Reconfigurable Holographic Surfaces (RHS)

• RHS is a metasurface-based, ultra-thin antenna system that synthesizes electromagnetic wavefronts via holographic beamforming instead of traditional phase shifting.
• They utilize dense, serial-fed metamaterial elements with tunable amplitude control to achieve high directive gains and scalable beamforming for ultra-massive MIMO.
• RHS architectures offer cost and power efficiencies with embedded transceivers, though they face challenges like frequency selectivity, mutual coupling, and limited amplitude resolution.

Reconfigurable holographic surfaces (RHSs) are metasurface-based, ultra-thin and lightweight surface antennas integrated with the transceiver, designed to synthesize electromagnetic wavefronts through holographic beamforming rather than conventional per-element phase shifting. Across the literature, an RHS is modeled as a leaky-wave aperture with densely packed sub-wavelength metamaterial radiation elements, a feed or multiple feeds that launch a reference wave into a waveguide, and electrically tunable element responses—most often amplitude responses in $[0,1]$—that convert guided energy into free-space radiation. This places RHSs between conventional phased arrays and reconfigurable intelligent surfaces (RISs): unlike phased arrays, RHSs avoid large phase-shifter networks; unlike RISs, RHSs act as transmit/receive apertures with embedded feeds rather than passive environmental reflectors (Deng et al., 2021, Deng et al., 2022, Di et al., 2024).

1. Conceptual position and distinguishing characteristics

RHSs are consistently presented as a practical route to holographic radio and ultra-massive MIMO because they offer large apertures without the cost and power burden of conventional phased arrays. The central hardware distinction is that phased arrays rely on many active components—especially phase shifters and power amplifiers—whereas RHSs use dense metamaterial elements with controllable radiation amplitudes and a series-fed surface-wave mechanism. In the foundational communication and holographic-radio treatments, the RHS is described as a PCB-based leaky-wave surface antenna, with the RF front end integrated into the surface itself and beam generation governed by a holographic interference pattern rather than direct phase programming (Deng et al., 2021, Deng et al., 2022).

The distinction from RIS is equally explicit. RIS is treated as a passive or semi-passive reflecting surface deployed in the propagation environment and usually configured through reflection-phase control, whereas RHS is a transceiver-integrated surface with an embedded connected feed or feeds. In the wideband DFRC formulation, the contrast is summarized as follows: RIS shapes propagation in the channel, while RHS shapes radiation at the transceiver (Wei et al., 2023). Related surveys and system papers similarly describe RHS as transmit/receive oriented and RIS as reflective, with RHS supporting direct beam generation and, in some settings, amplitude control in addition to phase-related propagation effects (Gadamsetty et al., 2024, Jamshed et al., 10 Sep 2025).

Several works also emphasize that RHSs are attractive because their inter-element spacing can be smaller than half a wavelength, yielding very dense apertures. This density supports high directive gain, narrow beams, and fine-grained beampattern control, but it also introduces practical issues such as mutual coupling, quantization sensitivity, and propagation loss along the serial-fed surface (Wang et al., 2 Jun 2026, Zeng et al., 9 Sep 2025).

2. Electromagnetic principle and hardware realization

The electromagnetic principle underlying RHS operation is the holographic interference between a desired object wave and a guided reference wave. In the standard formulation, the interference term is written as

$\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$

and the amplitude control law used for practical RHS implementations is commonly normalized as

$$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$

This mapping reflects the fact that the RHS does not generally provide arbitrary complex per-element weights; rather, elements aligned with the desired object wave radiate more strongly, and out-of-phase elements radiate less strongly (Deng et al., 2021, Deng et al., 2022).

In hardware terms, the canonical RHS consists of feeds, a waveguide layer, and many metamaterial radiation elements. The feed launches a reference electromagnetic wave into the guided structure; the wave propagates along the waveguide; engineered discontinuities and tunable metamaterial elements leak part of the guided energy into free space; and the superposition of these emissions forms the desired beam. This is why RHSs are repeatedly described as a special leaky-wave antenna (Zhang et al., 2022, Jamshed et al., 10 Sep 2025).

The most detailed implementations use complementary electric-LC or complementary-electric-resonator type elements loaded with PIN diodes. In the prototype-oriented holographic-radio works, each element is binary-controlled, and the OFF state radiates much more strongly than the ON state. Reported representative figures include radiation efficiencies of about $37\%$ in the OFF state and about $13\%$ in the ON state at resonance in one one-dimensional example, and a design target of about $30\%$–$40\%$ OFF-state efficiency with ON-state efficiency below $15\%$ in another (Deng et al., 2021, Deng et al., 2022). Other fabrication options cited for continuous tuning include varactor diodes and liquid crystals (Deng et al., 2021).

