Reconfigurable Distributed Antennas & Reflecting Surfaces

• RDARS is an integrated system combining DAS and RIS functionalities to support high-throughput communications, low energy consumption, and integrated sensing.
• RDARS employs dual-mode elements with RF switches for active transmission and tunable phase shifters for passive reflection, enabling real-time channel-aware reconfiguration.
• RDARS leverages advanced optimization algorithms for joint active/passive beamforming, achieving superior throughput, energy efficiency, and secure multi-user connectivity.

A reconfigurable distributed antennas and reflecting surface (RDARS) is an integrated wireless infrastructure that hybridizes the functionalities of a distributed antenna system (DAS) and a reconfigurable intelligent surface (RIS). In an RDARS architecture, each metasurface element can dynamically switch between two modes: (i) a connected ("direct-through") mode acting as a distributed antenna with active fronthaul to the base station, or (ii) a reflection mode behaving as a fully passive RIS element imparting a programmable phase shift to incident electromagnetic waves. This dual-mode capability provides independent control over distribution gain (from active elements) and reflection gain (from passive elements), as well as additional degrees of freedom for channel-aware, per-element reconfiguration. The RDARS concept fundamentally extends both DAS and RIS paradigms and supports simultaneous high-throughput transmission, low energy footprint, and integrated sensing over large-scale surfaces (Ma et al., 2023, Wang et al., 2024, Birari et al., 23 Jan 2025, Ji et al., 5 Jan 2026).

1. RDARS Architecture and Physical Implementation

An RDARS is typically implemented as a planar or conformal array comprising N elements, each equipped with:

• A RF switch to select between active and passive branches.
• In active (connected) mode: An electrical-to-optical interface, LNA, and fronthaul link (e.g., fiber, coax) connecting to the BS; the element acts as a distributed antenna for UL/DL signaling.
• In passive (reflection) mode: A tunable, low-cost phase shifter controlling the reflected phase $\theta_n$ (often 2–4 bits per element) (Ma et al., 2023).

The mode of each element is indicated by $a_n \in \{0,1\}$, forming a binary mode-selection matrix A. Control commands (mode flags, phase codes) are distributed via a centralized controller or FPGA and updated per coherence interval, typically over UDP or dedicated wired links (Ma et al., 2023, Wang et al., 2023, Ji et al., 2 Apr 2025).

Advanced RDARS hardware includes modular building-block panelization, scalable digital/optical control (e.g., per-block microcontrollers with distributed IR, Ethernet, or FPGA-based addressing), and varactor- or PIN-diode-based phase control (Sayanskiy et al., 2022, Ratajczak et al., 2023, Ming et al., 13 Apr 2025). Specialized cells may integrate hybrid reflection/transmission functions with tunable power splitters and dual-layer (reflection + transmission) aperture control (Ming et al., 13 Apr 2025).

2. Signal and Channel Models

The RDARS end-to-end link encompasses three primary paths: (i) direct, (ii) distributed-antenna (connected) mode, and (iii) radiation-reflection (passive) mode. For single-antenna endpoints,

$$\gamma_\text{RDARS} = \overline{\gamma} \left| h_{UB} + \mathbf{h}_{RB}^T B \mathbf{h}_{UR} \right|^2 + \overline{\gamma} \,\mathbf{h}_{UR}^H A \mathbf{h}_{UR}$$

where $h_{UB}$ is the direct channel, $\mathbf{h}_{UR}$ and $\mathbf{h}_{RB}$ are UE→RDARS and RDARS→BS channels, $A$ indicates connected-mode elements, and $B = (\mathbf{I}-A)\Theta$ captures RIS-phase effects (Ma et al., 2023, C et al., 27 Dec 2025).

With perfect CSI, optimal phase-shift design aligns all backscattered paths; in MIMO, the RDARS action generalizes to block-diagonal mode-selection and phase matrices, with joint signal vectorization and stacking for the active/passive domains (Ji et al., 1 Aug 2025, Ji et al., 5 Jan 2026). In sensing-communication coexistence, the SNR and mean-squared error (MSE) metrics jointly capture radar, localization, and communications performance under the RDARS channel model (Zhang et al., 2024, Wang et al., 2023).

3. Operating Principles and Joint Gains

Each element's per-symbol mode switching enables a joint exploitation of three distinct performance gains:

• Distribution gain: Linear in the number $a$ of connected elements ($\sim a$), leverages the diversity and proximity benefits of distributed antennas.
• Reflection gain: Quadratic in co-phased passive elements ($\sim (N-a)^2$), as with conventional RIS, but subject to multiplicative fading.
• Selection gain: Additional DoF due to dynamic mode selection, enabling channel-aware optimization of active/passive configuration for system objective (rate, SNR, secrecy, Pareto ISAC) (Ji et al., 5 Jan 2026, Wang et al., 2024, Pei et al., 18 Jan 2025).

Closed-form rate and SNR scaling reveal that, for moderate $N$, judiciously choosing $a \ll N$ can mitigate RIS fading bottlenecks while retaining substantial reflection gain, with full RIS quadratic scaling recovered for $a \ll N$, $N \to \infty$ (Ma et al., 2023, C et al., 27 Dec 2025). Flexible placement and sparsity patterns for connected elements further enhance beam directivity and suppress inter-user interference (Ji et al., 5 Jan 2026, Ji et al., 2 Apr 2025).

