Holographic MIMO Surfaces for 6G Wireless Networks:

Opportunities, Challenges, and Trends

Chongwen Huang, Sha Hu, George C. Alexandropoulos, Alessio Zappone,
Chau Yuen, Rui Zhang, Marco Di Renzo, Merouane Debbah

To cite this version:

Chongwen Huang, Sha Hu, George C. Alexandropoulos, Alessio Zappone, Chau Yuen, et al.. Holo-
graphic MIMO Surfaces for 6G Wireless Networks: Opportunities, Challenges, and Trends. IEEE
Wireless Communications, Institute of Electrical and Electronics Engineers, 2020, 27 (5), pp.118-125.
�10.1109/MWC.001.1900534�. �hal-03020417�

HAL Id: hal-03020417

https://hal.archives-ouvertes.fr/hal-03020417
Submitted on 24 Nov 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 1

Holographic MIMO Surfaces for 6G Wireless

Networks: Opportunities, Challenges, and Trends
Chongwen Huang, Member, IEEE, Sha Hu, Member, IEEE, George C. Alexandropoulos, Senior Member, IEEE,
Alessio Zappone, Senior Member, IEEE, Chau Yuen, Senior Member, IEEE, Rui Zhang, Fellow, IEEE, Marco Di
Renzo, Fellow, IEEE, and Mérouane Debbah, Fellow, IEEE

Abstract—Future wireless networks are expected to evolve Although massive MIMO, three-Dimensional (3D) beam-
towards an intelligent and software reconfigurable paradigm forming, and their hardware efficient hybrid analog and digital
enabling ubiquitous communications between humans and mobile counterparts [10] provide remarkable approaches to conquer
devices. They will be also capable of sensing, controlling, and
optimizing the wireless environment to fulfill the visions of low- signal attenuation due to wireless propagation via software-
arXiv:1911.12296v3 [cs.IT] 19 Apr 2020

power, high-throughput, massively-connected, and low-latency based control of the directivity of transmissions, they impose
communications. A key conceptual enabler that is recently mobility and hardware scalability issues. More importantly,
gaining increasing popularity is the Holographic Multiple Input the intelligent manipulation of the ElectroMagnetic (EM)
Multiple Output Surface (HMIMOS) that refers to a low-cost propagation is only partially feasible since the objects in
transformative wireless planar structure comprising of sub-
wavelength metallic or dielectric scattering particles, which is the deployment area, other than the transceivers, are passive
capable of impacting electromagnetic waves according to desired and uncontrollable. As a result, the wireless environment as
objectives. In this article, we provide an overview of HMIMOS a whole remains unaware of the ongoing communications
communications by introducing the available hardware archi- within it, and the channel model continues to be treated as
tectures for reconfigurable such metasurfaces and their main a probabilistic process, rather than a nearly deterministic one
characteristics, as well as highlighting the opportunities and key
challenges in designing HMIMOS-enabled communications. enabled through software-controlled techniques.
Following the recent breakthrough on the fabrication of
programmable metamaterials, reconfigurable intelligent sur-
I. I NTRODUCTION faces have the potential to fulfill the challenging vision for
Future wireless networks, namely beyond fifth Generation 6G networks, and materialize seamless connections and intel-
(5G) and sixth Generation (6G), are required to support ligent software-based control of the environment in wireless
massive numbers of end-users with increasingly demanding communication systems when coated on the otherwise passive
Spectral Efficiency (SE) and Energy Efficiency (EE) require- surfaces of various objects [5]–[8]. Holographic MIMO Sur-
ments [1]–[4]. In recent years, research in wireless commu- faces (HMIMOS) aim at going beyond massive MIMO, being
nications has witnessed rising interests in massive Multiple based on low cost, size, weight, and low power consumption
Input Multiple Output (MIMO) systems, where Base Stations hardware architectures that provide a transformative means of
(BSs) are equipped with large antenna arrays, as a way to the wireless environment into a programmable smart entity [3],
address the 5G throughput requirements. However, it is still a [5], [8], [9], [11], [12]. In this article, we overview the different
very challenging task to realize massive MIMO BSs with truly emerging HMIMOS architectures and their core functional-
large-scale antenna arrays (i.e., with few hundreds or more ities, and discuss their currently considered communication
antennas) mainly due to the high fabrication and operational applications as well as their future networking challenges.
costs, as well as due to the increased power consumption.
Future 6G wireless communication systems are expected II. HMIMOS D ESIGN M ODELS
to realize an intelligent and software reconfigurable paradigm, In this section, we present available hardware architectures,
where all parts of device hardware will adapt to the changes of fabrication methodologies, and operation modes of HMIMOS
the wireless environment [1], [3], [5]. Beamforming-enabled systems that render them a flexibly integrable concept for
antenna arrays, cognitive spectrum usage, as well as adaptive diverse wireless communication applications.
modulation and coding are a few of the transceiver aspects that
are currently tunable in order to optimize the communication A. Categorization based on the Power Consumption
efficiency. However, in this optimization process, the wire- 1) Active HMIMOS: To realize reconfigurable wireless
less environment remains an unmanageable factor; it remains environments, HMIMOS can serve as a transmitter, receiver,
unaware of the communication process undergoing within it or reflector. When the transceiver role is considered, and thus
[1], [3]–[9]. Furthermore, the wireless environment has in energy-intensive Radio Frequency (RF) circuits and signal
general a harmful effect on the efficiency of wireless links. The processing units are embedded in the surface, the term active
signal attenuation limits the connectivity radius of nodes, while HMIMOS is adopted [13], [14]. On another note, active
multipath propagation resulting in fading phenomena is a well- HMIMOS systems comprise a natural evolution of conven-
studied physical factor introducing drastic fluctuations in the tional massive MIMO systems, by packing more and more
received signal power. The signal deterioration is perhaps software-controlled antenna elements onto a two-Dimensional
one of the major concerns in millimeter wave and in the (2D) surface of finite size. In [4], where the spacing between
forthcoming TeraHertz (THz) communications [1]. adjacent surface elements reduces when their number increase,
 2

