Hybrid Reconfigurable Intelligent Metasurfaces: Enabling Simultaneous Tunable Reflections and Sensing for 6G Wireless Communications

04/10/2021 ∙ by George C. Alexandropoulos, et al. ∙ University of Athens Ben-Gurion University of the Negev 0

Current discussions on the sixth Generation (6G) of wireless communications are envisioning future networks as a unified communication, sensing, and computing platform that intelligently enables diverse services, ranging from immersive to mission critical applications. The recently conceived concept of the smart radio environment, enabled by Reconfigurable Intelligent Surfaces (RISs), contributes towards this intelligent networking trend, offering programmable propagation of information-bearing signals, which can be jointly optimized with transceiver operations. Typical RIS implementations include metasurfaces with nearly passive meta-atoms, allowing to solely reflect the incident wave in an externally controllable way. However, this purely reflective nature induces significant challenges in the RIS orchestration from the wireless network. For example, channel estimation, which is essential for coherent communications in RIS-empowered wireless networks, is quite challenging with the available RIS designs. This article introduces the concept of Hybrid reflecting and sensing RISs (HRISs), which enables metasurfaces to reflect the impinging signal in a controllable manner, while simultaneously sense a portion of it. The sensing capability of HRISs facilitates various network management functionalities, including channel estimation and localization. We discuss a hardware design for HRISs and detail a full-wave proof-of-concept. We highlight their distinctive properties in comparison to reflective RISs and active relays, and present a simulation study evaluating the HRIS capability for performing channel estimation. Future research challenges and opportunities arising from the concept of HRISs are presented.



There are no comments yet.


page 1

page 2

page 3

page 4

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

I Introduction

The potential of Reconfigurable Intelligent Surfaces for programmable ElectroMagnetic (EM) wave propagation has recently motivated extensive academic and industrial interests, as a candidate smart connectivity paradigm for the future sixth Generation (6G) wireless communications [1, 2, 3]. The RIS technology, which typically refers to artificial planar structures with nearly passive electronic circuitry (i.e., without any power amplification), is envisioned to be jointly optimized with conventional wireless transceivers [4] in order to significantly boost wireless communications in terms of coverage, spectral and energy efficiency, reliability, and security, while satisfying regulated EM field emissions.

The typical unit element of an RIS is the meta-atom, which is usually fabricated to realize multiple digital states corresponding to distinct EM responses. By externally controlling the states of sub-wavelength-spaced meta-atoms in such metasurfaces, various reflection and scattering profiles can be emulated [5]. The RIS technology up to date mainly includes meta-atoms of nearly passive nature, in the sense that they do not include any power amplifying circuitry. Such metasurfaces can only act as tunable reflectors, and thus, neither receive nor transmit on their own. While nearly passive RISs enable programmable wireless propagation environments, their purely reflective operation induces notable challenges when considered for empowering wireless networks. For instance, the inclusion of an RIS implies that a signal transmitted from a User Terminal (UT) to a Base Station (BS) undergoes at least two channels: the UT-RIS and RIS-BS channels. Estimating these individual channels is a challenging task due to the reflective nature of RISs, and can significantly limit the ability to reliably communicate in a coherent manner. The common approach in RIS-empowered wireless communications is to estimate the entangled combined effect of these channels, known as the cascaded channel [6]. However, this strategy limits the transmission scheme design and restricts network management flexibility. In particular, estimating each channel separately allows to use fewer pilot signals, exploits the knowledge that each channel changes in time at a different rate, and enables precoding schemes which rely on the knowledge of the individual channels [7]. Furthermore, reflective RISs impose challenges on wireless localization [8]. In fact, it was recently proposed in [9] to equip RISs with a single Radio Frequency (RF), enabling low-cost signal reception, and consequently individual channel estimation, at their side.

Dynamically radiating metasurfaces have recently emerged as a promising technology for realizing low-cost, low-power, and large-scale Multiple-Input Multiple-Output (MIMO) antennas [10, 11]. Dynamic Metasurface Antennas pack a large number of controllable radiative meta-atoms that are coupled to one or several waveguides, resulting in a MIMO transceiver with advanced analog processing capabilities [12]. While the implementation of DMAs differs from reflective RISs, the similarity in their unit element structures indicates the feasibility of designing hybrid reflecting and sensing meta-atoms. Metasurfaces consisting of such hybrid meta-atoms will be capable of reflecting their impinging signal, while simultaneously sensing a portion of it. Hybrid reflecting and sensing RISs (HRISs), which have not yet been treated in the wireless communication literature, bear the potential of significantly facilitating coherent communications, without notably affecting the energy efficiency and coverage extension advantages offered by purely reflective RISs.

