-known approach
in their impedance characteristic, where the impedance value to construct an artificial impedance surface is the use of an
approaches to infinity. In this case, X-parameter does not have array of quasi-periodic unit cells with variable patch size.
a valid amount any longer. Therefore, a proper solution is the In fact, quasi-periodic surfaces can locally meet the required
use of admittance factor instead of impedance. That is why input admittance or impedance of the hologram. So the proper
the input admittance value is zero at the resonance point. dimensions of the unit cell may be selected λ/5 × λ/5

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
KARIMIPOUR AND KOMJANI: HOLOGRAPHIC-INSPIRED MULTIBEAM RA 2873

Fig. 3. (a) Demonstration of the utilized unit cell. (b) and (c) Two
representations of the complementary counterpart of the unit cell. (Dark and
light areas represent metal and dielectric materials, respectively.)

metal-patch unit cells; however, in order to develop an

accurate and fast methodology for designing RA, the input
impedance quantity must be calculated in advance (according
to the incident wave angles and geometrical parameters of
the element) and stored in a lookup table. Obviously, this
is a very time-consuming process especially for analyzing
anisotropic structures because a lot of geometrical parameters
along with the impinging wave angle can potentially effect
on the input impedance values. As such, providing a lookup
table, including all variations of the design parameters requires
actually a lot of simulation time. In this paper, we propose a
fast, exact, and full-analytical method which enables everyone
to calculate the input scalar impedance of a predefined patch,
accurately. Section III-A has been dedicated for describing this
Fig. 2. Admittance distributions of scalar holograms for single- and
multibeam radiations. (a) Single beam at (θ, φ) = (0°, 0°) direction. method in detail.
(b) Single beam at (θ, φ) = (30°, 45°) direction. (c) Two beams at (θ, φ) =
(45°, 0°) and (45°, 180°) directions. (d) Three beams at (θ, φ) = (20°, 45°),
(30°, 135°), and (20°, 225°) directions. (e) Four beams at (θ, φ) = (30°, 45°), A. Analytical Approach for Calculating the Input
(30°, 135°), (30°, 225°), and (30°, 315°) directions. (f) Five beams at Impedance or Admittance of the Unit Cell
(θ, φ) = (20°, 45°), (30°, 135°), (20°, 225°), (30°, 315°), and (45°, 90°).
The admittance values are shown in millisiemens. In this paper, we utilize a subwavelength quasi-periodic
metal-patch structure to satisfy the input admittance distribu-
tion of the hologram. The constituent unit cell of the artificial
or smaller, where λ being the wavenumber associated with impedance surface is depicted in Fig. 3(a). The proposed
the operating frequency. The surface texture is quite dense method for extracting the input impedance of this unit cell
and appears as a continuous surface of impedance or admit- falls into two stages as: 1) metal patch sheet admittance
tance distribution by this amount of periodicity. The simplest extraction process and 2) grounded dielectric substrate admit-
structure to realize scalar impedance surface is to employ an tance extraction process. As will be shown later, these two
array of square patches with variable size. In order to calculate admittances completely determine the overall input admittance
the input impedance or admittance of a metal-patch unit cell of the desired unit cell shown in Fig. 3(a). The proposed
theoretically, it can be assumed that each patch is embedded approach to perform stage 1) comes from the scattering
in a locally uniform grid illuminated by an ordinary plane problem analysis of a periodic strip grid with constituent unit
wave with a given impinging angle and polarization and then cell shown in Fig. 3(b) or equivalently in Fig. 3(c).
solve the scattering problem to find the induced currents within We proceed our analysis method by expressing the relation
the patch and consequently extract the input impedance from among the overall input admittance of the unit cell depicted
these currents. in Fig. 3(a), the sheet admittance of metal patch, and grounded
As a rule of thumb, for an artificial impedance dielectric substrate input admittance via transmission line
surface including an array of quasi-periodic metal patches, method (TLM). Second, investigate the scattering problem of
the maximum obtainable input impedance depends on the a periodic strip grid embedded in free space as host medium
maximum metal patch size, dielectric permittivity, dielectric and find the sheet admittance of this grid, which is a nearly
thickness, and mesh density. One can use the full-wave simple task. Next, we use the logical relation between the
simulators such as CST or HFSS and calculate the input sheet impedance of the metal patch and strip grid structures
impedance or admittance of a unit cell embedded in a periodic shown in Fig. 3(a) and (b) according to Babinet’s principle [37]
environment. In general, applying the full-wave simulators and calculate the sheet admittance of the complementary
could determine the impedance value of every general-shape counterpart of the strip grid. As a matter of fact, the input

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
2874 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO. 6, JUNE 2018

Fig. 4. Schematic of the TLM applied for extracting the input admittance Fig. 5. Schematic of a perpendicular wire grid with spacing l and d along the
of unit cell with the square metal patch. x- and y-axes, respectively. m and u denote the number of the wires. (r, θ, ϕ)
have been defined in the spherical coordinate system.

impedance of these two couple structures is convertible to We assume that the incident electric field is a plane wave
each other. Finally, using the TLM, the input impedance of with arbitrary polarization as follows:
the unit cell in Fig. 3(a) will be obtained. We explain about
these steps in detail. E inc = E0 e j k0 (xsinθcosϕ+ysinθsinϕ+zcosθ) (6)
In general, the input impedance of a periodic structure where E0 is the incident electric field at the origin and k0 being
can be expressed by equivalent circuit model. Assuming only the free space wavenumber. According to the moment method,
fundamental-mode plane waves, it is convenient to model the the unknown currents belong to two wire branches along the
unit cell using the TLM where the input admittance of the unit x- and y-directions could be expanded in term of proper basis
cell is composed of the sheet admittance, which describes the functions with unknown coefficients as follows [39]:
EM behavior of the metal patch, parallel to the admittance