Figure 1: The two generic steps of holographic training and holographic communication [13].
an active HMIMOS is also termed as Large Intelligent Surface a limited surface area in order to form a spatially continuous
(LIS). A practical implementation of active HMIMOS can be transceiver aperture [13], [14]. For the better understanding of
a compact integration of an infinite number of tiny antenna the operation of contiguous surfaces and their communication
elements with reconfigurable processing networks realizing models, we commence with a brief description of the physical
a continuous antenna aperture. This structure can be used operation of the optical holography concept. Holography is
to transmit and receive communication signals across the a technique that enables an EM field, which is generally the
entire surface by leveraging the hologram principle [13], [14]. result of a signal source scattered off objects, to be recorded
Another active HMIMOS implementation is based on discrete based on the interference principle of the EM wave. The
photonic antenna arrays that integrate active optical-electrical recorded EM field can be then utilized for reconstructing the
detectors, converters, and modulators for performing transmis- initial field based on the diffraction principle. It should be
sion, reception, and conversion of optical or RF signals [13]. noted that wireless communications over a continuous aperture
2) Passive HMIMOS: Passive HMIMOS, also known as is inspired by the optical holography, which is sketched in
Reconfigurable Intelligent Surface (RIS) [3], [5]–[7], [9], or Fig. 1. In the training phase, the generated training signals
Intelligent Reflecting Surface (IRS) [8], [15], acts like a pas- from an RF source are split via a beamsplitter into two waves,
sive metal mirror or ‘wave collector,’ and can be programmed the object and reference waves. The object wave is directed
to change an impinging EM field in a customizable way [3], to the object, and some of the reflected wave mixed together
[5]. Compared with its active counterpart, a passive HMIMOS with the reference wave beam that does not impinge on the
is usually composed of low cost passive elements that do not object, are fed to the HMIMOS. In the communication phase,
require dedicated power sources. Their circuitry and embedded the transmitted signal is transformed into the desired beam to
sensors can be powered with energy harvesting modules, an the object user over the spatially continuous aperture of the
approach that has the potential of making them truly energy HMIMOS. Since the continuous aperture benefits from the
neutral. Regardless of their specific implementations, what integrated infinite number of antennas that is the asymptotic
makes the passive HMIMOS technology attractive from an limit of Massive MIMO, its potential advantages are to achieve
energy consumption standpoint, is their capability to shape ra- higher spatial resolution, and enable the creation and detection
dio waves impinging upon them and forwarding the incoming of EM waves with arbitrary spatial frequency components,
signal without employing any power amplifier nor RF chain, without undesired side lobes.
and also without applying sophisticated signal processing. 2) Discrete HMIMOS: The discrete HMIMOS is usually
Moreover, passive HMIMOS can work in full duplex mode composed of many discrete unit cells made of low power
without significant self interference or increased noise level, software-tunable metamaterials. The means to electronically
and require only low rate control link or backhaul connections. modify the EM properties of the unit cells range from off
Finally, passive HMIMOS structures can be easily integrated the shelves electronic components to using liquid crystals,
into the wireless communication environment, since their microelectromechanical systems or even electromechanical
extremely low power consumption and hardware costs allow switches, and other reconfigurable metamaterials. This struc-
them to be deployed into building facades, room and factory ture is substantially different from the conventional MIMO an-
ceilings, laptop cases, or even human clothing [3], [5]. tenna array. One embodiment of a discrete surface is based on
discrete ‘meta-atoms’ with electronically steerable reflection
B. Categorization based on the Hardware Structure properties [6]. As mentioned earlier, another type of discrete
1) Contiguous HMIMOS: A contiguous HMIMOS inte- surface is the active one based on photonic antenna arrays.
grates a virtually uncountably infinite number of elements into Compared with contiguous HMIMOS, discrete HMIMOS have
 3

