Theory, Analysis, and Design

of Metasurfaces for Smart

Radio Environments
This article provides an overview of the fundamental theory behind reconfigurable
intelligent surfaces, with focus on surface-wave and nonspecular reconfigurable
reflective surfaces.
By E. M ARTINI , Senior Member IEEE, AND S. M ACI , Fellow IEEE

ABSTRACT | The term “metasurface” (MTS) denotes an artifi-

I. I N T R O D U C T I O N
cial surface constituted by a distribution of electrically small The last years have experienced a continuous grow-
elements that collectively exhibit equivalent homogeneous ing request for massive and ubiquitous multimedia
boundary conditions (BCs) to an interacting electromagnetic information access. Together with these impressive data
field. MTSs are becoming increasingly popular due to the rates, future wireless networks are expected to offer sens-
technological simplification that they offer with respect to ing, localization, low-latency, and ultrareliable communi-
volumetric metamaterials. In this article, we review the basic cations. The fifth-generation (5G) network is currently
theory behind microwave MTSs seen as reconfigurable intel- faced with the challenge of limited data speed, exacerbated
ligent surfaces (RISs), oriented to the future visionary chal- by the proliferation of billions of data-intensive appli-
lenge of a smart radio environment. To this end, two different cations. Radically new communication paradigms, espe-
typologies of MTS are reviewed: surface-wave-based MTSs and cially at the physical layer, are required to handle this
nonspecular reflective MTSs. Both types can be effectively problem. In this context, the revolutionary concept of a
characterized using simplified problems that locally match the smart radio environment (SRE) has recently emerged as
homogenized, modulated BCs. A different use of these prob- a new paradigm for wireless communications, in which
lems allows for an accurate design of radiated and scattered the environment is controlled and programed jointly with
fields. An accurate ray representation is also suggested, which the transmitters and receivers, rather than something to
allows for an effective description of the scattered field also in compensate for.
the Fresnel region and for the insertion of the MTS description The sixth generation (6G) needs a new architectural
in ray-tracing tools for network planning. Several examples platform not only for transmission/reception but also
of practical implementation are shown, and the challenges in for sensing, communication, computing, and, above all,
applying electronic reconfigurability are discussed. an environment that can be cooperative and program-
KEYWORDS | Fifth generation (5G); metasurfaces; reflective mable; this visionary concept is denoted as SRE. The key
Intelligent surfaces; sixth generation (6G); smart radio envi- technology underpinning SREs is represented by metasur-
ronment (SRE); wireless communications. faces (MTSs) of next generation, which can be denoted
with the general acronym of reconfigurable intelligent sur-
faces (RISs) [1]–[5]. MTS-based RISs consist of electrically
Manuscript received 7 October 2021; revised 14 February 2022; accepted small elements whose electromagnetic (EM) behavior is
17 April 2022. Date of publication 25 May 2022; date of current version
19 September 2022. This work was supported in part by the European Space electronically controlled to properly tailor the macroscopic
Agency (ESA-ESTEC, The Netherlands) and in part by the Italian Ministry of the EM behavior. MTSs do not require signal processing nor
Research through a Project of Relevant Italian Interest (PRIN). The work of
E. Martini was supported by the University of Siena through the funding of a signal amplification, so they have extremely low power
Curiosity Driven (F-CUR) project in the framework of the program Piano di consumption, and they can process EM signals directly
Sostegno alla Ricerca (PSR) 2021. (Corresponding author: S. Maci.)
The authors are with the Department of Information Engineering and at the speed of light, dramatically reducing latency and
Mathematics, Università di Siena, 53100 Siena, Italy (e-mail: [email protected]; complexity with respect to the corresponding fully digital
[email protected]).
solutions. Hence, an RIS can be thought of as an inexpen-
Digital Object Identifier 10.1109/JPROC.2022.3171921
sive adaptive (smart) skin, which covers parts of walls,

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1227

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

In this article, we will review models, design, analysis,

and synthesis tools for MTSs and present several examples
of RIS and surface-wave-based antennas/sensors. The
perspective of the third generation of MTSs is discussed
with an emphasis on the performance, speed, losses of
the various configurations, and applicability in beyond 5G
scenarios.

Fig. 1. Three generations of MTSs since 2000 and relevant II. S P A C E W A V E V E R S U S S U R F A C E

applications in antenna and microwaves.
WAVE INT ER ACTI O N WI TH MT S
MTSs can be categorized depending on the different inter-
buildings, and ceilings, and can modify and control the action mechanisms between the wave and the MTS, and
radio waves’ paths, and sense the environment. A promi- can be designed for three different purposes, as illustrated
nent property of RISs is, therefore, the capability of being in Fig. 2: to achieve unusual reflection/transmission prop-
reconfigured after their deployment. erties of space waves [18]–[22], to modify the dispersion
MTSs have had and are having a strong impact on properties of surface/guided waves [23]–[25], or to oper-
antenna and microwave applications [6]–[9]. In the years ate a conversion between a surface wave (SW) and a space
2000–2010, MTSs were essentially uniform in space and wave (or vice versa) [26]–[29].
realized by periodic arrangements of printed elements. The first category is sometimes denoted as gradient
This was the first generation of MTSs. Several new MTS and can serve for realizing nonspecular reflective
applications in microwaves and antennas were born in surfaces [see Fig. 2(a)] or metadeflectors [see Fig. 2(b)].
that period; examples are those listed in Fig. 1. In the In this type of MTS, the output beam can be readdressed
second generation (2010–2020), MTSs were designed by changing the linear phase introduced by the MTS
in such a way to change boundary conditions (BCs) in while maintaining low levels of specular reflection or
space to control the field launched by an in-plane or direct transmission contributions. The cases illustrated in
external feed. Today, leveraging on the developments Fig. 2(c) and (d) are denoted as MTS lenses since they
of the second generation, we are facing a transition to
a third generation, where MTSs change BCs in space
and time, becoming controllable and intelligent. MTS
reconfigurability can be achieved by using electronics,
time-changing materials, or multiple switchable feed
points distributed over the MTS. The possibility to
independently control the individual elements gave rise
to the concept of “digital-coding MTSs” [10], [11], which
evolved into “programmable MTSs” after introducing a
field-programmable gate array to control different MTS
functionalities based on different digital states of the
elements [12], [13]. This concept can be further extended
to self-adaptive and cognitive MTSs [15].
Reconfigurability makes the MTS amenable to a mas-
sive MIMO and capable to reroute signals in nonspecular
directions. Therefore, reconfigurable MTSs become a key
enabling technology for the future generation of sensing
and communications, as the basic constituent of a fully
intelligent and cooperative wireless environment.
We should remark that an RIS can be also realized
with the well-established technology of reflectarrays [17].
Reflectarrays differ from MTSs essentially because the
printed scatterers are resonant and, therefore, comparable
with half the characteristic wavelength of the substrate
where they are printed. In MTSs, the elements are small
with respect to this wavelength, and therefore, they can
be globally modeled as a continuous impedance boundary.
This simple difference makes MTSs more capable than Fig. 2. Different types of MTS-based RIS. (a) and (b) Phase
gradient RIS. (a) Nonspecular reflective surface. (b) MTS deflector.
reflectarrays in manipulating the EM field even though
Metalenses with (c) broadside and (d) tilted beam and relevant
more challenging to be reconfigured at a low cost due to impedance holographies. (e) and (f) Surface-wave-based MTS with
the complexity of the control network. (e) broadside and (f) tilted beam.

1228 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

to different implementation technologies and adequately

simple to allow for the analysis of electrically large struc-
tures. A key point should be the capability to embed
the RIS EM description in propagation models for overall
system performance estimates. The model based on the
concept of equivalent impedance is well established for
static MTSs and can be generalized to reconfigurable ones;
it is presented in the following with an emphasis on the
typologies (a), (e), and (f) in Fig. 2.
Fig. 3. Layout of the impedance modulation for an MTS sensor,
which allows for SW coupling in receiving mode. Variation of the A. Equivalent Homogenized Boundary Conditions
(a) elevation angle and (b) azimuth angle. Thanks to the subwavelength size of the constituent
unit cells, MTSs can be conveniently described in terms
of homogenized BCs of impedance type, relating the tan-
perform the same function as optical lenses. The modu- gential components of the average electric and magnetic
lation of the MTS impedance changes from cylindrical to fields. This description allows one to disregard the elec-
elliptical. The cases in Fig. 2(e) and (f) represent SW-based trically small details of the single elements, thus signif-
antennas, where the feed is an in-plane monopole and icantly reducing the computational burden and avoiding
the exciting wave is an SW with a cylindrical wavefront. ill-conditioning problems in the numerical analysis.
The challenging problem in cases (a)–(d) is the cancella- For a modulated MTS, the definition of the effective BCs
tion of the specular reflections or direct transmission; in at each unit cell relies on a local periodicity assumption,
(e) and (f), the difficulty is, instead, the effective excitation i.e., it is derived by assuming the unit cell immersed
of the SW, along with the control of the leakage atten- in a microperiodic environment (see Fig. 4). With this
uation constant along the SW path. Indeed, the periodic terminology, we will refer to the subwavelength period of
modulation of the BCs transforms the SW launched by the the small elements, in contrast with the macroperiod of
feed into a curvilinear wavefront leaky wave (LW), thus the global modulation obtained by changing the printed
generating a radiating aperture with an amplitude control element dimension (see Figs. 5 and 6).
obtained through a proper design of the 2-D distribution For a generic MTS backed by a ground plane
of the attenuation constant. This results in a lightweight [see Fig. 4(b)], the BCs can be expressed in terms of
and low-profile structure, characterized by low losses and an “impenetrable” equivalent tensor impedance Z, which
simple low-cost manufacturing. Furthermore, by acting on relates the average tangential electric and magnetic fields
the BCs pattern, it is possible to obtain a unique control on the top interface of the MTS
also of the phase and polarization of the aperture field,
thus molding the radiation pattern.  
Et z = 0+ = Z · ẑ × Ht z = 0+ . (1)
We note that all the above configurations, and not only
(a) and (b), can be of use in SREs. In fact, configurations
(c) and (d) can be used as relay and (e) and (f) for From here on, bold characters will denote vectors, and bold
sensing the environment. In particular, the SW-based MTS characters underlined by double bars will indicate tensors.
in receiving mode [i.e., the reciprocal configuration of The tensor nature of the equivalent impedance accounts
Fig. 2(e) and (f)] can detect the direction of arrival of a for the possible coupling between different polarizations.
signal, as shown in Fig. 3. A variation of the BCs can gener- When the wavevector of the interacting wave lies in a
ate a different coupling with an inward traveling SW with
different amounts of power received at the port. The direc-
tion of arrival is detected when the maximum coupling
to SW is obtained at the output port. This phenomenon
is simple and extremely selective in angle for electrically
large MTSs. The various colored maps in Fig. 3 give an
idea of how MTS impedance modulation should change for
maximizing the coupling with a plane wave coming from
different elevation (a) and azimuth (b) angles.