M
2mπ
of the grounded dielectric substrate with a given thickness. Ixu = e j (kx x+k y du) · Am e j l x
(7a)
Fig. 4 shows the topology of the TLM to express the input m=−M
impedance or admittance of the scalar unit cell. Corresponding U
2uπ
the TLM, the input admittance can be written as I ym = e j (kx md+k y y) · Bu e j d y
(7b)
 in in   u=−U
Yx x Yx y Yx x 0 YTE 0
in Y in = + (5) where k x = k0 sinθ cosφ, k y = k0 sinθ sinφ and k0 is the
Y yx yy 0 Y yy sheet
0 YTM substrate free space wavenumber. The unknown complex coefficients
Am and Bu are the amplitude of the Floquet harmonic
where the first matrix on the right side of the equation
terms to be determined. Two parameters of d and l are the
represents the sheet admittance relevant to the metal patch
separation between x̂ and ŷ directed currents, respectively.
and the second one, which fully characterized in [38],
M and U determine the number of harmonics. According
represents the input admittance of the grounded dielectric
to [39], an extra function, f , can be added to the currents
substrate.
such that they meet Kirchhoffs law at the junction place of
It is clear that the sheet impedanceYsheet relates the average
each two orthogonal branches. This leads to better predict the
electric field tangential to the sheet Ē tan , and the surface
discontinuity of the currents at the junctions (e.g., the node
current density J¯. Therefore, it should be established an
at x = y = 0 for the center unit cell). This significantly
explicit expression for the induced current density in term
accelerates the convergence of the problem. In light of the
of overall electric field on the sheet when exposed under
above considerations, the current distributions for two orthog-
a plane wave with a given oblique incident angle. A well-
onal wire grids shown in Fig. 5 can be described based on the
known approach to determine the unknown surface currents
Floquet theorem by the following equations:
on metal strips is moment method. We will show that it is  
more convenient to the analysis of scattering characteristics 
M

of a wire grid instead of a strip grid. A perpendicular wire Ixu (x) = e j (kx x+k y ud) f 1 (x) + Am e j (2mπ/ l)x (8a)
grid possessing radii of r1 and r2 along the x- and y-axes  m=−M 
is assumed, as shown in Fig. 5. It is assumed that the radii U
j (kx ml+k y y)
I ym (y) = e − f 2 (y) + Bu e j (2uπ/d)y (8b)
of wires are very small in comparison of the wavelength and
u=−U
the lattice period. Therefore, the wires carry only the axial
 (1 − δm0 )  (1 − δu0 )
currents. The aim is to extract the sheet impedance of this Am = Am − ; Bu = Bu +
wire grid. 2π j m 2π j u

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
KARIMIPOUR AND KOMJANI: HOLOGRAPHIC-INSPIRED MULTIBEAM RA 2875

where A m and B  u are the new unknown coefficients, δi j is electric field on the wires along the x- and y-directions,
the Kronecker delta, and f is the sawtooth function defined respectively. E x 0 and E y 0 are the electric fields at x = y = 0.
for each unit cell as follows [39]: r1 and r2 show the radius of the wire grids along the x- and y-
  exp( j 2nπ x/l) directions, respectively, which are very small compared to a
f1 (x) = [U (x) − x/l − 0.5] = wavelength. For completeness, the defined quantities in (11)
2π j n
n,n =0 are read as follows:
  exp( j 2nπ y/d)
f2 (y) = [U (y) − y/d − 0.5] = . j ωμ0 2 2
 n
e− mu r1
2π j n Pm = k 0 − k 1,m (12a)
n,n =0 2k02 d mu
u=−n
(9)
j ωμ0 e− mu r1
In (9), −l/2 < x < l/2, −d/2 < y < d/2, and U Rm,u = k 1,m k 2,u (12b)
2k02l mu
is the step function with complex magnitude equal to  at
1 − δm0 j ωμ0
the cross junction of two orthogonal wires. After doing all Qm = Pm + k1,m
well-known steps for preparing matrix equation from moment j 2πm 2k02l
  n  
method, a matrix equation system will be obtained as follows 1  e− mu r1 ky
(the reader can refer to [39] for more details): × + S1,m (12c)
j d u=−n j 2π
⎡ 11   ⎤ ⎡ 11 ⎤ mu
ai j m×m ai12 [ai1
13 ]
m×1 bi1 m×1 j ωμ0 e− mu r2
⎢ a 21  ⎥ ⎢ 21
j m×u
⎥ Tu,m = k k (12d)
⎣ i j u×m ai j u×u [ai1 ]u×1 ⎦ ⎣ bi1 u×1 ⎦
22 23 2,u 1,m
2k02 d mu
a131j 1×m a132j 1×u a1133 31
b11
1×1 1×1 j ωμ0 2  n
e− mu r2
⎡ 11  ⎤ gu = k − k 2
(12e)
ci1 m×1 0 2,u
 2k02l mu
⎢ 21 ⎥ m=−n
= ⎣ ci1 u×1 ⎦ (10) δu0 − 1 j ωμ0
31
 γu = gu + k2,u
c11 1×1 j 2πu 2k02 d
  n  
where −M ≤ m ≤ M and −U ≤ u ≤ U . The indices i j at the 1  e− mu r2 kx
above coefficient matrix have been assigned to the position of × + S1,u (12f)
jl m=−n mu j 2π
each entry at each block. For example, in [ai11 j ]m×m block, i
and j start from 1 to M such that they represent the row and where S1,m , S1,u , ij , k1,i and k2, j are as follows:
column of each entry, respectively. It is worth pointing out that