I I I I I
C C C C C
C S C A C S C A C S

S A S A S

S A S A S

Figure 2: The two operation modes of HMIMOS systems along with their implementation and hardware structures. A schematic
view of the HMIMOS functions of EM field polarization, scattering, focusing, and absorption control is provided.

some essential differences from the perspectives of implemen- with software-defined metasurface antennas. A general log-
tation and hardware, as will be described in the sequel. ical structure (regardless of its physical characteristics) was
proposed in [6], as shown in Fig. 2(b). Its general unit cell
C. Fabrication Methodologies structure contains a metamaterial layer, sensing and actuation
There are various fabrication techniques for HMIMOS layers, shielding layer, computing layer, as well as an interface
including electron beam lithography at optical frequencies, and communications layer with different objectives. Specif-
focused-ion beam milling, interference and nanoimprint lithog- ically, the meta-material layer is implemented by graphene
raphy, as well as direct laser writing or printed circuit board materials for delivering the desired EM behavior through a
processes at microwaves. Usually, these fabrication techniques reconfigurable pattern, while the objective of sensing and
will be ascribed to produce two typical apertures, continuous actuation layer is to modify the behavior of the meta-material
or discrete apertures, as shown in Fig. 2. A fabrication layer. The shielding layer is made of a simple metallic layer
approach leveraging programmable metamaterials for approx- for decoupling the EM behavior of the top and bottom layers
imately realizing a continuous microwave aperture [13], [14] to avoid mutual interferences. The computing layer is used
is depicted in Fig. 2(a). This meta-particle structure uses the to execute external commands from the interface layer or
varactor loading technique to broaden its frequency response sensors. Finally, the interface and communications layer aims
range, and achieves continuous aperture and controllable re- at coordinating the actions of the computing layer and updating
flection phase. It is a continuous monolayer metallic structure, other external wireless entities via the reconfigurable interface.
and comprises an infinite number of meta-particles. Each meta- While HMIMOS is in its infancy, basic prototyping work
particle contains two metallic trapezoid patches, a central on different kinds of this technology is going on in the
continuous strip, and varactor diodes. By independently and world. A discrete HMIMOS was developed by the start-up
continuously controlling the bias voltage of the varactors, the company named “Greenerwave”, and which shows the basic
surface impedance of continuous HMIMOS can be dynami- feasibility and effectiveness of the HMIMOS concept using
cally programmed, and thus manipulate the reflection phase, the discrete metasurface antennas. In contrast, another start-
amplitude states, and the phase distribution at a wide range up company called “Pivotalcommware” with the investment
of frequency bands [1]. It should be highlighted that this of Bill Gates capital is developing the initial commercial
impedance pattern is a map of the hologram, and can be products of the contiguous HMIMOS based on the low-
calculated directly from the field distribution of the provided cost and contiguous metasurfaces, which further verifies the
reference wave and reflected object wave, as discussed in feasibility of the HMIMOS concept as well as advancement of
Fig. 1. Exploiting intelligent control algorithms, beamforming holographic technologies. Continued prototyping development
can be accomplished by using the hologram principle. is highly desired to both prove the HMIMOS concept with
In contrast to continuous aperture, another instance of even brand new holographic beamforming technologies and
HMIMOS is the discrete aperture that is usually realized discover potentially new issues that urgently need research.
 4

