Multi-Functional Reconfigurable Intelligent Surface

Updated 9 January 2026

MF-RIS is a metamaterial-based panel that enables simultaneous reflection, refraction, active amplification, and energy harvesting to overcome traditional RIS limitations.
Integrated with joint beamforming and power splitting, MF-RIS delivers significant improvements in spectral efficiency, energy efficiency, and physical layer security.
Advanced prototypes demonstrate real-time adaptive control and scalability for 6G and beyond, leveraging thin-film designs and microcontroller-driven integration.

A Multi-Functional Reconfigurable Intelligent Surface (MF-RIS) is a metamaterial-based panel whose elements simultaneously support wide-aperture signal reflection, transmission (refraction), active amplification, and energy harvesting. In contrast to conventional single-functional RIS (SF-RIS), which provides only passive reflection, MF-RIS introduces additional degrees of freedom via concurrent active and passive EM-wave manipulation, element-wise power splitting between reflection/refraction, and RF-to-DC energy conversion. These capabilities enable MF-RIS to overcome the double-fading attenuation and half-space coverage limitations of SF-RIS, delivering substantial performance improvements in spectral efficiency, network coverage, physical layer security, SWIPT, ISAC, and computation-centric wireless architectures (Zheng et al., 2023, Ni et al., 2024, Wang et al., 2023).

1. Architecture and Physical Layer Model

Each MF-RIS element consists of three highly integrated layers: (i) a reflective layer with a tunable impedance network (e.g., varactors, PIN diodes) for amplitude and phase control in reflection; (ii) an amplification layer including a low-noise amplifier (LNA), power divider, and DC biasing circuits; and (iii) a refractive (transmission) layer, symmetrically mirroring the reflective side. Key element design parameters are the amplitude-gain coefficients ( $\beta_m^r$ , $\beta_m^t$ ) and phase shifts ( $\theta_m^r$ , $\theta_m^t$ ) for reflection/transmission at each element $m$ , subject to local constraints: $\beta_m^r + \beta_m^t \leq \beta_{\max}$ , $0 \leq \beta_m^p \leq \beta_{\max}$ , $p\in\{r,t\}$ , with $\beta_{\max}\ge 1$ to allow amplification (Zheng et al., 2023, Ni et al., 2024).

The per-element block diagram synthesizes:

Signal splitting, with energy/power divider routing the LNA-amplified signal into reflection and refraction branches.
Active gain control via LNA settings, and independent phase shift via bias-tunable impedance networks.
Energy harvesting (in some variants) by integrating RF rectifier circuits that allow switching between harvesting, reflecting, or mixed modes at the element level (Wang et al., 2023, Wang et al., 2023, Mukherjee et al., 1 Jun 2025).

The outgoing MF-RIS reflection and transmission beamforming vectors are: $u_r = [\sqrt{\beta_1^r} e^{j\theta_1^r}, \ldots, \sqrt{\beta_M^r} e^{j\theta_M^r}]^T; \quad u_t = [\sqrt{\beta_1^t} e^{j\theta_1^t}, \ldots, \sqrt{\beta_M^t} e^{j\theta_M^t}]^T$ yielding two independent full-space beamformers, with direct amplitude and phase control per spatial region.

For analysis, the cascaded end-to-end channel for user $k$ (in reflection or transmission half-space) is given by: $\hat{h}_k^H = h_k^H + g_k^H \Theta_k H$ where $h_k$ , $g_k$ , $H$ denote the BS $\to$ user- $k$ , RIS $\to$ user- $k$ , and BS $\to$ RIS channels. $\Theta_k$ is the diagonal matrix encoding element-wise gains and phases for the corresponding space (Zheng et al., 2023, Ni et al., 2024).

2. Joint Optimization Frameworks

MF-RIS brings fundamentally new optimization dimensions for network control:

Amplitude and phase design: Each element’s amplitude split and phase shift must be jointly designed for the desired reflection/refraction/scattering characteristics under power, amplifier, and hardware constraints.
Active beamforming: The BS’s precoder must be tuned in tandem with MF-RIS coefficients, exploiting the new DoFs to maximize performance metrics (sum-rate, secrecy, or energy efficiency).
Self-sustainability: In designs that incorporate energy harvesting, element mode-indicators ( $\alpha_m$ for harvesting/active mode) become additional hybrid discrete-continuous control variables (Wang et al., 2023, Wang et al., 2023, Shen et al., 19 Jan 2025, Mukherjee et al., 1 Jun 2025).