n
e− mu r1 
n
e− mu r2
the wire lattice has been chosen instead of strip one at the first S1,m = , S1,u = (13a)
step because it can easily be derived the explicit expressions u mu m mu
u=−n,u =0 m=−n,m =0
for the scattered magnetic vector potential of a periodic wire 
grid, which is a required tool to be calculated for providing ij = 2 + k2 − k2
k1,i 2, j 0
matrix-vector equation system in (10). The matrix elements k1,i = k x + 2i π/l, k2, j = k y + 2 j π/d
in (10) are given as k x = k0 sin θ cos ϕ, k y = k0 sin θ sin ϕ. (13b)
j = Pi−(m+1) δi j
ai11
The parameter n should be chosen as large as possible
j = Ri−(m+1), j −(u+1)
ai12 to sure good convergence is met for each related summa-
13
ai1 = Q i−(n+1) tion. However, the summations in (12) converge rapidly by
some manipulations [39]. As observed, all entries of the
j = −Ti−(u+1), j −(m+1)
ai21
matrix coefficient in (10) can be obtained by performing
j = gi−(u+1) δi j
ai22 only several algebraic calculations, which is easily done by a
23
ai1 = γi−(u+1) simple MATLAB code thanks to matrix operation developed
in MATLAB for speed up the calculations.
a131j = j k1, j −(m+1)
The above-mentioned expressions can be used for the
a132j = j k2, j −(u+1) analysis of a periodic lattice including orthogonal strips having
33
a11 = j (1/l + 1/d) widths w1 and w2 , by employing the following well-known
substitution [39]:
11
bi1 = Ai−(m+1)
 d πw1 l πw2
21
bi1 = Bi−(u+1) r1 → sin , r2 → sin . (14)
2π 2d 2π 2l
31
b11 =
After truncating the system of equations (10) by a given
11
ci1 = δi−(m+1), 0 E xi number that guarantees the convergence, and then solving the
21
ci1 = δi−(u+1), 0 E iy finite matrix-vector equation using the inverse matrix method,
 
the unknown coefficients A , B , and  and consequently the
31
c11 = 0. (11)
current distributions, Ix , I y , will be determined through (8)
In (11), δi j is the Kronecker delta, θ E xi
= E x0 e j r1k0 cos and (9). On the other hand, the scattered electric fields along
and θ E ij = E x0 e j r 2 k 0 cos are the components of the incident the x- and y-directions, E xs and E sy , can be easily obtained

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
2876 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO. 6, JUNE 2018

through magnetic vector potential, which in turn related to

 
the current densities, in terms of A , B , and  coefficients.
As such, we can determine the sheet impedance a perpendic-
ular two-branch strip grids as follows:
   −1
Zxx Zxy E xi + E xs Ix /d
= . (15)
Z yx Z yy sheet
E iy + E sy I y /l

Note that the sheet impedance is not dependent on the

electric field amplitudes E x0 and E y0 , and without loss of
generality, one can set E x0 = E y0 = 1.
As mentioned above, the sheet impedance of a metal patch
grid, which is the complementary counterpart of a strip grid
can be obtained according to Babinet’s theory as follows [37]:
 
¯ patch   ¯ t ¯ strip ¯ −1 ¯ 0 −1
Z̄ sheet (ktan ) = (η0 /4) Ā Z̄ sheet Ā , Ā =
2
. (16)
1 0

In (16), η0 = 120π is the free space impedance and A is the

π/2 rotation matrix to convert the wave polarization. Finally,
using (5), (15), and (16), the input impedance of square
metal patch is found for any patch dimensions and incident
angles.
In order to benchmark the usefulness of the proposed
method, a λ/5 square metal patch as a constituent unit cell
of a 2-D periodic environment was simulated by the CST
software under normal and oblique incident plane waves and
the results compared with those obtained by our method.
The design parameters along with the input admittance are
shown in Fig. 6. We chose n = 8 in this example, which
is a reasonable amount to converge the summations defined
in (12a), (12c), (12e), (12f), and (13a). The yz plane has
been assumed as the incident plane, so two incident plane
waves with TE and TM modes shown in Fig. 6(b) and (c)
should be impinged on the unit cell to derive Y yy and Yx x
admittance components, respectively. For this simple symmet-
rical square patch, the acceptable results can be achieved by
only seven harmonics for describing the unknown currents.
However, if the unit cell dimension decreases lower than
λ/10 × λ/10, then the acceptable results will be obtained by
only single-harmonic currents. In this case, the calculations
considerably decrease and the final Z -matrix in method of
moments (MOM) will be reduced to a 2 × 2 matrix with
Fig. 6. Input admittances of the periodic square metal patch structure for
close form formulations [40]. Note that we do not use the (a) normal and (b) and (c) oblique incident plane waves. Design parameters
unit cells with dimensions in the order of λ/10, because the are f = 15 GHz, h 1 = 0.79 mm, p = 4 mm, h 2 = 1.58, εr1 = 3.55,
construction of an RA with this dimension involves the use εr2 = 2.94, and θ = 45°. L p is swept from 2 to 3.8 mm in case (a) and from
3.3 to 3.8 mm in cases (b) and (c).
of several thousand elements especially for large RAs, which
is a roughly impossible task to simulate RA with in hand
computers.
IV. C ONSTRUCTION OF THE
It is important to note that the Z impedance matrix dimen-
H OLOGRAPHIC R EFLECTARRAY
sions in (10) depend on the number of harmonics of the
unknown currents, i.e., Ix and I y , so enhancing the harmonic The design process of a typical HRA with linear polarization
numbers leads to improve the accuracy of the results. This becomes complete by putting the proper impedance elements
issue is depicted in Fig. 6. For instance, for M = U = on the hologram surfaces described in Section II. Several
1, 3, 6, and 7, Z -matrix in MOM is converted to 7×7, 15×15, examples are represented here to support the validity of our
27 × 27, and 31 × 31 matrices, respectively. The calculation design technique. The design frequency and the hologram
time for all cases is below of 3 s by a computer platform with dimensions are identical in all cases: f = 15 GHz and
CPU core i7 and 8 GB RAM. hologram area = 18 cm × 18 cm which contains 2025 square