D. Operation Modes point in the near or far field. The collimation (i.e., the
The following four operation modes for HMIMOS are reverse functionality) also belongs in this general mode
usually considered: 1) continuous HMIMOS as an active of beamforming operation.
transceiver; 2) discrete HMIMOS as a passive reflector; 3) • F4: EM Field Absorption, which implements minimal
discrete HMIMOS as an active transceiver; and 4) continuous reflected and/or refracted power of the incoming EM
HMIMOS as a passive reflector. Given the recent research field.
interests and due to space limitations, we next elaborate on
the first two representative modes of operation, which are also B. Characteristics
sketched within Fig. 2. Compared with currently used technologies in wireless
1) Continuous HMIMOS as Active Transceivers: Accord- networks, the most distinctive characteristics of the HMIMOS
ing to this mode of operation, the continuous HMIMOS concept lie in making the environment controllable by pro-
performs as an active transceiver. The RF signal is generated viding the possibility of fully shaping and controlling the
at its backside and propagates through a steerable distribution EM response of the environmental objects that are distributed
network to the contiguous surface constituted by the infinite throughout the network. An HMIMOS structure is usually
software-defined and electronically steerable elements that intended to perform as a signal source or ‘wave collector’
generate multiple beams to the intended users. A distinct with reconfigurable characteristics, especially for application
difference between active continuous HMIMOS and passively scenarios where it is used as a passive reflector with the
reconfigurable HMIMOS is that the beamforming process of objective of improving the EE. The fundamental properties of
the former is accomplished based on the holographic concept, HMIMOS systems1 and their core differences with massive
which is a new dynamic beamforming technique based on MIMO and conventional multi-antenna relaying systems are
software-defined antennas with low cost/weight, compact size, summarized as follows:
and a low-power hardware architecture. • C1: HMIMOS can be nearly passive. One significant
2) Discrete HMIMOS as Passive Reflectors: Another op- merit of passive HMIMOS is that they do not require
eration mode of HMIMOS is the mirror or ‘wave collector,’ any internally dedicated energy source to process the
where the HMIMOS is considered to be discrete and passive. incoming information-carrying EM field.
In this case, the HMIMOS include reconfigurable unit cells, as • C2: HMIMOS can realize continuous apertures. Re-
previously described, which makes their beamforming mode cent research activity focuses on low operational cost
resembling that of conventional beamforming [10], unlike methods for realizing spatially-continuous transmitting
continuous transceiver HMIMOS systems. It is worth noting and receiving apertures.
that most of the existing works (e.g., [5], [7], [8]) focus on • C3: Receiver thermal noise is absent in HMIMOS.
this HMIMOS operation mode which is simpler to implement Passive HMIMOS do not require to down-convert the
and analyze. received waveform for baseband processing. Instead they
implement analog processing directly on the impinging
III. F UNCTIONALITY, C HARACTERISTICS , AND EM field.
C OMMUNICATION A PPLICATIONS • C4: HMIMOS elements are tuned in software. Avail-

Different fabrication methods of HMIMOS systems result able architectures for metasurfaces enable simple repro-
in a variety of functionalities and characteristics, with most grammability of all settings of their unit elements.
of them being very relevant to late expectations for future 6G • C5: HMIMOS can have full-band response. Due to

wireless systems (e.g., Tbps peak rates). In this section, we recent advances in meta-materials’ fabrication, reconfig-
highlight the HMIMOS functions and key characteristics, and urable HMIMOS can operate at any operating frequency,
discuss their diverse wireless communications applications. ranging from the acoustic spectrum up to THz and the
light spectra.
• C6: Distinctive low latency implementation. HMIMOS
A. Functionality Types
are based on rapidly reprogrammable meta-materials,
Intelligent surfaces can support a wide range of EM in- whereas conventional relaying and massive MIMO sys-
teractions, termed hereinafter as functions. Ascribing to their tems include antenna array architectures.
programmable features and depending on whether they are
realized via structures with discrete or continuous elements, C. Communications Applications
HMIMOS have four common function types as illustrated in
The unique features of HMIMOS enabling intelligent and
the bottom part of Fig. 2:
rapidly reconfigurable wireless environments make them an
• F1: EM Field Polarization, which refers to the reconfig-
emerging candidate technology for the low-power, high-
urable setting of the oscillation orientation of the wave’s throughput, and low-latency vision of 6G wireless networks.
electric and magnetic fields. We next discuss representative communications applications
• F2: EM Field Scattering, where the surface redirects an
of HMIMOS for outdoor and indoor environments.
impinging wave with a given direction of arrival towards
1 It should be noted that not all HMIMOS architectures have all listed
a desired or multiple concurrent desired directions.
attributes. Few of them are inherent to passive HMIMOS, but not to active
• F3: Pencile-like Focusing, which takes place when a ones, and vice versa. However, we discuss HMIMOS properties here in a
HMIMOS acts as lens to focus an EM wave to a given broad scope, including all available types up to date.
 5