In this article, we introduce the concept of HRISs and discuss their possible applications for future wireless communications. We first review the possible configurations for hybrid meta-atoms, and then, present an illustrative implementation of an HRIS consisting of simultaneously reflecting and sensing meta-atoms, followed by full-wave EM simulations for its intended operation. To investigate the potential of HRISs for wireless communications, we highlight their distinctive properties with reflective RISs and active relays and, by introducing a simple model for their dual operation, we present a representative simulation study evaluating their ability for facilitating channel estimation. We conclude with a description of some research challenges and opportunities with HRISs, including key experimental directions which are expected to further unveil their potential role in 6G wireless communications.

Ii HRIS Design and Experimentation

A variety of RIS implementations has been lately proposed [5], ranging from metasurfaces that manipulate wave propagation in rich scattering environments to improve the received signal strength, to those that realize desired anomalous reflection beyond Snell’s law [3]. In all those efforts, the RISs are incapable of any sort of sensing of the impinging signal. Nonetheless, metasurfaces can be designed to operate in hybrid reflecting and sensing manner, while exhibiting comparable energy efficiency with their purely reflective counterparts. We next discuss how such HRISs can be designed in hardware, and present an implementation sketch along with a full-wave EM simulation proof-of-concept.

Configurations of Hybrid Meta-Atoms: From a hardware perspective, we can envision two different configurations that allow for meta-atoms to both reflect and sense. The first is comprised of hybrid meta-atoms, which simultaneously reflect a portion of the impinging signal, while another portion is sensed. The second implementation uses meta-atoms that reconfigure between near-perfect absorption and reflection. In this case, the metasurface has two modalities: for a short time, it senses the channel (absorption mode); once necessary information is extracted, it enters the second mode (reflection mode) and directs the signal towards a desired direction. We focus henceforth on the first case of hybrid meta-atom, as illustrated in Fig. 1. Such HRISs are realized by adding a waveguide to couple to each meta-atom. Each waveguide can be connected to a RF chain [9], allowing to locally process a portion of the received signals in digital domain. However, the coupling of the elements to the waveguides implies that the incident wave is not perfectly reflected. In fact, the ratio of the reflected energy to the absorbed one is determined by the coupling level. By keeping this waveguide near cutoff, we can reduce its footprint while also reduce coupling to the sampling waveguide.

Fig. 1: Illustration of the proposed HRIS and the constitutive hybrid meta-atom design. The layers of the meta-atom on the top right are artificially separated for better visualization. The cross section of the metasurface and the coupled wave signal path are depicted on the bottom right drawing.

Design Considerations: An important consideration in the HRIS design is the inter-element spacing. To accurately detect the phase front of the incident wave, the meta-atom spacing needs to be such that they are smaller than half a wavelength. However, this imposes constraints on the meta-atom size, which we accommodate via the proposed multi-layer structure in Fig. 1. Another important factor to consider is the necessary circuitry to detect the signal coupled from each meta-atom; in particular, each waveguide should be connected to an RF chain. Since the incident wave on the HRIS may couple to all sampling waveguides (with different amplitudes), we can think of the combination of the metasurface and the sampling waveguides as a receiver structure with equivalent analog combining. This detecting circuitry, however, adds another factor to consider when designing an HRIS: while closer element spacing improves the ability to direct the reflected signal to a desired direction (with smaller sidelobes) and increases the accuracy of detecting incident phase fronts, close element spacing increases the total number of meta-atoms, and consequently, cost and potentially power consumption. Nonetheless, this can be balanced by limiting the number of RF chains—which are connected to the waveguides and not directly to each meta-atom— and take advantage of the analog combining nature of the proposed structure. Such a configuration also reduces the cost and power consumption of the introduced sensing circuitry, which does not appear in conventional, incapable of sensing, RISs.