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
KARIMIPOUR AND KOMJANI: HOLOGRAPHIC-INSPIRED MULTIBEAM RA 2877

unit cell with dimension 4 mm × 4 mm. The substrate used

for all structures is RT6002 with εr = 2.94, a loss tangent
of 0.0012 and 1.58 mm thickness. ATM-62-441-6 model of
centered prime focus pyramidal horn antenna with linear
polarization as a standard feeder employed for all antennas,
which is appropriate for illuminating the hologram surfaces
at Ku-band. The far-field pattern of the applied horn can
be approximated by cosq function with q = 11 led to
f /D = 1.06 to obtain maximum achievable efficiency [2].
This ratio determines the H-parameter (see Fig. 1) to be equal
to 19.08 cm. For implementing the admittance distribution of
the holograms according to (1) and (3), we chose X = 0
and M = 0.0038 to cover all obtainable admittance values of
the unit cell (see Fig. 6).

A. Simulation Results and Discussion

All input admittances required for scalar holograms
described in Section II will be implemented by locally
Fig. 7. 3-D simulation results of four linearly polarized HRAs. (a) Single
changing the dimensions of each metal patch of the scalar unit beam at θ = 45°, φ = 0°. (b) Two beams at φ1 = 0°, φ2 = 180,
cell. We know that each unit cell is located along a straight and θ1,2 = 45. (c) Four beams at (θ, φ) = (30, 0), (30, 90), (30, 180),
line with given angle, (θ, φ), with respect to the feed location. and (30, 270). (d) Nine beams at (θ, φ) = (30, 0), (30, 45), (30, 90),
(30, 135), (30, 180), (30, 225), (30, 270), (30, 315), and (0, 0). Note that
Therefore, the input admittance for each cell must be deter- a local coordinate system specified by “prime” has been used in (b) and (c)
mined by considering this angle. In general, it is a nearly diffi- along one of the main beams for better understanding two orthogonal planes
cult process to find the best values for geometrical parameters of main beams depicted in Fig. 8.
of an impedance patch to satisfy the impedance distribution of
a hologram. This issue becomes more intricate in anisotropic unbalanced power between each pattern and also lossy nature
unit cells where each element should satisfy three impedance of the hologram surface, it is observed a further reduction
quantities (Z x x , Z x y , and Z yy ) for each cell on the hologram, in the power level of multibeam HRAs. In this example,
simultaneously. For this reason, we propose a novel procedure 3.3 dB of the directivity reduction was exhibited for both two
to solve this problem. The proposed procedure is not affiliated principal planes which is a very good achievement. It is clear
with the element type possessing scalar or tensor admittance from Fig. 8(a) that the SLL is below of −26.4 and −24 dB
characteristics. for single- and two-beam HRAs, respectively. The reason
We use an evolutionary algorithm named particle swarm behind increasing the SLL for two-beam HRA may be the
optimization (PSO) to facilitate the mapping process. This amplitude error, which is occurred in the design method based
algorithm locally finds the geometrical parameters of an on superposition concept. It may be further controlled by
element (the length of the square metal patch in this case) as introducing the amplitude factor in the design process of HRA.
best as possible by considering the desired locally impedances Furthermore, Fig. 8(b) shows a very good SLL for both single-
to be satisfied and their local position with respect to the feed and two-beam patterns in the orthogonal plane that they are
which determines the incident angle. Therefore, two quantities a very good achievement as same as those antenna developed
were considered as the input parameters of the PSO: 1) inci- in [29], where the authors utilized an optimization algorithm to
dent wave angle (θinc, φinc ) and 2) input admittance value of design a multiple-beam RA with decreased SLL. The method
each unit cell to be satisfied. As such, the proper element in [29] not only involves a proper search algorithm as an
could be selected easily by performing this auxiliary tool. auxiliary tool but also gives us no physical insight of how
After putting the proper impedance patches on the hologram the antenna performs. Furthermore, to design an RA with
surfaces, whole structures were simulated by CST software. asymmetric radiation patterns, one should employ sophisti-
The 3-D simulation results of the radiation patterns for four cated global evolutionary algorithm not to be trapped in local
HRAs including one-, two-, four-, and nine-beam patterns are minimums. The flexibility of the holographic technique to
presented in Fig. 7. We can see that all 3-D radiation patterns the radiation beam angles, which leads to design asymmetric
have reasonable SLL and back-lobe level. radiation patterns, and even polarization type of the beams
Close examination of the performance of the single- and without using any optimization algorithms, are two powerful
multiple-beam HRA is done by comparing the SLL, cross-pol advantageous of this technique in comparing those method
level, and the peak gain between them. The comparison results proposed in [29].
of single- and two-beam HRA are shown in Fig. 8. The peak Another important issue associated with the results
gains for single- and two-beam patterns were calculated in Fig. 8 is the cross-pol level. The cross-pol for both single-
26.5 and 23.2 dBi, respectively. It is well known that the power and two-beam HRAs are nearly low, and a good isolation has
of each beam in single feed N-beam reflectors is diminished been observed between desired and undesired polarizations.
by 10 × log (1/N) in decibel. Because of the nonsymmetric It seems that some explanations should be represented about
impinging wave from the feed in RA which leads to dividing the cross-pol level of an HRA. For illustration purposes of