Figure 3: Wireless communications applications of HMIMOS in outdoor and indoor environments.

1) Outdoor Applications: Consider the discrete passive locations via efficient beamforming that compensates
HMIMOS as an indicative example that comprises a finite for the signal attenuation from the BS or co-channel
number of unit elements, and intended for forwarding suitably interference from neighboring BSs.
phase-shifted versions of its impinging signals to users over • A3: Physical-layer security. HMIMOS can be deployed
different outdoor scenarios, such as typical urban, shopping for physical layer security in order to cancel out reflec-
malls, and international airports, as illustrated in the upper tions of the BS signals to eavesdroppers.
part of Fig. 3. We assume that HMIMOS are planar structures • A4: Wireless power transfer. HMIMOS can collect
of few centimeters thickness and variable sizes that can be ambient EM waves and direct them to power-hungry IoT
easily deployed onto nearly all environmental objects. devices and sensors enabling also simultaneous wireless
information and power transfer.
• A1: Building connections. HMIMOS can extend the 2) Indoor Applications: Indoor wireless communication is
coverage from outdoor BSs to indoor users, especially subject to rich multipath propagation due to the presence of
in cases where there is no direct link between the users multiple scatters and signal blocking by walls and furniture, as
and BS, or the link is severely blocked by obstacles. well as RF pollution due to the highly probable densification
• A2: Energy-efficient beamforming. HMIMOS are ca- of electronic devices in confined spaces. As such, providing
pable of recycling ambient EM waves and focusing them ubiquitous high throughput indoor coverage and localization
to their intended users via effective tuning of their unit is a challenging task. HMIMOS has the potential of being
elements. In such cases, surfaces are deployed as relays highly beneficial in indoor environments, leveraging from its
to forward the information bearing EM field to desired inherit capability to reconfigure EM waves towards various
 6

Table I: Some recent research results on HMIMOS-based wireless communication systems.

Related Works Applications Functions Characteristics Main Contributions
Presented a HMIMOS-based approach to combat the distance limita-
[1] A1, A2, A5 F2, F3 C1, C3-C6 tion in millimeter wave and THz systems; simulation results for an
indoor set up corroborated the merits of proposed approach.
Introduced an indoor signal propagation model and presented informa-
Indoor [4] A2, A5, A6 F2, F3 C2-C6 tion theoretical results for active and continuous HMIMOS systems.
Introduced the concept of programmable indoor wireless environments
[6] A1-A3, A5 F1-F4 C1, C3-C6 offering simultaneous communication and security; an indoor model
and a simulation set up for HMIMOS communication were presented.
Designed a 0.4m2 and 1.5mm thick planar metasurface consisting
[7] A1, A2 F2, F3 C1, C3-C6 of 102 controllable unit cells operating at 2.45GHz; demonstrated
increased received signal strength when deployed indoors.
Proposed free space pathloss models using the EM and physical prop-
[9] A1, A2, A5 F2, F3 C3-C6 erties of a reconfigurable surface; indoor field experiments validated
the proposed models.
Proposed HMIMOS for outdoor MIMO communications and presented
[5] A1, A2 F2, F3 C1, C3-C6 EE maximization algorithms; studied the fundamental differences
between HMIMOS and conventional multi-antenna relays.
Presented jointly active and passive beamforming algorithms for
[8] A2, A4 F2, F3 C1, C3-C6 HMIMOS-assisted MIMO communication; analyzed the interference
distribution and studied the power scaling law.
Derived the optimal HMIMOS phase matrix for the case of available
Outdoor [11] A2 F2, F3 C1, C3-C6 statistical channel information and presented a tight approximation for
the ergodic capacity.
[12] Studied compressive sensing and deep learning approaches for HMI-
A1, A2 F1-F4 C1, C3-C6 MOS channel estimation and online configuration.