Implementation: Most elements in available RISs designs are resonators, which can be easily modified to couple to a waveguide. For example, the meta-atom in [13] requires a via to deliver the DC signal to each element, in order to tune the switchable component. A radial stub was added to the path of this conductive trace to diminish RF coupling. One can envision a design where the via is attached to two copper traces, one to sample the signal and another one to transfer the DC signal (which is connected to a radial stub). Confining the coupled RF signal to such a copper trace may be challenging. As a result, we suggest a Substrate Integrated Waveguide (SIW), which is effectively a rectangular waveguide, to capture the sampled wave as shown in Fig.  1. By changing the annular ring around the coupling via or the geometrical size of the SIW, we can design HRISs with different coupling strengths.

Experimental Proof-of-Concept: The hybrid meta-atom illustrated in Fig. 1 is assumed to be loaded by a varactor, whose effective capacitance is changed by an external DC signal. The varying capacitance changes the phase of the reflected wave. By properly designing this phase variation along the HRIS, the reflected beam can be steered towards desired directions. For simplicity, we have considered an one-Dimensional (1D) configuration of an HRIS with such meta-atoms, and used the full-wave EM solver Ansys HFSS to demonstrate its reflection capability. Toward this goal, the incident wave was modeled as a Gaussian beam that impinges on this HRIS at normal angle. The beamwidth of the incident Gaussian beam wass selected to be smaller than the metasurface to alleviate edge effects. The 1D metasurface was designed to operate at GHz (corresponding to wavelength mm), and it consisted of elements that were placed at mm () apart. The design frequency is similar to that used in [13], and is considered here only for demonstration purposes. The proposed operation and design can be easily extended to other frequencies by just adjusting the geometrical dimensions of the HRIS. Two examples of a beam steered in different directions using this 1D HRIS are shown in Fig. 2. It is noted, however, that this tunable reflection is not new. The novelty of the proposed hybrid design lies in the metasurface’s capability to also estimate the phase front of the incident beam by analyzing the wave sampled by the sampling waveguides (i.e., the SIWs). The differential phase along the sampling waveguides for different incident angles are shown in Fig. 2. Note that we have analyzed the differential phase between adjacent SIWs, since the raw phase value depends on the phase reference plane which is not usually known. It is clear that the sampled differential phase is related to the incident Angle Of Arrival (AOA) (or incident phase front). By conducting a pre-characterization for different AOAs, one can use the differential phase to estimate the phase front at the HRIS and then compute any wireless channel’s characteristics of interest. It is worth noting that we have considered the case that the signal at each meta-atom is sampled by a dedicated RF chain separately. Alternatively, one may combine the sensing waveguides of many meta-atoms and then connect to an RF chain, giving rise to a hybrid analog and digital receiver [10].

Fig. 2: Full-wave EM simulation of an 1D HRIS with meta-atoms at GHz. Two programmable steered reflected beams from a normal incident beam at the HRIS are depicted on the left. The differential phases sensed along the sampling waveguides of the proposed HRIS for different incident angles are plotted on the right.

Iii HRISs for Wireless Networks

In this section, we first describe the envisioned HRIS-empowered wireless communication system. Next, we present a simple model that encapsulates the simultaneous tunable reflection and sensing operations of HRISs. Finally, some numerical results showcasing their potential for wireless channel estimation are presented.

Fig. 3: Two applications of RIS-empowered wireless communications focusing on the uplink of multi-user MIMO systems: using a conventional RIS which solely reflects the impinging UT signals in a controllable manner; and using the proposed HRIS that also senses a portion of the impinging UT signals. To avoid cluttering, we only visualize communications carried out through either the RIS or HRIS, while in practice there may also exist additional direct channels between any of the UTs and the BS.

Iii-a HRIS-Empowered Wireless Communications

The most common application of conventional RISs is to facilitate communication between UTs and the BS [4], by shaping the EM signal propagation allowing to improve coverage and overcome harsh non-line-of-sight conditions. In such setups, as illustrated in Fig. 3(a), the RIS tunes its elements to generate favourable propagation profiles. To achieve this, the BS maintains a control link with the digital controller of the RIS, where the latter sets the configuration of each element according to the control messages received by the BS. This setup can be also extended to multiple RISs and cloud-controlled networks [14]. Figure 3(a) also reveals some of the challenges associated with RIS-aided communications. For instance, in the absence of direct channels between the BS and UTs, the BS only observes the transmitted signals which propagated via the cascaded channels, namely, the composition of the UTs-RIS channel, the RIS reflection configuration, and the RIS-BS channel. The fact that one cannot disentangle this combined channel implies that the BS cannot estimate the individual channels, but only the cascaded ones [6]. This property does not only reflect on the ability to estimate the channels, but implies that the RIS must rely on the BS for configuration. It also limits the utilization of some network management tasks, such as wireless localization.