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
2878 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO. 6, JUNE 2018

Fig. 9. (a) Top view of two-beam RA hologram. (b) Side view of a portion
of the implemented hologram. The highlighted area shows the behavior of the
variations of the element lengths which follow up a sinusoidal template.

of the holographic concept. The first two factors are common

between all antennas specifically in RAs. It means that using
a feed horn with low cross-pol level and radiating elements
with low mutual coupling between each other are desired
to decrease the overall cross-pol of the antenna. About the
third factor which dedicates in holographic antennas, we can
say that the co-pol pattern is the only used quantity for
modeling the interference pattern and no information is given
about unwanted radiations or equivalently cross polarization.
Meanwhile, as seen with the conjugate beam, the interference
function was capable to predict the appearance of the conju-
gate object wave. Thus, it could once again also predict the
cross-pol radiation [41]. If the cross-pol specifications were
also used to model the interference plane, more control would
be performed on it. Another point to be mentioned is the
local periodicity approximation issue. In classical RAs with
the multibeam application, such as [29], the reflection phase
distribution becomes more complicate and has considerable
different values from one cell to another in comparison to
single-beam RAs where this discontinuity and phase wrap are
usually observed only at the edge of Fresnel zones. There-
fore, the geometrical parameters of the radiation elements
should suddenly be changed to meet this phase discontinuity.
Consequently, the local approximation used in the analysis of
the elements, which considers all elements to be identical,
is not valid any longer and includes more errors in the
multibeam case. To overcome this imperfection, one can use
smaller unit cells to more accurately sample the aperture phase
distribution by the proper elements and peruse the variations
with more gentleness. However, as we said the minimum unit
cell dimensions in classical RA designs approaches have the
limitation of meeting at least one cycle of 360° phase swing.
In HRA, we can not only diminish the unit cell dimension
below of λ/2 by an ordinary element which leads to sample
the hologram admittance distribution with high accuracy, but
Fig. 8. Comparison between the radiation patterns of single- and two- also the variation of the element dimension follow up a
beam HRA at two principal planes of the main beams. These two planes sinusoidal-type function and becomes more gentle variations
are highlighted by red (plane #1) and green (plane#2) in Fig. 7(b). (a) and (b) than classical RA. This issue is depicted in Fig. 9 for a two-
Co-pol radiated gains at plane #1 and plane #2, respectively. Note that the
pattern related to the multibeam antenna in (a) has been shifted by 45° to beam HRA.
match exactly with the pattern of single-beam HRA and make the comparison
easier. (c) and (d) Results of cross-pol radiated gains at plane #1 and plane #2, B. Bandwidth, Radiation Efficiency, and Beam Squint
respectively.
Up to this point, the advantages of the holographic theory to
the distinction between the co- and cross-polarizations, three design an efficient RA antenna with multiple-beam radiation
main factors must be considered, including the feeding type, patterns were investigated by theory and simulation tools.
the coupling between elements, and the fundamental nature The discussion about bandwidth, radiation efficiency, and

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
KARIMIPOUR AND KOMJANI: HOLOGRAPHIC-INSPIRED MULTIBEAM RA 2879

TABLE I
P EAK G AIN D ATA AND E FFICIENCY

Fig. 10. Peak gain and the aperture efficiency of two beams over the
frequency. Inset: gain difference at the main lobe of each beam.

beam squint phenomenon would be helpful to complete the

report.
As referred earlier, the main advantage of an HRA is
its frequency bandwidth. Since the subwavelength elements
have been utilized in HRA, one can expect that the band-
width enhances considerably. In order to show this interesting
feature, we investigate as a typical example the bandwidth
behavior of the two-beam HRA whose 3-D simulation results
are shown in Fig. 7(b), numerically and experimentally.
The results are shown in Fig. 10, wherein a small beam level
reduction is observed for the main beams with respect to each Fig. 11. Simulation results on plane #1 for two-beam HRA with the main
other. This is why the azimuthal radiation pattern of the feed beam directions at φ1 = 0°, φ2 = 180°, and θ1,2 = 45°. The far-field gains
have been shown for two extreme frequencies along with a sample one within
is asymmetric with respect to the hologram plane, and hence, the band ( f = 15.5 GHz).
the divided power between two beams is not equal. All simu-
lated and measured data for the peak gain of each beam are
listed in Table I (see columns 2–4). It is clear that 1 dB drop- efficiency at f = 15 GHz is about 53.09% which is much
gain bandwidth, which is a well-suited bandwidth definition higher than those obtained in [29] and [42]. The reasons
for RA with pencil beam radiation patterns, is calculated to be behind this enhancement in the whole band maybe come from
(17.5–14.17 GHz)/15 GHz × 100 = 22.2%, which is nearly being lower level of cross-pol, low loss (because of using only
14% higher than antenna proposed in [29]. This enhancement single-layer hologram), acceptable back lobe (see Fig. 8 for
definitely comes from using the subwavelength unit cells in the example), and nearly lower SLL than those structures designed
antenna design and considering all incident angle effects on in [29] and [42]. In general, multibeam antennas suffer from an
the input impedance calculations. unavoidable phenomenon named beam squint. By definition,
According to the modified definition of aperture efficiency beam squint is the main beam shifting process versus the
presented in [29], we have frequency. If the angle of the incident wave emitted from the
feed to the center of the RA aperture be equal to the angle of