communication objectives. An illustrative general example is systems operate over uncontrollable wireless environments,
sketched in the lower part of Fig. 3. In the left corner of this whereas HMIMOS-based systems will be capable of recon-
example where a HMIMOS is absent, the signal experiences figuring their EM propagation impact. This fact witnesses the
pathloss and multipath fading due to refraction, reflection, and need for new mathematical methodologies to characterize the
diffusion, which deteriorates its sufficient propagation to the physical channels in HMIMOS-based systems and analyze
target user. However, in the right corner of Fig. 3, signal their ultimate capacity gains over a given volume [14], as
propagation can be boosted using HMIMOS coated in the wall well as for new signal processing algorithms and networking
so as to assist the signal from the access point to reach the schemes for realizing HMIMOS-assisted communication. For
intended user with the desired power level. example, continuous HMIMOS is used for the reception and
• A5: Enhanced in-building coverage: As previously transmission of the impinging EM field over its continuous
discussed, indoor environments can be coated with HMI- aperture using the hologram concept. Differently from the
MOS to increase the throughput offered by conventional currently considered massive MIMO systems, HMIMOS op-
Wi-Fi access points. eration can be described by the Fresnel-Kirchhoff integral that
• A6: High accurate indoor positioning: HMIMOS has is based on the Huygens-Fresnel principle [13].
increased potential for indoor positioning and localiza-
tion, where the conventional Global Positioning System B. HMIMOS Channel Estimation
(GPS) fails. Large surfaces offer large, and possibly con- The estimation of possibly very large MIMO channels in
tinuous, apertures that enable increased spatial resolution. HMIMOS-based communication systems is another critical
There has been lately increasing research interest in wireless challenge due to the various constraints accompanying the
communication systems incorporating HMIMOS. In Table I, available HMIMOS hardware architectures. Most of the few
we list some of the recent works dealing with different currently available approaches mainly consider large time
combinations among the functionalities of HMIMOS, their periods for training all HMIMOS unit elements via pilots sent
characteristics, and communication applications. from the BS and received at the user equipment via generic
reflection. Another family of techniques employs compressive
sensing and deep learning via online beam/reflection training
IV. D ESIGN C HALLENGES AND O PPORTUNITIES
for channel estimation and design of the phase matrix [12].
In this section, we present some main theoretical and practi- However, this mode of operation requires large amounts of
cal challenges with HMIMOS-based communication systems. training data, and employs fully digital or hybrid analog and
digital transceiver architectures for HMIMOS, which will re-
sults in increased hardware complexity and operational power
A. Fundamental Limits consumption.
It is natural to expect that wireless communication systems
incorporating HMIMOS will exhibit different features com- C. Robust Channel-Aware Beamforming
pared with traditional communications based on conventional Channel dependent beamforming has been extensively
multi-antenna transceivers. Recall that current communication considered in massive MIMO systems. However, realizing
 7

106

Traditional MIMO

104

Massive MIMO

2
10

HMIMOS
100

10-2

10-4
10-4 10-2 100 102 104

(b)
(a)

Figure 4: (a) CRLBs of positioning with an active HMIMOS of a radius R for the case where a single user is located z = 4m
away from the center of surface. The wavelength λ is 0.1m, and τ represents the normalized surface-area [4]. (b) Average EE
with HMIMOS-assisted communication versus the maximum BS transmit power Pmax in dB.