We envision HRISs to be deployed in a similar manner to conventional reflective RISs. This includes, for instance, the typical application of coverage extension and connectivity boosting, by modifying the propagation profile of information-bearing EM waves, as illustrated in Fig. 3(b). Similar to the setup in Fig. 3(a) involving a conventional RIS, the BS maintains a dedicated control link with the HRIS. However, there are two main differences between wireless networks empowered by HRISs and purely reflective RISs. First, the HRIS does not reflect all the energy of the incident signal, since a portion of it is absorbed and sensed by its local processor; nevertheless, this may degrade the Signal-to-Noise Ratio (SNR) at the BS. Second, the link between the BS and HRIS is not used solely for unidirectional control messages from the BS to the metasurface, but now the latter can also use this control link to convey some sensed information to the BS. The fact that the UTs’ transmissions are also captured by the HRIS can notably improve channel estimation, as discussed in the sequel, which in turn facilitates coherent communications. Furthermore, the sensed signal can be used for localization, RF sensing and radio mapping, as well as AOA estimation and AOA-based phase configuration searching.

The potential benefits of HRISs over conventional ones come with additional energy consumption and cost. While the proposed meta-atom design in Fig. 1 is nearly passive, as that of purely reflective RISs, HRISs require additional power for locally processing the sensed signals. The utilization of RF chain(s) and the additional analog combining circuitry, which are not required by conventional RISs, is translated into increased cost. While explicitly quantifying this increased cost and power consumption is highly subject to the specific implementation, it is emphasized that an HRIS is still an RIS, and not a relay. Namely, it does not involve a power-consuming wireless transmission mechanism for amplifying and transmitting its received signals, and is thus expected to maintain the cost and power gains of RISs over traditional relays [4, 15]. An additional challenge associated with HRISs stems from the optional need for a bi-directional control link between itself and the BS. This link can be used for making HRIS’s observations available to the BS (otherwise, they remain locally) to support higher throughput compared to the unidirectional control link utilized by conventional RISs.

Iii-B Model for Simultaneous Reflection and Sensing

A simple, yet generic, model for the HRIS operation can be obtained by considering its reflective and sensing capabilities individually. As discussed in Section II, the HRIS tunable reflection profile is similar to conventional RISs, where it is typically modeled as controllable phase shifters. However, unlike conventional RISs, in HRISs the reflectivity amplitude for each meta-atom is determined by a design knob that can be adjusted by changing the amount of coupling to the sensing circuitry. This design knob is modeled by a parameter , representing the portion of the signal energy which is reflected by the proposed meta-atom. This parameter is related to the amount of coupling between the meta-atom and the sampling waveguide, which essentially determines how much loss the HRIS introduces compared to solely reflective RISs. Note that if the coupling is too small, the sensed signal at HRIS becomes susceptible to noise, while when it is too high, the signal reflected is considerably attenuated. As a result, it is useful to consider the reflectivity amplitude a design knob and optimize its acceptable range given a desired wireless system.

The second part of the HRIS model describes the relationship between the impinging waveform at the HRIS and the field values picked up by its sensing circuitry. The exact details of this model are heavily dependent on the specific implementation (see Section IV for more details). For simplicity, one can consider this relationship to be as simple as the non-reflected part (i.e., ) multiplied by a phase component. Finally, as in DMAs [12], the RF chain(s) via which the sensed signals are forwarded to the digital processor are connected to the waveguides, and thus the sensed signal path can be viewed as a hybrid analog and digital receiver.

The resulting HRIS operation model is illustrated in Fig. 4, where one can configure the portion of the reflected energy for each meta-atom, as well as the phase shifting carried out at the reflect and receive signal paths. In practice, these parameters are typically coupled, implying that one cannot tune these parameters arbitrarily. However, this relatively simple model allows to provide an understanding of what one can gain by properly tuning these HRIS parameters, as detailed in the following section.

Fig. 4: The HRIS operation model with hybrid meta-atoms. The incident signal at each meta-atom is split into a portion which is reflected (after tunable phase shifting), while the remainder of the signal is sensed and processed locally. The received portion undergoes the controllable element response and is collected at the RF chains connected to the sampling waveguides, whose digital outputs are processed by a digital processor.