N
λ2 G i the outgoing wave from the antenna, i.e., θi = θo , the beam
ηa = (17)
4π A squint will be minimized in RA. Clearly, this condition does
i=1
not meet in multibeam applications. Accordingly, it seems
where ηa is the aperture efficiency, N represents the number logical that the beam squint can limit the frequency bandwidth
of beams, G i is the gain of i th beam, λ is the wavelength, of the antenna if it has a considerable amount. In fact, the beam
and A is the aperture area. According to (17), the aperture squint will be tolerable if its value across the 1 dB drop-gain
efficiency of two-beam HRA shown in Fig. 7(b) was calculated bandwidth being lower than 3 dB beamwidth of the main
for multiple frequencies. Fig. 10 shows the aperture efficiency beams located on the principal planes. Otherwise, the beam
and the peak gain for two beams over the frequency. All data squint is a crucial limiting factor in the operating bandwidth
about the simulated efficiency are listed in Table I for the of RA.
convenience of the reader (see column 5). As can be seen, We investigated this factor for the previous example,
the aperture efficiency varies from 32% to 54% across the i.e., two-beam HRA, in order to make sure that the beam
frequency from 13 to 17.5 GHz. Specifically, the aperture squint does not deteriorate the bandwidth of this antenna. By

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
2880 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO. 6, JUNE 2018

Fig. 12. Photograph of two-beam linearly polarized HRA. Two scan planes
and their beams are clearly depicted for convenient.

TABLE II
C OMPARISON B ETWEEN O UR W ORK A CHIEVEMENTS AND O THER W ORKS

Fig. 13. Measurement results of two-beam HRA with the main beam
directions at φ1 = 0°, φ2 = 180°, and θ1,2 = 45°. (a) Plane #1. (b) Plane #2.

C. Prototype Fabrication and Measurement Results

A sample of two-beam linearly polarized HRA composed

of 2025 elements, which was designed and simulated
in Section IV-B and whose 3-D simulation results are shown
changing the frequency, the beam direction will be shifted in Fig. 7(b), was fabricated using commercial printed circuit
on plane #1 [see Fig. 7(b)] such that the direction of the board technology and tested at f = 15 GHz in an anechoic
beams move toward the normal axis of the hologram as the chamber with a far-field measurement setup. The geometrical
frequency is increased. In turn, the main beam direction has features of the antenna including the type and thickness of the
a very small variation in the orthogonal plane (plane #2). substrate layer were proposed at the beginning of Section IV.
As such, we only investigated these variations on plane #1. Again, the antenna has been designed to radiate two pencil
The −3 dB beamwidth for two main beams placed on this beams at φ1 = 0°, φ2 = 180°, and θ1,2 = 45° which causes
plane was obtained about 9.6°. This is due to the direction of the feed blockage effect not seen over the results because
two main beams varied about 9.1° across the 1 dB drop-gain the feed positioned along the center axis of the antenna
bandwidth which verified the beam squint has a little effect aperture. Fig. 12 depicts the photograph of the manufactured
on the bandwidth. (All the main beam direction variations at antenna along with the alignment tools consisting of antenna
each frequency are summarized in Table I.) Fig. 11 shows feed holder and a right-angle wedge beneath the antenna.
the variation of the main beam directions at three sample The subtended angle of the wedge is equal to 45° to align
frequencies, including extreme frequencies of the expected one of the antenna main beams with the transmitter antenna
bandwidth. However, a small enhancement in SLLs observed in the testing room. This leads to facilitate the far-filed pattern
at extreme frequencies. It is worthwhile to point out that if measurement process for two principal planes of the main
the beam squint value is smaller than the beamwidth, then beams. The co-pol measurement results for one of the main
the gain reduction at the design direction will be lower as the beams placed on φ = 0° are presented in Fig. 13 and compared
frequency is changed. with the simulation results of two principal planes. This beam