environment-aware designs in HMIMOS-based communica- V. C ASE STUDIES

tion systems is extremely challenging, since the HMIMOS unit In this section, we summarize the performance of HMIMOS
cells that are fabricated from metamaterials impose demanding in two typical application scenarios: indoor positioning with
tuning constraints. The latest HMIMOS design formulations an active continuous HMIMOS and outdoor downlink com-
(e.g., [5], [8]) include large numbers of reconfigurable pa- munication assisted by a passive discrete HMIMOS.
rameters with non-convex constraints rendering their optimal
solution highly non-trivial. For the case of continuous HMI-
MOS, intelligent holographic beamforming is the objective A. Indoor Positioning with an Active Continuous HMIMOS
in order to smartly target and follow individual or small We assume an active HMIMOS where the distance between
clusters of devices, and provide them with high-fidelity beams any of each two adjacent unit elements is λ/2, with λ being the
and smart radio management. However, self-optimizing holo- carrier wavelength. In such a discretized manner, traditional
graphic beamforming technologies that depend on complex MIMO, massive MIMO, and HMIMOS are unified, and the
aperture synthesis and low level modulation are not available differences lie in the number of antenna elements used, i.e., the
yet. surface area. It was shown in [4] that the number of antennas
in a traditional massive MIMO system for a given surface area
D. Distributed Configuration and Resource Allocation πR2 is equal to πR2/(λ2 /4) = πτ z 2 /(λ2 /4) ∼= 20106τ , when
z = 4m, λ = 0.1m, and τ , (R/z)2 (the normalized surface
Consider a HMIMOS-based communication system com- area). A typical massive MIMO array comprising of N = 200
prising multiple multi-antenna BSs, multiple HMIMOS, and antennas results in τ ≈ 0.01, while an active HMIMOS
massive number of users, where each user is equipped with typically increases the surface-area (so as τ ) by 10 ∼ 20 [4].
a single or multiple antennas. The centralized configuration In Fig. 4a, the Cramér–Rao Lower Bounds (CLRBs) of user
of HMIMOS will require massive amounts of control infor- positioning in the presence of phase uncertainty are sketched.
mation to be communicated to a central controller, which As depicted, the CRLB of positioning decreases linearly with
is prohibitive both in terms of computational overhead and τ for traditional MIMO, while massive MIMO falls short in
energy consumption. Hence, distributed algorithms for the reaching the cubic decreasing slope that is achieved by the
EE-optimal resource allocation and beamforming, HMIMOS active HMIMOS, yielding significant gains in user positioning.
configurations, and users’ scheduling need to be developed.
Additional optimization parameters complicating the network
optimization are anticipated to be the power allocation and B. EE Maximization with a Passive Discrete HMIMOS
spectrum usage, as well as the users’ assignment to BSs and We consider an outdoor 16-antenna BS simultaneously serv-
distributed HMIMOS. Naturally, the more HMIMOS are incor- ing 16 single-antenna users in the downlink communication
porated in the network, the more challenging the algorithmic using a discrete passive HMIMOS with 32 unit elements that
design will becomes. is attached to a surrounding building’s facade [5], and the
 8

Table II: Simulation Parameters for the Average EE Performance Results in Fig. 4b.

Parameters Values Parameters Values

HMIMOS central element placement: (100m, 100m) Circuit dissipated power at BS: 9dBW
BS central element placement: (0m, 0m) Circuit dissipated power coefficients at BS and AF relay: 1.2
Small scale fading model: Rayleigh Maximum transmit power at BS and AF relay Pmax : 20dBW
Large scale fading model at distance d: 10−3.53 d−3.76 Dissipated power at each user: 10dBm
Transmission bandwidth: 180KHz Dissipated power at each HMIMOS element: 10dBm
Algorithmic convergence parameter:  = 10−3 Dissipated power at each AF relay transmit-receive antenna: 10dBm