Iii-C Channel Estimation with HRISs

As an indicative application of HRISs, we investigate the contribution of their sensing capability in estimating the channel among the UTs and the BS via the metasurface in Fig. 3(b), which is comprised of two individual channels: the composite UTs-HRIS channel and the HRIS-BS channel. Recall that the estimation of those individual channels is challenging when using a purely reflective RIS, as one only observes the channel outputs at the BS. In this case, the effect of the individual channels is highly entangled, and the common practice is to resort to estimating the combined channel effect, i.e., the cascaded channel [6], using pilot reflective patterns at the RIS. The fact that HRISs can also process their observations, rather than solely reflect them, allows to disentangle the individual channels. Doing so, notably facilitates channel estimation compared to conventional RIS-empowered wireless systems, enabling the usage of precoding methods relying on the knowledge of the individual channels, as well as improving the spectral efficiency by reducing the number of required pilots symbols [7].

To demonstrate the channel estimation capability of HRISs, we consider the uplink system of Fig. 3(b) with UTs and an HRIS with meta-atoms and RF chains, as in Fig. 4. A simple strategy to estimate the individual channels is to have the HRIS estimate the UTs-RIS channel based on its sensed observations, and forward this estimate to the BS over the control link (wired or wireless), while changing the phase configuration between pilot symbols as in [6]. The fact that the HRIS also reflects while sensing allows the BS to estimate the HRIS-BS channel from its observed reflections, effectively re-using the transmitted pilots for estimating both channels. In this setting, the HRIS can estimate the coefficients of which the UTs-RIS channel is comprised; the measurements are obtained via digital signal paths. At the absence of noise, the HRIS is able to recover the UTs-RIS channel from pilots, while the BS requires at least pilots to recover the RIS-BS channel from its observations and the estimate provided by the HRIS via the control link. For instance, when UTs communicate via an HRIS comprised of meta-atoms all attached to RF chains, both individual channels can be recovered from merely pilots. For comparison, state-of-the-art methods for conventional RISs would require over pilot transmissions for recovering solely the cascaded channel for a fully digital -antenna BS [6]. This simplistic strategy for channel estimation is beneficial not only in noise-free channels, but also in noisy setups. However, one need to also account for the fact that the power splitting between the reflected and sensed waveforms carried out by the HRIS affects the resulting SNRs at the metasurface and BS. In particular, in noisy setups where the HRIS estimates the composite UTs-RIS channel locally and forwards it to the BS, there is an inherent trade-off between the accuracy in estimating each of the individual channels, which is dictated by how the HRIS is configured to split the power of the incident wave between its reflected and absorbed portions.

Fig. 5: Normalized MSE performance in recovering the combined UTs-HRIS channel at the HRIS and the HRIS-BS channel at the BS for dB transmit SNR as well as different power splitting values and phase configurations.

In Fig. 5, we depict the aforedescribed trade-off between the MSE performances when estimating the UTs-RIS channel at the HRIS and the RIS-BS channel at the BS for different values of the power splitting parameter . Each curve corresponds to a different random setting of the individual phase shift at each meta-atom. In particular, the simulation setup consists of a BS with antennas serving UTs, assisted by an HRIS with elements and RF chains. The UTs

are uniformly distributed in a cell of

meter radius, where the HRIS is located at the top edge of the cell, and at distance meters from the BS. The UTs transmit pilot symbols for channel estimation, which are received at the BS after being reflected by the HRIS with SNR dB. It is observed in Fig. 5 that there is a clear trade-off between the accuracy in estimating each of the individual channels, which is dictated by how the HRIS splits the power of the impinging signal. While the MSE values depend on the HRIS phase configuration, we observe that increasing the portion of the signal that is reflected in the range of up to notably improves the ability to estimate the RIS-BS channel, while having only a minor effect on the MSE in estimating the UTs-RIS channel. However, further increasing the amount of power reflected, notably degrades the MSE in estimating the UTs-RIS channel, while hardly improving the accuracy of the RIS-BS channel estimation.

The ability to sense the individual channels at the HRIS side facilitates channel estimation, especially in wideband multi-user multi-antenna systems where cascaded channel estimation requires prohibitive overhead [6]. In addition, it enables RF sensing, localization, and radio mapping whose role is envisioned to be prominent in 6G wireless communications [1]. In Fig. 6, the MSE in estimating the cascaded channel for the setup in Fig. 5 with an HRIS, which reflects on average of the received signal, is compared to the method [6] for reflective RISs. The latter method requires over pilots and we have used . As shown, the HRISs sensing capability is translated into improved cascaded channel estimation accuracy, compared to the state of the art. In particular, the considered HRIS with RF chains achieves an SNR gain of over dB, though at the cost of higher power consumption and hardware complexity.