In NOMA-based downlink, the optimization seeks to maximize the sum-rate over beamformers $\{w_k\}$ and MF-RIS coefficients $\{\Theta_k\}$ , subject to BS and RIS power constraints, amplitude/phase limits, per-user QoS, and, if present, energy causality at the RIS (Zheng et al., 2023, Wang et al., 2023). The core alternating optimization approach decomposes the problem into two subproblems:

Beamforming update (fixed MF-RIS): SDPs via variable lifting (e.g., $W_k=w_k w_k^H$ ), employing SCA or penalty-based rank-one relaxation.
MF-RIS coefficient update (fixed beamformer): SDPs over amplitude/phase variables, penalizing non-rank-one solutions, and enforcing energy or hardware constraints (Zheng et al., 2023, Wang et al., 2023).

For systems with imperfect CSI, robust beamforming leverages S-procedure LMIs (bounded errors) or Bernstein-type convex reformulations (statistical errors), enforcing worst-case or probabilistic QoS (Zheng et al., 2024, Wang et al., 2023).

In communication-sensing union (ISAC), the objective is multi-objective: maximize communication sum-rate while also maximizing aggregate target sensing SINR. This yields three-block AO—over transmit beamforming, RIS coefficients, and sensing filters—typically solved by sequential convex/quadratic programming, eigen-decomposition, and SCA (Han et al., 2024, Ni et al., 2024).

3. Performance Analysis and Numerical Insights

MF-RIS consistently outperforms passive RIS, STAR-RIS (dual-functional but passive), and classical active RIS across all major metrics:

Spectral efficiency: In NOMA networks (N=16, M=100, K=6), MF-RIS achieves ~59% sum-rate gain over passive RIS, 16% over active-only RIS, and 44% over STAR-RIS at 10 dBm BS power (Zheng et al., 2023).
Energy efficiency: MF-RIS attains 24–90% higher EE than dual-functional/passive RIS for reasonable CSI error bounds (N=6, M=32, K=6). Over-provisioning $M$ can degrade EE due to amplified overhead (Zheng et al., 2024).
Physical layer security: In NOMA, MF-RIS improves secrecy outage/diversity order for external eavesdroppers, with full-space active beamforming and jamming capability exceeding both active RIS and STAR-RIS. For internal Eves, diversity benefits collapse, underscoring deployment and power allocation subtleties (Pei et al., 2024).
SWIPT: By optimal partitioning into energy-harvesting, reflection, and transmission sub-surfaces and tuning DCSK waveform/correlation parameters, MF-RIS achieves Pareto-optimal harvested energy vs. BER frontiers unavailable to passive RIS (Mukherjee et al., 1 Jun 2025).
ISAC performance: For multi-user multi-target ISAC, MF-RIS with energy splitting (ES) achieves 50–75% higher sensing SINR than active/passive/STAR-RIS at fixed power (Han et al., 2024).
Scalability: Performance scales favourably with $M$ until amplifier power budget or CSI errors dominate; in multi-RIS deployments, DRL-based control of positions and coefficient settings is required for near-optimal system energy efficiency (Kuo et al., 2 Jan 2026, Shen et al., 19 Jan 2025).

Key deployment findings:

RIS placement: For most MF-RIS variants, optimal location is closer to the transmitter to maximize available incident RF for amplification/harvest, in contrast to the passive-RIS optimum (at BS or users) (Wang et al., 2023, Wang et al., 2023).
Self-sustainability: There exists an optimal fraction of active/harvesting elements; too many active elements starve the RIS of power, while too many harvesting elements limit spatial DoFs (Wang et al., 2023, Wang et al., 2023, Mukherjee et al., 1 Jun 2025).
Thermal noise: Amplifier noise per element sets a cap on feasible amplification gains; at high SNR or large $M$ , residual amplifier noise and imperfect SIC dominate performance degradation (Pan et al., 18 Jan 2025, Pei et al., 2024).

4. Advanced MF-RIS Prototypes and Realization

Recent hardware realizations validate the multi-functional paradigm:

Thin-film, microcontroller-driven MF-RIS: A 1×4 CPW-based thin-film MF-RIS on PET achieves independent per-element per-harmonic reflection/refractive phase control using inkjet-printed silver NP circuits, PIN diodes, and MCU control. Simultaneous front/rear beam steering and harmonic multiplexing are demonstrated with 5–6 dBi array gain and <15 dB conversion losses (Xie et al., 2024).
Partitioned surfaces: EH/relay clusters, with dynamic OOK control logic and biasing, realize practical time/frequency or mode switching among MF-RIS functions in real-time (Mukherjee et al., 1 Jun 2025, Xie et al., 2024).
Amplification hardware: Integration of LNA, power divider, and dual-feed impedance networks per element, with low-noise bias/amp circuits and controller modules for adaptive phase/amplitude, is required for true simultaneous multi-functionality (Ni et al., 2024, Wang et al., 2023).