III. M E T A S U R F A C E M O D E L S A N D
CANONICAL PROBLEMS
The effective design of RIS devices requires the use of
realistic models that are, at the same time, accurate
enough to describe practical characteristics (losses, fre- Fig. 4. Geometry for the periodic MTS and corresponding
quency dependency, MTS discretization, coupling among equivalent transmission line model in homogenization assumption.
elements, and so on), sufficiently general to be applicable (a) Opaque impedance model. (b) Penetrable impedance model.

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1229

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

from the equivalent transmission line model as

 −1  −1  −1

Z ee Z eh Zsee Zseh e
Zsc 0
= +
Z he Z hh Zshe Zshh 0 h
Zsc
(3)

e/h
where Zsc is the TM/TE reactance of the grounded slab.
We note that Zs possesses the same property of Z in the
absence of losses.
The penetrable impedance model can be generalized to
a generic multilayer environment and applied also to MTSs
operating in transmission. In the presence of multiple
printed layers, interlayer interactions can be accounted for
Fig. 5. Local canonical macroperiodic homogenized problems for
by means of a multiport network based on the concept of
(a) space-wave illumination or (b) surface-wave illumination. “accessible modes” [30]. This model neglects ports associ-
ated with MTS-layer Floquet’s modes that are sufficiently
symmetry plane of the MTS, there is no coupling, and attenuated at contiguous layers.
scalar equivalent impedances equal to the eigenvalues of Furthermore, it is noted that the penetrable impedance,
the tensors can be used for each polarization. which only represents the contribution of the patterned
For static MTSs operating in the microwave and metallic layer, is only weakly dependent on the transverse
millimeter-wave ranges, losses can normally be neglected, wavenumber of the interacting field, as opposed to the
and the equivalent impedance tensor Z is anti-Hermitian impenetrable impedance, which models the whole finite
(namely, with purely reactive eigenvalues). For reconfig- thickness MTS. For this reason, the first model offers a
urable MTSs, however, losses are in general higher, and significantly higher accuracy whenever the same BC has
the tensor matrix loses the anti-Hermitian properties. to be used for different wavenumbers [31], [32].
In the common case in which the MTS is realized by
printing metallic patches on a grounded slab, a more B. Infinite Canonical Problem: Periodic
accurate model can be defined relying on an equivalent Homogenized MTSs
transmission line description. According to this model,
a penetrable impedance BC is introduced, which relates Despite the apparent difference among them, various
the tangential electric field to the discontinuity of the applications of spatially modulated MTSs share the same
average tangential magnetic field across the metallization objective: tailoring the space spectrum of the excitation
as follows: field, being it either a space wave or a surface wave. More
specifically, two fundamental applications for SRE aim at
   generating fields with a wavenumber different from the
Et (z = 0) = Zs ẑ × Ht z = 0+ − Ht z = 0− . (2)
one of the exciting wave. The first one is the case of
anomalous reflection (case (a) in Fig. 2), where, given an
The relationship between Z and Zs expressed in the TM/TE incident space wave, we desire to generate a nonspecular
reference system of the interacting wave can be derived reflected field while canceling the specular reflection. The
second application is relevant to antennas and sensors
based on SWs, where either the incident (cases (e) and (f)
in Fig. 2) or the generated surface wave (see Fig. 3)
propagates with phase velocity less than the speed of light.
In both these cases, the objective is reached by introducing
a (local) macroperiodicity in the homogenized MTS so
that different space harmonics, or curvilinear-wavefront
Floquet waves (FWs), can be generated. It is, therefore,
important to formulate in a parametric way the analysis of
the interaction of an EM plane wave with a periodically
modulated unidirectional-gradient homogenized surface
impedance, which matches the local BCs even in cases
where the modulation is curvilinear (see Fig. 5).
Fig. 6. Geometry for (a) nonhomogenized and (b) homogenized
This unidirectional-gradient, macroperiodic impedance
canonical problems of a periodically modulated MTS with
macroperiod d. In the inset on the right: printed elements and
problem represents, therefore, the “canonical” problem for
microperiod d0 (red arrow denotes SW, and blue arrow denotes the the MTS design. (Note that we use here the term “canon-
homogeneous incident plane wave). ical” even if the solution of the problem is not analytical

1230 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

due to its numerically simple formulation and solution.) is a constant representing the complex amplitude of the
We stress again that we will refer to the periodicity of incident field. Both sides of (7) are periodic functions of
the impedance profile as a macroperiodicity d, as opposed the spatial coordinate y ; therefore, they can be expanded
to the microperiodicity d0 related to the subwavelength in the Fourier series. After equating the corresponding
unit cell used for the MTS implementation (see Fig. 6). coefficients of the series, for any n-index, we get
An approach for a rigorous solution of the problem, which

can be seen as a generalization of the approach proposed inc
Im Zn−m = GEJ (kyn ) In + Ey0 δn,0 (8)
in [33], is described in the following. m
Consider the geometry depicted in Fig. 6, where a

periodically modulated MTS is excited by an EM plane where m Im Zn−m = [{Z̃} ⊗ {I}] ˜ n is the nth term of the
wave field characterized by a transverse wavenumber ky0 . convolution between the impedance and current expan-
For the sake of simplicity, we will consider here a scalar sion coefficients, and δi,j is the Kronecker delta. For all n,
impedance and a TM-polarized incident field, propagating (8) can be written in the form of an infinite linear system
along a direction orthogonal to the periodicity variation; in the unknowns Im
however, the procedure illustrated here can be readily
generalized to the case of tensor impedance, arbitrary 
inc
polarization, and skew incidence [34]. ζn,m Im = Ey0 δn,0 (9)
m
This problem can be rigorously solved through a mode-
matching method using FWs as modal functions. To this
with
end, the Fourier series of the impedance profile and the
FW expansion of the currents induced in the equivalent
impedance surface are introduced ζn,m = Zn−m − GEJ (kyn ) δn,m . (10)

 2πm y
Z (y) = Zm e−j d (4) In numerical evaluation, the Fourier expansions can be
m truncated to a finite number of terms (generally conver-

I (y) ŷ = ẑ × Ht = In e−jkyn y ŷ (5) gence is reached with few tens of terms). This leads to a
n finite linear system, whose solution provides the current
coefficients Im and, hence, the scattered fields.
where kyn are the FW wavenumbers given by Before proceeding further, we observe that the approach
described here is very different from a localized physical
2π optics (PO)-like approach, as the one used for reflectarray
kyn = ky0 + n (6)
d analysis [17] or in the so-called generalized reflection law
[35]. The latter assumes that the field reflected at any
and d is the macroperiodicity. Notice that the currents point can be approximated with the one reflected by the
in (5) may represent either the currents flowing in the infinite uniform surface, which locally approximates the
metallic cladding or currents ideally flowing in an equiva- modulated MTS [36], and therefore, it fails to correctly
lent impenetrable impedance depending on the impedance account for the macromodulation and the MTS spatial
model adopted. dispersion. The approach indicated here, instead, accounts
The following step consists of the imposition of the BCs; for these effects, and therefore, it provides much more
the specific equations are slightly different depending on accurate results [37]. This aspect is quantitatively inves-
the type of problem, as illustrated in the following. tigated in Section V.