A distinctive practical constraint appears in the ultra-massive MIMO overview: because the RHS is serial-fed and radiates through leakage, the sum of radiated powers is bounded by the feed power through a leakage power constraint,

$$\sum\limits_{n_y}\sum\limits_{n_z} \eta_{n_y,n_z} \left(m(\mathbf{r}_{n_y,n_z}^k,\theta_0, \varphi_0)\right)^2 \leq P_t.$$

This couples the effective contribution of different elements and makes large-aperture design fundamentally different from independent-element phased-array synthesis (Di et al., 2024).

3. Beamforming architectures and mathematical models

The dominant architectural pattern in RHS systems is hybrid beamforming. In the earliest communication formulations, the base station performs digital beamforming across a small number of RF chains or feeds, and the RHS performs analog or holographic beamforming by setting element amplitudes. The corresponding optimization is typically a joint design of a digital beamformer $\mathbf{V}$ and a holographic beamformer $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$0, driven by sum-rate, beampattern, secrecy, or sensing criteria (Deng et al., 2021, Di et al., 2024).

More elaborate models incorporate additional layers. In tri-hybrid holographic ISAC, the transmit chain comprises a digital precoder $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$1, an analog phase-shifter layer $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$2, and an electromagnetic RHS beamformer $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$3, with effective cascade

$\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$4

Here the analog layer provides subarray-level phase control through a small number of phase shifters, while the RHS layer provides element-level amplitude control. This architecture is explicitly motivated by the need to retain low cost for very large apertures while still allowing coherent subarray alignment (Zhang et al., 3 Feb 2026).

The element model used in serial-fed holographic apertures reflects the same separation between inherent propagation phase and controllable amplitude. For the $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$5-th element,

$\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$6

where $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$7 is the normalized amplitude response, $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$8 is an energy coefficient, and $\Psi_{intf} = \Psi_{obj}\Psi_{ref}^{\ast},$9 captures the surface-wave phase progression. This makes explicit that many RHS formulations optimize amplitudes, while the phase is largely induced by on-surface propagation (Zhang et al., 3 Feb 2026).

In some communication and radar formulations, the feed-to-surface response is represented by a fixed propagation matrix. In the limited-amplitude ISAC model, the transmitted signal is

$$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$0

with $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$1 the RHS amplitude matrix and $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$2 the feed-to-element propagation matrix (Zeng, 16 Jun 2026). In wideband DFRC, the holographic beamformer is written as a diagonal amplitude-control matrix

$$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$3

while the RIS-assisted path is handled through a separate passive beamformer $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$4 (Wei et al., 2023).

A recurrent analytical consequence of these models is that RHS beamforming problems are coupled across digital, analog, and electromagnetic layers, and that the feasible set is often a box constraint $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$5, occasionally with binary or quantized variants. This leads to alternating optimization, fractional programming, semidefinite relaxation, majorization-minimization, projected gradient descent, or Riemannian methods depending on the application (Li et al., 2024, Sheemar et al., 3 Mar 2025, Gadamsetty et al., 2024, Zhang et al., 3 Feb 2026).

4. Communications, sensing, and integrated applications

In communications, RHSs have been studied for multi-user downlink beamforming, cell-free uplink reception, secrecy enhancement, wideband dual-function radar-communications, and RIS-assisted MU-MISO. The shared premise is that the RHS offers a large effective aperture with fewer costly RF components. In the original multi-user RHS communication scheme, the digital beamformer is optimized at the base station and the holographic beamformer at the RHS through alternating optimization to maximize sum rate (Deng et al., 2021). In cell-free near-field networks, each distributed base station uses local CSI to configure the RHS and the CPU uses MMSE combining; the analysis shows that the ergodic spectral efficiency based on the near-field channel model is higher than that based on the far-field channel model assumption (Li et al., 2024).

Security-oriented work has extended this model to joint design of digital beamforming, artificial noise, and analog holographic beamforming. In the secrecy-rate formulation, the analog RHS beamformer is parameterized by amplitudes $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$6 multiplying fixed reference-wave phases, while artificial noise is projected into the null space of the legitimate channel and aligned with the unintended receiver channel (Sheemar et al., 3 Mar 2025).