simulation parameters are shown in Table II [5]. The ob- [3] M. Di Renzo, M. Debbah, D.-T. Phan-Huy, A. Zappone, M.-S.
tained EE performance using an approach based on Sequential Alouini, C. Yuen, V. Sciancalepore, G. C. Alexandropoulos, J. Hoydis,
H. Gacanin, J. de Rosny, A. Bounceur, G. Lerosey, and M. Fink, “Smart
Fractional Programming (SFP), as well as a gradient descent radio environments empowered by reconfigurable AI meta-surfaces: An
approach, to tune the HMIMOS system is sketched in Fig. 4b idea whose time has come,” EURASIP J. Wireless Commun. Netw., vol.
as a function of the maximum BS transmit power Pmax . 2019, no. 1, pp. 1–20, 2019.
[4] S. Hu, F. Rusek, and O. Edfors, “Beyond massive MIMO: The potential
We have also numerically evaluated the EE of conventional of positioning with large intelligent surfaces,” IEEE Trans. Signal
Amplify-and-Forward (AF) relaying. It is shown that the Process., vol. 66, no. 7, pp. 1761–1774, Apr. 2018.
HMIMOS-assisted system achieves a three-fold increase in [5] C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and
C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in
EE compared to the AF relaying case when Pmax ≥ 32dBm. wireless communication,” IEEE Trans. Wireless Commun., vol. 18, no.
When Pmax ≥ 32dBm, EE performance saturates, which 8, pp. 4157–4170, Aug. 2019.
reveals that the excess BS transmit power should not be used [6] C. Liaskos, S. Nie, A. Tsioliaridou, A. Pitsillides, S. Ioannidis, and I. F.
Akyildiz, “A new wireless communication paradigm through software-
because it would decrease EE. controlled metasurfaces,” IEEE Commun. Mag., vol. 56, no. 9, pp. 162–
169, Sep. 2018.
[7] N. Kaina, M. Dupré, G. Lerosey, and M. Fink, “Shaping complex mi-
crowave fields in reverberating media with binary tunable metasurfaces,”
VI. C ONCLUSION Sci. Rep. 4, pp. 1–7, No. 076401, 2014.
[8] Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless
In this article, we investigated the emerging concept of network via joint active and passive beamforming,” IEEE Trans. Wireless
Commun., vol. 18, no. 11, pp. 5394–5409, Nov. 2019.
HMIMOS wireless communication, and in particular the avail- [9] W. Tang, M. Z. Chen, X. Chen, J. Y. Dai, Y. Han, M. Di Renzo,
able HMIMOS hardware architectures, their functionalities Y. Zeng, S. Jin, Q. Cheng, and T. J. Cui, “Wireless communications with
and characteristics, as well as their recent communication reconfigurable intelligent surface: Path loss modeling and experimental
measurement,” [online] https://arxiv.org/abs/1911.05326, 2019.
applications. We highlighted their great potential as a key [10] S. Abu-Surra Z. Pi O. E. Ayach, S. Rajagopal and Jr. R. W. Heath,
enabling technology for the physical layer of future 6G wire- “Spatially sparse precoding in millimeter wave MIMO systems,” IEEE
less networks. HMIMOS technology offers huge advantages Trans. Wireless Commun., vol. 13, no. 3, pp. 1499–1513, Mar. 2014.
[11] Y. Han, W. Tang, S. Jin, C. Wen, and X. Ma, “Large intelligent surface-
in terms of SE and EE, yielding smart and reconfigurable assisted wireless communication exploiting statistical CSI,” IEEE Trans.
wireless environments. It reduces the cost, size, and en- Veh. Tech., vol. 68, no. 8, pp. 8238–8242, Aug. 2019.
ergy consumption of network devices, providing ubiquitous [12] A. Taha, M. Alrabeiah, and A. Alkhateeb, “Enabling large intelli-
gent surfaces with compressive sensing and deep learning,” [online]
coverage and intelligent communication in both indoor and https://arxiv.org/abs/1904.10136, 2019.
outdoor scenarios. Benefiting from its merits, HMIMOS can [13] O. Yurduseven, D. L. Marks, T. Fromenteze, and D. R. Smith, “Dynam-
be compactly and easily integrated into a wide variety of appli- ically reconfigurable holographic metasurface aperture for a mills-cross
monochromatic microwave camera,” Opt. Express, vol. 26, no. 5, pp.
cations. Representative use cases are the extension of coverage, 5281–5291, 2018.
physical-layer security, wireless power transfer, and position- [14] A. Pizzo, T. L. Marzetta, and L. Sanguinetti, “Spatial
ing. However, there are still challenges ahead to achieve the characterization of holographic MIMO channels,” [online]
https://arxiv.org/abs/1911.04853, 2019.
full potential of this emerging technology. Among them belong [15] Q. Wu and R. Zhang, “Towards smart and reconfigurable environment:
the realistic modeling of metasurfaces, the analysis of the Intelligent reflecting surface aided wireless network,” IEEE Commun.
fundamental limits of wireless communications with multiple Mag., vol. 58, no. 1, pp. 106–112, Jan. 2020.
HMIMOS, the implementation of intelligent environment-
aware adaptation, and the channel estimation with nearly
passive surfaces. These challenges provide academic and in-
dustrial researchers with a gold mine of new problems and
challenges to tackle.

R EFERENCES

[1] I. F. Akyildiz, C. Han, and S. Nie, “Combating the distance problem

in the millimeter wave and terahertz frequency bands,” IEEE Commun.
Mag., vol. 56, no. 6, pp. 102–108, Jun. 2018.
[2] M. Xiao P. Yang, Y. Xiao and S. Li, “6g wireless communications:
vision and potential techniques,” IEEE Netw., vol. 33, no. 4, pp. 70–75,
Jul., 2019.