2) SW-Coupling Model: When considering the

1) Scattering Problem: In a scattering problem, the
SW-coupling model [see Fig. 5(b)], the objective is
transverse wavenumber ky0 is set by the incident field,
to find the modes supported by the periodic structure
and the objective is the control of the intensity of the
in transverse resonance conditions and control the
FWs, at least the one falling in the visible region. After
relevant complex propagation constant. The problem is,
introducing Green’s function to represent the scattered
therefore, formulated as an eigenvalue problem, where
field, the equation imposing the BCs can be written as
the transverse wavenumber ky0 is an unknown of the
 2π(m+q)  problem, and it is, in general, complex to account for the
2πn y
Zm Iq e−j d
y
= GEJ (kyn ) In e−j d
inc
+ Ey0 leakage of the −1 indexed (leaky)-mode. This represents
m,q n the main difference with the scattering problem, where
(7) ky0 is imposed by the incident wave. The equation to be
satisfied in the SW coupling is
where a common dependency exp(−jky0 y) has been sup-
 2π(m+q)  2πn y
pressed, GEJ (.) is the spectral Green’s function relating Zm Iq e−j d
y
= GEJ (kyn ) In e−j d (11)
inc
the electric current to the tangential electric field, and Ey0 m,q n

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1231

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

which, similar to what has been shown in Section III-B1,

can be cast in the form of the following homogeneous
linear system:


ζn,m Im = 0 (12)
m

where the matrix ζn,m is still given by (10). The complex

wavenumber of the supported 0-indexed mode is then
determined by setting to zero the determinant of the
system matrix [32]. The relevant current distribution is
given by the eigenvector corresponding to the zero eigen-
value. This is a fundamental piece of information, e.g., for
studying field polarization in the case of anisotropic MTSs.
Fig. 7. Geometry of the polygonal contoured RIS and cone of
IV. M T S A N A L Y S I S singularities in the far-field associated with the mn-indexed FW

A. Scattering From a Finite Polygonal, estimated by the infinite canonical problem.

Nonspecular Reflecting RIS

An MTS designed to redirect an impinging beam in

one or more predefined directions will consist of a finite
repetition of the macroperiods analyzed in Section III-B. definition. The field Ef ar at a distance r > 2D2 /λ, where
The equivalent current distribution is, therefore, obtained D is the maximum dimension of the RIS and λ is the free
by arranging the FW-based current modes over the aper- space wavelength, is given by
ture and formulating the radiation integral of it over the  
N
truncated structure. e−jkr  j Knm · r(i)  (i)
Ef ar = e v
Dv,nm · Einc (15)
For the sake of simplicity, we refer to a planar aperture r i=1 n,m
with arbitrary polygonal contour, illuminated by a plane 
wave propagating in the direction e(i) × e(i−1) · n̂
(i)
Dv,nm = −GE,J (knm ) · Ynm
Knm · e(i) Knm · e(i−1)

1 2 2
 (16)
k̂inc = kx0 x̂ + ky0 ŷ + ẑ k2 − kx0 − ky0 (13)
k
where n̂ is the normal to the RIS surface,
where k is the free space wavenumber, in which the electric
currents are given by a superposition of plane waves com-
ing from the solution of the problem in (9), with current Knm = (knm − kr̂) (17)
amplitudes Inm,x and Inm,y along directions x and y , for
the FW contributions with transverse-to-z wavenumbers and GE,J (k) is the pertinent spectral dyadic Green’s
function.
knm = kxn x̂+kym ŷ Equation (15) represents the summation of field
  contributions radiated by the FW-induced current sheets
2π 2π
= kx0 + n x̂ + ky0 + m ŷ. (14) associated with the canonical infinite problem, windowed
dx dy
on the RIS aperture. It is represented by diffraction contri-
butions from the vertexes of the polygonal rim. Each vertex
The FW-current coefficients Inm,x , Inm,y are proportional (i)
diffraction coefficient is a summation of contributions Dnm
to the electric field Einc of the incident plane wave and,
associated with the nm-indexed FW-induced current. These
therefore, can be rewritten as Inm = (Inm,x x̂ + Inm,y ŷ) =
coefficients exhibit singularities in direction r̂ located on
Ymn · Einc .
the surface of cones with axes on the two edges conflu-
1) Far-Field: Following the procedure in [38] and [39], ent to the corners and semiaperture equal to the angles
one can find a form of the far-field pattern in the direction between the two edges and the vectors Knm = (knm − kr̂).
r̂ = ux̂+v ŷ that eventually only depends on the position of We observe that the singularities do not exist for all the
the vertexes. To this end, let us assume that the ith vertex indexes nm, but only for those nm-indexed FWs whose
of the polygonal RIS is in position r(i) v at the confluence phase velocity along the edge is faster than the speed of
of two edges aligned along e(i−1) , e(i) , respectively, both light (normally only few of them). The singularities of
of them with amplitude equal to the side of the polygon the vertex diffraction coefficients of each couple of con-
(see Fig. 7). The index i = 1, . . . , N denotes both edges tiguous vertexes compensate for each other, thus giving a
and vertexes, with r(i)
v = e
(i+1)
− e(i) and e(N+1) = e(1) by well-behaved field everywhere. The numerical impairment

1232 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

associated with the cancellation of singularities can be

overcome by an a priori summation of contributions in the
critical directions. For instance, the nm-indexed diffraction
coefficient at edge i can be combined with the one at edge
i + 1 giving a well-behaved field everywhere in space [38],
[39]. For rectangular RISs, the combination of the infinities
recovers the expression given in [38].

2) Ray Description in the Fresnel Region: For observations

in the Fresnel region (r < 2D2 /λ), and far from the
reactive region, the field is obtained by summation of
contributions of the direct “visible” FWs (namely, those
whose wavenumbers are in the visible spectral region) plus
FW-induced edge and vertex diffracted ray contributions
from the contour, i.e., Fig. 8. Ray contributions for the scattering by a nonspecular
reflective RIS in the Fresnel region. The existence region of FW
contribution and the shadow boundary cones of edge i are also
E (r) indicated.
N (i)
 e −jk|r−rv | 
(i) (i)
∼ Dv,nm Tv,nm · Einc diffraction point r(i)
e,nm slides out from the edge at the
i=1 k|r − r(i)
v | n,m
  vertexes i or i + 1, namely, when the observation point
N (i)
  e−jk|r−re,nm | crosses the shadow-boundary cones (SBCs) depicted in
(i)
+ 
Fe,nm D(i) U (i) 
e,nm e,nm
· Einc (i)
Fig. 8. The transition functions Fv,nm (i)
and Tv,nm compen-
(i)
i=1 n,m k|r − re,nm | sate for the discontinuities of the FW and FW-induced edge
 
 (i) diffracted field contributions, respectively, appropriately
−jk|r−rnm |
+ GE,J (knm ) · Ynm Unm e · Einc
changing their spreading factor and leading to a uniform
n,m
(18) field everywhere in space.

(i)
where Dv.nm and D(i) are vertex and edge diffraction B. Analysis of SW-Based MTS Sensors and
e,nm
coefficients, respectively, associated with each FW, and Antennas
(i) (i)
Fv,nm and Tv,nm are transition functions based on Fres- In the case of SW-based MTS antennas and sensors,
nel functions [41], [42] and generalized Fresnel func- the equivalent impedance is generally modulated along
tions [44]–[49]. Equation (18) casts the scattering from the radial direction. It is convenient to vary the mod-
the RIS in a ray-type format, which is well suited to be ulation parameters across the aperture to improve the
introduced in a conventional ray-tracer for the network performance. This implies that the overall structure is
planning of a smart environment. We note that, although not rigorously periodic but only “locally periodic” [see
the summation in (18) can be extended to all the FW asso- Fig. 5(a)]. Accordingly, the analysis of these structures
ciated with the infinite problem, only the ones in the visible can be performed by using an adiabatic extension of the
region can be included in the approximation since the rigorous FW expansion introduced in Section III-B, which
others are attenuated exponentially in the direction normal is referred to the 1-D problem which locally matches the
to the surface and also produce negligible diffraction effect 2-D structure. This approach, named the flat optics (FO)
at the edges. method, is described in [51], where its accuracy in the
The last term in (18) represents the direct FW ray fields representation of the local currents flowing in the MTS is
emanating from the “nonspecular” reflection points rnm demonstrated.
inside the surface. In a “perfect” design of the RIS, only We assume that there are no losses in the grounded
the FW associated with the anomalous reflection has to be dielectric characterized by constant relative permittivity
considered since all the Inm coefficients are zero except the εr and thickness h, the elements are also lossless, and
one associated with the nonspecular desired direction. the available circular aperture has radius a. The MTS top
This contribution exists only when the observation point film constituted by the subwavelength metallic elements
is inside a certain region of space delimited by the exis- is modeled as a penetrable BC in the transverse transmis-
tence function Unm , which is the projection of the RIS sion line model [see Fig. 4(b)] and is represented by a
polygon in the direction of propagation of the nmth FW. modulated equivalent capacitive reactance that is locally
The FW-induced edge diffracted rays emanate from edge periodic, as shown in Fig. 5(b).
diffraction points r(i)e,nm whose position can be found It is convenient to adopt a cylindrical coordinate refer-
through the generalized Fermat principle. Their existence ence system (ρ, φ) with unit vectors (ρ̂, ϕ̂) [see Fig. 5(b)],
(i)
field is regulated by the unit-step function Ue,nm that whose origin is centered at the feeder position, usually
intervenes forcing to zero the diffracted ray when the edge in the geometrical center of the aperture. The metallic

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1233

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

film top covering the MTS imposes the BC in (2) with

a homogenized impedance tensor Zs = j X; the currents
flowing into it are denoted by J. The tensor X is expressed
as

X = ρ̂ρ̂Xρρ + (ρ̂ϕ̂ + ϕ̂ρ̂) Xρϕ + ϕ̂ϕ̂Xϕϕ (19)