A second major application area is sensing and integrated sensing and communication (ISAC). Holographic ISAC was introduced as an RHS-enabled alternative to phased-array ISAC, with digital beamforming at the base station and analog holographic beamforming at the RHS (Zhang et al., 2022). Subsequent work extended the model in several directions. Multi-band RHS-based ISAC was proposed to address the high frequency selectivity of metamaterial elements and the difficulty of supporting ultra-wide bandwidth; instead of forcing UWB behavior, the system uses multiple narrower OFDM bands and minimizes the average CRLB over the region of interest by jointly optimizing the RHS analog codes and digital beamforming matrices. Reported simulation results show $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$7 less positioning error and reduce $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$8 communication capacity loss relative to the directional benchmark (Hu et al., 2023).

Radar and localization studies use the same amplitude-controlled aperture principle. In RHS-enabled radar for adaptive multi-target detection, a waveform and amplitude optimization algorithm jointly optimizes the radar waveform and the Tx/Rx RHS amplitudes using a relative-entropy objective over multi-hypothesis distributions. Under the same hardware cost, the proposed RHS-enabled radar increases the probability of detection by $$M(\mathbf{r},\theta_0,\varphi_0)=\frac{\mathrm{Re}[\Psi_{intf}(\mathbf{r},\theta_0,\varphi_0)]+1}{2}.$$9 compared to phased array radars when six iterations of adaptive detection are performed (Zhang et al., 2024). In wireless SLAM, RHS replaces phased arrays in radar front ends, with offline beam-pattern optimization, matched-filter range estimation, compressed-sensing angle estimation, and an RBPF-based mapping pipeline; simulation results report about $37\%$0 m lower RMSE than the phased-array baseline and trajectory estimation error one-third of that obtained by the phased-array-based scheme with the same cost (Zhang et al., 2024).

ISAC formulations also reveal how RHS design changes with application objectives. In tri-hybrid holographic ISAC, the optimization minimizes sensing waveform error while satisfying minimum user rate, and the proposed design achieves a controllable performance trade-off between communication and sensing tasks (Zhang et al., 3 Feb 2026). In electromagnetic-compliant ISAC with mutual coupling, the objective becomes maximizing the minimum user data rate while imposing explicit sidelobe constraints, relative sensing-beam balance constraints, minimum sensing gain constraints, and minimum user rate constraints (Zeng et al., 9 Sep 2025).

Beyond terrestrial links, RHS has been proposed for non-terrestrial networks and near-field communication. In the NTN context, RHS is combined with satellites, HAPS, and UAVs to support precise beamforming, intelligent wavefront control, and near-field beam focusing. The public-safety UAV use case uses a $37\%$1 GHz carrier, $37\%$2 MHz bandwidth, four RHS panels each $37\%$3 elements, and shows that the UAV + RHS system consistently outperforms the UAV-only system in energy efficiency as circuit power varies (Jamshed et al., 10 Sep 2025). A distinct THz cooperative-network formulation uses a miniature UAV empowered by an absorptive energy-harvesting RHS rather than a reflective surface, and the proposed method reports EE gains of $37\%$4, $37\%$5, $37\%$6, $37\%$7, and $37\%$8 over five baselines (Song et al., 2024).

5. Practical constraints, impairments, and algorithmic themes

The practical RHS literature is dominated by five recurring constraints: frequency selectivity, mutual coupling, limited amplitude resolution, faulty elements, and channel estimation burden.

Frequency selectivity arises because metamaterial elements are highly frequency selective. In multi-band ISAC, this is identified as the main reason that UWB operation is hard to realize in practice on RHS hardware, which motivates sequential operation over multiple narrower bands rather than true UWB (Hu et al., 2023). Wideband models also expose beam-squint effects and motivate frequency-selective digital beamforming combined with fixed or slowly varying holographic structure (Wei et al., 2023).

Mutual coupling is especially important in sensing-centric and ISAC settings. The coupling-aware holographic ISAC model represents each RHS element as a magnetic dipole and derives a closed-form beamformer

$37\%$9

which reduces to the ideal no-coupling beamformer when $13\%$0. The paper shows that ignoring mutual coupling can produce strong sidelobes, including a representative sidelobe level around $13\%$1 dB, whereas the proposed method improves it to about $13\%$2 dB in one scenario and from $13\%$3 dB to $13\%$4 dB in a close-angle clutter scenario (Zeng et al., 9 Sep 2025).

Limited amplitude resolution is another central issue. In practical hardware, continuous control is replaced by $13\%$5-bit amplitude quantization over

$13\%$6

Closed-form lower bounds derived for communication rate and sensing SINR show that both losses scale with $13\%$7, and the required minimum quantization bits increase as the performance objective becomes more communication-centric (Zeng, 16 Jun 2026). A related tri-hybrid result shows that optimized amplitude responses cluster near boundary values, suggesting that 1-bit amplitude control is a practical approximation and that 1-bit performance is close to continuous control in simulation (Zhang et al., 3 Feb 2026).