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
KARIMIPOUR AND KOMJANI: HOLOGRAPHIC-INSPIRED MULTIBEAM RA 2881

is placed along the z’-direction of the defined local coordinate [13] U. Jakobus and F. M. Landstorfer, “Improved PO-MM hybrid formula-
system. [See Fig. 7(b).] Similar results were obtained for tion for scattering from three-dimensional perfectly conducting bodies
of arbitrary shape,” IEEE Trans. Antennas Propag., vol. 43, no. 2,
another beam that does not show here to conserve space. pp. 162–169, Feb. 1995.
Some discrepancies are observed between simulation and [14] M. Karimipour and N. Komjani, “Bandwidth enhancement of electrically
measurement data specifically in the sidelobe region, which large shaped-beam reflectarray by modifying the shape and phase
distribution of reflective surface,” Int. J. Electron. Commun., vol. 70,
may be caused by fabrication error. However, quantitative no. 5, pp. 530–538, May 2016.
review of the measurement and simulation results show accept- [15] M. Karimipour, A. Pirhadi, and N. Ebrahimi, “Accurate method
able agreements between them and meet the expectations. for synthesis of shaped-beam non-uniform reflectarray antenna,”
A comprehensive comparison has been done between our work IET Microw., Antennas Propag., vol. 7, no. 15, pp. 1247–1253,
Dec. 2013.
and similar recent works and represented in Table II. One [16] M. Karimipour and A. Pirhadi, “A novel approach to synthesis of
can infer that our proposed method leads to design of a more non-uniform conformai reflectarray antennas,” Source Appl. Comput.
effective multibeam antenna than the previous works. Electromagn. Soc. J., vol. 28, no. 11, p. 1040, Nov. 2013.
[17] J. A. Zornoza, R. Leberer, J. A. Encinar, and W. Menzel,
“Folded multilayer microstrip reflectarray with shaped pattern,”
V. C ONCLUSION IEEE Trans. Antennas Propag., vol. 54, no. 2, pp. 510–518,
Feb. 2006.
A novel approach based on the holographic theory to [18] D. R. Prado, M. Arrebola, M. R. Pino, and F. Las-Heras, “Complex
design an effective multibeam RA was developed in this reflection coefficient synthesis applied to dual-polarized reflectarrays
paper. The proposed method can potentially use for several with cross-polar requirements,” IEEE Trans. Antennas Propag., vol. 63,
no. 9, pp. 3897–3907, Sep. 2015.
different applications of RA. In this method, the interference
[19] P. Nayeri, F. Yang, and A. Z. Elsherbeni, “Design of single-feed
plane or hologram records the electromagnetic characteristics reflectarray antennas with asymmetric multiple beams using the particle
of both feed pattern and desired far-field radiations. To imple- swarm optimization method,” IEEE Trans. Antennas Propag., vol. 61,
ment the hologram surface practically, we used a single- no. 9, pp. 4598–4605, Sep. 2013.
[20] O. M. Bucci, A. Capozzoli, G. D’Elia, and S. Musto, “Power
layer quasi-periodic surface as an artificial impedance surface. pattern synthesis of reflectarrays: Comparison between two approaches,”
We showed that this method allows decreasing the size of the in Proc. URSI Int. Symp. Electromagn. Theory (EMTS), May 2004,
unit cells bellow of wavelength considerably. The simulation pp. 1053–1055.
[21] M. Zhou, S. B. Sørensen, O. S. Kim, E. Jørgensen, P. Meincke, and
and experimental results showed a great enhancement in oper- O. Breinbjerg, “Direct optimization of printed reflectarrays for contoured
ational bandwidth and antenna efficiency. beam satellite antenna applications,” IEEE Trans. Antennas Propag.,
vol. 61, no. 4, pp. 1995–2004, Apr. 2013.
R EFERENCES [22] M. Zhou et al., “The generalized direct optimization technique for
printed reflectarrays,” IEEE Trans. Antennas Propag., vol. 62, no. 4,
[1] J. Shaker, M. R. Chaharmir, and J. Ethier, Reflectarray Antennas: pp. 1690–1700, Apr. 2014.
Analysis, Design, Fabrication, and Measurement. Norwood, MA, USA: [23] B. Van Ha, P. Pirinoli, M. Mussetta, and R. E. Zich, “Direct synthesis
Artech House, 2013. of dual-parameter concentric ring RA with enhanced bandwidth,” Int.
[2] J. Huang and J. A. Encinar, Reflectarray Antenna, 1st ed. Hoboken, NJ, J. Antennas Propag., vol. 2012, Jun. 2012, Art. no. 535673.
USA: Wiley, 2008. [24] J. A. Encinar, “Design of two-layer printed reflectarrays using patches
[3] J. Ethier, M. R. Chaharmir, and J. Shaker, “Reflectarray design of variable size,” IEEE Trans. Antennas Propag., vol. 49, no. 10,
comprised of sub-wavelength coupled-resonant square loop elements,” pp. 1403–1410, Oct. 2001.
Electron. Lett., vol. 47, no. 22, pp. 1215–1217, Oct. 2011. [25] J. A. Encinar, M. Arrebola, L. F. de la Fuente, and G. Toso, “A transmit-
[4] K. K. Karnati, Y. Yusuf, S. Ebadi, and X. Gong, “Theoretical analysis receive reflectarray antenna for direct broadcast satellite applications,”
on reflection properties of reflectarray unit cells using quality factors,” IEEE Trans. Antennas Propag., vol. 59, no. 9, pp. 3255–3264, Sep. 2011.
IEEE Trans. Antennas Propag., vol. 61, no. 1, pp. 201–210, Jan. 2013. [26] L. S. Ren, Y. C. Jiao, F. Li, J. J. Zhao, and G. Zhao, “A dual-layer
[5] K. K. Karnati, Y. Yusuf, S. Ebadi, and X. Gong, “Q factor analysis of T-shaped element for broadband circularly polarized reflectarray with
reflectarray elements investigating the effects from angle of incidence linearly polarized feed,” IEEE Antennas Wireless Propag. Lett., vol. 10,
using floquet modes,” IEEE Trans. Antennas Propag., vol. 62, no. 10, pp. 407–410, 2011.
pp. 5017–5028, Oct. 2014. [27] M. Niroo-Jazi, M. R. Chaharmir, J. Shaker, and A. R. Sebak, “On radi-
[6] D.-C. Chang and M.-C. Huang, “Multiple-polarization microstrip reflec- ation performances of reflectarray antennas constructed with subwave-
tarray antenna with high efficiency and low cross-polarization,” IEEE length unit cells,” IEEE Antennas Wireless Propag. Lett., vol. 14,
Trans. Antennas Propag., vol. 43, no. 8, pp. 829–834, Aug. 1995. pp. 1258–1262, 2015.
[7] D. M. Pozar, S. D. Targonski, and H. D. Syrigos, “Design of millimeter [28] L. Guo, P.-K. Tan, and T.-H. Chio, “On the use of single-layered
wave microstrip reflectarrays,” IEEE Trans. Antennas Propag., vol. 45, subwavelength rectangular patch elements for broadband folded reflec-
no. 2, pp. 287–296, Feb. 1997. tarrays,” IEEE Antennas Wireless Propag. Lett., vol. 16, pp. 424–427,
[8] A. W. Robinson, M. E. Bialkowski, and H. J. Song, “A passive reflect 2017.
array with dual?feed microstrip patch elements,” Microw. Opt. Technol. [29] P. Nayeri, F. Yang, and A. Z. Elsherbeni, “Design and experiment of
Lett., vol. 23, no. 5, pp. 295–299, Dec. 1999. a single-feed quad-beam reflectarray antenna,” IEEE Trans. Antennas
[9] P. Nayeri, A. Z. Elsherbeni, and F. Yang, “Radiation analysis approaches Propag., vol. 60, no. 2, pp. 1166–1171, Feb. 2012.
for reflectarray antennas,” IEEE Antennas Propag. Mag., vol. 55, [30] Y. Li, Q. Zhu, and R. Mo, “Studies on the holographic antenna: Theories
pp. 127–134, 2013. and experiments,” in Proc. APMC, Dec. 2011, pp. 654–657.
[10] M. Arrebola, J. A. Encinar, and M. Barba, “Multifed printed reflec- [31] Handbook of Antenna Technologies, Z. N. Chen, D. Liu, H. Nakano,
tarray with three simultaneous shaped beams for LMDS central station X. Qing, and T. Zwick, 1st ed., Springer Verlag, Singapore, 2015.
antenna,” IEEE Trans. Antennas Propag., vol. 56, no. 6, pp. 1518–1527, [32] C. Rusch, J. Schäfer, H. Gulan, P. Pahl, and T. Zwick, “Holo-
Jun. 2008. graphic mmW-antennas with TE0 and TM0 surface wave launchers
[11] H. Li, B.-Z. Wang, L. Guo, W. Shaoand, and P. Du, “A far field pattern for frequency-scanning FMCW-radars,” IEEE Trans. Antennas Propag.,
analysis technique for reflectarrays including mutual coupling between vol. 63, no. 4, pp. 1603–1613, Apr. 2015.
elements,” J. Electromagn. Waves Appl., vol. 23, no. 1, pp. 87–95, 2009. [33] S. Pandi, C. A. Balanis, and C. R. Birtcher, “Design of scalar
[12] S. R. Rengarajan, “Reciprocity considerations in microstrip reflectar- impedance holographic metasurfaces for antenna beam formation with
rays,” IEEE Antennas Wireless Propag. Lett., vol. 8, pp. 1206–1209, desired polarization,” IEEE Trans. Antennas Propag., vol. 63, no. 7,
2009. pp. 3016–3024, Jul. 2015.