Xρρ = X̄ρ [1 + mρ (ρ) cos (Ks (ρ) + Φρ (ρ))] (20)
Xρϕ = X̄ρ mϕ (ρ) cos (Ks (ρ) + Φϕ (ρ)) (21)
Xϕϕ = X̄ϕ [1 − mρ (ρ) cos (Ks (ρ) + Φρ (ρ))] (22)

where the components of X feature a constant average

value on the aperture. The average reactance in front of the
square parentheses is capacitive and homogeneous on the
aperture, and K is a large constant that implies the condi-
tion on the transverse gradients K|∇t s|  |∇t Φρ,ϕ |. Note
that the function s is slowly variant w.r.t. the wavelength.
The different modulation parameters in (19)–(22) are
Fig. 9. Flowchart of the synthesis procedure for reflective MTSs.
employed and tailored to control the SW interaction with
the surface; in particular, it can be noticed that the factor
KS (ρ) represents a quickly varying phase contribution that
ing for the LW propagation and its radial power leakage,
determines the phase of the −1 indexed FW, whereas the
which reads k(0) = βsw + βΔ − jα. The wave attenuation
term Φρ,ϕ (ρ) has much slower variation and controls the
parameter α = α(ρ) accounts for the SW-to-LW power
aperture field polarization; finally, the modulation index
transfer due to progressive radiation, while the field travels
mϕ,ρ determines the leakage rate of the radiating mode
toward the structure edge, βΔ = βΔ (ρ), accounts for a
and has a major impact on the aperture illumination and
small deviation in the dominant mode wavenumber due
efficiency. It is important to note that all these parameters
to the local reactance. Notice that both these two terms
enable an independent control on the desired aperture
directly depend on the local modulation tensor X. It is also
field.
important to notice that k(0) is a spatially variable complex
The reactance modulation functions that we are dealing
wavenumber, and the global phase of the dominant mode
with are characterized by a constant average value X̂ρ ; this
can then be retrieved by integrating k(0) . The adiabatic
ensures that the average current flowing in the coating
Floquet modes (AFMs) for the currents flowing into the
can be treated as a perturbation of the SW current that
coating can, thus, be obtained using local modes
would be excited by the source in a nonmodulated surface
characterized by its average. It is, thus, introduced a proper   ρ
(2)
modification of the SW current to represent the zeroth J(n) = Jρ(n) ρ̂ + Jϕ(n) φ̂ e−jnKs(ρ) H1 k(0) dρ .
indexed term of the adiabatic Floquet modal expansion. 0
(24)
On an inductive nonmodulated “opaque” surface, the SW
current can be expressed as
By considering the asymptotic form of the Hankel function,
(2) it is clear that any nth mode in (24) is characterized by
J0 = J0 H1 (βsw ρ) ρ̂ (23)
a curvilinear wavefront that propagates with the local n-
indexed Floquet mode wavevector
where H1(2) is the Hankel function of second kind and first
order. In (23), βsw is the wavenumber of the characteristic  ρ 
mode determined by solving a dispersion equation relevant β (n) = Re∇t k(0) dρ + nKs
0
to the isotropic constant penetrable coating having an = (β sw + β Δ ) ρ̂ + nK∇t s. (25)
impedance equal to X̄ρ ; this can be expressed as βsw =
2
k(1 + X̄op /ζ 2 )1/2 , where X̄op is the opaque reactance at the
resonance, namely, the ratio between total magnetic and This leaks energy at the rate dictated by α(ρ).
electric fields seen at the top interface, and k and ζ are
the free space wavenumber and impedance, respectively. V. M T S S Y N T H E S I S
A closed-form approximation of the resonant reactance is A. Synthesis of Nonspecular Reflectors
available in [50]. When a reactance is added to the average The synthesis procedure for nonspecular reflectors is
reactance, the dominant zeroth Floquet mode is derived shown in Fig. 9. It starts from the coverage requirements,
from the vectoral form of (12) by a local transformation of which dictate the modulation direction and period, and
the unperturbed βsw into a complex wavenumber account- the desired scattered beam amplitudes. The approach

1234 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

field polarization, as explained in [51]. Starting from

these distributions and from the knowledge of the average
impedance, it is possible to define at each aperture point
the local impedance modulation able to excite the LW with
the desired characteristics. Once the impedance profile
is known, the last step consists of its practical imple-
mentation through microperiod printed elements (see
Section V-C).

C. MTS Implementation Through Printed Patches

In patch-based MTSs, modulation is realized by gradu-
ally changing the size, shape, or orientation of the patches
in adjacent unit cells. A local “micro”periodicity approxi-
mation is used for the practical MTS implementation. More
Fig. 10. Flowchart of the synthesis procedure for SW-based MTS specifically, different unit cells are studied in a periodic
antennas and sensors. environment to construct a database relating the equiva-
lent impedance to one or more parameters characterizing
the unit cell. For static MTSs, geometrical parameters with
presented in Section IV-A represents an efficient tool for a continuous variation can be used. For reconfigurable
the accurate evaluation of the scattered field for a generic MTS, a given geometry is normally analyzed for differ-
piecewise polygonal periodic surface and an arbitrary ent status of a tunable material or active component(s)
observation domain (either in the Fresnel or far-field embedded in the unit cell. This provides a discrete set
zone), and therefore, it can be effectively embedded in a of impedance values available for the synthesis of the
synthesis procedure for the determination of the required desired profile. In both cases, the basic tool for the MTS
FW coefficients for each polygonal patch. implementation is represented by the (continuous or dis-
Once the desired FW coefficients are known, one can crete) maps relating some unit cell parameters to the
proceed to the design of the impedance profile using corresponding impedance values. The desired impedance
an iterative procedure based on the model proposed in profile is sampled at the center of the unit cells, and
Section III-B. This step must be done having in mind the for each unit cell, the element that best fits the desired
structure for the MTS practical implementation. In fact, impedance value is selected. For anisotropic metasurfaces
depending on the available status for the single MTS ele- described by a tensor impedance, maps are generated for
ments, this profile will be continuous or step-like. Typically, all the tensor entries, and the synthesis is performed based
continuous profiles can be used for static MTSs or MTSs on a least-squares minimization involving all the entries.
with continuously tunable elements (e.g., varactors), while
step-like profiles are the only option for reconfigurable VI. N U M E R I C A L R E S U L T S F O R
elements based on switches (e.g., diodes). Finally, the ANOMALOUS REFLECTION
resulting impedance profile will be implemented through A. Far-Field Example
practical elements, as explained in Section V-C. This section presents some numerical results illustrating
the improved accuracy of the analysis approach introduced
B. Synthesis of MTS Antennas and Sensors in Section III with respect to a local model of the MTS.
The synthesis of MTS antennas and sensors follows According to the local approach, the impedance profile of
the steps described in Fig. 10. For the sake of simplic- an anomalous reflector is designed to provide a reflection
ity, we refer here to the case of a transmitting antenna, phase linearly varying to compensate for the phase mis-
implying that the same procedure can be also applied to match between the incident and reflected waves, assuming
receiving antennas and sensors. that the local reflection coefficient is the same as the cor-
For a transmitting antenna, the requirements are nor- responding homogeneous MTS. This leads to the following
mally given for the radiation pattern, which is associated impenetrable impedance expression for incidence in the yz
with a certain aperture distribution. Several techniques plane:
are available in the literature to make this step, and we  
will not describe them here. The successive step con- ky
Z (y) = jZ0 cot (sin θi − sin θr ) (26)
sists of the identification of the radiating field with the 2
−1 indexed term of the generalized FW expansion asso-
ciated with the periodic modulation of the MTS. Based on where Z0 is the wave impedance of the impinging
this identification, it is possible to translate the aperture wave, while θi and θr are the incidence and reflection
field distribution into the corresponding distributions of angles, respectively. According to the local description, this
the LW attenuation constant, propagation constant, and impedance profile provides a single beam in the desired

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1235

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

As a matter of fact, it has been shown in [53]–[55] that

perfect nonspecular reflection (with no other FW modes
excited) without cross-polarization can only exist in the
case of retroreflection, i.e., wave reflected toward the inci-
dence direction. On the other hand, it is possible to define
a continuous tensor impedance profile performing perfect
reflection along with polarization conversion for any arbi-
trary couple of incidence and reflection angles [54], [55].
In particular, Yepes et al. [55] provide the expression of
the penetrable impedance over a grounded slab exhibiting
such behavior, which can be readily implemented through
printed elements.
An example corresponding to normal TE-polarized inci-
dence and TM-polarized reflection at −40◦ is illustrated
in Fig. 12. The dielectric slab and the design frequency
are the same as in the previous example, and the period
is discretized into eight unit cells, each containing a
double-anchor-shaped metallic element. As can be seen,
the specular reflection (θ = 0◦ ) is very weak. The residual
lobe at +40◦ , corresponding to the +1 FW, is due to the
impedance profile discretization and to some approxima-
Fig. 11. Far-field scattered by an MTS characterized by the tions in the reconstruction of the tensor impedance entries
impedance modulation in (26) implemented through a penetrable profiles.
impedance on a grounded slab: (a) local approximation and
(b) comparison between the model of Section III-B and full-wave
simulations associated with the elements in the bottom-right inset.
B. Fresnel Region Ray-Field Example
Representing the ray field in the Fresnel zone is an
important issue in smart environment since, not always,
the observer is in the far zone of the RIS. In order to show
direction, corresponding to the −1-indexed FW, as shown the accuracy of the ray description in the Fresnel region,
in Fig. 11(a), for the case of incidence angle θi = 0◦ and the formulation in (18) has been applied to a hexagonal
reflection angle θr = 60◦ . array of small dipoles in the free space illuminated by
The results are relevant to the far-field scattered in a plane wave. The electric dipoles are oriented along x
the incidence plane for a TE-polarized wave impinging and have a periodicity of 0.6λ along two axes squinted of
on a 20-period-long MTS, assuming a windowing of the 60◦ . The incident plane wave is coming from the direction
currents over a rectangular area. We have then analyzed √
k̂inc = ((1/2)ŷ + ẑ)( 3/2) with an electric field polarized
this scattering problem with the procedure presented in along x. For simplicity, the currents induced on the dipoles
Section III-B. The impedance profile in (26) has been
obtained through a modulated penetrable impedance over
a grounded dielectric slab of 1.524-mm thickness and
a relative permittivity of 9.8 at the operative frequency
of 10 GHz. The resulting scattered far-field, as reported
in Fig. 11, exhibits two undesired lobes, corresponding to
the 0-indexed (specular reflection) and +1-indexed FWs,
thus revealing the limits of the local design approach.
To verify the accuracy of the analysis in the description
of real problems, the penetrable impedance has been syn-
thetized through 12 rectangular metallic patches, as shown
in the inset of Fig. 11. The constituent unit cells have the
size of 2.9 mm along the periodicity direction and 11.2 mm
in the orthogonal direction. The width of all the rectangu-
lar patches is equal to 2 mm, while the length is varied
to reproduce the desired impedance profile. The resulting
structure has been analyzed with the commercial full-
wave solver CST. The results are compared in Fig. 11(b).
Fig. 12. Full-wave results for the far-field scattered by an MTS
As can be seen, the actual behavior of the structure is well performing anomalous reflection at −40◦ for normal incidence
predicted by the proposed model, as opposed to the local designed according to the theory presented in [55] and
description. implemented through metallic patches.