Faults and impairments have also become a major topic. In fault-aware RHS-aided ISAC, faulty RHS elements are modeled as having uncontrollable amplitudes uniformly distributed on $13\%$8, and performance degradation is quantified through communication SINR and misspecified CRB for AoD sensing. Joint optimization of digital beamforming, sensing covariance, and functional RHS amplitudes yields an average $13\%$9 performance gain compared to the fault-unaware benchmark (Wang et al., 2 Jun 2026). In cell-free near-field analysis, phase shift error at RHS elements and hardware impairment at RF chains create high-SNR saturation effects; the paper shows that phase shift error at the RHS elements and the HWI at the RF chains of BSs can be compensated by increasing the number of BSs, but UE hardware impairment cannot (Li et al., 2024). In an energy-efficiency study of switch-controlled RHS architectures, explicitly accounting for HWI in beamforming design improves maximum EE by about $30\%$0 Mbit/J relative to ignoring HWI when $30\%$1 (Li et al., 2024).

Finally, channel estimation is repeatedly identified as a bottleneck because RHS-based beamforming normally requires CSI for very large apertures. The recordable and reconfigurable metasurface (RRM) proposal addresses this by adding a power sensor to each metamaterial radiation element so that the surface records interference power during an uplink recording process and reconstructs the beam without pilot-based channel estimation. The reported mutual-information results are close to those of the RHS benchmark with perfect CSI (Wang et al., 24 Jun 2025).

6. Prototypes, measurements, and emerging directions

A defining feature of RHS research is the unusually strong emphasis on hardware prototypes. One of the earliest holographic-radio platforms implements a one-dimensional 16-element RHS of size $30\%$2 at $30\%$3 GHz, controlled by an FPGA through PIN-diode biasing. Full-wave simulations report about $30\%$4 dBi directivity for a 16-element 1D RHS, about $30\%$5 dBi for a 32-element 2D RHS, and about $30\%$6 dBi for a 64-element 2D RHS, with transmit and receive beam patterns nearly identical by reciprocity. A point-to-point prototype built with USRP LW-N321 devices reports an SNR of about $30\%$7 dB in the experimental environment and successful real-time data transmission (Deng et al., 2022).

The holographic ISAC prototype literature extends this to simultaneous communication and sensing. In the 2022 holographic ISAC testbed, the implemented RHS is a 1D surface of size $30\%$8 with 16 metamaterial elements and binary ON/OFF control. In beam-pattern comparison, the RHS and a conventional phased array are both configured to generate two beams, while power consumption is reported as about $30\%$9 W for the phased array and about $40\%$0 W for the RHS. In anechoic-chamber ISAC experiments, one beam is steered toward a target direction and another toward a communication user, the estimated target ranges are close to the actual simulated ranges, and the communication link supports a data rate of $40\%$1 Mbit/s (Zhang et al., 2022).

Later overview and prototype papers report more application-specific measurements. A 2D RHS wireless-communication prototype demonstrates real-time video transmission with QPSK modulation at $40\%$2 resolution and $40\%$3 frames/s, with received SNR higher than $40\%$4 dB (Di et al., 2024). In tri-hybrid holographic ISAC, measured RHS beam gain shows that adding phase-shifter-based subarray alignment increases peak beam gain by $40\%$5 dB for a beam directed at $40\%$6, experimentally validating the role of subarray-level phase correction (Zhang et al., 3 Feb 2026).

The forward-looking literature identifies several directions. Near-field beam focusing and spherical-wavefront beamforming are emphasized because large apertures and high frequencies push many users into the near field, where beam design depends on both angle and distance rather than angle alone (Jamshed et al., 10 Sep 2025, Li et al., 2024). Multi-band and wideband operation remain constrained by metamaterial frequency selectivity (Hu et al., 2023). Additional research themes include multiscale metasurface architectures, mobility-aware holographic relays, integrated sensing and communication, wireless SLAM, secure communication, and low-overhead channel estimation or channel-free operation (Jamshed et al., 10 Sep 2025, Zhang et al., 2024, Sheemar et al., 3 Mar 2025, Wang et al., 24 Jun 2025).

Taken together, these results define RHS as a hardware-centric beamforming paradigm in which a dense, serial-fed metamaterial aperture replaces much of the active complexity of phased arrays. The literature shows consistent gains in power consumption, hardware cost, aperture scalability, and multifunctionality, while also making clear that practical RHS design is inseparable from electromagnetic constraints such as leakage, coupling, quantization, frequency selectivity, and faults (Di et al., 2024, Wang et al., 2 Jun 2026).