Fig. 6: Normalized MSE performance as a function of the transmit SNR in dB of the cascaded channel estimation using the proposed HRIS and a purely reflective RIS via the scheme of [6].

Iv Research Challenges and Opportunities

The design of HRISs, discussed in Section II, and their representative effect on channel estimation, evaluated in Section III, provide an initial understanding on concept’s implementation and application potential, respectively. The ability of HRISs to carry out reflection and sensing simultaneously gives rise to a multitude of research challenges and opportunities. Some important studies, which we expect to play a key role in unveiling the potential of such metasurface architectures in HRIS-empowered wireless communications, are next discussed.

Fundamental Limits: The inclusion of RF sensing capabilities at metasurfaces considerably changes the operation of metasurface-assisted wireless communication systems. This motivates the characterization of the fundamental limits of HRIS-empowered communications in terms of achievable rate. Such an analysis can quantify the added value of HRISs compared to purely reflective RISs, in a manner which is invariant of the operation of the communicating end entities.

Operation Protocols:

HRISs provide additional network design degrees of freedom compared to reflective RISs, motivating novel algorithmic designs to exploit them. Their RF sensing capability enables application processing at the metasurface side (e.g., localization, AOA estimation, and radio mapping), which can be used locally or shared to the network for management purposes. In the latter case, efficient information exchange protocols are needed (possibly among multiple HRISs and BSs), while considering rate-limited in- or out-of-band control links.

HRIS Modeling: The performance evaluation of HRIS-assisted channel estimation in Figs. 5 and 6 is based on the presented simplified model for its dual operation. In practice, each meta-atom EM response is expected to exhibit a more complex configurable profile, including coupling between its parameters (i.e., power splitting and phase shifting coefficients) as well as between different elements. Such a physics-driven characterization will allow to more faithfully evaluate HRISs. An additional critical aspect of HRISs, which should be characterized, is its excessive power consumption and cost compared to reflective RISs. While such an analysis is expected to be highly implementation dependent, it will help understanding the price associated with the gains of HRISs.

Hardware Designs: The presented proof-of-concept is an important first step in demonstrating the feasibility of HRIS hardware. Nonetheless, realizing such metasurfaces for wireless communications still requires a large body of experimental efforts and hardware designs, from low up to THz frequencies. Akin to reflective RISs, the HRIS design requires investigations into the number of meta-atoms and their the tuning mechanism, as well as the size of the metasurface and its phase configuration capabilities (e.g., beam steering span, grating lobes, and sidelobes). All other studies related to forming desired reflected beams (as done for reflective RISs) are also relevant to HRISs.

There is also an additional study unique to the hybrid meta-atom design in Fig. 4. The complex amplitude sensed at the sampling waveguides does not necessarily exhibit an one-to-one relationship with that at each meta-atom, which is related to the channel at that particular element. This is due to the fact that the wave coupled by the meta-atom into the substrate layer travels along it, gets reflect from boundaries, and can couple to all sampling waveguides. This overall effect can be considered as information multiplexing, and can be pre-characterized into a sensing matrix. This matrix can be then used to retrieve the complex field at the surface of the HRIS, i.e., the required channel information. In an alternative implementation to Fig. 4, each meta-atom is directly connected to a sampling waveguide; the wave incident on different meta-atoms will now have minimal coupling. In this case, we will have a nearly one-to-one relationship between the wave at the meta-atom and the one at the sampling waveguide; we still need to pre-characterize, however, the complex amplitude of the coupling. While the latter design seems simpler from sensing perspective, it may require a large number of RF chains to sample the received signal, which directly increases the cost and power consumption. The multiplexing factor can in fact be used to reduce the number of receiving RF chains and also mitigate the impact of noise.

V Conclusion

In this article, we introduced the novel concept of HRISs, which, in contrast to purely reflective RISs, are capable to simultaneously reflect a portion of the the impinging signal in a controllable manner, while sensing the other portion of it. We presented an HRIS design and discussed a full-wave EM simulation proof-of-concept, showcasing its hardware feasibility. We highlighted the possible operations of HRIS-empowered wireless communication systems, provided a simplified model for the HRIS reconfigurability, and evaluated their capability to facilitate channel estimation, in comparison with reflective RISs. We discussed several research challenges and opportunities which arise from the HRIS concept, and which are expected to pave the way in unveiling the potential of this technology for 6G wireless systems.