1236 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

continuous BCs [61], and by a fast multiple method (FMM

analysis) [63], which accounts for the actual pixelated
layout. Fig. 14(a) and (b) also shows excellent agreement
of the measured and simulated results predicted by the
mentioned analysis methods.

B. Tilted Beam
Fig. 15 presents a comparison between the simulated
and measured gain patterns at 20 GHz of a modulated
MTS antenna designed to radiate an RHCP beam that is
tilted 30◦ off from boresight. The antenna has a radius
of 10 λ at the working frequency. An excellent agreement
between the measured gain patterns and the calculated
ones is obtained, corroborating the effectiveness of the
design procedure. The achieved peak gain is equal to
33 dBi and corresponds to an aperture efficiency of 0.58.
The polarization purity is extremely good. Fig. 15 also
Fig. 13. Far-field scattered in the Fresnel region by an array of
shows the overall antenna layout and some details of the
dipoles in the near zone calculated by summation of individual
scatterers (dotted line) and by FW and FW-induced diffracted rays
feeder, which was composed essentially of a coaxial guide
(continuous line) with uniform transition functions. The calculations coupled to some patches of annular shape for matching
have been performed by A. Cucini (WaveCom). purposes.

are just taken proportional to the incident electric field C. Very High Gain
(no coupling is assumed between the dipoles, namely, An example of a very-high-gain MTS antenna prototype
the matrix Ynm in (18) is assumed to be the identity). obtained by an appropriate shape of the modulation is
Fig. 13 compares the results obtained by simply summing
the dipole contributions and using a ray-field description,
showing excellent accuracy.

VII. P R A C T I C A L E X A M P L E S O F
S W- B A S E D M T S
In this section, we present examples of high-performance
MTS antennas that have been designed with the method
seen before, realized and measured.

A. High Efficiency
An MTS antenna with maximum aperture efficiency
can be designed by implementing the modulation profile
introduced in [60]. A Ka-band prototype antenna has been
designed and fabricated to verify this approach.
The antenna features a diameter of 27λ at the center
frequency of 29.75 GHz. It was designed to radiate a
boresight beam with a product between the tapering and
spill-over efficiencies of 85%. The prototype was realized
by using a Rogers RO3003 laminate characterized by a
thickness of 0.762 mm and a relative dielectric constant
equal to 3.0 with a loss factor equal to tan δ = 0.001. The
measured peak gain was equal to 37 dBi corresponding to
an overall efficiency εtot ≈ 0.70, including the launching
efficiency of SW by the dipole and the losses. This antenna
currently constitutes a benchmark in terms of the maxi-
mum efficiency attained by MTS apertures. The simulated Fig. 14. Layouts and performance of a highly efficient MTS
antenna. (a) Copolar and (b) Cross-polar gain patterns. The curves
and measured copolar and cross-polar gain patterns are
provide a comparison of the measurements (blue lines) with three
shown in Fig. 14(a) and (b). The measured patterns are different numerical analysis results. Green lines: AFM method. Gray
compared with those produced by the analysis based on lines: GR-basis function continuous impedance BCs MoM. Red line:
the AFMs [51], by the Gaussian ring MoM code based on full-wave FMM for the textured layout.

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1237

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

Fig. 17 illustrates these mechanisms and the annular

active regions of a wideband MTS antenna. The directivity
patterns are color-coded to the active regions that produce
them. Note that the active region slides inward on the
aperture away from its edge as the working frequency
increases. This phenomenon is due to a local resonance
that couples the SW power with the LW power. Con-
versely, the aperture area outside of an active region barely
interacts with the SW at certain frequencies, and hence,
it radiates very weakly at them.
A functional wideband design, for instance, for a
broadside radiating aperture is ensured by matching
the modulation phase Ks(ρ) in (19)–(22) with the
frequency-dependent SW phase through radially shaping
the periodicity of the modulation function. Note that the
vector position of the modulation phase will be assumed
to be independent of the azimuthal angle throughout
this section. As the SW wavenumber βsw (ω, ρ) depends
dominantly on the “opaque” reactance X̄op (ω) of the MTS
Fig. 15. Comparisons between the measured and simulated gain antenna, any wideband design needs to take into consid-
patterns of a tilted beam high-efficiency MTS antenna realized by eration both the dispersion of the metallic coating and the
the “grain of rice” elements. The antenna has been designed to contribution from the grounded slab over the design band-
radiate a tilted RHCP beam pointed at 30◦ at 20 GHz. (a) Copolar
width. Because the dispersion of the capacitive reactance
component. (b) Cross-polar component. Insets: photographs of the
of the metallic coating is quite weak when compared with
antenna that includes the details of the aperture implementation
and the feeder. the one of the grounded slabs, it can be evaluated at a
single frequency (usually the central one). The in-band
frequency behavior can then be extrapolated with a linear
approximation of the capacitive admittance. When needed,
shown in Fig. 16. It attained the largest gain that has
a more refined estimation of the texture reactance can be
been measured until now with this type of antenna. The
obtained by using the method illustrated in [50]. We note
prototype was designed to operate at 30 GHz and was
that βsw depends, in the second order, on the amplitude
fabricated using a Rogers RO3003 substrate of 0.762 mm
thickness with a 46-cm diameter. The feeder consisted of
a coaxial square-flanged connector placed at the center of
the antenna that was surrounded by metallic rings on the
MTS side. This configuration achieved both a low input
reflection coefficient and a high SW launching efficiency.
The measured gain patterns of the antenna are shown in
Fig. 16. The peak value reached 39.5 dBi at 30 GHz. The
corresponding peak directivity was 41.2 dBi. The losses of
this antenna did not exceed 1.7 dB across the operational
bandwidth of 500 MHz (5%). It is an unprecedented value
for an MTS antenna. This value also includes the losses in
the SMA connector. The achievement of this performance
is obtained by using the optimal mathematical modulation
function presented in [60] as an initial step.

D. Large Bandwidth
A technique to enlarge the operative bandwidth of an
MTS antenna consists of radially modifying the modulation
periodicity, i.e., the periodicity becomes a function of the
radial coordinate. Its values are chosen according to the
SW frequency dispersion. The antenna can then be con-
ceived as an active region device where different annular
rings are characterized by their own local period and are Fig. 16. Measured gain pattern of the very-high-gain MTS antenna
activated alternately as the frequency is changed to radiate prototype. (a) Copolar. (b) Cross polar. Inset: details of the feeder
a beam in the desired direction. (design and measurement done by Wave Up).

1238 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

diagram. They are able to have not only up to a 40-dB

peak gain with few percent bandwidths but also a large
bandwidth with medium gain.

F. Aperture Sharing: Multiplexing by Multiport

A major practical goal for the wireless network is to
obtain multiple directive beams from a single aperture
simultaneously. For SW-based antennas, this aperture shar-
ing concept foresees, in essence, the superposition of two
or more modulation functions point by point on the entire
aperture. An example of such an MTS antenna was devel-
oped in [62] and is illustrated in Fig. 19. In these specific
cases, the reactance modulation tensor assumes a more
general form of a summation of modulation, which is even-
tually implemented by printed elements. Assigning one
feed point per beam instead of a single, unique feed, allows
for multiplexing by a simple switching network. One could
consider dividing the aperture in distinct angular sectors,
each excited by a different source. This approach requires a
feed network scheme in which each source illuminates only
its corresponding sector. Consequently, the overall system
is more complex in comparison to the single feed ver-
sion. Nevertheless, one can still apply the superposition of
the modulation patterns to obtain one independent beam
per source. The main difference is that each modulation
Fig. 17. Ka-band large bandwidth MTS antenna. (a) Active region
will have to be referenced to a local coordinate system
principle. (b) Photograph of the fabricated prototype with insets
showing the large bandwidth feed and the elliptical patches used to
whose origin is located at the corresponding feed point
synthesize the impedance tensor. (c) Input impedance. location. A practical example with three ports is shown
(d) Directivity versus frequency response: measured (blue line) and in Fig. 20 where the results of three partially overlapped
simulated directivities. The simulated values were computed with aperture modulations are realized by printed elliptical
the IBC-MoM in [61] (red line) taking into consideration the solution
subwavelength patches at 4.5 GHz.
for the ideal homogenized reactance and with the commercial
software CST.
A further example of the multibeam antenna at 4GHz is
shown in Fig. 21; it consists of a 19-port antenna printed
on the same substrate and realized by individual MTS ten-
and phase of the impedance modulation, which changes sors with moderately overlapped impedance modulation.
point by point over the surface. The apertures are circular and about 3.5 wavelength diam-
Fig. 17 shows the results for an aperture designed to eter each; they are distributed with centers over a hexag-
work at the Ka-band. Its substrate has the thickness h = onal lattice. Each aperture has a different layout in such
0.635 mm and the relative permittivity εr = 6.15. The a way to have beams that overlap in the far zone at about
modulation is characterized by an exponential profile of
the radial periodicity that is bounded at the center and
the rim by the values d1 = 7.0 mm and d2 = 13.7 mm,
respectively. The antenna has a 3-dB bandwidth ranging
from 23.0 to 29.1 GHz and exhibits ±2-dB in-band gain
oscillation. Even though the present design is quite effec-
tive in extending the operability of MTS antennas in terms
of their pattern bandwidths, these gain oscillations with
respect to its average are still nonnegligible.