4. Algorithmic Optimization and Mode Selection

RDARS-enabled networks require joint optimization over active and passive beamforming matrices and per-element mode selection under non-convex, mixed-integer constraints. State-of-the-art algorithmic solutions include:

• Block coordinate descent (BCD) and majorization-minimization (MM) for alternating updates of active beamforming, passive phase matrices, and binary mode-selection vectors (Ji et al., 1 Aug 2025, Wang et al., 2024).
• Greedy/augmented Lagrangian search, projected gradient ascent (PGA), and Riemannian manifold algorithms for efficient handling of unit-modulus phase constraints (Ma et al., 2023, Wang et al., 2024).
• Model-driven deep learning (unfolding) to accelerate convergence and circumvent local minima, with trainable penalty coefficients and hyperparameters (Ji et al., 1 Aug 2025).
• Joint ISAC formulations leveraging Pareto-optimal trade-off surfaces between communication and sensing objectives, supported by penalty and surrogate optimization (Zhang et al., 2024, Birari et al., 23 Jan 2025).

Sparsity-driven designs exploit physical aperture expansion for narrow beams and low inter-user correlation; selection of optimal placement for active elements is tractable analytically for $K=1,2$, and efficiently solved for larger user sets (Ji et al., 5 Jan 2026).

5. Comparative Performance and Prototyping

Measurements and simulations across the literature confirm that RDARS architectures surpass both pure RIS (reflection-gain-only, suffers massive power loss due to multiplicative fading) and DAS (distribution-gain-only, incurring high hardware/RF cost). Representative findings include:

• Uplink throughput enhancements of +21% (over DAS) and +170% (over RIS) for $N=256$ elements, $a=1$ (Ma et al., 2023).
• Multi-user sum-rate and MSE improvements over fixed-index and random-index designs, robustly holding across SNR, number of elements, and user clustering (Wang et al., 2024, Ji et al., 5 Jan 2026).
• Energy efficiency: RDARS offers maximum coverage and EE for moderate $N$ and UEs in sub-6 GHz bands versus active RIS architectures, with optimal element count and placement depending on channel distances and frequency (C et al., 27 Dec 2025).
• Integrated sensing and communication: Sub-meter user localization accuracy with minimal degradation of communication rates; RMSE $\approx 0.3$ m and $\approx 10$–15% trade-off in throughput as $a$ is varied (Wang et al., 2023).
• 3D beamforming and sustainable prototyping: Large-scale varactor-RIS-based RDARS achieves $>10$ dB gain, modular IR/digital block-scale implementations enable 2D holographic phase control and field-reconfigurable surfaces (Ratajczak et al., 2023, Sayanskiy et al., 2022).

6. Extensions: Sensing, Security, and Hybrid Surfaces

A growing body of literature generalizes RDARS for multimodal operation:

• Integrated Sensing and Communication (ISAC): RDARS elements facilitate target/range localization (DoA, RSSI-based ranging) by leveraging both direct RF sampling and adaptive reflection, surpassing RIS-only and DAS-only benchmarks in joint SNR and estimation error (Zhang et al., 2024, Birari et al., 23 Jan 2025).
• Physical layer security: Channel-aware mode selection and joint beamforming maximize secrecy rates against eavesdroppers, outperforming both pure RIS and DAS, with AO and SCA-based penalty optimization (Pei et al., 18 Jan 2025).
• Hybrid transmitting/reflecting surfaces: Cells combining phase-reconfigurable antennas and splitters, e.g., BD-RIS, support independent simultaneous beam steering in two spatial domains, enabling beyond-diagonal MIMO, relaying, and full-space communications (Ming et al., 13 Apr 2025).
• Sustainable and scalable designs: Refurbishment of classic reflectarrays as RDARS surfaces, energy-harvesting modules, and modular control architectures support green and large-area deployments (Ratajczak et al., 2023, Sayanskiy et al., 2022).

7. Practical Considerations and Future Perspectives

RDARS platforms introduce specific challenges:

• Hardware cost and energy: Only $a \ll N$ active RF chains required, minimizing total cost and operational power compared to DAS (Ma et al., 2023, C et al., 27 Dec 2025).
• Control and synchronization: High-speed mode/phase updates per element, phase calibration, and tight synchronization with fronthaul and BS baseband are necessary for full performance (Ma et al., 2023, Wang et al., 2023).
• CSI and channel estimation: Active elements boost pilot-based CSI acquisition, partially alleviating the pilot contamination and overhead burden of large passive RISs (Ma et al., 2023).
• Open research topics: Real-time large-$N$ optimization, scalable mode-selection/control, full-duplex and time-varying switching, wideband OFDM ISAC extensions, and robust distributed processing architectures remain largely open (Ji et al., 5 Jan 2026, Zhang et al., 2024).

RDARS is expected to serve as a foundational component for cell-free massive MIMO, ultra-dense network infrastructure, and next-generation ISAC deployments, further catalyzed by advances in scalable surface engineering, model-driven control, and intelligent optimization (Ma et al., 2023, Ji et al., 5 Jan 2026).