  • [1] “The next hyper- connected experience for all,” White Paper, Samsung 6G Vision, Jun. 2020.
  • [2] C. Liaskos, S. Nie, A. I. Tsioliaridou, A. Pitsillides, S. Ioannidis, and I. F. Akyildiz, “A new wireless communication paradigm through software-controlled metasurfaces,” IEEE Commun. Mag., vol. 56, no. 9, pp. 162–169, Sep. 2018.
  • [3] M. Di Renzo, M. Debbah, D.-T. Phan-Huy, A. Zappone, M.-S. Alouini, C. Yuen, V. Sciancalepore, G. C. Alexandropoulos, J. Hoydis, H. Gacanin, J. de Rosny, A. Bounceu, G. Lerosey, and M. Fink, “Smart radio environments empowered by reconfigurable AI meta-surfaces: an idea whose time has come,” EURASIP J. Wireless Commun. Net., vol. 2019, no. 1, pp. 1–20, May 2019.
  • [4] C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in wireless communication,” IEEE Trans. Wireless Commun., vol. 18, no. 8, pp. 4157–4170, Aug. 2019.
  • [5] C. Huang, S. Hu, G. C. Alexandropoulos, A. Zappone, C. Yuen, R. Zhang, M. Di Renzo, and M. Debbah, “Holographic MIMO surfaces for 6G wireless networks: Opportunities, challenges, and trends,” IEEE Wireless Commun., vol. 27, no. 5, pp. 118–125, Oct. 2020.
  • [6] Z. Wang, L. Liu, and S. Cui, “Channel estimation for intelligent reflecting surface assisted multiuser communications: Framework, algorithms, and analysis,” IEEE Trans. Wireless Commun., vol. 19, no. 10, pp. 6607–6620, Oct. 2020.
  • [7] C. Hu and L. Dai, “Two-timescale channel estimation for reconfigurable intelligent surface aided wireless communications,” [Online] arxiv.org/abs/1912.07990, 2019.
  • [8] H. Wymeersch, J. He, B. Denis, A. Clemente, and M. Juntti, “Radio localization and mapping with reconfigurable intelligent surfaces: Challenges, opportunities, and research directions,” IEEE Veh. Technol. Mag., vol. 15, no. 4, pp. 52–61, Dec. 2020.
  • [9] G. C. Alexandropoulos and E. Vlachos, “A hardware architecture for reconfigurable intelligent surfaces with minimal active elements for explicit channel estimation,” in Proc. IEEE ICASSP, Barcelona, Spain, May 2020, pp. 9175–9179.
  • [10] N. Shlezinger, G. C. Alexandropoulos, M. F. Imani, Y. C. Eldar, and D. R. Smith, “Dynamic metasurface antennas for 6G extreme massive MIMO communications,” IEEE Wireless Commun., to appear, 2021.
  • [11] M. Boyarsky, T. Sleasman, M. F. Imani, J. N. Gollub, and D. R. Smith, “Electronically steered metasurface antenna,” Scientific reports, vol. 11, no. 1, pp. 1–10, 2021.
  • [12] N. Shlezinger, O. Dicker, Y. C. Eldar, I. Yoo, M. F. Imani, and D. R. Smith, “Dynamic metasurface antennas for uplink massive MIMO systems,” IEEE Trans. Commun., vol. 67, no. 10, pp. 6829–6843, Oct. 2019.
  • [13] T. Sleasman, M. F. Imani, J. N. Gollub, and D. R. Smith, “Microwave imaging using a disordered cavity with a dynamically tunable impedance surface,” Physical Review Applied, vol. 6, no. 5, p. 054019, 2016.
  • [14] G. C. Alexandropoulos, S. Samarakoon, M. Bennis, and M. Debbah, “Phase configuration learning in wireless networks with multiple reconfigurable intelligent surfaces,” in Proc. IEEE GLOBECOM, Taipei, Taiwan, Dec. 2020, pp. 1–6.
  • [15] E. Björnson, Ö. Özdogan, and E. G. Larsson, “Intelligent reflecting surface versus decode-and-forward: How large surfaces are needed to beat relaying?” IEEE Wireless Commun. Lett., vol. 9, no. 2, pp. 244–248, Feb. 2020.