E. Range of Applicability in a Diagram

Gain–Bandwidth
The range of operability of MTS antennas in terms of
their main performance characteristics is compared with
those of other flat microwave antennas in Fig. 18. The
diagram represents the obtained gain versus the relative Fig. 18. Gain versus relative bandwidth diagram for several flat
bandwidth. The MTS antennas cover a large region of this antenna typologies.

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1239

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

Fig. 20. Subwavelength elliptical patch implementation of

three-beam/three-port MTS antenna at 4 GHz realized by a partial
overlap of the modulation functions (designed by Wave-Up and
realized by Ultimetas).

devices and tunable materials yield an increase in the

antenna losses with a consequent reduction of gain and
increase in power demand; losses become more important
when working at higher frequencies (e.g., the Ka-band):
frequency scalability is indeed another challenge. Losses
become less critical when dealing with mechanical recon-
Fig. 19. Four-beam MTS antenna realized by superimposing the
modulation functions. (a) Aperture sharing to generate the four
beams using four monopoles and (top left) the superposition of their
modulated impedance patterns. The colors on the aperture
represent the amplitude of the impedance modulation. (b) Far-field
directivity patterns in the uv plane that was obtained with the
homogenized-impedance MoM simulations in [61].

−3 dB. The beams realized by each port are reported in the

uv plane in the bottom part of Fig. 21. Each beam exhibits a
directivity of about 15-dB spanning with −3 dB over a con-
ical angle of ±45◦ angle. The described structure allows
for implementing a very simple MIMO for sensing the
environment with extremely low cost. We stress that every-
thing can be obtained by printing all the MTS on the same
substrate and connectorizing the back of the substrate.

VIII. C H A L L E N G E S F O R S M A R T R A D I O
ENVIRONMENT
Fast reconfigurability of the beam and beam scanning
based on dynamic, electronically reconfigurable MTSs
are the most impactful challenges for rendering RIS
an enabling technology for SRE scenarios. This can be
Fig. 21. Example of multiport/multibeam antenna with MTS
achieved through the inclusion in the MTS of active printed over a unique substrate. The antenna can realize 19 beams
devices or tunable materials; this way, the electric features of about 15-dB directivity each. Right corner: layout of one of the
of the inclusions become voltage controlled, and hence, entries of the impedance tensor. Bottom: radiated beams directivity
the BCs offered by the MTS can be properly adjusted by maps in the uv plane for each excited port, obtained by a
homogenized-impedance full-wave analysis over the entire
an external control.
19-aperture structure. (Zoomed-in view of one u-v map with the
Although this concept has been already demonstrated scale is reported on the right.) The position of each u-v contour map
(see [57], [64], and [69]), existing solutions still need is associated with the analogous position of the space aperture
to be improved in terms of efficiency. In fact, active layout (see the arrows.)

1240 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

Fig. 22. Pictorial representation of the possible beam scanning

approaches for SW-based antennas. (a) Change of SW wavelength
for a fixed period. (b) Change of period for a fixed surface-wave
wavelength.

figuration. The use of micromechanical systems or piezo-

electric devices has been proposed, but they may suffer
from low reliability, and an MTS antenna based on such
devices may be too sensitive to the external vibrations if
installed on moving vehicles.
Beam scanning in SW-based reconfigurable MTS
antennas can be obtained through two different
approaches, illustrated in Fig. 22: the first one is Fig. 23. Implementation of the approach in Fig. 22(a). Measured
based on the control of the SW wavenumber and the radiation pattern from a liquid crystal-based reconfigurable MTS
antenna under different bias conditions. The inset shows a picture
second one on the control of the periodicity. Both allow
of the prototype and the geometry.
for a 1-D scanning of the radiated beam. In particular,
the first approach allows beam scanning at a fixed
for RIS in 5G scenarios. In that case, the pointing angle
frequency with a single voltage control. This approach
was precisely controlled by piloting a network of p-i-n
has been exploited in [70] using liquid crystals as tunable
diodes connected to metallic dipoles printed on top of a
substrates, making use of the formulation presented in
PCB. The basic idea is presented in Fig. 24. MTS with
[71]. The radiating part is constituted by a parallel plate
integrated active devices, instead of discrete ones, would
waveguide partially filled by liquid crystals with gradually
bring advantages in terms of reliability, losses, and per-
modulated, electrically small slots etched on the upper
formance. In this framework, the power consumption for
wall, while the feeding network is realized through a
the control network can be decreased by using field-effect
pillbox network (see the inset in Fig. 23). The concept
transistors instead of p-i-n diodes as switches. However,
was validated through the realization and experimental
realizing integrated devices in a wide area will require
characterization of an antenna prototype in the Ka-band
significant improvement in the accuracy and repeatability
although only a limited scan range (around 10◦ ) was
of the fabrication processes.
measured due to an imperfect alignment of the LC layer.
Special liquid crystals for microwaves could be
extremely beneficial, and some companies are investing in
this direction although they are limited in terms of recon-
figuration speed. Alternative tunable materials are ferro-
electric or phase-changing materials, such as vanadium
dioxide. However, phase-changing materials suffer from
temperature instability. Advances in phase-changing mate-
rials would bring a significant benefit to reconfigurable
MTSs, in terms of losses and switching time reduction,
as well as an increase in temperature stability.
The second approach [see Fig. 22(b)] is based on the
tuning of the modulation period, and therefore, it requires
multiple control signals across the MTS. As a difference
from the previous approach, these controls can also be
binary. If all the MTS unit cells are individually piloted,
2-D scanning can be also achieved through the pointwise
Fig. 24. Implementation of the approach in Fig. 22(b). Dipole type
control of the MTS modulation. This strategy was applied MTS constituted by printed dipoles with p-i-n diodes in on (red
and experimentally validated in [72] to implement ele- dipoles) and off (gray dipoles) configurations. Different periods give
vation scanning in an array of modulated MTS channels different radiated beams.

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1241

Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

Alternative strategies for reconfigurability of the RIS Furthermore, the operation principle of MTS antennas
consist of optical pumping photosensitive substrate with allows for tailoring, in real time, the MTS with active
MTSs printed on top. Silicon or gallium arsenide sub- devices or tunable materials. While the initial concern of
strates can be used, together with checkerboard-type self- sometimes ago about bandwidth is a challenge in fast
complementary MTSs’ CB-MTS. The vertexes of CB-MTS progress, losses, agility, and dynamic speed are still open
are disconnected by a small gap in off-conditions, and they issues for both design and fabrication technology aspects.
are virtually connected by the incident optical power den- Finally, the possibility to easily customize RIS features in
sity [73]. Since the gaps are very small (for instance at the the Fresnel zone and describe it in terms of rays opens the
level of few optical wavelengths), this allows to modify the possibility of a fast deterministic analysis of the SRE.
MTS configuration with an extremely small power [74].
Acknowledgment
3-D printing processes are also extremely appealing for
The authors wish to acknowledge the contribution of sev-
MTS antennas, creating the possibility to move beyond the
eral people to the results shown in this article: Francesco
classic PCB processes, offering chances to build up the MTS
Caminita (Wave Up Srl, Siena, Italy) for the design of the
by adding specific features to the inclusions, and, hence,
high efficiency antenna in Fig. 15 and the FMM results,
improving performance at low cost.
Gabriele Minatti (Wave Up Srl) for the design of the
Other challenges arise from the complexity and large
very high-gain antenna in Fig. 16 and the multibeam
size of the MTSs required for SREs. Indeed, despite the
antenna in Fig. 21, Cristian Della Giovampaola (Wave
significant advances done in the development of the design
Up Srl) for the design of the multibeam antennas in
procedures described here, MTS optimization is still a
Fig. 20 and the measurements in Fig. 16, David Gonzalez-
future challenge topic. Whenever strong requirements on
Ovejero (Centre National de la Recherche Scientifique-
performance are demanded, it is necessary to optimize
Institut d’Electronique et des Technologies du numéRique
the final MTS layout, which may be composed of up to
(CNRS-IETR), Rennes, France) for the realization and
100 000 elements. Managing the high number of inclusions
measurements of the broadband antenna in Fig. 17,
with numerical methods, including electronic devices and
the multibeam antenna in Fig. 19, and the continuous
losses, requires smart strategies, which make combined
impedance boundary condition method of moments (IBC-
use of homogenized BCs and fast integral-equation solvers,
MoM) results, Marco Faenzi (University of Siena) for the
embedding an accurate model of the tunable elements.
design of the broadband antenna in Fig. 17, Prof. Charlotte
Tripon Canseliet (Ultimetas, Paris, France) for the realiza-
IX. C O N C L U S I O N tion of the multibeam antenna in Fig. 20, Alessio Cucini
This article has reviewed modeling, design, and simulation (WaveComm Srl, Siena) for the results in Fig. 13, Cristina
strategies for MTS-based reflecting surfaces and anten- Yepes (University of Siena, Siena) for the implementation
nas/sensors, which represents the key enablers for future of the anomalous reflector in Fig. 12, Romeo Beccherelli
SREs. Several practical realizations have been shown to [Istituto per la Microelettronica e Microsistemi (IMM),
illustrate the promising potential of this technology, includ- Consiglio Nazionale delle Ricerche (CNR)] and Valerio
ing overlapping apertures for multibeaming. Fast beam Martorelli (Ingegneria Dei Sistemi) for the realization of
hopping and multibeam operations with low power con- the antenna in Fig. 23, and Giorgio Giordanengo (Links)
sumption and low cost are interesting features for SRE. for the measurements of the antenna in Fig. 23.