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.
2882 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 66, NO. 6, JUNE 2018

[34] B. H. Fong, J. S. Colburn, J. J. Ottusch, J. L. Visher, and Majid Karimipour was born in Shahrekord, Iran,
D. F. Sievenpiper, “Scalar and tensor holographic artificial in 1986. He received the M.Sc. degree in elec-
impedance surfaces,” IEEE Trans. Antennas Propag., vol. 58, no. 10, trical engineering from Shahid Beheshti University,
pp. 3212–3221, Oct. 2010. Tehran, Iran, in 2011. He is currently pursuing the
[35] A. M. Patel and A. Grbic, “A printed leaky-wave antenna based on Ph.D. degree in communication engineering with the
a sinusoidally-modulated reactance surface,” IEEE Trans. Antennas Iran University of Science and Technology, Tehran.
Propag., vol. 59, no. 6, pp. 2087–2096, Jun. 2011. His current research interests include the develop-
[36] G. Minatti, F. Caminita, M. Casaletti, and S. Maci, “Spiral leaky-wave ment and design of reflectarray and transmitarray
antennas based on modulated surface impedance,” IEEE Trans. Antennas antennas, phased array antennas, tracking radars,
Propag., vol. 59, no. 12, pp. 4436–4444, Dec. 2011. complex electromagnetic (EM) media, optimization
[37] S. Samaddar, “Babinet’s principle for an anisotropic resistive surface algorithms, and numerical methods in EM.
using different approaches,” J. Mod. Opt., vol. 33, no. 9, pp. 1177–1192, Mr. Karimipour has been a member of the Applied Computational Electro-
Oct. 1986. magnetic Society’s Reviewer Group since 2013.
[38] S. M. Hashemi, S. A. Tretyakov, M. Soleimani, and C. R. Simovski,
“Dual-polarized angularly stable high-impedance surface,” IEEE Trans.
Antennas Propag., vol. 61, no. 8, pp. 4101–4108, Aug. 2013.
Nader Komjani received the B.Sc., M.Sc., and
[39] D. A. Hill and J. R. Wait, “Electromagnetic scattering of an arbitrary Ph.D. degrees in electrical engineering from the Iran
plane wave by two nonintersecting perpendicular wire grids,” Can. University of Science and Technology, Tehran, Iran,
J. Phys., vol. 52, no. 3, pp. 227–237, 1974. in 1988, 1991, and 2000, respectively.
[40] N. Chiotellis, and A. Grbic, “Analytical modeling of tensor metasur- He has been teaching microstrip antennas, commu-
faces,” J. Opt. Soc. Amer. B, vol. 33, no. 2, pp. A51–A60, 2016. nication systems, advanced engineering mathe-
[41] M. Chen, “Manipulating antenna radiation with angle holography,” matics, and electromagnetic courses since 1998.
M.S. thesis, Dept. Electron. Eng, Univ. Toronto, Toronto, Canada, 2015. He is currently an Associate Professor with the Iran
[42] H.-X. Xu, T. Cai, Y.-Q. Zhuang, Q. Peng, G.-M. Wang, and J.-G. Liang, University of Science and Technology. His current
“Dual-mode transmissive metasurface and its applications in multi- research interests include UWB and multiband
beam transmitarray,” IEEE Trans. Antennas Propag., vol. 65, no. 4, microstrip antennas, numerical methods in electro-
pp. 1797–1806, Apr. 2017. magnetic, phased array antennas, and passive and active microwave systems.

Authorized licensed use limited to: Indian Institute of Technology Indore. Downloaded on December 19,2024 at 10:35:59 UTC from IEEE Xplore. Restrictions apply.