Fabrication leverages additive manufacturing and off-the-shelf MCU/SMD components for significant cost reduction ( $\sim$ $10–15 per 4-element array) and rapid prototyping. Remaining engineering challenges include element coupling, real-time adaptive control, thermal management, and DC power delivery (Xie et al., 2024, Ni et al., 2024).

5. Methodologies for Adaptive and Intelligent Configuration

Given the large-dimensional, hybrid discrete-continuous MF-RIS control space, algorithmic advances are crucial:

Alternating optimization (AO): Decomposition into convex subproblems for beamforming, MF-RIS configuration, and (when present) energy harvesting/control, using semidefinite programming and penalty methods to enforce rank constraints (Zheng et al., 2023, Wang et al., 2023, Zheng et al., 2024).
SCA and MM: Sequential convex approximation and majorization-minimization to linearize non-convex constraints (rate, EH, SINR).
Robust optimization: S-procedure (LMI) and Bernstein-inequalities ensure QoS/outage under bounded/statistical CSI error (Wang et al., 2023, Zheng et al., 2024).
Hybrid DRL/FL schemes: Multi-agent deep RL (with parametrized sharing, federated DDPG) for multi-MF-RIS and multi-agent scenarios, handling both continuous (amplitude/phase/EH ratios/location) and discrete (mode switching) controls, with centralized or federated aggregation (Kuo et al., 2 Jan 2026, Shen et al., 19 Jan 2025).
Channel estimation: DFT-structured pilot design and AO-based RIS beam pattern selection to achieve CRLB-optimal channel estimation under significant thermal noise (Pan et al., 18 Jan 2025).

These algorithmic frameworks are shown to converge rapidly and to deliver performance that closely tracks theoretical outer bounds across a range of MF-RIS architectures (Han et al., 2024, Kuo et al., 2 Jan 2026, Shen et al., 19 Jan 2025).

6. Applications and Future Challenges

MF-RIS are pivotal in emerging 6G and beyond strategies, spanning:

Downlink NOMA/RSMA/OFDMA: Full-space, flexible DoFs, enabling optimal user decoding order and spectral reuse (Zheng et al., 2023, Ni et al., 2024, Wang et al., 2023).
Physical layer security: Simultaneous directional amplification and jamming, outperforming all prior passive and dual-mode schemes (Pei et al., 2024, Zhao et al., 22 Dec 2025).
Non-terrestrial platforms (LEO, UAVs): Dynamic joint optimization of MF-RIS configuration, relay path selection, and power control to maximize energy- and spectral-efficiency in LEO constellations and drone-assisted access (Shen et al., 19 Jan 2025, Zhao et al., 22 Dec 2025).
SWIPT/ISAC/AirComp: Joint waveform, RIS coefficient, and resource optimization for ultra-reliable multi-modal RF services (Han et al., 2024, Ni et al., 2024, Ni et al., 2024, Mukherjee et al., 1 Jun 2025, Jin et al., 2024).
Cell-free and mmWave networks: MF-RIS as distributed virtual LoS relays, reducing base station array size while preserving ultra-rich spatial coverage (Ni et al., 2024).

Outstanding research frontiers include robust near-field ISAC models, integration of AI-based real-time control, wireless power self-sustainability (fully battery-less MF-RIS), holistic electromagnetic–IT co-design, mitigation of mutual coupling and non-idealities, and global RIS deployment strategies in massive multi-agent and non-stationary scenarios (Ni et al., 2024, Kuo et al., 2 Jan 2026, Ni et al., 2024).

References:

(Zheng et al., 2023, Ni et al., 2024, Wang et al., 2023, Wang et al., 2023, Pan et al., 18 Jan 2025, Mukherjee et al., 1 Jun 2025, Xie et al., 2024, Zhao et al., 22 Dec 2025, Zheng et al., 2024, Han et al., 2024, Ni et al., 2024, Jin et al., 2024, Pei et al., 2024, Shen et al., 19 Jan 2025, Kuo et al., 2 Jan 2026)