REFERENCES
[1] C. Liaskos, S. Nie, A. Tsioliaridou, A. Pitsillides, pp. 1–72, May 2016. [14] L. Li and T. J. Cui, “Information
S. Ioannidis, and I. Akyildiz, “A new wireless [7] K. Achouri, M. A. Salem, and C. Caloz, “General metamaterials—From effective media to real-time
communication paradigm through metasurface synthesis based on susceptibility information processing systems,” Nanophotonics,
software-controlled metasurfaces,” IEEE Commun. tensors,” IEEE Trans. Antennas Propag., vol. 63, vol. 8, no. 5, pp. 703–724, Mar. 2019.
Mag., vol. 56, no. 9, pp. 162–169, Sep. 2018. no. 7, pp. 2977–2991, Jul. 2015. [15] Q. Ma, G. D. Bai, H. B. Jing, C. Yang, L. Li, and
[2] M. D. Renzo et al., “Smart radio environments [8] F. Yang and Y. Rahmat-Samii, Surface T. J. Cui, “Smart metasurface with self-adaptively
empowered by reconfigurable AI meta-surfaces: Electromagnetics, With Applications in Antenna, reprogrammable functions,” Light, Sci. Appl.,
An idea whose time has come,” EURASIP J. Microwave, and Optical Engineering. Cambridge, vol. 8, no. 1, p. 98, Dec. 2019.
Wireless Commun. Netw., vol. 2019, no. 1, U.K.: Cambridge Univ. Press, 2019. [16] V. S. Asadchy, M. Albooyeh, S. N. Tcvetkova,
pp. 1–20, Dec. 2019. [9] A. Li, S. Singh, and D. Sievenpiper, “Metasurfaces A. Díaz-Rubio, Y. Ra’di, and S. A. Tretyakov,
[3] M. Di Renzo et al., “Smart radio environments and their applications,” Nanophotonics, vol. 7, “Perfect control of reflection and refraction using
empowered by reconfigurable intelligent surfaces: no. 6, pp. 989–1011, Mar. 2018. spatially dispersive metasurfaces,” Phys. Rev. B,
How it works, state of research, and the road [10] T. J. Cui, M. Q. Qi, X. Wan, J. Zhao, and Q. Cheng, Condens. Matter, vol. 94, no. 7, Aug. 2016,
ahead,” IEEE J. Sel. Areas Commun., vol. 38, “Coding metamaterials, digital metamaterials and Art. no. 075142.
no. 11, pp. 2450–2525, Nov. 2020. programmable metamaterials,” Light, Sci. Appl., [17] P. Nayeri, F. Yang, and A. Z. Elsherbeni,
[4] E. Basar, M. D. Renzo, J. De Rosny, M. Debbah, vol. 3, no. 10, p. e218, Oct. 2014. Reflectarray Antennas: Theory, Designs, and
M. Alouini, and R. Zhang, “Wireless [11] C. D. Giovampaola and N. Engheta, “Digital Applications. Hoboken, NJ, USA: Wiley, 2018.
communications through reconfigurable metamaterials,” Nature Mater., vol. 13, no. 12, [18] C. Pfeiffer and A. Grbic, “Millimeter-wave
intelligent surfaces,” IEEE Access, vol. 7, pp. 1115–1121, Sep. 2014. transmitarrays for wavefront and polarization
pp. 116753–116773, 2019. [12] T. J. Cui, M. Q. Qi, X. Wan, J. Zhao, and Q. Cheng, control,” IEEE Trans. Microw. Theory Techn.,
[5] F. Tariq et al., “A speculative study on 6G,” IEEE “Coding metamaterials, digital metamaterials and vol. 61, no. 12, pp. 4407–4417, Dec. 2013.
Wireless Commun., vol. 27, no. 4, pp. 118–125, programmable metamaterials,” Light, Sci. Appl., [19] N. Yu et al., “Flat optics: Controlling wavefronts
Aug. 2020. vol. 3, no. 10, p. e218, 2014. with optical antenna metasurfaces,” IEEE J. Sel.
[6] S. B. Glybovski, S. A. Tretyakov, P. A. Belov, [13] L. Li et al., “Electromagnetic reprogrammable Topics Quantum Electron., vol. 19, no. 3,
Y. S. Kivshar, and C. R. Simovski, “Metasurfaces: coding-metasurface holograms,” Nature Commun., p. 4700423, May/Jun. 2013.
From microwaves to visible,” Phys. Rep., vol. 634, vol. 8, no. 1, p. 197, Dec. 2017. [20] A. M. H. Wong and G. V. Eleftheriades, “Perfect

1242 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

 Martini and Maci: Theory, Analysis, and Design of MTSs for SREs

anomalous reflection with a bipartite Huygens’ Propag., vol. 54, no. 12, pp. 3659–3668, A, Gen. Phys., vol. 8, no. 5, Nov. 2017,
metasurface,” Phys. Rev. X, vol. 8, no. 1, Dec. 2006. Art. no. 054048.
Feb. 2018, Art. no. 011036. [40] A. Diaz-Rubio and S. A. Tretyakov, “Macroscopic [57] D. F. Sievenpiper, “Forward and backward leaky
[21] A. Epstein and G. V. Eleftheriades, “Huygens’ modeling of anomalously reflecting metasurfaces: wave radiation with large effective aperture from
metasurfaces via the equivalence principle: Design Angular response and far-field scattering,” IEEE an electronically tunable textured surface,” IEEE
and applications,” J. Opt. Soc. Amer. B, Opt. Phys., Trans. Antennas Propag., vol. 69, no. 10, Trans. Antennas Propag., vol. 53, no. 1,
vol. 33, no. 2, p. A31, 2016. pp. 6560–6571, Oct. 2021. pp. 236–247, Jan. 2005.
[22] V. G. Ataloglou, M. Chen, M. Kim, and [41] F. Capolino, M. Albani, S. Maci, and L. B. Felsen, [58] O. Yurduseven, D. L. Marks, T. Fromenteze, and
G. V. Eleftheriades, “Microwave Huygens’ “Frequency-domain Green’s function for a planar D. R. Smith, “Dynamically reconfigurable
metasurfaces: Fundamentals and applications,” periodic semi-infinite phased array. I. Truncated holographic metasurface aperture for a mills-cross
IEEE J. Microw., vol. 1, no. 1, pp. 374–388, floquet wave formulation,” IEEE Trans. Antennas monochromatic microwave camera,” Opt. Exp.,
Jan. 2021. Propag., vol. 48, no. 1, pp. 67–74, Jan. 2000. vol. 26, no. 5, pp. 5281–5291, 2018.
[23] S. Maci, G. Minatti, M. Casaletti, and [42] F. Capolino, M. Albani, S. Maci, and L. B. Felsen, [59] M. Li, S. Xiao, J. Long, and D. F. Sievenpiper,
M. Bosiljevac, “Metasurfing: Addressing waves on “Frequency-domain Green’s function for a planar “Surface waveguides supporting both TM mode
impenetrable metasurfaces,” IEEE Antennas periodic semi-infinite phased array. II. Diffracted and TE mode with the same phase velocity,” IEEE
Wireless Propag. Lett., vol. 10, pp. 1499–1502, wave phenomenology,” IEEE Trans. Antennas Trans. Antennas Propag., vol. 64, no. 9,
2011. Propag., vol. 48, no. 1, pp. 75–85, Jan. 2000. pp. 3811–3819, Sep. 2016.
[24] R. Quarfoth and D. Sievenpiper, “Artificial tensor [43] A. Cucini and S. Maci, “Macro-scale basis [60] G. Minatti, E. Martini, and S. Maci, “Efficiency of
impedance surface waveguides,” IEEE Trans. functions for the method of moment analysis of metasurface antennas,” IEEE Trans. Antennas
Antennas Propag., vol. 61, no. 7, pp. 3597–3606, large periodic microstrip arrays,” Appl. Comput. Propag., vol. 65, no. 4, pp. 1532–1541, Apr. 2017.
Jul. 2013. Electromagn. Soc. J., vol. 21, no. 3, pp. 256–266, [61] D. González-Ovejero and S. Maci, “Gaussian ring
[25] R. Quarfoth and D. Sievenpiper, “Surface wave Nov. 2006. basis functions for the analysis of modulated
scattering reduction using beam shifters,” IEEE [44] F. Mariottini, F. Capolino, S. Maci, and L. B. Felsen, metasurface antennas,” IEEE Trans. Antennas
Antennas Wireless Propag. Lett., vol. 13, “Asymptotic high-frequency Green’s function for a Propag., vol. 63, no. 9, pp. 3982–3993, Sep. 2015.
pp. 963–966, 2014. large rectangular planar periodic phased array of [62] D. González-Ovejero, G. Minatti,
[26] G. Minatti et al., “Modulated metasurface dipoles with weakly tapered excitation in two G. Chattopadhyay, and S. Maci, “Multibeam by
antennas for space: Synthesis, analysis and dimensions,” IEEE Trans. Antennas Propag., metasurface antennas,” IEEE Trans. Antennas
realizations,” IEEE Trans. Antennas Propag., vol. 53, no. 2, pp. 608–620, Feb. 2005. Propag., vol. 65, no. 6, pp. 2923–2930, Feb. 2017.
vol. 63, no. 4, pp. 1288–1300, Apr. 2015. [45] F. Capolino, S. Maci, and L. B. Felsen, “Asymptotic [63] F. Caminita, E. Martini, G. Minatti, and S. Maci,
[27] B. H. Fong, J. S. Colburn, J. J. Ottusch, high-frequency Green’s function for a planar “Fast integral equation method for metasurface
J. L. Visher, and D. F. Sievenpiper, “Scalar and phased sectoral array of dipoles,” Radio Sci., antennas,” in Proc. URSI Int. Symp. Electromagn.
tensor holographic artificial impedance surfaces,” vol. 35, no. 2, pp. 579–593, Mar. 2000. Theory (EMTS), Aug. 2016, pp. 480–483.
IEEE Trans. Antennas Propag., vol. 58, no. 10, [46] S. Maci, F. Capolino, and L. B. Felsen, [64] F. Liu et al., “Intelligent metasurfaces with
pp. 3212–3221, Oct. 2010. “Three-dimensional Green’s function for planar continuously tunable local surface impedance for
[28] C. Pfeiffer and A. Grbic, “A printed, broadband rectangular phased dipole arrays,” Wave Motion, multiple reconfigurable functions,” Phys. Rev. A,
Luneburg lens antenna,” IEEE Trans. Antennas vol. 34, no. 3, pp. 263–279, Sep. 2001. Gen. Phys., vol. 11, no. 4, Apr. 2019,
Propag., vol. 58, no. 9, pp. 3055–3059, Sep. 2010. [47] F. Capolino and S. Maci, “Uniform high-frequency Art. no. 044024.
[29] M. Faenzi et al., “Metasurface antennas: New description of singly, doubly, and vertex diffracted [65] K. Chen et al., “A reconfigurable active Huygens’
models, applications and realizations,” Sci. Rep., rays for a plane angular sector,” J. Electromagn. metalens,” Adv. Mater., vol. 29, no. 17, May 2017,
vol. 9, no. 1, pp. 1–14, Jul. 2019. Waves Appl., vol. 10, no. 9, pp. 1175–1197, Art. no. 1606422.
[30] S. Maci and A. Cucini, “FSS-based EBG surface,” Jan. 1996. [66] M. Kim and G. V. Eleftheriades,
in Electromagnetic Metamaterials: Physics and [48] F. Capolino and S. Maci, “Simplified closed-form “Guided-wave-excited binary Huygens’
Engineering Aspects, N. Engheta and R. Ziolkowski, expressions for computing the generalized Fresnel metasurfaces for dynamic radiated-beam shaping
Eds. Hoboken, NJ, USA: Wiley, 2006. integral and their application to vertex with independent gain and scan-angle control,”
[31] M. A. Francavilla, E. Martini, S. Maci, and diffraction,” Microw. Opt. Technol. Lett., vol. 9, Phys. Rev. A, Gen. Phys., vol. 15, no. 5, May 2021,
G. Vecchi, “On the numerical simulation of no. 1, pp. 32–37, May 1995. Art. no. 054037.
metasurfaces with impedance boundary condition [49] M. Albani, F. Capolino, G. Carluccio, and S. Maci, [67] S. Foo, “Liquid-crystal reconfigurable metasurface
integral equations,” IEEE Trans. Antennas Propag., “UTD vertex diffraction coefficient for the reflectors,” in Proc. IEEE Int. Symp. Antennas
vol. 63, no. 5, pp. 2153–2161, May 2015. scattering by perfectly conducting faceted Propag. USNC/URSI Nat. Radio Sci. Meeting,
[32] E. Martini, F. Caminita, and S. Maci, “Double-scale structures,” IEEE Trans. Antennas Propag., vol. 57, Jul. 2017, pp. 2069–2070.
homogenized impedance models for periodically no. 12, pp. 3911–3925, Dec. 2009. [68] J. Shabanpour, S. Beyraghi, and A. Cheldavi,
modulated metasurfaces,” EPJ Appl. Metamater., [50] M. Mencagli, Jr., E. Martini, and S. Maci, “Ultrafast reprogrammable multifunctional
vol. 7, p. 12, Jan. 2020. “Transition function for closed-form vanadium-dioxide-assisted metasurface for
[33] A. A. Oliner and A. Hessel, “Guided waves on representation of metasurface reactance,” IEEE dynamic THz wavefront engineering,” Sci. Rep.,
sinusoidally-modulated reactance surfaces,” IRE Trans. Antennas Propag., vol. 64, no. 1, vol. 10, no. 1, Dec. 2020, Art. no. 8950.
Trans. Antennas Propag., vol. 7, no. 5, pp. 136–145, Jan. 2016. [69] A. Li, Z. Luo, H. Wakatsuchi, S. Kim, and
pp. 201–208, Dec. 1959. [51] G. Minatti, F. Caminita, E. Martini, and S. Maci, D. F. Sievenpiper, “Nonlinear, active, and tunable
[34] F. Caminita, E. Martini, and S. Maci, “Exact “Flat optics for leaky-waves on modulated metasurfaces for advanced electromagnetics
solution for anisotropic, periodically modulated metasurfaces: Adiabatic Floquet-wave analysis,” applications,” IEEE Access, vol. 5,
boundary conditions excited by a surface wave,” IEEE Trans. Antennas Propag., vol. 64, no. 9, pp. 27439–27452, 2017.
in Proc. 9th Eur. Conf. Antennas Propag. (EuCAP), pp. 3896–3906, Sep. 2016. [70] E. Martini et al., “Reconfigurable antenna based
Jan. 2015. [52] G. Minatti, F. Caminita, E. Martini, M. Sabbadini, on liquid crystals for continuous beam scanning
[35] N. Yu et al., “Light propagation with phase and S. Maci, “Synthesis of modulated-metasurface with a single control,” in Proc. IEEE Int. Symp.
discontinuities: Generalized laws of reflection and antennas with amplitude, phase, and polarization Antennas Propag. USNC-URSI Radio Sci. Meeting,
refraction,” Science, vol. 334, no. 6054, control,” IEEE Trans. Antennas Propag., vol. 64, Jul. 2019, pp. 449–450.
pp. 333–337, Oct. 2011. no. 9, pp. 3907–3919, Sep. 2016. [71] S. C. Pavone, E. Martini, F. Caminita, M. Albani,
[36] Q. Wu, S. Zhang, B. Zheng, C. You, and R. Zhang, [53] N. M. Estakhri and A. Alù, “Wave-front and S. Maci, “Surface wave dispersion for a
“Intelligent reflecting surface-aided wireless transformation with gradient metasurfaces,” Phys. tunable grounded liquid crystal substrate without
communications: A tutorial,” IEEE Trans. Rev. X, vol. 6, no. 4, Oct. 2016, Art. no. 041008. and with metasurface on top,” IEEE Trans.
Commun., vol. 69, no. 5, pp. 3313–3351, [54] V. S. Asadchy, M. Albooyeh, S. N. Tcvetkova, Antennas Propag., vol. 65, no. 7, pp. 3540–3548,
May 2021. A. Díaz-Rubio, Y. Ra’di, and S. A. Tretyakov, Jul. 2017.
[37] J. Budhu and A. Grbic, “Perfectly reflecting “Perfect control of reflection and refraction using [72] C. D. Giovampaola, F. Caminita, G. Labate,
metasurface reflectarrays: Mutual coupling spatially dispersive metasurfaces,” Phys. Rev. B, E. Martini, and S. Maci, “Experimental verification
modeling between unique elements through Condens. Matter, vol. 94, no. 7, Aug. 2016, of a 5G active metasurface antenna with beam
homogenization,” IEEE Trans. Antennas Propag., Art. no. 075142. steering capabilities,” in Proc. 15th Eur. Conf.
vol. 69, no. 1, pp. 122–134, Jan. 2021. [55] C. Yepes, M. Faenzi, S. Maci, and E. Martini, Antennas Propag. (EuCAP), 2021.
[38] A. Cucini and S. Maci, “A compact formula for the “Perfect non-specular reflection with polarization [73] D. González-Ovejero et al., “Basic properties of
array factor of planar phased arrays with control by using a locally passive metasurface checkerboard metasurfaces,” IEEE Antennas
polygonal shape and skewed grid,” IEEE Antennas sheet on a grounded dielectric slab,” Appl. Phys. Wireless Propag. Lett., vol. 14, pp. 406–409, 2014.
Wireless Propag. Lett., vol. 1, pp. 138–141, 2002. Lett., vol. 118, no. 23, Jun. 2021, Art. no. 231601. [74] H. Krraoui et al., “Characterization of
[39] A. Cucini, “High-frequency Green’s function for [56] D. R. Smith, O. Yurduseven, L. P. Mancera, optically-reconfigurable metasurfaces by a free
planar periodic phased arrays with skewed grid P. Bowen, and N. B. Kundtz, “Analysis of a space microwave bistatic technique,” Appl. Sci.,
and polygonal contour,” IEEE Trans. Antennas waveguide-fed metasurface antenna,” Phys. Rev. vol. 10, no. 12, p. 4353, Jun. 2020.

Open Access funding provided by ‘University of Siena’ within the CRUI CARE Agreement

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1243